Predicting Body Fat and Blood Lipids with Sugars Intake

Objective: The primary objective of this study was to determine if caloric intake of fructose sugars (free fructose plus sucrose) predicts body fat percentage in young adults. The secondary objective was to determine if caloric intake of fructose sugars predicts total cholesterol (TC) and low density lipoprotein cholesterol (LDL-C). Participants and Methods: Men (n=55, body fat=16.3±14.0%) and women (n=281, body fat=26.9±7.5%), 18 to 24 years of age, were recruited for an ongoing, crosssectional study, The Nutrition Assessment Study. Anthropometric, biochemical and dietary data were collected. Linear modeling was used to assess predictions of body fat percentage and blood lipids with sugars intake, and multiple regressions were used to control for possible covariates. Results: In a linear model, a 1% increase in caloric intake of fructose sugars predicted a 0.56% higher body fat in men (β=0.311, R=0.097, p=0.037). This prediction remained significant when adjusting for BMI and alcohol intake (β=0.260, R=0.505, p=0.036). In women, no predictions were seen with caloric intake of fructose sugars and body fat. Fructose sugars did not predict TC or LDL-C in this sample. Conclusion: In this population of healthy young adults, higher consumption of fructose sugars is associated with higher body fat in men but not in women. Longitudinal research is needed to determine if these predictions are observed over


Predicting body fat and blood lipids with sugars intake
Objective: The primary objective of this study was to determine if caloric intake of fructose sugars (free fructose plus sucrose) predicts body fat percentage in young adults. The secondary objective was to determine if caloric intake of fructose sugars predicts total cholesterol (TC) and low density lipoprotein cholesterol (LDL-C).
Participants and Methods: Men (n=55, body fat=16.3±14.0%) and women (n=281, body fat=26.9±7.5%), 18 to 24 years of age, were recruited for an ongoing, crosssectional study, The Nutrition Assessment Study. Anthropometric, biochemical and dietary data were collected. Linear modeling was used to assess predictions of body fat percentage and blood lipids with sugars intake, and multiple regressions were used to control for possible covariates.
Results: In a linear model, a 1% increase in caloric intake of fructose sugars predicted a 0.56% higher body fat in men (β=0.311, R 2 =0.097, p=0.037). This prediction remained significant when adjusting for BMI and alcohol intake (β=0.260, R 2 =0.505, p=0.036). In women, no predictions were seen with caloric intake of fructose sugars and body fat. Fructose sugars did not predict TC or LDL-C in this sample.

Conclusion:
In this population of healthy young adults, higher consumption of fructose sugars is associated with higher body fat in men but not in women.
Longitudinal research is needed to determine if these predictions are observed over time.

INTRODUCTION
A preventable chronic disease, obesity, affects 600 million adults aged 18 years and older worldwide 1 . In the United States (US), obesity generates health care costs ranging from $147 billion to nearly $210 billion per year 2 . The US has one of the highest overweight and obesity rates, with over 60% of adults defined as overweight or obese (BMI>25) 1 .
Young adults (18 to 24 years old) have experienced increases in obesity 3 , with weight gain in early adulthood linked to increased obesity 4 and cardiovascular disease (CVD) risk later in life 5 . One dietary factor, consumption of fructose sugars (sucrose plus free fructose), may lead to adverse metabolic outcomes, such as dyslipidemia [6][7][8] , cardiovascular diseases [9][10][11] and obesity [12][13][14] through stimulation of de novo lipogenesis 15,16 . Adolescents and young adults are among the highest consumers of sugars 17 , making them a critical group on which to focus. Research shows that the prevalence of dyslipidemia early in life is a strong predictor of the obesity later in life [18][19][20] and that adverse lipid profiles in young adults accelerates the development of atherosclerosis 17,21 .
Cross-sectional research concludes no significant associations between fructose consumption and body mass index (BMI) 22 . However, BMI does not take into account body fat percentage, which experimental research suggest may be increased with fructose sugars intake 12,23 . Despite this, experimental studies utilize consumption levels exceeding the 95 th percentile (14.5% daily energy) for fructose consumption 24 .
Thus, a gap exists in the literature as to whether fructose sugars have deleterious associations with body fat percentage when consumed in free-living individuals.
Therefore, this study aimed to determine if consumption of sugars, with emphasis on fructose sugars, predict body fat percentage and fasting blood lipids in young men and women. Primarily, it was hypothesized that caloric intake of fructose sugars would predict body fat percentage. Secondarily, it was hypothesized that Fructose sugars intake would predict total cholesterol (TC) and low density lipoprotein cholesterol (LDL-C). Further, predictions with consumption of non-fructose sugars (free glucose plus lactose) and total sugars were explored. Lastly, predictions of the metabolic risk factors high-density lipoprotein cholesterol (HDL-C) and triacylglycerol (TAG) were explored with consumption of sugars.

