Determining the Effect of Small Doses of Fructose and Its Epimers on Glycemic Control

Determining the Effect of Small Doses of Fructose and its Epimers on Glycemic Control

Jarvis Clyde Noronha

A thesis submitted in conformity with the requirements for the degree of Master of Science Department of Nutritional Sciences University of Toronto

Determining the Effect of Small Doses of Fructose and its Epimers on Glycemic Control

Jarvis Clyde Noronha

Master of Science

Department of Nutritional Sciences University of Toronto

2017 Abstract

Given that sugars have emerged as the dominant nutrient of concern in diabetes, there is a need for the development of alternative sweeteners. To assess the role of small doses (5g, 10g) of fructose and allulose on postprandial glucose metabolism, we conducted an acute randomized controlled trial in individuals with type 2 diabetes. We found that small doses of allulose, but not fructose, modestly reduced the postprandial glycemic response to a 75g oral glucose load. To assess whether low-dose (< 50-g/day) fructose and all its epimers (allulose, tagatose and sorbose) lead to sustainable improvements in long-term glycemic control, we conducted a systematic review and meta-analysis of controlled feeding trials. The available evidence suggested that fructose and tagatose led to significant reductions in HbA1c and fasting glucose. Our findings highlight the need for long-term randomized controlled trials to confirm the viability of fructose and its epimers as alternative sweeteners.

iii

Table of Contents

Contents

Abstract ...... ii

Table of Contents ...... iii

List of Tables ...... vii

List of Figures ...... ix

Chapter 1 : Introduction ...... 1

Chapter 2 : Literature Review ...... 3

2.1 Diabetes ...... 4

2.1.1 Definition & Complications of Diabetes ...... 4

2.1.2 Classification of Diabetes ...... 4

2.1.3 Prevalence and Economic Burden of Diabetes ...... 4

2.1.4 Diagnosis of Diabetes ...... 5

2.1.5 Postprandial Hyperglycemia & Type 2 Diabetes ...... 5

2.1.6 OGTT-Derived Indices of Insulin Secretion, Sensitivity & β-cell compensation ...... 6

2.2 Fructose & its Epimers (Allulose, Tagatose, Sorbose) ...... 10

2.2.1 Background ...... 10

2.2.2 Absorption, Metabolism & Excretion ...... 11

2.2.3 Potential for Improving Postprandial Hepatic Glucose Metabolism ...... 18

2.3 Gaps in the Literature ...... 29

Chapter 3 : Rationale, Objectives & Hypotheses ...... 30

3.1 Rationale ...... 31

3.2 Objectives ...... 31

3.3 Hypotheses ...... 32

Chapter 4 : Effect of Small Doses of Fructose and Allulose on Postprandial Carbohydrate Metabolism in Type 2 Diabetes: A Randomized Crossover Trial ...... 33

4.1 Abstract ...... 34

4.2 Introduction ...... 35

4.3 Research Design and Methods ...... 36

4.3.1 Participants ...... 36

4.3.2 Trial Design ...... 36

4.3.3 Treatments ...... 37

4.3.4 Protocol ...... 37

4.3.5 Plasma Glucose and Insulin Analysis ...... 38

4.3.6 Outcomes ...... 38

4.3.7 Data Analysis ...... 38

4.4 Results ...... 40

4.4.1 Flow of Study Participants ...... 40

4.4.2 Participant Characteristics ...... 42

4.4.3 Effect of Small Doses of Fructose and Allulose on Plasma Glucose and Insulin Responses ...... 43

4.4.4 Dose-Response Analysis ...... 45

4.4.5 Equivalence Test ...... 46

4.4.6 Effect of Small Doses of Fructose and Allulose on Secondary Outcomes ...... 47

4.4.7 Subgroup Analysis...... 50

4.5 Discussion ...... 53

4.5.1 Summary of Findings ...... 53

4.5.2 Findings in Context of Previous Research ...... 53

4.5.3 Mechanism of Action ...... 54

4.5.4 Clinical Implications ...... 55

4.5.6 Strengths & Limitations ...... 55

4.6 Conclusions ...... 56

Chapter 5 : Effect of Small Doses of Fructose and its Epimers on Markers of Long- term Glycemic Control: A Systematic Review and Meta-Analysis of Controlled Feeding Trials ...... 57

5.1 Abstract ...... 58

5.2 Introduction ...... 60

5.3 Methods ...... 61

5.3.1 Data Sources and Searches ...... 61

5.3.2 Study Selection ...... 64

5.3.3 Data Extraction ...... 64

5.3.4 Data Synthesis and Analysis ...... 64

5.3.5 Quality of Evidence Assessment ...... 65

5.4 Results ...... 66

5.4.1 Search Results ...... 66

5.4.2 Trial Characteristics ...... 67

5.4.3 Effect of Low-dose Fructose & its Epimers on HbA1c ...... 70

5.4.4 Effect of Low-dose Fructose & its Epimers on Fasting Glucose ...... 70

5.4.5 Effect of Low-dose Fructose & its Epimers on Fasting Insulin ...... 71

5.4.6 Publication Bias ...... 73

5.4.7 GRADE Assessment ...... 73

5.5 Discussion ...... 75

5.5.1 Summary of Findings ...... 75

5.5.2 Findings in Context of Previous Research ...... 75

5.5.3 Mechanism of Action ...... 76

5.5.4 Clinical Implications ...... 77

5.5.5 Consideration of a Dose Threshold for Harm ...... 78

5.5.6 Strengths & Limitations ...... 79

5.6 Conclusions ...... 79

Chapter 6 : General Discussion ...... 87

6.1 Summary ...... 88

6.2 Clinical Implications ...... 88

6.3 Strengths & Limitations ...... 89

6.4 Future Directions ...... 90

6.5 Conclusions ...... 91

References ...... 93

vii

List of Tables Chapter 2

Table 2.1 Classification of diabetes…………………………………………………...... ……4

Table 2.2. Canadian Diabetes Association diagnostic criteria for diabetes………….…...5

Table 2.3. Evidence evaluating absorption, metabolism and excretion of allulose…… 13

Table 2.4. Evidence evaluating absorption, metabolism and excretion of tagatose…..16

Table 2.5. Proof-of-concept studies evaluating the ‘catalytic’ mechanism of fructose…20

Table 2.6. Proof-of-concept studies evaluating the ‘catalytic’ mechanism of allulose…22

Table 2.7. Proof-of-concept studies evaluating the ‘catalytic’ mechanism of tagatose..23

Table 2.8. Proof-of-concept studies evaluating the ‘catalytic’ mechanism of sorbose…24

Table 2.9. Acute clinical studies evaluating the efficacy of low-dose fructose on postprandial carbohydrate metabolism……………………………………………………...26

Table 2.10. Acute clinical studies evaluating the efficacy of low-dose allulose on postprandial carbohydrate metabolism……………………………………………………...27

Table 2.11. Acute clinical studies evaluating the efficacy of low-dose tagatose on postprandial carbohydrate metabolism……………………………………………………...28

Chapter 4

Table 4.1. Participant Characteristics……………………………………...……….……….42

Table 4.2. Individual and Pooled Dose Analysis for the Effect of ‘Catalytic’ Doses of Fructose on Primary, Secondary and Exploratory Outcomes …………...………….…...48

Table 4.3. Individual and Pooled Dose Analysis for the Effect of ‘Catalytic’ Doses of Allulose on Primary, Secondary and Exploratory Outcomes….…………....…………….49

viii

Chapter 5

Table 5.1. Search strategy for studies assessing the effect of fructose and its epimers (allulose, tagatose and sorbose) on markers of long-term glycemic control…….…...…62

Table 5.2 Characteristics of controlled feeding trials investigating the effect of low-dose fructose, allulose and tagatose on glycemic control………………………….………...….68

Table 5.3. GRADE Quality of Evidence Assessment for the effect of low-dose fructose, allulose and tagatose on markers of long-term glycemic control……….……….…..…...74

List of Figures Chapter 2

Figure 2.1. Open chain chemical structures of fructose, allulose, tagatose and sorbose………………………………………………………………………………………….10

Figure 2.2. Short-term regulation of GK activity by fructose and its epimers….….…….19

Chapter 4

Figure 4.1. Flow of participants through the various phases of the trial…………..……...41

Figure 4.2 Effect of small doses (5g, 10g) of fructose and allulose on glucose and insulin responses to a 75g-oral glucose tolerance test (75g-OGTT)……………………….…….44

Figure 4.3 Linear dose-response analysis of small doses of fructose and allulose on plasma glucose (top) and insulin iAUC (bottom)………………..……….……….………...45

Figure 4.4 Equivalence test comparing the effect of allulose to fructose on plasma glucose iAUC……………………………………………………………………….…………..46

Figure 4.5. Subgroup analysis for the effect of small doses of fructose on plasma glucose iAUC……………………………………………………………………....……………...……..51

Figure 4.6. Subgroup analysis for the effect of small doses of allulose on plasma glucose iAUC……………………….………………………………………………...…………...……..52

Chapter 5

Figure 5.1. Flowchart of literature search for the effect of small doses of fructose and its epimers (allulose, tagatose and sorbose) on glycemic endpoints (HbA1c, fasting glucose and fasting insulin)……………………………………………………………...…..………….66

Figure 5.2. Summary of the pooled effect estimates from controlled feeding trials assessing the effect of low-dose fructose, allulose and tagatose on markers of long-term glycemic control………………………………………………………………………………..72

Supplementary Figure 5.2. Forest plot for the effect of low-dose fructose (< 50g/day) on

HbA1c…………………………………………………………………….….………...………...82

Supplementary Figure 5.3. Forest plot for the effect of low-dose fructose (< 50g/day) on fasting glucose……………………………………………………..……….……...…………..83

Supplementary Figure 5.4. Forest plot for the effect of low-dose fructose (< 50g/day) on fasting insulin………………………………………………………...………….…..…………84

Supplementary Figure 5.5. Forest plot for the effect of low-dose allulose (< 50g/day) on HbA1c, fasting glucose and fasting insulin……….……………………………….…….….85

Supplementary Figure 5.6. Forest plot for the effect of low-dose tagatose (< 50g/day) on HbA1c, fasting glucose and fasting insulin…………………………………………..….86

Chapter 1 : Introduction

Diabetes is a global health problem and is on the rise. The number of people living with diabetes in the world has risen from 108 million in 1980 to 415 million in 20151, 2. It was estimated that approximately 642 million people will be living with diabetes worldwide by 20402. Poorly managed diabetes can result in serious medical complications and increase the overall risk of premature death. In 2015, 1.6 million people died from diabetes making it the sixth leading cause of death worldwide1. Although multiple dietary and lifestyle factors have been implicated in the etiology of diabetes, sugars have emerged as the dominant nutrient of concern. This has created a need for the development of caloric and low-caloric alternative sweeteners.

The fructose moiety of sucrose and high-fructose corn syrup has particularly come under attack as a potent driver of type 2 diabetes3-5. This is a result of undue emphasis on ecological observations, animal models of fructose overfeeding and select human studies assessed in isolation. The important biological mechanisms by which small doses of fructose may assist in the metabolic handling of glucose have largely been ignored. An emerging literature has also suggested that the low-caloric epimers of fructose (allulose, tagatose and sorbose) act in a similar way to fructose by enhancing postprandial hepatic glucose metabolism and reducing the glycemic response to high glycemic index foods. Considering that this body of knowledge is less recognized and appreciated, this thesis will aim to explore the effect of small doses of fructose and its epimers on glycemic control.

Specifically, this thesis will: (1) provide a brief overview of diabetes, describe various characteristics of fructose and its epimers and summarize their potential in improving glycemic control, (2) report results from an acute randomized double-blind crossover trial assessing the acute effect of small doses of fructose and its c-3 epimer, allulose on postprandial glucose metabolism in individuals with type 2 diabetes, and (3) report results from a systematic review and meta-analysis synthesizing the evidence of the long-term effect of low-dose fructose and all its epimers (allulose, tagatose and sorbose) on markers of glycemic control.

Chapter 2 : Literature Review

2.1 Diabetes

2.1.1 Definition & Complications of Diabetes

Diabetes is a group of metabolic disorders characterized by chronic hyperglycemia. As insulin acts to normalize glucose in the blood by increasing uptake into cells, dysregulation of insulin (defective insulin secretion, defective insulin action, or both) leads to chronic hyperglycemia. Chronic hyperglycemia is associated with increased risk of microvascular complications (diabetic nephropathy, neuropathy and retinopathy) and macrovascular complications (coronary artery disease, peripheral arterial disease, and stroke)6. Macrovascular complications are the major cause of death in people with diabetes7.

2.1.2 Classification of Diabetes

The classification of diabetes into type 1 diabetes, type 2 diabetes and gestational diabetes is summarized in Table 2.1.

Table 2.1 Classification of diabetes6

Classification Description

Result of an autoimmune process that destroys pancreatic β- Type 1 diabetes cells for which the cause is currently unknown Begins with insulin resistance. As the disease progresses, a Type 2 diabetes lack of insulin secretion may also develop as a result of β-cell failure A temporary condition that occurs during pregnancy where Gestational diabetes women develop glucose intolerance

2.1.3 Prevalence and Economic Burden of Diabetes

The International Diabetes Federation has declared diabetes as one of the largest global health emergencies of the 21st century. In 2015, it was estimated that 415 million people worldwide, or 8.8% of adults aged 20-79, had diabetes. If trends continued, it was predicted that 642 million people, or one in 10 adults, will have diabetes by 20402. In high-income countries, type 2 diabetes is the most common form affecting approximately 87% to 91% of all people with diabetes. 7 to 12% are estimated to have type 1 diabetes and 1% to 3% have other types of diabetes2. High prevalence rates of

5 diabetes place an economic burden on the healthcare system. In Canada, it was estimated that the economic burden of diabetes in 2010 was $12.2 billion and was projected to rise by another $4.7 billion by 20208.

2.1.4 Diagnosis of Diabetes

The diagnostic criteria for diabetes according to the Canadian Diabetes Association guidelines are summarized in Table 2.26. These criteria are based on venous blood samples.

Table 2.2. Canadian Diabetes Association diagnostic criteria for diabetes6

Diagnostic criteria for Parameter diabetes* Fasting plasma glucose (mmol/L) > 7.0 mmol/L

HbA1c (%) > 6.5% (in adults) 2-hour plasma glucose in a 75g-oral glucose tolerance test > 11.1 mmol/L (mmol/L)

Random plasma glucose (mmol/L) > 11.1 mmol/L

*If a single laboratory test results in the diabetes range in the absence of symptoms associated with hyperglycemia, a repeat confirmatory laboratory test (fasting plasma glucose, HbA1c, 2h-oral glucose tolerance test) must be done on another day. It is preferable that the same test be repeated for confirmation. If symptoms of hyperglycemia are present with a single laboratory test rest in the diabetes range, the diagnosis has been made and a confirmatory test is not required before treatment is initiated. 2.1.5 Postprandial Hyperglycemia & Type 2 Diabetes

Postprandial hyperglycemia refers to excess blood glucose circulating in the blood after a meal. Individuals with type 2 diabetes commonly experience significant and prolonged hyperglycemia in response to carbohydrate ingestion, typically returning to baseline glucose levels after two to three hours or longer. Postprandial hyperglycemia is harmful in people with type 2 diabetes as it has been associated with markers of cardiovascular disease (oxidative stress, carotid intima-media thickness and endothelial dysfunction), as well as cardiovascular risk and outcomes9. Postprandial hyperglycemia has also been linked to retinopathy and cognitive dysfunction in the elderly9. Factors which contribute to postprandial hyperglycemia include: (1) excessive rates of endogenous glucose production that fail to suppress after eating; (2) delayed insulin secretion; (3) impaired glucagon suppression; (4) impaired insulin-induced stimulation of hepatic and

6 extra-hepatic glucose uptake; and (5) lower rates of hepatic glycogen synthesis due to decreased hepatic glucokinase activation10.

Although there is compelling evidence from large randomized clinical trials that improving glycemic control in type 2 diabetes reduces the risk of microvascular complications but not macrovascular complications, achieving optimal glycemic control 11-15 is still fundamental in diabetes management . Glycated hemoglobin or HbA1c is the most robust clinical marker for evaluating the three-month average plasma glucose concentration. Hemoglobin is a protein found within red blood cells and undergoes a 16 non-enzymatic glycation reaction upon exposure to glucose . The HbA1c test is limited to a three-month average because the lifespan of a red blood cell is ~120 days. In diabetes, the higher amounts of glycated hemoglobin indicate poorer control of blood glucose levels. The relative contribution of postprandial glycemia to overall glycemia has been shown to increase as the HbA1c level decreases. One study showed that postprandial glycemia contributed ~70% to HbA1c when people had HbA1c levels <7.3% 17 and ~40% when HbA1c was >9.3% . Similarly, another study showed that postprandial glycemia accounted for ~80% of HbA1c when HbA1c was <6.2% and ~40% when HbA1c was >9.0%18. Taken together, management of postprandial hyperglycemia through non- pharmacologic and pharmacologic modalities are important, especially early in the evolution of type 2 diabetes.

2.1.6 OGTT-Derived Indices of Insulin Secretion, Sensitivity & β-cell compensation

Given that insulin resistance is a hallmark of type 2 diabetes, methods to accurately measure insulin secretion, insulin sensitivity and β-cell compensation are important in evaluating the management of type 2 diabetes in clinical and research settings. The hyperinsulinemic-euglycemic clamp test is considered the “gold standard” for in vivo quantification of insulin sensitivity however, this technique is time consuming, costly and experimentally demanding19, 20. Frequently sampled intravenous glucose tolerance tests (FSIVGTT) can be utilized to measure insulin secretion and β-cell compensation. Although less labor intensive than the hyperinsulinemic-euglycemic clamp technique, the FSIVGTT can also be expensive as glucose and insulin measurements are usually made from multiple blood samples taken over a 3-h period20, 21. The oral glucose

7 tolerance test (OGTT) is a more widely used procedure to evaluate carbohydrate tolerance and is less expensive than the hyperinsulinemic-euglycemic clamp test and the FSIVGTT20, 22. Because plasma glucose and insulin responses during the OGTT reflect the ability of pancreatic β-cells to secrete insulin and the sensitivity of tissues to insulin, several OGTT-derived indices have been validated against “gold standard” techniques to assess insulin secretion, insulin sensitivity and β-cell function. Three of these indices include: Early Insulin Secretion Index (∆PI30-0/∆PG30-0), Matsuda Insulin

Sensitivity Index (Matsuda ISIOGTT) and Insulin Secretion-Sensitivity Index-2 (ISSI-2).

Early Insulin Secretion Index (∆PI30-0/∆PG30-0)

Insulin secretion may be measured by the FSIVGTT21. Insulin release following an acute glucose stimulus (i.e. intravenous glucose) occurs in two phases. A rapid first- phase insulin release is observed within 10-15min after glucose administration followed by a slower second phase insulin release over the next 1-2h. The timing of insulin release with a rapid first phase peak is important in normalizing blood glucose. Early defects in β-cell function and insulin release that occur with the development of impaired glucose tolerance and type 2 diabetes result in the reduction or even absence of first phase insulin release allowing hyperglycemia to develop over the following 2-h. As type 2 diabetes progresses over time, the second phase insulin release declines causing blood glucose levels to rise and remain above the normal range23, 24.

The Early Insulin Secretion Index (also known as the Insulinogenic Index) is a validated OGTT-derived measure of insulin secretion. It is calcuIated as the change in plasma insulin (PI) from 0 minutes to 30 minutes divided by the change in plasma glucose (PG) over the same period (∆PI30-0/∆PG30-0). Phillips et al. validated the early insulin secretion index by demonstrating a modest correlation (r=0.61, p<0.001) with the first phase insulin release during the intravenous glucose tolerance test in 85 normal subjects and 23 subjects with impaired glucose tolerance25, 26.

Matsuda Insulin Sensitivity Index (Matsuda ISIOGTT)

The hyperinsulinemic-euglycemic clamp test is considered the “gold standard” method to assess insulin sensitivity19, 27. In this test, a continuous intravenous infusion of insulin is provided to acutely raise and maintain plasma insulin concentration at 100µU/ml

(“hyperinsulinemic”). Meanwhile, the plasma glucose concentration is held constant at basal levels by a variable glucose infusion (“euglycemic”). When a steady-state is achieved, the glucose infusion rate equals glucose uptake by all the tissues in the body19, 27. As the physiologic regulation of glucose homeostasis requires the simultaneous and coordinated roles of pancreatic β-cells, liver and peripheral tissues (i.e. muscle), this test can provide a reliable measure of tissue insulin sensitivity. Incorporation of radioactive-labeled glucose during this test can further measure glucose metabolism in individual organs27.

The Matsuda Insulin Sensitivity Index (Matsuda ISIOGTT) is a validated OGTT-derived measure of whole-body insulin sensitivity. It is calculated using plasma glucose and insulin concentrations following the 75g-OGTT as follows: 10,000/square root of [fasting glucose X fasting insulin X mean glucose X mean insulin during OGTT]. Matsuda and DeFronzo validated this whole-body insulin sensitivity index with the hyperinsulinemic- euglycemic clamp. They found the Matsuda ISIOGTT to be highly correlated (r=0.73, p<0.0001) with the rate of whole-body glucose disposal during the hyperinsulinemic- euglycemic clamp in 153 subjects with varying degress of glucose tolerance (62 subjects with normal glucose tolerance, 31 subjects with impaired glucose tolerance and 60 subjects with type 2 diabetes)28.

Insulin Secretion-Sensitivity Index-2 (ISSI-2)

The maintenance of normal glucose homeostasis requires that pancreatic β-cells adjust for changes in whole-body insulin sensitivity through a proportionate and reciprocal change in insulin secretion29, 30. It has been proposed that this resultant negative feedback loop between insulin secretion and insulin sensitivity mathematically characterizes as a hyperbolic function (i.e. y= constant/x)31. This negative feedback loop was confirmed during the FSIVGTT using measures of insulin secretion (acute-insulin- 32 response-to-glucose, AIRg) and sensitivity (insulin sensitivity index, SI) . The product of

AIRg and SI are termed the disposition index and has been used to provide a measure of β-cell compensation.

ISSI-2 is a validated OGTT-derived measure of β-cell compensation. It is calculated by taking the product of 1) insulin secretion as measured by the ratio of total area-under-

the-insulin-curve (AUCins) to the total-area-under-the-glucose curve (AUCglu) and 2) insulin sensitivity as measured by Matsuda ISIOGTT, such that ISSI-2 = (total AUCins/glu x

Matsuda ISIOGTT). In the search for an OGTT-derived index analogous to the disposition index, Retnakaran et al. compared the ISSI-2 and (∆PI30-0/∆PG30-0)/fasting insulin with the disposition index. They demonstrated that the ISSI-2 showed modestly stronger association with the disposition index compared to (∆PI30-0/∆PG30-0)/fasting insulin (r=0.24, p=0.0003 vs. r=0.21, p=0.0022) in 213 children without diabetes (122 boys and 91 girls)33.

2.2 Fructose & its Epimers (Allulose, Tagatose, Sorbose) 2.2.1 Background

The open chain chemical structures of fructose, allulose (also known as psicose), tagatose and sorbose are illustrated in Figure 2.1. Fructose is a ketohexose which contains three stereocenters at the 3rd, 4th and 5th carbon. A switch in the position of the -H and -OH group at each of these stereocenters results in a unique epimer. As a result, an inversion at the 3rd, 4th and 5th carbon results in allulose (c-3 epimer of fructose), tagatose (c-4 epimer of fructose) and sorbose (c-5 epimer of fructose), respectively. Due to its configurational stereochemistry, fructose, allulose, tagatose and sorbose exist either in the D- isomeric form or the L- isomeric form. Since naturally occurring sugars predominantly exist as D- isomers, our discussion about fructose and its epimers throughout this thesis will be focused on D-fructose, D-allulose (also known as D- psicose), D-tagatose and D-sorbose.

Figure 2.1. Open chain chemical structures of D-fructose, D-allulose, D-tagatose and D-sorbose.

Fructose exists in foods either as free fructose (i.e. honey, high fructose corn syrup) or fructose bound to glucose (i.e. sucrose). Some food sources of fructose include: agave nectar, honey, maple syrup, fruit and fruit juices34. It is the sweetest of all naturally occurring carbohydrates (~120% as sweet as surcrose) which is one of the reasons why it is commercially used in foods35. Fructose also elicits a lower postprandial glycemic and insulinemic response compared to most carbohydrates36-39.

Allulose, the c-3 epimer of fructose, is found naturally in small amounts in foods such as raisins, caramel sauce, maple syrup, brown sugar, processed cane, beet molasses and wheat40. Allulose is generally recognized as safe (GRAS) by the FDA as a food ingredient under GRAS registry number (GRN) 400 since 2012 and GRN 498 since 2014. It has been reported to be ~70% as sweet as sucrose and provides 0.4kcal/g41. Studies have shown that oral ingestion of allulose does not have a significant effect on blood glucose and insulin levels in healthy individuals and those with type 2 diabetes42- 45.