Subjects and Research Design:
This cross-sectional study in college students used data from the Nutrition Assessment Study (NAS). The NAS is an ongoing observational study of health risk factors in college students enrolled in an introductory nutrition course and a senior level nutrition course. The NAS was approved by the University of Rhode Island Institutional Review Board (IRB) (IRB HU1112-069). Demographic survey, anthropometric measures and biochemical indices were extracted from the NAS database.
Consenting students from the fall 2013 through the spring 2015 semester were given the opportunity to complete the validated Comprehensive Nutrition Assessment Questionnaire (CNAQ) 25 as a dietary assessment. Study staff informed students of their eligibility to participate in the study, described the study design and collected a signed consent form of those agreeing to participate.

Demographic and Anthropometric Measures
Participants completed a brief demographic survey, the Nutrition Assessment  26,27 , with predicted thoracic volume.

Biochemical Measures
After an overnight fast, a full lipid profile including TC, LDL-C, HDL-C and TAG were collected. The lipid profile was measured using the validated Alere Cholestech LDX® System 28,29 (Alere Inc., Waltham MA). To calculate LDL-C the Friedewald equation was used. Researchers drew 40uL aliquots of blood, via finger stick, from participants using capillary tubes. Measured outcomes were provided immediately from this system and participants were provided with an explanation of and a copy of their results.

Dietary Measures
The CNAQ is a semi-quantitative 297-item online food frequency questionnaire validated in 2010 for use in adults to evaluate intake of 52 nutrients 25 .
The CNAQ was designed to analyze macronutrients, micronutrients, and indigestible carbohydrates. Responses to the CNAQ were processed using the food composition database, created and maintained by Monash University in Melbourne, Australia 25 .
This questionnaire generates immediate feedback including estimated intake of energy (kJ), total sugars (g), fructose (g), sucrose (g), glucose (g), and lactose (g) 25 . Fructose sugars in this paper will refer the amount of fructose plus the amount of sucrose consumed. Non-fructose sugars will refer to the amount of glucose plus the amount of lactose consumed.
The CNAQ could be saved, stopped and continued over multiple intervals if necessary. Participants were prompted to evaluate their average intakes over a one- year period (responses include, but are not limited to "daily", "weekly", "monthly" or "never or rarely"). The CNAQ provided brief instructions on how to document food items that are consumed only in specific seasons. Prompts encourage participants to identify quantities of foods consumed, while an unanswered question prevented the participants from submitting the CNAQ. In order to navigate differences in food terminology between the US and Australian citizens, study staff developed a translation sheet. An example: what Australians refer to as "rocket", the US refer to as "arugula".

Statistical Analysis
All data analyses were conducted using SPSS version 22.0. For demographic variables, dependent variables, and independent variables, descriptive statistics were used to analyze means, standard deviations, and medians. Frequencies were conducted for categorical variables. All variables were normal according to Shapiro-Wilk after eliminations of outliers greater than three standard deviations from the mean for intake of fructose sugars, non-fructose sugars and total sugars, as well as for body fat percentage and lipid values of TC, LDL-C, HDL-C and TAG.
A total of 17 women and 5 men were eliminated for one or more of the categories for sugars intake. In total, 7 men and 20 women did not complete the full lipid panel, but were included for comparisons with body fat. Two men were missing body fat analysis, and were eliminated as outliers for LDL-C and TAG. One man was eliminated as an outlier for LDL-C and TAG, and one man was eliminated as an outlier for only TAG. Five men did not complete body fat analysis, but completed the full lipid profile. Lastly, one male participant was eliminated as an outlier with a body fat percentage >65%. Among women, two were eliminated as outliers for both TC and LDL-C, and one women was eliminated as an outlier for TC. Three women were eliminated as outliers for both HDL-C and LDL-C. One women was eliminated as an outlier for HDL-C and did not have readable LDL-C and TAG by the Cholestech. In addition, 12 women did not have readable levels of TAG and LDL-C. One women was an outlier for only TAG, and three women for only LDL-C. Lastly, 19 women did not complete body fat percentage and 4 were eliminated as outliers.
To determine associations between independent and dependent variables, Pearson correlations were applied. To determine if there were relationships between potential covariates, such as alcohol, saturated fat intake and BMI with outcome variables, Spearman's Rho was applied to non-normal covariates. To address our hypotheses, linear modeling was used to determine if caloric intake of fructose sugars, non-fructose sugars and/or total sugars predict body fat percent, TC, LDL-C, HDL-C and TAG. To avoid overfitting in regression models, 5-10 participants are required per predictor when assumptions of normality are met, and 10-20 participants are required per predictor when assumptions of normality are not met. 30 .

Subject Characteristics and Dietary Intakes
In this cross-sectional analysis data from 414 participants were collected, but

Prediction of Body Fat Percentage with Caloric Intake of Sugars
The correlations among fructose intake, non-fructose intake and body fat percentage are presented in Table 2. Gram intake of fructose sugars and total sugars negatively correlated with body fat percentage in women. However, they did not correlate with body fat percentage when analyzed as percentage of caloric intake, in women. Caloric intake of fructose sugars, non-fructose sugars and total sugars positively correlated with body fat percentage in men.
Caloric intake of fructose sugars, non-fructose sugars and total sugars did not predict body fat percentage in any linear models in women. In men, a 1% increase in caloric intake of fructose sugars predicted a 0.56% higher body fat percentage in men (β=0.311, R 2 =0.097, p=0.037), Figure 1.