Tagatose, the c-4 epimer of fructose, is also found naturally in small amounts but mainly in dairy products such as sterilized, ultra-high-temperature and powdered milk. Hot cocoa, various cheeses, certain kinds of yogurt, and milk-based infant formula also contain small amounts of tagatose46, 47. Tagatose is also GRAS as a food ingredient under GRN 78 since 2001 and GRN 352 since 2010. It has been reported to be ~92% as sweet and provides 1.5 kcal/g48. Studies have reported that oral ingestion of tagatose does not have a significant effect on plasma glucose and insulin levels in healthy individuals and those with type 2 diabetes49, 50.

Sorbose, the c-5 epimer of fructose, is the least studied of the three epimers and its presence in foods, caloric content and postprandial glucose and insulin responses have yet to be determined.

2.2.2 Absorption, Metabolism & Excretion

Fructose

The ability of the small intestine to absorb free fructose varies among healthy individuals and is dose dependent. It has been suggested that the absorptive capacity for free

12 fructose in healthy individuals ranges from less than 5g to more than 50g, with a greater likelihood of malabsorption above 25g51-54. Malabsorption of fructose can be largely overcome by co-ingestion with glucose, such as in the form of sucrose or HFCS51, 55. In the classical model of absorption, fructose enters the apical membrane of the enterocyte through GLUT5, a cotransporter specific to fructose. Fructose then diffuses into blood vessels through the GLUT2 transporter located on the basolateral membrane6. There is some evidence that presence of glucose with fructose in the small intestine recruits additional GLUT2 transporters to the apical membrane which can further facilitate diffusion of fructose56, 57. The additional recruitment of GLUT2 to the apical membrane could potentially explain why fructose malabsorption does not occur with co-ingestion of glucose58, 59.

After absorption, fructose enters the hepatic portal vein where it is directed to the liver. Studies in humans have demonstrated that dietary fructose is largely converted to glucose (28.9–54%) which can either be stored as glycogen or released as plasma glucose. Part of the absorbed fructose is converted into lactate (~28%) and a small portion is converted into triglycerides (<1%)60, 61.

Any fructose that is not absorbed by the small intestine ends up in the colon where it undergoes colonic fermentation by the gut microbiota. Colonic fermentation of unabsorbed fructose typically produces carbon dioxide, organic acids and trace gases which causes flatulence, abdominal discomfort and/or diarrhea in some individuals54.

Allulose

The evidence evaluating absorption, metabolism and excretion of allulose is summarized in Table 2.3. Majority of the evidence from animal and human metabolic studies suggest that a large quantity of allulose is absorbed in the small intestine and is mostly excreted in the urine without significantly being metabolized41, 62-64. Some evidence from animal studies suggest that a small amount of allulose may be metabolized by the liver63, 65. Unabsorbed allulose passes into the large intestine where it is not readily fermented by intestinal bacteria41, 64.

Table 2.3. Evidence evaluating absorption, metabolism and excretion of allulose

Study Subjects Methods Relevant findings Whistler et 8 fasted  Experiment 1: allulose was intravenously  Experiment 1: 97-98% of intravenously administered 62 14 al. 1974 rats (24h) injected into rats. Urine and exhaled [ C] CO2 allulose was recovered in urine, 1% in liver glycogen and 14 were collected for 6h. 0.6% in [ C] CO2  Experiment 2: allulose was orally administered  Experiment 2: Within 7h, 4% of orally administered 14 14 to rats. [ C] CO2, urine, fecal samples and allulose was recovered in [ C] CO2 and 35.4% in urine. carcass were collected at 7h and 72h Within 72h, 15.1% of orally administered allulose was 14 recovered in [ C] CO2, 37.3% in urine, 12.5% in feces and 38.7% in carcass Matsuo et al. 52 Male  Experiment 1: 8 rats ingested 5g/kg BW  Experiment 1: Over 24h, 11-15% of ingested allulose was 200365 Wistar rats followed by urine and fecal collections at 24, 48 recovered in the urine and 8-13% in feces and 72h  Allulose was not detected in urine and feces after 24h,  Experiment 2: 18 rats ingested 5g/kg BW. Six suggesting that a large amount stayed in the rat body or rats were sacrificed at 1, 3 and 7h after allulose was metabolized ingestion. Serum, stomach, small intestine and  Experiment 2: Serum concentrations of allulose peaked at cecum samples were collected at these time 1-h and decreased progressively after that. At 1h, allulose points. concentrations were 26-37% in the stomach, 6-10% in the  Experiment 3: SCFAs in the cecum were small intestine and undetected (0%) in the cecum analyzed in 26 rats following consumption of 0,  At 3h, concentrations were 0.4-0.6% in the stomach, 2-3% 10, 20 or 30% allulose diets for 34 days. in the small intestine and 11-18% in the cecum. At 7h, concentrations were undetected (0%) in the stomach, 1- 3% in the small intestine and 10-19% in the cecum  Experiment 3: ↑ SCFAs (acetic, propionic and butyric acid) were reported with ↑ amounts of allulose in the diets Tsukamoto Male Wistar  Rats were administered radiolabelled allulose  Following oral administration, blood concentration of et al. 201463 rats either orally or intravenously at 100mg/kg. allulose peaked at 1h (48.5 + 15.6 μg/g) and 53% of Concentrations of allulose in whole blood, urine ingested allulose was detected in urine within 2h. and organs were assessed at 10, 30, 60 and  Following intravenous administration, half-life of allulose 120 mins after administration concentration in blood was 57 mins and excretion to urine was ~50% within 1h  After oral and intravenous administration, accumulation of allulose in organs was only detected in the liver  <1% of allulose was detected 7 days after allulose administration (Continued)

Table 2.3. (Continued)

Study Subjects Methods Relevant findings Iida et al. 20 healthy  Study 1: 6 participants consumed starch  Study 1: CEE and RQ did not increase after allulose or 201041 human hydrolysate, allulose (20g) or water on three water ingestion (indicating little to no metabolism of participants separate occasions. allulose)  Study 2: urinary excretion was measured after  Study 2: excretion rates over 48 h were 78.8% + 11.7% 14 participants consumed allulose at 5g, 10g or after 5g allulose ingestion, 78.6% + 10.6% after 10g 20g and 66.2% + 12.6% after 20g allulose ingestion  Study 3: breath hydrogen rates were compared  Study 3: AUC breath hydrogen after allulose ingestion in 14 participants after allulose vs. FOS was significantly ↓ than after FOS ingestion at all doses consumption (5g, 10g or 20g) (5g, 10g and 20g), suggesting low fermentability of  Study 4: 8 participants consumed 5g x 3 per allulose day for 8 weeks. Breath hydrogen gas excretion  Study 4: there was no significant difference in breath was measured before and after allulose hydrogen excretion after allulose ingestion before and treatment after the 8-week adaption period  Study 5: in an in vitro study, 35 typical intestinal  Study 5: 4/35 strains were identified as capable of bacterial strains were assessed for allulose fermenting allulose: Bacteroides thetaiotaomicron, fermentability Bacteroides uniformis, Bifidobacterium dentium and Ruminococcus productus) Williamson et 8 healthy  14C radiolabelled allulose was consumed with  Allulose concentration in plasma peaked ~1.5h after al. 201464 male 15g of unlabeled sweetener following a light administration human breakfast.  ~89% of the total radioactive allulose on average was participants  Blood, urine, fecal and breath samples were recovered over 168h in all subjects collected at various timepoints through 168h  Of the ~89% recovered, 86.26% was detected in the urine, 2.76% in feces and and 0.02% in breath samples BW, body weight; SCFAs, short chain fatty acids; CEE, carbohydrate energy expenditure; RQ, respiratory quotient

Tagatose

The evidence evaluating absorption, metabolism and excretion of tagatose is summarized in Table 2.4. The majority of evidence from animal and human metabolic studies suggest that ~25% of ingested tagatose is absorbed in the small intestine, out of which up to 20% (5% of that ingested) is excreted in the urine66-69. The rest of the absorbed tagatose (~20% of that ingested) is metabolized by the liver48, 70. Almost all the ~75% unabsorbed ingested tagatose enters the large intestine and is mainly fermented by intestinal bacteria66-68, 71, 72. Unfermented tagatose is excreted in feces (~11-29% of that ingested)67. Some studies have reported that tagatose has a prebiotic effect by promoting the growth of beneficial microorganisms and increasing butyrate production in the large intestine68, 71, 72.

Sorbose

To date, there have been no studies evaluating the absorption, metabolism and excretion of sorbose.

Table 2.4. Evidence evaluating absorption, metabolism and excretion of tagatose

Study Subjects Methods Relevant findings

Buemann 8 healthy  In a randomized crossover study, breath H2  24h urinary excretion rates in the different subjects on et al. 19981 human production was measured in a respiration the different days ranged from 0.7-5.3% of tagatose participants chamber on the first and last days of a 15-day ingested (3 male and period of 30g tagatose or 30g sucrose  24h H2 production significantly ↑ by 35% and 38% on 5 female) consumption day 1 and day 15, respectively, after 30g tagatose consumption compared to sucrose consumption Saunders 7 Sprague-  Rats with no prior exposure to tagatose  In unadapted rats, 49.4% of the 14C dose was 2 et al. 1999 Dawley rats (unadapted) and rats administered 100 g/kg recovered in CO2, 5.8% in urine and 28.7% were in tagatose for 28 days (adapted) orally consumed feces 14C tagatose (220-380 kBq)  In adapted rats, 67.9% of the 14C dose was recovered  CO2, urine and fecal samples were collected at in CO2, 5.2% in the urine and and 11.4% in feces regular intervals over 72h Laerke & 16 pigs  8 pigs were fed a diet containing 15% sucrose  25.8% + 5.6% of tagatose was absorbed in the distal Jensen. and 8 pigs were fed a diet containing 5% third of the small intestine 19993 sucrose+10% tagatose for 18 days  ↑ SCFA (propionate, butyrate and valerate) concentrations were found in the cecum and colon of rats fed tagatose Bertelsen 16 human  Volunteers delivered fecal samples before any  Production of total SCFA (mainly butyric acid) was ↑ in et al. 19994 participants intake of tagatose (unadpated) and another the adapted state compared to the unadapted state fecal sample after consuming tagatose 3x10g  The number of pathogenic bacteria (i.e. coliform per day for 14 days (adapted) bacteria) ↓ and the number of beneficial bacteria (i.e.  Fecal samples were incubated for 48 with or lactobacilli and lactic acid bacteria) ↑ in fecal samples without 1% tagatose from the adapted state compared to the unadapted state Buemann 8 healthy  Participants consumed either 30g tagatose or  1.45% of tagatose was detected in the urine et al. male 30g fructose in water 20005 human participants Normen et 6 ileostomy  Participants were served a controlled diet for  Median tagatose excretion in 24h illeal effluent was al. 20016 patients (4 two 2-day periods. In one of the periods, 15g 19%, corresponding to an apparent absorption of 81% male, 2 tagatose was added to the diet daily female) (Continued)

Table 2.4. (Continued)

Study Subjects Methods Relevant findings Venema 30 healthy  In this five-way randomized double-blind  12.5g tagatose consumption resulted in ↑ lactobacilli et al. 20057 human crossover study, changes in fecal microbiota in men, but not in women, compared to 7.5g tagatose participants and SCFA production were assessed in consumption (12 males subjects who consumed 30g raspberry jam  ↑ in vitro butyrate production was reported after and 18 containing 7.5 or 12.5g tagatose, 7.8g FOS, tagatose incubation with fecal samples from females) 7.6g tagatose+7.5g FOS and 15g sucrose at participants who consumed 7.5g (33 vs. 17 mol% breakfast for 2 weeks butyrate in SCFAs) and 12.5g tagatose (28 vs. 17 mol% butyrate in SCFAs) compared to butyrate production in fecal samples from the same participants who consumed 15g sucrose SCFAs, short chain fatty acids; FOS, fructo-oligosaccharide

2.2.3 Potential for Improving Postprandial Hepatic Glucose Metabolism

Evidence from Proof-of-concept Studies

As mentioned earlier, people with type 2 diabetes tend to have excessive rates of hepatic glucose output, impaired hepatic glucose uptake, and decreased hepatic glycogen synthesis rates in part due to decreased hepatic glucokinase activity10, 73, 74.

The current mechanism proposed by which small doses (< 10-g/meal) of fructose and its epimers may improve postprandial hepatic glucose metabolism is illustrated in Figure 2.2. In the fasting state, hepatic glucokinase is inactive and is bound to its regulatory protein and fructose-6-phosphate within the nucleus. In the postprandial state, ketohexokinase catalyzes the conversion of fructose, allulose and tagatose to fructose-1-phosphate, allulose-1-phosphate and tagatose-1-phosphate, respectively. These metabolites go on to displace fructose-6-phosphate from the regulatory protein enabling glucokinase to translocate to the cytosol. This enhancement of glucokinase activity results in the conversion of glucose to glucose-6-phosphate, increased hepatic glucose uptake, decreased hepatic glucose output and increased rate of glycogen synthesis, thereby reducing plasma glucose. The role of sorbose in this ‘catalytic’ mechanism is speculative due to the structural similarity with fructose, allulose and tagatose. However, future studies are needed to confirm its role in this mechanism.

Evidence evaluating the plausibility of fructose, allulose, tagatose and sorbose in this ‘catalytic’ mechanism is highlighted in Tables 2.5, 2.6, 2.7 and 2.8, respectively.

Figure 2.2 Short-term regulation of GK activity by fructose and its epimers. In the fasting state (top), hepatic GK is inactive and is bound to its regulatory protein (GKRP) and F6P within the nucleus. In the postprandial state, KHK catalyzes the conversion of fructose, allulose and tagatose to F1P, A1P and T1P, respectively. These metabolites go on to displace F6P from GKRP enabling GK to translocate to the cytosol. This enhancement of GK activity results in the conversion of glucose to G6P, increased hepatic glucose uptake and increased rate of glycogen synthesis. The role of sorbose in this mechanism is speculative due to the structural similarity with fructose, allulose and tagatose. Future studies are needed to confirm its role in this mechanism. GK, glucokinase; GKRP, glucokinase regulatory protein; F6P, fructose-6-phosphate; KHK, ketohexokinase; F1P, fructose-1-phosphate; A1P, allulose-1-phosphate; T1P, tagatose- 1-phosphate; G6P, glucose-6-phosphate.

Table 2.5. Proof-of-concept studies evaluating the ‘catalytic’ mechanism of fructose

Study Subjects Methods Relevant findings Shiota et al. 5 fasting (42h)  [3-3H] glucose and insulin were infused at a  Hyperglycemia (0-120 min) switched net 199875 conscious dogs constant rate to induce hyperglycemic hepatic glucose balance from output (11 + 2 hyperinsulinemia (0-390 min). μmol · kg–1 · min– 1) to uptake (14 + 2 μmol ·  Fructose was infused into the portal vein at kg–1 · min– 1) constant rates of 1.7 (120-210 min), 3.3 (210-300  Fructose infusion at 1.7, 3.3 and 6.7 μmol · min) and 6.7 μmol · kg–1 · min– 1 (300-390 min) kg–1 · min– 1 led to significantly ↑ net hepatic glucose uptake (41 + 3, 54 + 5 and 69 + 8 μmol · kg–1 · min– 1, respectively) compared to hyperglycemia alone.  69% + 3% of infused [3-3H] glucose incorporated into hepatic glycogen throughout the experiment Shiota et al. 18 fasting (42h)  [1-13C] glucose was infused intraduodenally with  ↑ net hepatic glucose uptake (28 + 5 vs. 17 + 200276 conscious dogs or without fructose (2.2 μmol · kg–1 · min– 1) for 6 μmol · kg–1 · min– 1), ↑ net glycogen 240 min deposition (3.68 vs. 2.44 mmol glucose equivalent/kg BW), and ↑ glycogen synthesis via the direct pathway (68% vs. 49%) were reported in dogs infused with fructose compared to those without Wolf et al. 39 Zucker fatty  Experiment 1: 10 female Zucker fatty fa/fa rats  Experiment 1: fructose ↓ iAUC glucose 200277 fa/fa rats received glucose (1.0 g/kg BW) alone or glucose response by 34% when supplemented to a supplemented with fructose (0.16 g/kg BW) in a glucose challenge randomized crossover experiment  Experiment 2: fructose ↓ iAUC glucose  Experiment 2: 10 male Zucker fatty fa/fa rats response by 32% when supplemented to a received maltodextrin (1.0 g/kg BW) alone or maltodextrin challenge maltodextrin supplemented with fructose (0.16  Experiment 3: low-dose fructose ↓ iAUC g/kg BW) in a randomized crossover experiment glucose response by 18% when  Experiment 3: 19 male Zucker fatty fa/fa rats supplemented to a maltodextrin challenge received maltodextrin (1.0 g/kg BW) alone or maltodextrin supplemented with low-dose fructose (0.075 g/kg BW) in a randomized crossover experiment (Continued)

Table 2.5 (Continued)

Study Subjects Methods Relevant findings Petersen et 6 healthy  13C nuclear magnetic resonance spectroscopy  Subjects receiving low-dose fructose infusion al. 200178 participants was used to assess rates of hepatic glycogen had ↑ rate of net hepatic glycogen synthesis synthesis under euglycemic (5 mmol/L) (0.54 + 0.12 vs. 0.14 + 0.05 mmol/L per min) hyperinsulinemic conditions (400 pmol/L) with compared to subjects without infusion or without infusion of low-dose fructose (3.5 μmol · kg–1 · min– 1)

Hawkins et 10 participants  Under a euglycemic-hyperglycemic clamp,  Participants with type 2 diabetes: glucose al. 200279 with moderately subjects received: (1) low-dose fructose production failed to decrease in response to controlled type 2 infusion (0.6 mg · kg–1 · min– 1), (2) higher-dose hyperglycemia. Infusion of low- and higher-dose diabetes and 7 fructose infusion (1.8 mg · kg–1 · min– 1), or (3) fructose ↓ glucose production in response to healthy no fructose infusion hyperglycemia (-27% + 6% and -33% + 8 %, participants respectively), approaching the 44% decline in glucose production observed in healthy subjects without fructose infusion  Healthy participants: glucose production was not altered in healthy subjects with low-dose (- 47% + 7%) and higher-dose (42% + 5%) fructose infusion  Contrary to previous findings, glucose uptake was not altered in healthy subjects and those with type 2 diabetes during the hyperglycemic period Perreault 17 participants  Glucose production and flux were assessed  During euglycemia, glucose production and total et al. 201480 with impaired during a hyperglycemic-euinsulinemic glucose output were not different between fasting glucose somatostatin clamp before and after infusion of groups. Glucose cycling (marker of hepatic (IFG) and 17 with low-dose fructose (0.6 mg· kg–1 · min– 1) glucokinase activity in post-absorptive state) normal glucose was ↓ in subjects with IFG than NGT (p = 0.04) tolerance (NGT)  Hyperglycemia suppressed endogenous glucose production more in NGT than IFG (p < 0.01)  Addition of fructose ↓ endogenous glucose production (p < 0.01) and total glucose output (p = 0.01) in response to hyperglycemia in subjects with IFG. No such changes were reported in subjects with NGT

Table 2.6. Proof-of-concept studies evaluating the ‘catalytic’ mechanism of allulose

Study Subjects Methods Relevant findings Hossain 45 OLETF rats  15 OLETF rats consumed drinking water  Before the glucose load, glucokinase was et al. 201181 (control), allulose in drinking water (5% total predominantly present in the nucleus of energy intake) or glucose in drinking water (5% hepatocytes in all three groups total energy intake) for 13 weeks  At 30 min after glucose load, intensity of nuclear  At the end of the treatment period, staining was ↓ in the allulose and glucose groups immunofluorescence staining was used to (suggesting glucokinase translocation). Extent of assess hepatic glucokinase translocation in 15h glucokinase translocation was less marked in the fasted rats followed by glucose load in the three control group. groups  At 90 min after glucose load, intensity of nuclear staining was further ↓ in the allulose group compared with the glucose and control groups (suggesting enhanced glucokinase translocation activity in allulose treated rats) Shintani et 36 Wistar rats  12 Wistar rats were treated with drinking water  Hepatic glycogen content in the rare sugar syrup al. 201782 (control), rare sugar syrup containing allulose or group was 3 times ↑ than in the water and high- high-fructose corn syrup (HFCS) in drinking fructose corn syrup group at 0 min. water for 10 weeks  Before glucose loading, distribution of glucokinase  At the end of the treatment period, hepatic in the cytoplasm was ↑ in the rare sugar syrup glycogen content and glucokinase translocation group (46%) compared to the water (28%) and (via immunofluorescence) were analyzed HFCS group (21%). before and 30 min after a glucose load in all  After glucose loading, the distribution of three groups glucokinase in the cytoplasm was ↑ in the rare sugar syrup group compared to the other groups

Table 2.7. Proof-of-concept studies evaluating the ‘catalytic’ mechanism of tagatose

Study Subjects Methods Relevant findings Kruger et al. 20 Sprague-  In a 90-day study, rats consumed  ↑ relative liver weights were observed in rats fed 10, 15 199983 Dawley rats tagatose at dietary doses of 5, 10, 15 and 20% tagatose compared to the dietary control and 20%. The control group only (presumably due to ↑ liver glycogen) received chow  No gross pathological findings correlated with these increased liver weights Bär et al. 210 adult male  Study 1: rats received two separate  Study 1: on day 28, liver glycogen content was 199984 Sprague-Dawley control diets (Purina and SDS) either significantly ↑ in rats fed Purina + 20% tagatose rats alone or with the addition of 20% compared to Purina alone tagatose for 28 days  Study 2: liver weights were significantly ↑ in linear  Study 2: rats received a control diet or relation to tagatose dose intakes. Qualitative the same diet with 5%, 10% or 20% observation revealed ↑ amounts of glycogen in the liver tagatose added for 29-31 days of tagatose-fed rats compared to control, without  Study 3: rats received a control diet alterations in cellular components (i.e. mitochondria, (Group A and C) or the same diet with endoplasmic reticulum, and Golgi apparatus) 5% tagatose (Groups B and D) for 28  Study 3: livers of tagatose-treated rats (Group D) had ↑ days. Groups A and B were fasted for glycogen content compared livers of control rats (Group 24h before sacrifice, whereas Groups C C) and D had food available until sacrifice Buemann et 5 healthy human  31P-magnetic resonance spectroscopy  A peak at 5.2 ppm in the spectrum suggested that al. 200085 males was used to assess tagatose-1- ~1mmol/L of tagatose-1-phosphate was found in the phosphate levels in the liver after liver within 30 mins of tagatose administration in all fives consumption of 30g tagatose subjects. This peak disappeared after 150 mins Boesch et al. 12 healthy male  In a double-blind crossover study,  Tagatose and sucrose treatment did not alter liver 200186 subjects participants consumed tagatose (3 x 15g volume or glycogen concentration daily) and sucrose (3 x 15g daily) for 28  A small ↑ in liver volume, independent of tagatose or days each sucrose intake, was observed over the study in parallel  Liver volume and glycogen concentration with a slight increase in body weight were determined by magnetic resonance  Tagatose did not affect biochemical and hematological (MR) imaging and spectroscopy, along parameters, including liver enzymes and uric acid with routine medical examinations

Table 2.8. Proof-of-concept studies evaluating the ‘catalytic’ mechanism of sorbose

Study Subjects Methods Relevant findings Oku et al. 6 male Wistar  In a crossover experiment, rats were fed  Oral administration of the sucrose solution 201487 rats 300mg/mL sucrose alone or with 33.3 with sorbose significantly ↓ plasma glucose mg/mL sorbose in 1.5 mL solution on two and insulin levels, in comparison of the separate occasions sucrose solution without sorbose, 30 and 60  Plasma glucose and insulin were analyzed minutes after administration

Evidence from Acute Clinical Trials

Tables 2.9, 2.10 and 2.11 summarize the acute clinical evidence regarding the role of small doses of fructose, allulose and tagatose on postprandial carbohydrate metabolism, respectively. To date, no clinical trials have investigated the role of small doses of sorbose on postprandial carbohydrate metabolism.

Two separate studies conducted in people with and without type 2 diabetes demonstrated that 7.5g fructose significantly reduced plasma glucose iAUC response by 14% and 19% to a 75g-oral glucose tolerance test (75g-OGTT), respectively88, 89. However, two follow-up studies failed to demonstrate a significant glycemic response reduction in healthy individuals when 5g or 10g fructose were consumed with a meal load90, 91. One study only found significant glycemic reductions of 25% and 27% when 10g fructose was consumed 60min and 30min prior to a meal load, respectively90.

Studies evaluating the role of small doses of allulose on postprandial carbohydrate metabolism in healthy individuals have produced inconsistent results. One study found that 5g and 10g allulose significantly reduced the plasma glucose iAUC response to 75g maltodextrin by 28% and 31%, respectively42. However, in another study, 5g allulose failed to reduce the glycemic response to a meal load in healthy participants. In this study, 5g allulose significantly reduced the glycemic response to a meal load by ~14% only in participants with borderline diabetes92.

Studies evaluating the role of small doses of tagatose on postprandial carbohydrate metabolism have generally been consistent. In healthy individuals, administration of tagatose at doses ranging from 5-15g have failed to demonstrate significant glycemic response reductions when administered with or prior to glucose and meal loads48, 93. However, in subjects with impaired glucose metabolism (prediabetes or type 2 diabetes), doses ranging from 5-10g administered with or prior to glucose and meal loads have demonstrated significant glycemic reductions of 4-19%49, 50, 93.