Prediction of Blood Lipids with Caloric Intake of Sugars
In this sample, no correlations were detected for TC and sugars did not predict TC in any linear models. Consumption of fructose sugars, non-fructose sugars and total sugars did not predict LDL-C in women. Among men, non-fructose sugars in grams correlated positively and moderately with LDL-C, Table 2. In a linear model, a 20gram increase in non-fructose sugars predicted a 6.76mg/dL higher LDL-C level in men (β=0.317, R 2 =0.100, p=0.041). When adjusted for body fat percent and alcohol intake, non-fructose sugars no longer predicted LDL-C (β=0.313, R 2 =0.148, p=0.080).
The associations between HDL-C, TAG and sugars are presented in Table 2.
There were significant inverse associations between fructose and total sugars with HDL-C, in men. In linear models, a 1% increase in caloric intake of fructose sugars predicted a 1.10mg/dL lower HDL-C level in men (β=-0.407, R 2 =0.165, p=0.005), Figure 2. A 1% increase in caloric intake of total sugars predicted a 0.71mg/dL lower HDL-C level in men (β=-0.400, R 2 =0.160, p=0.006). When adjusted for intake of saturated fat, BMI, and TAG, a 1% increase in caloric intake of fructose sugars predicted a 0.77 mg/dL lower HDL-C in men (β=-0.326, R 2 =0.442, p=0.034).
Caloric intake of non-fructose sugars did not predict HDL-C in men.
Among men, a 20gram increase in non-fructose sugars predicted a 9.74mg/dL higher TAG level (β=-0.398, R 2 =0.158, p=0.010). When adjusted for body fat percentage and alcohol intake a 20gram increase in non-fructose sugars predicted a 7.38mg/dL higher TAG (β=0.332, R 2 =0.270, p=0.046). There were no significant associations with respect to HDL-C and TAG in women.

DISCUSSION
A recent cross-sectional study using NHANES 1999-2006 data observed relationships with fructose and non-fructose sugars intake with respect to BMI. They concluded no significant associations with BMI in 25,506 participants 22 . Our study filled a research gap by exploring predictions with fructose sugars and body fat percentage, a more accurate way to assess weight status. We found predictions of body fat percentage with fructose and total sugars intake in men. However, we did not see these predictions in women. A possible explanation could be increases in visceral fat, which may be specific to fructose 23 . Visceral fat is stored to a greater extent in men when compared to women 32 . However, this was not assessed in the present study, and is an area for future research.
In line with previous cross-sectional research 17,22 , no significant predictions of TC and LDL-C with consumption of fructose sugars were observed, when consumed in free-living young adults. Despite this, in men non-fructose sugars predicted LDL-C.
Previous research has shown a relationship between added sugars and LDL-C in women, but not men 34 . Glucose has a high glycemic index, which when consumed elicits an insulin response to help deliver glucose to the muscles and surrounding tissues 35 . Insulin is believed to have an indirect stimulation of HMG-CoA reductase in favor of cholesterol biosynthesis 35 . However, this was no longer significant when adjusting for significant covariates, such as body fat and alcohol intake.
Increased consumption of sugars may also lead to elevations in TAG through stimulation of de novo lipogenesis 6 , leading to increased production of VLDL 6 Secondly, this study is limited because physical activity data were not collected. Physical activity has been shown to have beneficial effects on body fat as well as on blood lipids 38 . Despite our inability to adjust for physical activity, we were able to adjust for saturated fat, cholesterol intake and TAG which have been shown to related to HDL-C. We were also able to adjust for intake of alcohol, body fat percentage and BMI.

CONCLUSIONS
In conclusion, this study helps to fill a gap in the fructose research by exploring predictions with body fat percentage rather than BMI in free-living young adults. Daily energy intake from fructose sugars predicted body fat percentage and HDL-C in men, not in women. When consumed in free-living young adults, consumption of fructose sugars did not appear to be predictors of TC, LDL-C, or

Figure 2b:
A one percent increase in caloric intake of non-fructose sugars did not predict HDL-C in men (p=0.382).

Figure 2c:
A one percent increase in caloric intake of total sugars predicts a 0.71mg/dL lower HDL-C in men (p=0.006).

REVIEW OF THE LITERATURE
Overview: This literature review will discuss the consumption of sugars, specifically fructose sugars (fructose and sucrose), as well as non-fructose sugars (glucose plus lactose) and sugar sweetened beverages and their potential relationships with markers of weight status and blood lipids. First we will discuss definitions, sources and tools for measuring sugars. Then we will discuss the relationships between sugars and body composition, specifically analyzing the relationships with body mass index (BMI) and body fat percentage. Lastly, the relationships between consumption of sugars and blood lipids will be analyzed, with emphasis on total cholesterol (TC), low density lipoprotein cholesterol (LDL-C), high density lipoprotein cholesterol (HDL-C) and triglycerides (TAG).