Table 2.9. Acute clinical studies evaluating the efficacy of low-dose fructose on postprandial carbohydrate metabolism

Study Subjects Methods Relevant findings Moore et al. 11 healthy participants  2-h plasma glucose and insulin were  7.5g fructose significantly ↓ plasma glucose 200088 (5 men and 6 women) measured after 75g-OGTT or 75g OGTT + iAUC response to the 75g-OGTT by 19%, 7.5g fructose without significantly affecting plasma insulin iAUC response Moore et al. 5 obese participants with  3-h plasma glucose and insulin were  7.5g fructose significantly ↓ plasma glucose 200189 DM2 (1 man and 4 measured after 75g-OGTT or 75g-OGTT + response by 14% to the 75g-OGTT, without women) 7.5g fructose significantly affecting plasma insulin iAUC response Heacock et al. 31 healthy participants  Capillary finger-stick blood samples were  Contrary to previous findings, fructose 200290 (13 men and 18 women) collected at -60, -30, 0, 15, 30, 45, 60, 90 administration with meal load (0 min) failed and 120min after consumption of an instant to lower the glycemic response mashed potato meal load (50g available  Fructose administration at -60 and -30min carbohydrate) alone or meal load significantly ↓ iAUC glucose response to supplemented with 10g fructose at -60, -30 the instant mashed potato meal by 25% or 0min and 27%, respectively Wolf et al. 30 healthy participants  3-h finger-prick capillary glucose samples  Addition of 5g fructose failed to reduce 200391 (13 men and 17 women) were analyzed after MTT (50g available iAUC capillary blood glucose response to carbohydrate from maltodextrin and white the meal tolerance test bread) alone or MTT + 5g fructose DM2, diabetes mellitus type 2; OGTT, oral glucose tolerance test; MTT, meal tolerance test; iAUC, incremental area under the curve

Table 2.10. Acute clinical studies evaluating the efficacy of low-dose allulose on postprandial carbohydrate metabolism

Study Subjects Methods Relevant findings Iida et al. 20 healthy  A fasting finger-prick blood sample was  2.5g allulose failed to reduce the glycemic and 200842 participants collected followed by samples at 30, insulinemic response to 75g maltodextrin (11 males and 9 60, 90 and 120 min after consumption  Addition of 5g and 7.5g allulose significantly ↓ AUC females) of 75g-maltodextrin alone or 75g- plasma glucose response to 75g maltodextrin by maltodextrin + 2.5g, 5g or 7.5g allulose 22% and 32%, respectively  Addition of 5g and 7.5g allulose significantly ↓ AUC plasma insulin response to 75g maltodextrin by 28% and 31%, respectively Hayashi et al. 11 healthy  Participants consumed tea (200ml)  In healthy participants, 5g allulose failed to reduce 201092 participants and sweetened with 10mg aspartame or 5g the glycemic and insulinemic response to the meal 15 participants allulose with meal load load compared to 10mg aspartame with borderline  Meal load involved a bun filled with  In participants with borderline diabetes, 5g allulose diabetes adzuki bean paste (425 kcal; 84.5g significantly ↓ AUC blood glucose response by ~14% carbohydrate, 3.7g fat, 13.3g protein) to the meal load compared to 10mg aspartame  A fasting venous blood sample was (control), without significantly affecting AUC blood collected followed by samples at 30, insulin response 60, 90 and 120 min

Table 2.11. Acute clinical studies evaluating the efficacy of low-dose tagatose on postprandial carbohydrate metabolism

Study Subjects Methods Relevant findings Donner et al. 10 participants  3-h plasma glucose was measured after 0, 10, 15,  10, 20 and 30g tagatose administration 30 199950 with DM2 (6 men 20 or 30g tagatose were consumed 30 min prior to min prior to the 75g-OGTT significantly ↓ and 4 women) a 75g-OGTT iAUC plasma glucose response by 13%, 22% and 26%, respectively  15g tagatose failed to reduce the iAUC plasma glucose response to the 75g- OGTT Madenokoji et 12 participants  3-h blood samples were measured after 75g-  7.5g tagatose significantly ↓ the AUC al. 200349 (10 men and 2 OGTT or 75g-OGTT + 7.5g tagatose blood glucose response to the 75g-OGTT women) with IFG by ~19% (110-126 mg/dL) Lu et al. 200848 30 healthy  In 7 separate meal tolerance tests, subjects  Treatment with 10g or 15g tagatose, participants (16 consumed 90g instant mashed potato load either administered before or with the meal or men and 14 alone, or with the addition of 10g or 15g tagatose glucose load, blunted the rise of plasma women) at 60, 30 or 0 min prior to the meal load glucose, although reductions in iAUC,  In 4 separate glucose tolerance tests, subjects when compared with corresponding underwent a 75g-OGTT alone or with the addition controls, were not statistically significant of 15g tagatose at 60, 30 or 0 min prior to the glucose load  Finger-prick capillary blood glucose samples were obtained at -60, -30, 0, 30, 45, 60, 90 and 120 min after the start of the meal and glucose load Kwak et al. 52 healthy  2-h blood samples were collected after  In healthy subjects, 5g tagatose failed to 201393 participants and participants consumed one of two experimental reduce the whole blood glucose response 33 participants drinks right before a meal to the meal tolerance test compared to the with  The drinks consisted of: (1) erythritol + 0.004g placebo drink hyperglycemia sucralose (placebo), or (2) 5g tagatose  In subjects with hyperglycemia, the 5g (IFG or DM2)  The meal consisted of a sandwich (356.4 kcal – tagatose drink significantly ↓ iAUC glucose 60% carbohydrate, 18% fat and 22%) made from response by 4% to the meal tolerance test bread (75g), ham (20g), lettuce (20g), strawberry compared to the placebo drink jam (20g), crab stick (30g), and cheddar cheese (10g). DM2, diabetes mellitus type 2; IFG, impaired fasting glucose; OGTT, oral glucose tolerance test; AUC, area under the curve

2.3 Gaps in the Literature

The characteristics of fructose and its epimers to taste like sugar while eliciting lower postprandial glucose and insulin responses when consumed alone make them attractive candidates as alternative sweeteners. Several lines of evidence have also suggested that small doses of fructose and its epimers reduce the postprandial glycemic response to high glycemic index meals by having a ‘catalytic’ effect on postprandial glucose metabolism via enhancement of hepatic glucokinase activity. Though promising, a few gaps remain:

1. Although there is compelling evidence from animal models and human infusion studies that small doses of fructose could improve the postprandial hepatic handling of glucose75-80, clinical translation in people with type 2 diabetes is limited to only one acute trial in 5 individuals89. There is a need for replication studies in a larger sample to confirm the effect of small doses of fructose on postprandial carbohydrate metabolism in individuals with type 2 diabetes.

2. Only two trials have assessed the role of small doses of allulose on postprandial carbohydrate metabolism in healthy individuals and those with borderline diabetes42, 92 . To our knowledge, there are no acute clinical trials that have investigated the role of small doses of allulose on postprandial carbohydrate metabolism in individuals with type 2 diabetes.

3. Whether the effects of fructose and allulose are equivalent is also unknown. This is of interest as allulose serves as a low-caloric alternative to fructose.

4. The minimum effective dose at which improvements in carbohydrate metabolism are observed also remains to be determined for fructose and allulose in individuals with type 2 diabetes.

5. Whether the acute ‘catalytic’ effects of fructose and its epimers translate into sustainable improvements in glycemic control under chronic feeding conditions also remains to be determined.

Chapter 3 : Rationale, Objectives & Hypotheses

3.1 Rationale

Diabetes is a global health problem and is on the rise. Sugars have emerged as the dominant nutrient of concern in diabetes creating a need for caloric and low-caloric alternative sweeteners. The fructose moiety of sugar has been implicated as a potent driver of type 2 diabetes due to its unique set of potential biochemical, metabolic and endocrine responses3-5. This a result of undue emphasis on ecological observations, animal models of fructose overfeeding, and select human studies assessed in isolation. The important biological mechanisms by which small doses of fructose may assist in the hepatic handling of glucose have largely been ignored. An emerging literature has also suggested that the low-caloric epimers of fructose (allulose, tagatose and sorbose) may act in a similar way to fructose by enhancing postprandial hepatic glucose metabolism and reducing the glycemic response to high glycemic index foods. A careful investigation into this less appreciated body of research is warranted.

3.2 Objectives

The overall objective of this thesis is to assess the effect of small doses of fructose and its epimers (allulose, tagatose and sorbose) on glycemic control.

The specific objectives are:

1. To assess the acute effects of small doses (5g, 10g) of fructose and its c-3 epimer, allulose compared with control (0g) on glucose and insulin responses to a 75g-oral glucose tolerance test (75g-OGTT)

2. To assess whether there is a dose response over the proposed dose range (0g, 5g, 10g) for the effects of fructose and allulose on glucose and insulin responses to a 75g-OGTT

3. To assess whether the effects of allulose and fructose are equivalent on the incremental area under the curve (iAUC) for plasma glucose across the 2 dose levels (5g and 10g) compared with control (0g)

4. To synthesize the evidence of the long-term effect of low-dose fructose and its

epimers (allulose, tagatose and sorbose) on HbA1c, fasting glucose and fasting insulin

Objectives 1-3 were addressed in an acute, randomized, double-blind crossover trial assessing the effect of small doses (5g, 10g) of fructose and allulose on postprandial carbohydrate metabolism in individuals with type 2 diabetes. Objective 4 was addressed in a systematic review and meta-analysis synthesizing the effect of low-dose fructose and its epimers on glycemic control in chronic feeding trials.

3.3 Hypotheses

We hypothesized that:

1. Small amounts of fructose and allulose at 5g and 10g would significantly reduce the glucose and insulin responses to a 75g-OGTT compared with control

2. Fructose and allulose would show a dose response across the proposed dose range (0g, 5g and 10g)

3. Allulose would be shown to be equivalent to fructose in its effects on the primary endpoint of iAUC for plasma glucose compared with control

Chapter 4 : Effect of Small Doses of Fructose and Allulose on Postprandial Carbohydrate Metabolism in Type 2 Diabetes: A Randomized Crossover Trial

4.1 Abstract

Objective: To assess the effect of small, single doses (5g, 10g) of fructose and its c-3 epimer, allulose, on the postprandial glycemic response to a 75g-oral glucose tolerance test (75g-OGTT) in participants with type 2 diabetes.

Research Design and Methods: In this acute, randomized, double-blind, crossover trial, twenty-four participants with well-controlled type 2 diabetes ([mean + s.e.m] age, 2 66 + 1.2 years; BMI, 27.0 + 0.9 kg/m ; HbA1c, 6.7% + 0.1%) were analyzed. Each participant randomly received six treatment drinks, separated by a one-week washout. The treatment drinks consisted of a 75g-OGTT to which fructose or allulose at 0g (control), 5g or 10g were added. A standard 75g-OGTT protocol was followed with blood samples drawn at -30, 0, 30, 60, 90, and 120 minutes. Separate analyses for fructose and allulose were conducted with data meaned for the two control doses for comparisons with the other doses (5g and 10g). The primary outcome was the difference in incremental area under the curve (iAUC) for plasma glucose.

Results: iAUC for plasma glucose after consumption of the 75g-OGTT without addition of fructose or allulose (control) was 777.5 + 39.9 mmol·min/L. The iAUC for plasma glucose after addition of 5g and 10g fructose to the 75g-OGTT were 779.4 + 40.7 and 767.9 + 39.3 mmol·min/L, respectively. The iAUC for plasma glucose after addition of 5g and 10g allulose to the 75g-OGTT were 729.3 + 44.9 and 717.4 + 38.3 mmol·min/L, respectively. Pairwise comparisons indicated that 5g and 10g fructose did not have a significant effect on the plasma glucose iAUC response (p=1.00 and p=0.91, respectively). Pairwise comparisons indicated that 10g allulose significantly reduced the plasma glucose iAUC response to the 75g-OGTT by 8% compared to the 75g-OGTT alone (p=0.04). There was no significant effect of 5g allulose on the plasma glucose iAUC response (p=0.11).

Conclusion: 10g allulose modestly reduced the postprandial glycemic response to oral glucose in individuals with type 2 diabetes. There is a need for long-term randomized controlled trials to confirm whether these acute reductions in postprandial glycemia will lead to sustainable improvements in glycemic control.

4.2 Introduction

Diabetes is a global health problem affecting approximately 415 million people worldwide and is estimated to affect 642 million people by 20402. Sugars have emerged as the dominant nutrient of concern in diabetes creating a need for the development of caloric and low-caloric alternative sweeteners. The fructose moiety of sugar has been implicated as a potent driver of type 2 diabetes due to its unique set of potential biochemical, metabolic and endocrine responses3-5. This a result of undue emphasis on ecological observations, animal models of fructose overfeeding, and select human studies assessed in isolation. The important biological mechanisms by which small doses of fructose may assist in the hepatic handling of glucose have largely been ignored. Contrary to the concerns that fructose may have adverse metabolic effects, an emerging literature has suggested that small doses of fructose at a level obtainable from fruit (<10-g/meal) and its c-3 epimer, allulose (a low-calorie sugar found naturally in small amounts in foods such as dried fruits, brown sugar, and maple syrup), may improve the hepatic handling of glucose.

In vitro studies have demonstrated that fructose and allulose, through their metabolites fructose-1-phosphate and allulose-1-phosphate, respectively, elicit ‘catalytic’ effects on hepatic glucose metabolism by enhancing glucokinase activity in hepatocytes94-97. This mechanism has been shown to relate to decreased hepatic glucose output and increased glycogen synthesis rates in both animal and human infusion studies75, 76, 78-80, 82. Clinical translation of this ‘catalytic’ mechanism has also proven promising. Addition of 7.5g fructose and 5g allulose have been shown to reduce the glycemic response to high glycemic index meals (oral glucose and meal load) in individuals with type 2 diabetes and borderline diabetes, respectively89, 92.

Although these findings provide a compelling proof of concept, there is a need for replication studies to confirm the ‘catalytic’ effect of fructose in individuals with type 2 diabetes. Up to date, the effect of small doses of allulose on postprandial carbohydrate metabolism in individuals with type 2 diabetes is unknown. Whether fructose and allulose display equivalent effects on glycemic response is also of interest, as allulose represents a low-caloric alternative to fructose. The minimum ‘catalytic’ dose of fructose

36 and allulose at which improvements in carbohydrate metabolism are observed also remains to be determined in individuals with type 2 diabetes.

To address these gaps, we conducted an acute randomized crossover trial in individuals with type 2 diabetes. The objective of our trial was to: (1) assess the ‘catalytic’ effects of fructose and allulose at 2 dose levels (5g, 10g) compared with control (0g) on glucose and insulin responses to a 75g-oral glucose tolerance test (75g- OGTT); (2) assess whether there was a dose response or threshold over the proposed range (0g, 5g, 10g) of the effect of fructose and allulose on glucose and insulin responses; and (3) to assess whether the effect of fructose and allulose were equivalent on the primary endpoint of incremental area under the curve (iAUC) for plasma glucose across the 2 dose levels (5g and 10g) compared with control (0g).

4.3 Research Design and Methods

4.3.1 Participants

The study protocol was approved by the St. Michael’s Hospital Ethics Review Board. Participants were included in the study if they met the following eligibility criteria: age 18-75 years, non-pregnant, non-smoker, BMI 18.5-35 kg/m2, well-controlled type 2 diabetes (HbA1c <7.5%) on diet and/or antihyperglycemic agents, not taking insulin and free of other major illnesses. Eligible participants gave written informed consent.

4.3.2 Trial Design

This was a randomized, double-blind, multiple-crossover “equivalence” trial. Randomization of the sequence of treatments was performed by computer randomization and stratified by sugar (fructose, allulose) and control (0g fructose, 0g allulose) to ensure a balanced randomization between the first three visits and the last three visits. The study statistician performed the randomization of treatments while being blinded to the identity of the participants and not having contact with the participants or the data. Neither the participants nor the investigators knew the identity of the treatments. The power calculation was based on the sample size needed for equivalence testing to compare the effect 10g allulose to 10g fructose on iAUC for

37 plasma glucose. Assuming equivalence margins (+ δ) set at + 20%98, with a significance level of α=0.05, power to detect a significant difference set at 1-β=80% and intra-subject standard deviation of 16.25%99, a total recruitment of 24 participants were needed.

4.3.3 Treatments

Participants received a total of 6 treatment drinks (Tate & Lyle, London, UK) in random order: 4 test drinks and 2 control drinks. The four test drinks consisted of 75g glucose dissolved in 500 ml of water to which fructose or allulose at 5g or 10g were added. The two control drinks consisted of 75g glucose dissolved in 500 ml of water without the addition of fructose (0g) or allulose (0g). The drinks were matched as much as possible in appearance, sweetness, texture and packaging. Flavor and color enhancements were used to mask any differences.

4.3.4 Protocol

The protocol followed the World Health Organization guidelines for the administration of an oral glucose tolerance test100. This study was conducted in an outpatient setting at the Applied Health Research Centre, St. Michael’s Hospital. Participants arrived at the study center on 6 separate mornings following a 10-12h overnight fast. They were instructed to maintain the same dietary and exercise patterns the evening before each study visit. Participants were also requested to consume a minimum of 150g of carbohydrate each day over the 3 days prior to the study visit. Diabetes medication use was discontinued on the morning of each study visit. At the beginning of each study visit, fasting blood glucose was measured via finger prick blood sample (CONTOUR®NEXT EZ blood glucose monitor, Bayer, NJ, USA) to ensure that values were within + 2 mmol/L of their screening fasting glucose. If the fasting glucose value fell outside + 2 mmol/L of their screening values, participants were requested to come back for another visit. If participant fasting glucose values fell outside + 2 mmol/L of their screening values after two consecutive visits, the average fasting values of previous study visits were used as the new cut-off. If participants fell within + 2 mmol/L of their screening or average fasting glucose values from previous study visits, participants proceeded to have a catheter inserted into a forearm vein. Two samples were collected

38 in the fasting state: one at -30 min and the other at 0 min. One of the six treatment drinks were then administered with instructions to consume it at a constant rate over five minutes. Additional blood samples were drawn at 30, 60, 90 and 120 minutes after the start of the treatment.

4.3.5 Plasma Glucose and Insulin Analysis

Plasma samples for glucose and insulin were separated by centrifuge and were immediately frozen at -720C for analysis. Plasma glucose was measured with the Roche/Hitachi MODULAR P analyzer (Roche Diagnostics, Indianapolis, IN, USA) using the hexokinase method101, 102. Plasma insulin was measured with the MODULAR ANALYTICS E170 (Roche Diagnostics, Indianapolis, IN, USA) immunoassay analyzer and electrochemiluminescence immunoassay kit103.

4.3.6 Outcomes

The primary outcome of this study was the difference in incremental area under the curve (iAUC) for plasma glucose upon addition of fructose or allulose at 5g or 10g to the 75g- OGTT compared with the control (0g fructose and 0g allulose). Secondary outcomes included: iAUC for plasma insulin, total maximum concentrations (Cmax) for plasma glucose and insulin, the Matsuda whole body insulin sensitivity index (Matsuda ISIOGTT), the early insulin secretion index (∆PI30-0/∆PG30-0) and mean incremental plasma glucose and insulin concentrations. Exploratory outcomes included: total AUC for plasma glucose and insulin, incremental Cmax for plasma glucose and insulin, insulin secretion-sensitivity index-2 (ISSI-2) and total mean plasma glucose and insulin concentrations. An exploratory pooled dose analysis was also conducted by comparing the average effect fructose and allulose across the two doses (5g, 10g) with control (0g) on the primary, secondary and exploratory outcomes.

4.3.7 Data Analysis

Participants were excluded from analysis if fasting plasma glucose values at one or more study visits fell outside + 2mmol/L of their average value from all six study visits. This criterion was applied across all participants who completed the study. Separate

39 analyses for fructose and allulose were conducted with data meaned for the two control doses for comparisons with the other doses (5g and 10g). The -30min and 0min samples were meaned to provide a single precise measurement of fasting glucose (0min) and fasting insulin (0min). Plasma glucose iAUC was calculated geometrically for each participant, ignoring areas below the fasting value104. Incremental glucose and insulin concentrations were used to control for baseline/fasting differences between the treatments.

The early insulin secretion index (∆PI30-0/∆PG30-0) is a measure of insulin secretion derived from the early period of the OGTT. It was calculated as the change in plasma insulin (PI) from 0 minutes to 30 minutes divided by the change in plasma glucose (PG) 25 over the same period . The Matsuda Insulin Sensitivity Index (Matsuda ISIOGTT) is an OGTT-derived measure of whole-body insulin sensitivity that has been validated against the euglycemic insulin clamp technique28. It was calculated using the 75g-OGTT PG and PI concentrations as follows: 10 000 / √ (fasting PG x fasting PI x mean PG x mean PI), where PG was expressed in mg/dL (1/18 mmol/L) and PI in U/ml (6 pmol/L). ISSI- 2 is an OGTT-derived measure of β-cell function that has been validated against the disposition index from frequently sampled intravenous glucose tolerance tests105. It was calculated by taking the product of 1) insulin secretion as measured by the ratio of total area-under-the-insulin-curve (AUCins) to the total-area-under-the-glucose curve (AUCglu) and 2) insulin sensitivity as measured by Matsuda ISIOGTT, such that ISSI-2 = (total

AUCins/glu x Matsuda ISIOGTT). ISSI-2 was calculated using SI units for AUCins, AUCglu and Matsuda ISIOGTT.

Repeated measures one-way analysis of variance (ANOVA) assessed the differences in the primary outcome and secondary outcomes. Planned pairwise comparisons comparing each of the 2 catalytic doses (5g, 10g) with the mean of the two controls (0g) were conducted using the Dunnett’s test for both primary and secondary endpoints. This test was done irrespective of the results of the repeated measures one-way ANOVAs. Repeated measures two-way ANOVA assessed the interactive effects of treatment and time (-30, 0, 30, 60, 90, 120min) on mean incremental changes in plasma glucose and

40 insulin. Significant interactions were explored with repeated measures one-way ANOVA at individual time points.

Dose–response analyses across the full dose range (0g, 5g, and 10g) were carried out using random effects generalized least squares trend (GLST) estimation models. Equivalence testing was done by determining whether the upper and lower bounds of the 90% CI for the effect of allulose on iAUC for plasma glucose fell within the equivalence margins (+ δ) set at + 20%. These equivalence margins were chosen because a minimum 20% decrease in the average iAUC for glucose in comparison to the reference food is considered a physiologically relevant change and is required to support a claim related to the reduction of glycemic response106. If the 90% CIs completely fell within the equivalence margins, then allulose was considered equivalent to fructose. If either the upper or lower bound of the 90% CI fell outside the equivalence margins, then the assessment was considered inconclusive. If the 90% CIs fell either completely above or completely below the equivalence margins, then allulose was considered inferior or superior to fructose, respectively. Significance for subgroup analyses was tested using mixed model test for interaction.

Statistical analyses were performed using STATA 13.1 (StataCorp LP, College Station, TX, USA) and SAS 9.2 (SAS Institute Inc, Cary, NC, USA). All data are represented as mean + SEM, unless specified otherwise.

4.4 Results

4.4.1 Flow of Study Participants

Figure 4.1 shows the progression of study participants throughout the trial. 238 participants were assessed for eligibility out of which 27 were randomized. 25 of 27 participants completed the trial. Two participants were unable to complete the trial due to work conflict. One participant was excluded from analysis due to fasting plasma glucose values at one or more study visits exceeding + 2 mmol/L of their average value from all six study visits.

Assessed for eligibility (n=238)

Excluded (n=211) • Declined to participate/did not meet inclusion criteria via telephone screening (n=188) • Eligible via telephone screening, but did not attend screening visit (n=14) • Screening blood work and/or anthropometric measurements did not meet inclusion criteria (n=9)

Randomized (n=27)

Did not complete trial (n=2) • Work conflict

Completed Trial (n=25)

Excluded from analysis (n=1) • Fasting plasma glucose values at one or more study visits fell outside + 2 mmol/L of their average value from all six study visits

Participants Analyzed (n=24)

Recruitment period: November 2015 – July 2016 First patient to start trial: 8-March-2016 Last patient to finish trial: 18-August-2016

Figure 4.1. Flow of participants through the various phases of the trial.

4.4.2 Participant Characteristics

Table 4.1 outlines the participant characteristics. 24 participants with type 2 diabetes 2 were analyzed (mean age, 66 + 1.2 years; mean BMI, 27.0 + 0.9 kg/m ; mean HbA1c, 6.7 + 0.1%). Participants were treating their diabetes using diet alone (n=5), metformin only (n=8) or metformin plus a second-line therapy (n=11). Second-line therapies included DPP-4 inhibitors (n=6), sulfonylureas (n=3), thiazolidinediones (n=1) and SGLT-2 inhibitor (n=1).