Defining Sugars:
The American Heart Association (AHA) and Academy of Nutrition and Dietetics (AND) defined sugars in various contexts 1 . First, the AHA and AND defined sugars as monosaccharides and disaccharides including glucose, galactose and fructose 1 . Similarly, they defined sugars as both naturally occurring (intrinsic) in fruits, vegetables and dairy foods, or as added (extrinsic) to foods during processing, or in preparation for consumption 2 . In contrast the term sugar refers to sucrose, which is derived from sugar cane or beets 2 .
Fructose is the most common naturally occurring monosaccharide found in fruits and vegetables 1,2 . In nature fructose is linked as the disaccharide sucrose (glucose plus fructose), but is also used as a caloric sweetener 2 . Other disaccharides include lactose (glucose plus galactose), which is found in milk products and maltose (glucose plus glucose), which is found in malt and molasses 1,2 .
Lastly, the AHA and AND define total sugars and high-fructose corn syrup

Tools for measuring nutrient intake levels
There are multiple ways to determine a person's nutrient intake. These include 24-hour dietary recalls, food frequency questionnaires, and food records. Large data sets such as those seen in the National Health and Nutrition Examination Survey (NHANES) utilize 24-hour recalls as a way to collect information on nutrient intakes 9 .
Some studies have gone further and have analyzed differences between tools for measuring nutrient intakes.
The Observing Protein and Energy Nutrition (OPEN) study involving 484 participants assessed intake using a food frequency questionnaire, two 24 hour recalls, urinary sucrose and fructose as a predictive biomarker of total sugars and doubly labelled water to adjust for grams per 1000kcals 10,11 . Compared to the predictive biomarkers, self-reported intake of total sugars by food frequency questionnaire was 13.5% lower in men and women and 24-hour recall were biased high in men and nearly identical with women 10 . Table 1 and Table 2 summarizes the differences between the food frequency questionnaire used (DHQ), 24 hour recalls and total energy expenditure. Compared with the total energy expenditure, men underreported energy intake by 12-14% on 24hour recalls and 31-34% on food frequency questionnaires 11 . In contrast, women underreported 16-20% on 24 hour recalls and 34-38% on food frequency questionnaires 11 . This suggests that women might be greater under reporters of energy intake when compared to men.
Despite previous research suggesting underreporting by food frequency questionnaires, a food frequency questionnaire was recently developed to quantify intake of fermentable oligosaccharides, disaccharides, monosaccharides and polyols 12 .
Barrett et al. 12 conducted a validation paper comparing a food frequency questionnaire to four sets of seven day recalls taken three months apart 12 .

Total and added sugar consumption levels
According to the United States Department of Agriculture, in 1970 roughly 2,109kcals/day were consumed on average per person 13 . By 2010, that number increased to 2,568kcals/day per person 13 . This is the equivalent to an increase of about 22%, or an additional 459kcals daily 13 . Of this increase, about 4% (20kcals) comes from an increase in added sugars with the rest coming from an increase in flour, cereal products and added fats 13 . While this is only a small increase in added sugars over 40 years, some data suggest that sugars consumption actually decreased in recent years 14 .
In 1999, annual sugars intake was on average 89.3 lbs/person/year, in America 14 .
However, in 2013 annual sugars intake was at 75.4 lbs/person/year 14 . This is equivalent to a 16% decrease in annual sugars intake 14 .
Using data from 1971 and 1994 (NHANES I and III), Chun et al. 18 estimated total and added sugars intake in participants <18 years old, participants >19 years old and among all participants 18 . Total and added sugars intake was higher among participants <18 years old 18 . On average participants <18 years old consumed 138g total sugar with 88g added according to NHANES I and 139g total sugars with 92 grams added according to NHANES III 18 . This translates to an increase in one gram of total sugars, but a 3g decrease in natural sugars and a 4g increase in added sugars, corresponding with the rise in refined carbohydrates 18 . Among participants 19 and older, total sugars intake was 110g with 71g added sugars in NHANES I and was 126g and 84g in NHANES III, respectively 18 . This age group experienced a 14-gram increase in total sugars with an increase seen in both natural and added sugars.
According to NHANES III data, free fructose accounted for 21% of sugars intake among both age categories and sucrose accounted for 39% and 43% of total sugars intake in <18 year olds and >19 year olds, respectively 18 9 . This suggests that added sugars intake has remained relatively constant over the last decade. Of these adolescents, 88% consumed ≥10% energy from added sugars and 5.5% had a usual intake above 25% total caloric intake 9 . This suggests that of the adolescents surveyed, only a small percent of them are meeting current guidelines for consumption of sugars set by the AHA.  25 . These costs stem largely from metabolic consequences of excess adiposity 25 .

Metabolic Risk Factors
Metabolic Syndrome is a cluster of interrelated risk factors of metabolic origin that directly promote the development of cardiovascular diseases and other metabolic diseases 26 . These risk factors include elevated waist circumference, elevated TAG, reduced HDL-C, elevated blood pressure and elevated fasting glucose, which are defined in Table 4   and about 3% of women had metabolic syndrome 28 . They further analyzed the number of risk factors by BMI categories (18.5-24.9, 25-29.9, and ≥30kg/m 2 ) and stated that both men and women with a BMI ≥30kg/m 2 had significantly more metabolic criteria than those in the other BMI categories 28 . Although the focus of this paper was to look at risk factors, they did include that women had a greater intake of total sugars in relation to calories as compared to men (21.9% vs 20.0%, p<0.001) 28 . Despite this, they concluded that overweight/obese college aged men present with a greater prevalence of risk when compared to college aged women 28 .
A similar study observed metabolic risk factor criterion across large (>10,000students) diverse universities to examine the relationship with weight status and adiposity 33 . Overall, more than half of the sample had at least one metabolic syndrome criterion, with men twice as likely to have metabolic syndrome when compared to women (12% vs 6%, respectively) 33 . Metabolic syndrome was five times more prevalent among overweight and obese participants when compared to normal weight (16% vs 3%, p<0.001 respectively) 33 . Lastly, overfat (≥20% body fat for men and ≥33% for females) 33,34 participants had significantly more metabolic syndrome criteria than participants with normal levels of body fat (1.7 vs 0.7; p<0.001) 33 .

Metabolism of sugars
The Academy of Nutrition and Dietetics stated that HFCS and sucrose are similar in composition 2 . Similarly, the metabolic effects of HFCS and sucrose do not differ making it essential to observe fructose sugars (sucrose and free fructose) to assess metabolic impacts 16 . However, there are differences in the metabolism of the two monosaccharides that make up sucrose and HFCS, glucose and fructose 35 .
Glucose metabolism occurs in all tissues of the body with about 30-40% of metabolism occurring in the liver. 36 Glucose is a high glycemic index non-fructose containing sugar 35 . The glycemic index is a physiological classification of the available carbohydrate content in foods, which was first proposed in 1981 37 . It reflects the capacity of a carbohydrate containing food to raise blood glucose 38 36 . The consumption of a sugar-rich meal saturates SGLT1 and GLUT5 can result in recruitment of GLUT2 to transport sugars across apical membrane 36 . This can triple the sugar uptake by enterocytes 36 .
monosaccharides within the lumen, glucose and galactose are transported into the enterocyte by the sodium/glucose cotransporter 1 (SGLT1) 2,36 . The SGLT1 is located on the apical membrane of the enterocyte, and has a high affinity for glucose and galactose 2,36 . The lumen of the intestines has a higher concentration of sodium when compared to the enterocyte, which allows for an inward gradient into the enterocyte 36 .
This drives glucose and galactose absorption against their own concentration gradients with help from two sodium ions 36 .
Once inside the enterocyte the glucose and galactose part from the sodium ion 36 . Fructose, on the other hand, is not transported into the enterocyte via SGLT1, but is instead transported by the facilitated fructose transporter 5 (GLUT5) 36 . This transporter is also located on the apical membrane, but this transporter has a low affinity but high capacity for transporting fructose 36  Once absorbed, fructose metabolism occurs preferentially and primarily in the liver and does not elicit the same insulin response as glucose, identifying fructose as a lower glycemic index sugar 23,42 . The liver has a high level of glucokinase, the enzyme responsible for phosphorylation of glucose in the liver. 36 Once glucose becomes phosphorylated, glucose 6-phosphate can continue through glycolysis. 36 However, this enzyme does not phosphorylate fructose. 36 Thus, the liver utilizes fructokinase, an enzyme that catalyzes the reaction of fructose to fructose 1-phosphate, instead of the glycolytic intermediate glucose 6-phosphate. 36 Fructose 1-phosphate can then enter fructolysis, bypassing the regulated phosphofructokinase enzymatic reaction in glycolysis. 36 This inadvertently provides fructose with a less regulated metabolism 36,42 .
Thus, high fructose consumption can result in unchecked carbon flow towards acetyl CoA 36 . If this exceeds the demands of the Krebs cycle, then the carbon will be directed to fatty acid biosynthesis via acetyl CoA carboxylase 36 .
Chronic over consumption of fructose sugars increases de novo lipogenesis, resulting in elevated serum triglyceride (TAG) concentrations in adults of 80-200% 43-46 . Excessive fructose consumption may lead to adverse metabolic effects, such as increased visceral adiposity or dyslipidemia 47 .