Table 4.1. Participant Characteristics

Characteristics Type 2 Diabetes Sex, M/F 12/12

Age, years 66 + 1.2

Weight, kg 76.2 + 3.7

BMI, kg/m2 27.0 + 0.9

HbA1c, % 6.7 + 0.1

Fasting blood glucose, mmol/L 6.9 + 0.2

Diabetes therapy - Diet alone 5 - Metformin only 8 - Metformin + DPP-4 inhibitor 6 - Metformin + sulfonylurea 3 - Metformin + thiazolidinedione 1 - Metformin + SGLT-2 inhibitor 1 Data reported as mean + s.e.m

4.4.3 Effect of Small Doses of Fructose and Allulose on Plasma Glucose and Insulin Responses

Figure 4.2. (A) shows incremental changes and iAUC for plasma glucose and plasma insulin following consumption of 75g-OGTT to which 0g, 5g or 10g fructose were added. Plasma glucose iAUC after consumption of 75g-OGTT alone, 75g-OGTT+5g fructose and 75g-OGTT+10g fructose was 777.5 + 39.9, 779.4 + 40.7 and 767.9 + 39.3 mmol·min/L, respectively. Plasma insulin iAUC after consumption of 75g-OGTT alone, 75g-OGTT+5g fructose and 75g-OGTT+10g fructose was 38405 + 6917, 39963 + 7285 and 39345 + 6353 pmol·min/L, respectively. Pairwise comparisons indicated that 5g and 10g fructose did not have a significant effect on the plasma glucose iAUC response (p=1.00 and p=0.91, respectively) and the plasma insulin iAUC response (p=0.78 and p=0.35, respectively) to the 75g-OGTT.

Figure 4.2. (B) shows incremental changes and iAUC for plasma glucose and plasma insulin following consumption of 75g-OGTT to which 0g, 5g or 10g allulose were added. Plasma glucose iAUC after consumption of 75g-OGTT alone, 75g-OGTT+5g allulose and 75g-OGTT+10g allulose were 777.5 + 39.9, 729.3 + 44.9 and 717.4 + 38.3 mmol·min/L, respectively. Plasma insulin iAUC after consumption of 75g-OGTT alone, 75g-OGTT+5g allulose and 75g-OGTT+10g allulose were 38405 + 6917, 38054 + 6599 and 34065 + 5292 pmol·min/L, respectively. Pairwise comparisons indicated that 10g allulose significantly reduced the plasma glucose iAUC response to the 75g-OGTT by 8% compared to the 75g-OGTT alone (p=0.04). There was no significant effect of 5g allulose on the plasma glucose iAUC response (p=0.11), and no significant effect of both 5g and 10g allulose on the plasma insulin iAUC response (p=1.00 and p=0.30, respectively) to the 75g-OGTT.

Two-way repeated measures ANOVA indicated that there was a significant effect of time (p<0.0001), but no significant effect of fructose and allulose on mean incremental changes in plasma glucose and plasma insulin, with no interaction (p>0.05).

Figure 4.2 Effect of small doses (5g, 10g) of fructose and allulose on glucose and insulin responses to a 75g-oral glucose tolerance test (75g- OGTT). (A) The line plots and bars represent the incremental change and incremental area under the curve (iAUC) for plasma glucose and insulin following consumption of 75g-OGTT with the addition of of 0g (), 5g (▲) or 10g () fructose in 24 participants with type 2 diabetes. (B) The line plots and bars represent the incremental change and iAUC for plasma glucose and insulin following consumption of 75g-OGTT with the addition of 0g (), 5g (▲) or 10g () allulose in 24 participants with type 2 diabetes. * represents statistically significant difference (p<0.05, Dunnett’s Test) compared to 0g (control). Data are mean + s.e.m.

4.4.4 Dose-Response Analysis

Figure 4.3 shows the linear dose-response relationship across the three doses (0g, 5g and 10g) for fructose and allulose. There was no linear dose-response observed for fructose and allulose on plasma glucose iAUC (p=0.89 and p=0.48, respectively) and plasma insulin iAUC (p=0.94 and p=0.69, respectively).

Figure 4.3 Linear dose-response analysis of the effect of small doses (0g, 5g, 10) of fructose (left) and allulose (right) on plasma glucose iAUC (top) and insulin iAUC (bottom). Each circle represents the iAUC of plasma glucose or insulin at a given dose for each sugar. The dotted line represents the trend in plasma glucose iAUC and insulin iAUC as dosage increases for each sugar.

4.4.5 Equivalence Test

Figure 4.4 shows results from the equivalence test comparing the effect of allulose to fructose on plasma glucose iAUC. Although allulose showed statistically significant reductions compared to fructose at individual and pooled doses, fructose and allulose displayed equivalent effects on iAUC for plasma glucose based on the equivalence margins set at + 20%.

Figure 4.4 Equivalence test comparing the effect of allulose to fructose on plasma glucose iAUC. % difference plasma glucose iAUC = [(alluloseiAUCglucose/controliAUCglucose) – (fructoseiAUCglucose/controliAUCglucose)] x 100%. Equivalence margins (+δ, - δ) were set at -20%, +20%. If the 90% CIs completely fell within the equivalence margins, then allulose was considered equivalent to fructose. If either the upper or lower bound of the 90% CI fell outside the equivalence margins, then the assessment was considered inconclusive. If the 90% CIs fell either completely above or completely below the equivalence margins, then allulose was considered inferior or superior to fructose, respectively.

4.4.6 Effect of Small Doses of Fructose and Allulose on Secondary and Exploratory Outcomes

Table 4.2 shows results of the individual and pooled dose analysis of the effect of small doses of fructose on primary, secondary and exploratory outcomes. There was no significant effect (p>0.0125) of fructose at individual and pooled doses on any of the outcomes.

Table 4.3 shows results of the individual and pooled dose analysis of the effect of small doses of allulose on primary, secondary and exploratory outcomes. Addition of 5g allulose to the 75g-OGTT significantly reduced total mean plasma glucose response (control – 13.6 + 0.5 vs. 5g allulose – 13.0 + 0.6 mmol/L; p<0.0125) and total AUC plasma glucose response (control – 1694.1 + 57.8 vs. 5g allulose – 1615.7 + 67.6 mmol·min/L; p<0.0125) to the 75g-OGTT. Addition of 10g allulose significantly reduced total mean glucose (control – 13.6 + 0.5 vs. 10g allulose – 12.9 + 0.5 mmol/L; p<0.0125), total AUC plasma glucose response (control – 1694.1 + 57.8 vs. 10g allulose – 1607.7 + 59.3 mmol·min/L; p<0.0125), total Cmax for plasma glucose (control

– 17.5 + 0.6 vs. 10g allulose – 16.1 + 0.7 mmol/L; p<0.0125) and incremental Cmax for plasma glucose (control – 9.8 + 0.5 vs. 10g allulose – 8.7 + 0.5 mmol/L; p<0.0125) in response to the 75g-OGTT. There was no effect of allulose on any of the OGTT-indices.

Pooled doses of allulose significantly reduced iAUC for plasma glucose (control – 777.5

+ 39.9 vs. pooled allulose – 723.4 + 39.4 mmol·min/L; p<0.0125), total Cmax for plasma glucose (control – 17.5 + 0.6 vs. pooled allulose – 16.5 + 0.7 mmol/L; p<0.0125), mean incremental plasma glucose (control – 6.0 + 0.3 vs. pooled allulose – 5.6 + 0.3 mmol/L; p<0.0125), total AUC for plasma glucose (control – 1694.1 + 57.8 vs. pooled allulose – 1607.7 + 59.3 mmol·min/L; p<0.0125) and mean plasma glucose (control – 13.6 + 0.5 vs. pooled allulose – 13.0 + 0.5 mmol/L; p<0.0125) in response to the 75g-OGTT.

Table 4.2. Individual and Pooled Dose Analysis for the Effect of ‘Catalytic’ Doses of Fructose on Primary, Secondary and Exploratory Outcomes

Pooled doses Pooled doses 5g vs. 0g 10g vs. 0g Outcome 0g fructose 5g fructose 10g fructose vs. 0g (5g + 10g) p-value** p-value** p-value**

Primary Outcome Glucose iAUC (mmol·min/L) 777.5 + 39.9 779.4 + 40.7 767.9 + 39.3 773.7 + 37.2 0.996 0.910 0.854 Secondary Outcomes

Cmax glucose (mmol/L) 17.5 + 0.6 17.3 + 0.7 17.4 + 0.7 17.3 + 0.7 0.868 0.930 0.602

Tmax glucose (mins) 88.8 + 4.0 91.3 + 3.8 90.0 + 4.8 90.6 + 4.0 0.652 0.895 0.479 Mean incremental glucose (mmol/L) 6.0 + 0.3 5.9 + 0.3 5.9 + 0.3 5.9 + 0.9 0.986 0.944 0.788 Insulin iAUC (pmol·min/L) 38405 + 6917 39963 + 7285 39345 + 6353 39654 + 6734 0.776 0.348 0.114

Cmax insulin (pmol/L) 649.0 + 126.2 655.2 + 117.7 624.9 + 95.5 640.0 + 104.7 0.829 0.718 0.325

Tmax insulin (mins) 103.1 + 3.9 96.3 + 4.8 98.8 + 4.2 97.5 + 3.9 0.540 0.850 0.142 Mean incremental insulin (pmol/L) 302.1 + 57.1 312.4 + 58.6 307.4 + 50.6 309.9 + 53.8 0.754 0.330 0.108

∆PI30-0/∆PG30-0 49.3 + 8.2 40.4 + 6.0 15.3 + 29.7*** 27.9 + 15.9*** 0.129 0.932*** 0.373***

Matsuda ISIOGTT 2.9 + 0.4 2.8 + 0.4 2.8 + 0.5 2.8 + 0.4 0.468 0.489 0.212 Exploratory Outcomes Glucose total AUC (mmol·min/L) 1694.1 + 57.8 1692.9 + 63.5 1667.1 + 62.6 1680.0 + 61.1 0.999 0.563 0.566

Incremental Cmax glucose (mmol/L) 9.8 + 0.5 9.7 + 0.5 9.9 + 0.6 9.8 + 0.5 0.902 0.995 0.826 Mean glucose (mmol/L) 13.6 + 0.5 13.6 + 0.5 13.4 + 0.5 13.5 + 0.5 0.954 0.570 0.500 Insulin total AUC (pmol·min/L) 47934 + 7537 49883 + 8112 49115 + 7046 49499 + 7489 0.599 0.220 0.101

Incremental Cmax insulin (pmol/L) 569.6 + 122.9 572.5 + 112.3 543.5 + 90.7 558.0 + 99.7 0.909 0.783 0.338 Mean insulin (pmol/L) 381.5 + 61.9 395.1 + 65.2 388.8 + 56.2 391.9 + 59.9 0.596 0.220 0.098 ISSI-2 188.9 + 25.3 182.8 + 20.4 194.6 + 24.6 188.7 + 21.4 0.988 0.678 0.511 Data reported as mean + SEM. N=24. *Primary outcome: p-values <0.05 are significant. Secondary and exploratory outcomes: p-values <0.0125 are significant due to adjustment for multiple comparisons. **P-values for insulin iAUC, insulin total AUC, incremental Cmax insulin, Cmax insulin, Matsuda ISIOGTT, ∆PI30-0/∆PG30-0, ISSI-2, mean incremental insulin and mean insulin correspond to log-transformed data. ***P-values represent comparisons in n=23 due to one participant value being negative.