Fructose in relation to body composition
When compared to glucose, consumption of fructose does not attenuate circulating levels of ghrelin, an appetite stimulating hormone 48,49 . Thus, fructose intake may lead to increased food intake 50 . In fact, several studies examining fructose consumption have reported increased caloric intake with higher energy consumption of fructose, which may be responsible for changes in body composition [51][52][53] . Many of these studies have focused on fructose in beverages, although a cross-sectional study in college students found elevated fasting hunger with higher total dietary fructose intake 54 .
To observe the effects of fructose consumption on body composition, Lowndes et al. 51 conducted a randomized, prospective, parallel group, blinded study in which participants consumed 3 different levels (8%, 18% and 30% kcals/day) of HFCS or sucrose. There were no significant differences when looking at sucrose intake vs HFCS intake, therefore the participants were pooled for analyses 51

Total and added sugars in relation to body composition
Research from Rikkers et al. 59 attempted to estimate Australian refined sucrose consumption over decades and concluded that it was not possible to produce reliable data. In response to this Barclay and Brand-Miller released a report in which they demonstrate the "Australian paradox", where sugars consumption declined over the same period that obesity rates increased 60 . This is true in Americans as well, where from 1977 to 2012 obesity rates have increased and sugars intake increased through 1998, but has since dropped to similar consumption levels seen in 1991 13,14 . Figure  Several reviews have explored the relationship of total and/or added sugars and body composition [63][64][65][66] . In 2003, Saris et al. 64 concluded that there is little evidence that sugars have direct negative effects on body weight control. However, the combination of frequent consumption of sugar sweetened beverages (SSB) with an inactive lifestyle, reduces the metabolic need for fat as fuel, potentially leading to considerable increases in weight 64 .
In contrast, 10-years later Te Morenga et al. 63 conducted a systematic review and meta-analysis of randomized control trials and cohort studies and concluded among free living people, intake of free sugars or SSB was a determinant of body weight. They reviewed 30 randomized control trials and concluded that by reducing intake of free sugars in ad libitum diets, there was an average of 0.8kg reduction in weight 63 . When increasing free sugar intake there was an association with a comparable 0.75kg increase in weight 63 . Increases in SSB at a one year follow up in prospective studies, concluded a higher odds ratio of being overweight or obese with higher consumption of SSB when compared to lower consumption 63 .
In 2009 van Baak et al. 67 concluded that there are inverse associations between content of sugars and body adiposity and weight using randomized control trials that replaced fat in the diet to increase carbohydrate intake 65,67 . In another review, Ruxton et al. 66 examined whether sugar consumption is detrimental to health. They concluded similar findings that sugars intake in place of fat intake increases body weight 65,66 .
Cross-sectional studies suggest there may not be a relationship between BMI and sugars intake. To investigate if the uptrend of obesity prevalence in the USA was associated with dietary sugar intake, Song et al. 68 used NHANES I and III data to compare intakes. They concluded that the primary contributor to BMI in all age groups was energy intake 68 . Total sugars intake was a non-predictor for BMI in all age groups 68 . A similar cross-sectional study examining data from NHANES 1999-2006 categorized sugars intake of participants ≥18years old into five categories: <5%, 5%-<10%, 10%-<17.5%, 17.5%-<25%, ≥25% of total calories 69 . There were no significant differences in BMI or WC among the groups 69 . Similarly, a cross-sectional study used NHANES data from 1999-2004 to categorize sugars intake of 12-18year olds 19 using the same five categories above 69 . There were no significant differences in BMI z-score among the categories 19 .

Sugar Sweetened Beverages and Body Composition:
Sugar-sweetened beverages include soft drinks, fruit drinks, energy and vitamin waters and are composed of naturally derived caloric sweeteners such as sucrose, HFCS and fruit juice concentrates 23  To assess disparities in calorie intake between SSB consumers and nonconsumers and determine associations with obesity and overweight-obesity, a New York City population study was conducted in 488 adults 53 . Consumers of SSBs consumed on average 193kcals/day from SSB, equating to roughly 10% of total caloric intake 53 . When compared to non-consumers, adults who consumed SSBs consumed on average 572 kcals more, possibly due to greater SSB consumption 53 .
However not all of these calories can be attributed to SSB consumption, so it is believed that higher intake of fructose may disrupt regulating hormones, as previously discussed 48,49,54 . Lastly, this study concluded that each 10oz serving of SSBs increased obesity likelihood and increased overweight-obesity likelihood 53 . A cross-sectional study using NHANES data suggests that there is also an increased likelihood of being overweight or obese, defined by BMI, with consumption of SSBs 71 .
To compare consumption of SSB with consumption of an isocaloric milk and with consumption of a non-caloric SSB on changes in total fat mass, Maersk et al. 72 conducted a 6-month randomized intervention with four groups (control was given water). On average, SSB consumption had significantly higher liver fat (132%-143% change) and visceral fat (24-31% change), when compared to the non-caloric SSB and isocaloric milk 72 . However, total fat mass was not different across the four groups 72 .
To explain the possible associations between SSB and increase overweight and obesity, Bachman et al. 73 reviewed four possible mechanisms: (1) excess calories, (2) glycemic index and glycemic load, (3) lack of effect of liquid calories on satiety, and (4) displacement of milk. The evidence on whether liquid and solid foods differ in the effects on caloric compensation is conflicting, and research needs to more carefully consider the many factors that influence satiety 73 . However, evidence is inconsistent about whether this displacement has implications on obesity 73 . This review concluded that the evidence regarding SSB consumption and obesity remains inconclusive 73 .
A positive relation between added sugars consumption and total energy intake is observed in many cross-sectional studies 52,[74][75][76][77][78][79] . Despite this, some cross-sectional studies have shown inverse associations between added sugars consumption and body weight or BMI [80][81][82] . Next, a review of 31 short term studies (<1day) found that only 15 studies show an association between low glycemic index meals and greater satiety and reduced hunger; while 16 studies reported reduced satiety or no differences with low glycemic index foods 83 . During the same time that SSB consumption increased, milk consumption in children decreased 74,84 . Similarly, other reviews have also deemed SSB consumption and obesity as inconclusive based on current evidence 85,86 .
Despite the previous review 73 , research conducted in children shows a relationship between the consumption of SSB and BMI 75 . A prospective observational study has shown that for each additional serving of SSB, there is an average increase in BMI of 0.24kg/m 2 in children 75 . There was also an increased frequency of obesity observed in this sample of 548 ethnically diverse school children 75 . Another review observed the relation between SSB and body weight in children and adults 87 . Cohort studies and randomized control trials were included, 20 in children 12 in adults 87 . In cohort studies, 1 serving increment of SSB was associated with a 0.06 and 0.05-unit increase in BMI in children and a .22kg and .12kg increase in weight in adults over one year in random and fixed effects models, respectively 87  Despite all previous research on SSB, added sugars and fructose consumption, a review by Dolan et al. 89 concluded that there is no convincing evidence from long term studies that fructose ingestion up to 100 g/day instead of glucose or sucrose is associated with an increase in body weight. Similarly, they did not find any associations with blood lipids when consuming fructose up to 100g/day 89 .

Fructose and blood lipids
As discussed earlier, higher intake of fructose may be associated with increased hepatic de novo lipogenesis, 43,45,46,90 however, more evidence is needed. 91 The metabolic fate of fructose is described in Figure 3 92 .  respectively. Both the 7.5% fructose diet and the 15% energy fructose diet produced statistically higher TC and LDL-C then when participants consumed 0% energy from fructose 97 .
However one year later, in 1984, Crapo et al. 100 did not have these same findings when observing fructose consumption in healthy individuals. The diet consisted of between 63 to 99grams fructose or roughly 24% of total carbohydrates consumed (with carbohydrates consumed at 55% of total caloric intake) 100 . When consumed for 2-weeks, TC decreased from baseline to 14 days (188mg/dL to 173mg/dL, p<.05) 100 . According to the sample diet, roughly 13% of calories were consumed as fructose, similar to the upper level used in a study one year prior 97,100 .
One main difference was that the participants used in the studies were different in terms of health status, one study observing changes in hyperinsulinemic participants 97 and the other in healthy individuals 100 . The other is that one study, which found significance, was 5-weeks in length 97 , while the other, which did not find significance, was only 2-weeks 100 .
In 1985, Crapo et al. 98 repeated this 2-week trial in diabetic subjects with the diet consisting of between 63 to 99grams of fructose or roughly 24% of total carbohydrates consumed (with carbohydrates consumed at 55% of total caloric intake). 98 He concluded that there was no significant difference in TC after 14-days of consuming a fructose in diabetic subjects. 98 However, the length of the study was rather short, limiting the potential for changes to occur.
Bantle et al. 99 conducted a cross-over study in which participants consumed, in random order, a fructose diet at 17% energy needs and an isoenergetic diet with glucose at 14% and with fructose at less than 3% energy needs 99 . Despite consuming fructose at higher levels and for a longer duration than previous studies 98,100 , no longterm changes were observed in TC or LDL-C 99 . However, TAG were higher in men consuming fructose at 17% of energy (p<.05) 99 .
Jameel et al. 93 conducted a randomized, single blind, cross-over study in 14 men and women. Three different isocaloric beverages (50 grams fructose, 50 grams glucose and 50 grams sucrose dissolved in water) were served on three different occasions. 93 Consumption of fructose led to an initial significant increase in TC at 30min and 60min, when compared to glucose and sucrose. 93 However, at 120min the increase was no longer significantly different from glucose and sucrose. 93 Similarly, LDL-C was significantly increased with consumption of fructose at 30 min and 60 min, but not at 120 minutes. 93 Overall, plasma TC area under the curve (AUC) and LDL-C AUC were higher when consuming fructose. 93 Next, Aeberli et al. 94 conducted a randomized, double blind, cross-over study in which nine males consumed a medium fructose diet (40g/day), a high fructose diet (80g/day), a high glucose diet (80g/day) and a high sucrose diet (80g/day) for three weeks each with a four-week wash-over between diets. Compared with the high glucose group, the all fructose containing diets had higher TC and LDL-C at 3weeks. 94 To compare the effects of fructose, glucose and HFCS on risk factors for CVD, 48 adults enrolled in a three tiered study. 95 First, participants completed a 3.5day inpatient baseline testing, while consuming an energy balanced diet. 95 Next, participants completed 12 outpatient days consuming 25% of their energy requirements via a glucose, fructose or HFCS sweetened-beverage (n=16 per group) with an ad libitum diet. 95 Lastly, participants completed a final 3.5day inpatient testing, in which they consumed an energy-balanced diet containing 25% energy from sugar-sweetened beverages. 95 Fasting LDL-C was significantly increased during fructose consumption (+0.29 ±0.082 mmol/L) and HFCS (+0.42±0.11mmol/L) but not in the glucose sweetened beverages group. 95 Participants consuming fructose and HFCS also had higher postprandial concentrations of LDL-C 95 .
To determine if fructose had adverse effects on metabolic outcomes, 14 healthy participants consumed an isoenergetic diet consisting of either 20% energy from fructose or < 3% of energy from fructose with the remaining carbohydrate source from starch for 28 days each 96  A review by Zhang et al. 47 concluded that higher consumption of fructose (>100 grams) in place of other carbohydrate sources caused higher levels of TC and LDL-C (13.0 mg/dL and 11.6 mg/dL respectively). These results did not emerge when participants were consuming less than 100 grams of fructose daily 47 .
Contradictory to this, Silbernagel et al. 56 conducted a study using very high fructose (150g) and very high glucose (150g) and did not observe any significant changes in TC or LDL-C. This could be due to increased absorption of sugars seen when consumed in a sugar rich meal 36 .
Using a within subjects cross-over design, healthy participants consumed three meals a day with 30% of caloric intake from either glucose or fructose on two separate visits 101 . They concluded that plasma TAG concentrations were elevated after ingestion of fructose-sweetened beverages with meals when compared with glucosesweetened beverages 101 .
In a longer 4-week long randomized single blind intervention trial the effects of a very high fructose and a very high glucose hyperenergetic diet on plasma lipids and body fat were explored 56  In a randomized single blind cross-over study, participants were tested on three different occasions where they consumed either 50grams of fructose, glucose or sucrose with blood collection at 0, 30, 60, and 120 minutes no differences in TAG were observed after consumption of all three sugars 93 . Despite no differences observed in TAG, postprandial HDL-C significantly increased 93 , suggesting that HDL-C can change regardless of a lack of change in TAG.

Total and added sugars in relation to cholesterol levels
To conduct a systemic review and meta analyses of dietary free sugars and lipids, Te Morenga et al. 106 analyzed 39 randomized control trials. When comparing high and low intakes of sugars, higher intake raised TC (mean difference 0.16mmol/L) and LDL-C (mean difference 0.12mmol/L) 106 . Cross-sectional studies on NHANES data suggest that added sugars may be detrimental to health 19,69 . NHANES data from 1999-2006 in adults aged 18 and older show a linear trend between added sugars intake and LDL-C in women, but not men 69 . NHANES data from 1999-2004 in children ages 12-18 years old show a positive correlation between LDL-C and added sugars intake 19 . Among the lowest and highest consumers of added sugars, LDL-C was 9% higher in participants with higher intake of added sugars after controlling for several covariates 19 . Using more recent NHANES data, 2005-2010, no association with TC or LDL-C were observed 9 .

SSB in relation to cholesterol levels
Data from the Coronary Artery Risk Development in Young Adults (CARDIA) study were used to assess the relations of low-and whole-fat milk, fruit juice and SSB consumption with cardio metabolic risk factors 107 . Higher SSB consumption was associated with higher risk of elevated LDL-C and TAG, while intake of whole-fat milk was associated with lower risk of elevated TAG 107 .
To observe the effects of SSB on LDL particle size, researchers conducted a prospective, randomized control trial in twenty-nine subjects 108 . Subjects participated in six three week interventions where they consumed 600mL SSB of 40 and 80grams of glucose, fructose or sucrose 108 . LDL particle size was reduced after high fructose and high sucrose and a more atherogenic LDL subclass distribution was seen when consuming any of the fructose-containing SSB, when compared to glucose 108 . Similar shifts in LDL particle size and subclass were seen in a 10-week long investigation where participants consumed 25% energy from with fructose sweetened or glucose sweetened beverages, with only the fructose sweetened beverages altering LDL particle size 109 .

Summary
In conclusion, higher consumption of fructose sugars may alter both body fat and blood lipids in healthy participants, although more research is needed. Current consumption levels of sugars are higher than recommendations among all age groups.
With high consumption levels, and high rates of obesity and dyslipidemia, more research is needed to determine if there is a relationship between consumption of sugars, specifically fructose, and markers of obesity and dyslipidemia.   Table 1a: Random sampling of women and the associated differences by gender, Sample 1.