Table 4.3. Individual and Pooled Dose Analysis for the Effect of ‘Catalytic’ Doses of Allulose on Primary, Secondary and Exploratory Outcomes

Pooled doses Pooled doses 5g vs. 0g 10g vs. 0g Outcome 0g allulose 5g allulose 10g allulose vs. control (5g + 10g) p-value** p-value** p-value**

Primary Outcome Glucose iAUC (mmol·min/L) 777.5 + 39.9 729.3 + 44.9 717.4 + 38.3 723.4 + 39.4 0.112 0.040* 0.0121* Secondary Outcomes

Cmax glucose (mmol/L) 17.5 + 0.6 16.9 + 0.8 16.1 + 0.7 16.5 + 0.7 0.131 0.0002* 0.0007* Tmax glucose (mins) 88.8 + 4.0 91.3 + 4.9 95.0 + 5.6 93.1 + 4.8 0.816 0.949 0.110 Mean incremental glucose (mmol/L) 6.0 + 0.3 5.6 + 0.4 5.5 + 0.3 5.6 + 0.3 0.093 0.035 0.009* Insulin iAUC (pmol·min/L) 38405 + 6917 38054 + 6599 34065 + 5292 36059 + 5833 0.998 0.304 0.450

Cmax insulin (pmol/L) 649.0 + 126.2 595.1 + 96.4 531.6 + 77.3 563.4 + 85.8 0.455 0.040 0.083 Tmax insulin (mins) 103.1 + 3.9 96.3 + 4.0 96.3 + 5.4 96.3 + 3.8 0.954 0.500 0.113 Mean incremental insulin (pmol/L) 302.1 + 57.1 295.6 + 51.8 268.2 + 42.6 281.9 + 46.6 0.994 0.375 0.489

∆PI30-0/∆PG30-0 49.3 + 8.2 47.5 + 7.5 43.7 + 6.3 45.6 + 6.4 0.988 0.915 0.985

Matsuda ISIOGTT 2.9 + 0.4 3.1 + 0.6 3.2 + 0.6 3.2 + 0.6 0.794 0.133 0.181 Exploratory Outcomes Glucose total AUC (mmol·min/L) 1694.1 + 57.8 1615.7 + 67.6 1607.7 + 59.3 1611.7 + 61.7 0.012* 0.005* 0.0005* Incremental Cmax glucose (mmol/L) 9.8 + 0.5 9.5 + 0.6 8.7 + 0.5 9.1 + 0.5 0.481 0.0018* 0.009* Mean glucose (mmol/L) 13.6 + 0.5 13.0 + 0.6 12.9 + 0.5 13.0 + 0.5 0.008* 0.004* 0.0002* Insulin total AUC (pmol·min/L) 47934 + 7537 47727 + 7213 42957 + 5739 45342 + 6370 1.000 0.132 0.144

Incremental Cmax insulin (pmol/L) 569.6 + 122.9 514.5 + 92.4 457.6 + 74.6 486.0 + 82.3 0.462 0.090 0.118 Mean insulin (pmol/L) 381.5 + 61.9 376.2 + 56.7 342.3 + 46.2 359.3 + 50.8 0.997 0.147 0.316 ISSI-2 188.9 + 25.3 206.4 + 26.1 200.9 + 24.8 203.6 + 25.0 0.119 0.124 0.033 Data reported as mean + SEM. N=24. *Primary outcome: p-values <0.05 are significant. Secondary and exploratory outcomes: p-values <0.0125 are significant due to an adjustment for multiple comparisons. **P-values for insulin iAUC, insulin total AUC, incremental Cmax insulin, Cmax insulin, Matsuda ISIOGTT, ∆PI30-0/∆PG30-0, ISSI-2, mean incremental insulin and mean insulin correspond to log-transformed data.

4.4.7 Subgroup Analysis

Figure 4.5 shows the sub-group analysis of the effect of pooled doses of fructose on the difference in plasma glucose iAUC compared to control (0g). There were no statistically significant effect modifiers (p>0.0125).

Figure 4.6 shows the sub-group analysis of the effect of pooled doses of allulose on the difference in plasma glucose iAUC compared to control (0g). There were no statistically significant effect modifiers (p>0.0125).

Figure 4.5. Subgroup analysis for the effect of pooled doses of fructose on plasma glucose iAUC compared to control. The diamonds represent the effect size for each subgroup. The dashed line represents the overall effect size. Significance for subgroup analyses was tested using mixed model test for interaction. P<0.0125 was considered significant due to adjustment for multiple comparisons.

Figure 4.6. Subgroup analysis for the effect of pooled doses of allulose on plasma glucose iAUC compare to control. The diamonds represent the effect size for each subgroup. The dashed line represents the overall effect size. Significance for subgroup analyses was tested using mixed model test for interaction. P<0.0125 was considered significant due to adjustment for multiple comparisons.

4.5 Discussion

4.5.1 Summary of Findings

Our study demonstrated that 10g allulose resulted in a modest lowering in the postprandial glycemic response to a 75g oral glucose load in 24 individuals with well- controlled type 2 diabetes (HbA1c < 7.5%). 5g allulose, 5g fructose and 10g fructose did not have a significant effect on the postprandial glycemic response to a 75g oral glucose load.

4.5.2 Findings in Context of Previous Research

Moore et al. previously showed that 7.5g fructose significantly reduced 3-h plasma glucose iAUC response to a 75g oral glucose challenge by 14%89. We were unable to reproduce these findings with 5g and 10g fructose. Potential sources of discrepancy between the previous trial and our trial include: sample size (n=5 vs. n=24), handling of medications (discontinued 5 days prior to treatment vs. discontinued on the morning of the treatment), participant age (mean + SEM: 42 + 5 vs. 66 + 1.2 years), participant BMI 2 (mean + SEM: 42 + 4 vs. 27 + 0.9 kg/m ) and HbA1c (mean + SEM: 8.5 + 0.5 vs. 6.7 + 0.1%). In a study conducted in 11 healthy participants, 7.5g fructose reduced the iAUC plasma glucose response by 19% to a 75g oral glucose challenge88. A follow-up study which assessed timing of fructose administration in 31 healthy participants failed to demonstrate postprandial glycemic reduction when 10g fructose was consumed with an instant mashed potato meal (50g available carbohydrate). Instead, glycemic reductions of 25% and 27% were observed only when fructose was consumed 60 or 30 min prior to the meal load, respectively90. It could be possible that although we didn’t observe a ‘catalytic’ effect from fructose in our trial, fructose administration prior to, instead of with the 75g oral glucose challenge may reduce the postprandial glycemic response in individuals with type 2 diabetes.

Our trial was the first to evaluate the effect of small doses of allulose on postprandial glucose metabolism in individuals with type 2 diabetes. The majority of studies evaluating the effect of small doses of allulose on postprandial carbohydrate

54 metabolism have been conducted in healthy participants. A study in 20 healthy subjects found that 5g and 7.5 allulose reduced the postprandial glycemic response by ~22% and ~32% and insulinemic response by ~28% and ~31%, respectively, to a 75g maltodextrin challenge42. Follow-up studies have failed to reproduce these findings. In 25 healthy participants, 5g allulose and 10g allulose did not significantly have an effect on glucose and insulin reponses to a 75g-oral glucose load107. Similarly, in another study when 11 healthy participants consumed 5g allulose-sweetened tea with a standard meal load, no significant differences were found in postprandial glucose and insulin responses when compared to consumption of the same meal load with 10mg aspartame sweetened tea92. However, in this same study when 15 participants with borderline diabetes were analyzed, 5g allulose-sweetened tea resulted in ~14% reduction in postprandial glycemic response to the standard meal load compared to aspartame-sweetened tea. These studies in conjunction with our study suggest that small doses of allulose have the potential to elicit a greater ‘catalytic’ effect in individuals with impaired glucose metabolism (i.e. borderline diabetes and type 2 diabetes) compared to those with normal glucose tolerance.

4.5.3 Mechanism of Action

Our study was not designed to examine the mechanism(s) by which allulose reduced the glycemic response to an oral glucose load, but the following are two possibilities: 1) enhanced insulin secretion by allulose, and 2) allulose-induced stimulation of hepatic glucose uptake. The first mechanism is not likely since allulose did not have a significant effect on the plasma insulin iAUC response. The most likely explanation is that small doses of allulose stimulated hepatic glucose uptake. There is some evidence that glucokinase activity is decreased in individuals with type 2 diabetes10, 73. Phosphorylation of glucose by glucokinase is a rate-determining step in hepatic glucose metabolism. Glucokinase is inhibited by glucokinase regulatory protein (GKRP), and this action is enhanced in the presence of fructose-6-phosphate96, 108-112. Under fasting conditions, hepatic glucokinase is localized primarily in the nucleus, where it is bound to the glucokinase regulatory protein (GKRP) and fructose-6-phosphate 95, 96, 113. In the postprandial state (presence of allulose and glucose), allulose is rapidly phosphorylated

55 to allulose-1-phosphate by an enzyme called ketohexokinase114. Allulose-1-phosphate competes with fructose-6-phosphate binding to GKRP, and in so doing, releases glucokinase from GKRP. This enables the liberated and activated glucokinase to translocate from the nucleus to the cytosol where it can drive hepatic glucose uptake, promote glycogen synthesis, suppress hepatic glucose output and reduce plasma glucose levels115, 116. In support of this hypothesis, immunohistochemical analysis in allulose-fed rats have showed induction of glucokinase translocation from the nucleus to the cytoplasm and increased amount of hepatic glycogen content after glucose loading81, 82. No studies have been conducted in humans to confirm this mechanism.

4.5.4 Clinical Implications

Clinical implications of these findings are that allulose could serve as an alternative sweetener in the management of type 2 diabetes. Allulose is classified as a rare sugar since it is found in small quantities in foods such as, raisins, caramel sauce, maple syrup, brown sugar, processed cane, beet molasses and wheat40. It tastes 70% as sweet as sucrose but contains 90% fewer calories41. When consumed alone, allulose does not raise blood glucose and insulin levels in healthy individuals and those with type 2 diabetes42-45. Our study along with a previous study in participants with borderline diabetes have shown that addition of small doses of allulose to a source of carbohydrate (i.e. 75g-OGTT or standard meal load) helps to lower the postprandial glycemic response by ~8-14%92. This decrease is modest when compared to an oral antihyperglycemic agent such as, acarbose. A study conducted in 14 individuals with type 2 diabetes demonstrated that acarbose significantly reduced the postprandial glycemic response to a meal load (450 kcal; 51.4% carbohydrate, 33.3% fat and 15.3% protein) by ~31% compared to placebo117.

4.5.6 Strengths & Limitations

This trial had many strengths: (1) it was a randomized double-blind controlled trial which is considered the gold standard for a clinical trial; (2) the crossover design of the trial reduced the between-subject variation by allowing each subject to act as his/her own control; (3) the control drink (75g-OGTT) was administered twice providing a more

56 reliable estimate of postprandial glycemic and insulinemic response; and (4) based on a power analysis, the sample size of the trial was sufficient for equivalence testing of the effect of fructose and allulose on plasma glucose iAUC.

The limitations of our trial were: (1) it was not designed to examine the mechanism(s) by which allulose reduced the postprandial glycemic response to an oral glucose load; (2) the doses examined in this study may have been insufficient to detect a dose- response/threshold; and (3) the acute design of the study creates uncertainty as to whether reductions in the postprandial glycemic response will manifest as sustainable improvements in glycemic control over the long-term.

4.6 Conclusions 10g allulose modestly reduced the postprandial glycemic response to oral glucose in individuals with type 2 diabetes. There is a need for long-term randomized trials to confirm whether these acute reductions in postprandial glycemia will lead to sustainable improvements in glycemic control.

Chapter 5 : Effect of Small Doses of Fructose and its Epimers on Markers of Long-term Glycemic Control: A Systematic Review and Meta-Analysis of Controlled Feeding Trials

5.1 Abstract

Objective: To synthesize the evidence from controlled feeding trials assessing the effect of small, ‘catalytic’ doses (< 50g/d) of fructose and its epimers (allulose, tagatose and sorbose) on HbA1C, fasting glucose and fasting insulin.

Methods: We searched MEDLINE, EMBASE, and Cochrane Library through Jan 31, 2017. We included controlled feeding trials of > 1 weeks investigating the effect of small, ‘catalytic’ doses (< 50-g/day) of fructose and its epimers in comparison to control diets. Two independent reviewers extracted relevant data. Risk of bias was assessed using the Cochrane Risk of Bias Tool. Data were pooled using the generic inverse variance method and expressed as mean differences (MD) with 95% confidence intervals (CIs). Heterogeneity was assessed using the Cochran Q statistic and quantified using the I2 statistic. The overall quality of the evidence was assessed using the Grading of Recommendations, Assessment, Development, and Evaluation (GRADE) approach.

Results: 14 low-dose (22.5–46 g/day) fructose trials (n=339), 1 low-dose (15 g/day) allulose trial (n=17) and 3 low-dose (30–45 g/day) tagatose trials (n=376) were identified. No trials were available for sorbose. ‘Catalytic’ doses of fructose significantly reduced HbA1c (MD= -0.38% [95% CI= -0.64%, -0.13%]) and fasting glucose (MD= - 0.14 mmol/L [95% CI= -0.25 mmol/L, -0.04 mmol/L]), without having a significant effect on fasting insulin (MD= 2.22 pmol/L [95% CI= -5.3 pmol/L, 9.73 pmol/L]). The available evidence suggested that ‘catalytic’ doses of allulose did not have a significant effect on

HbA1c (MD= 0.1% [95% CI= -0.04%, 0.24%]), fasting glucose (MD= -0.06 mmol/L [95% CI= -0.30 mmol/L, 0.18 mmol/L]) and fasting insulin (MD= 0.70 pmol/L [95% CI= -4.20 pmol/L, 5.60 pmol/L]). ‘Catalytic’ doses of tagatose significantly reduced HbA1c (MD = - 0.20% [95% CI = -0.34%, -0.06%]) and fasting glucose (MD = -0.30 mmol/L [95% CI= - 0.57 mmol/L, -0.04 mmol/L]), without having a significant effect on fasting insulin (MD= - 1.59 pmol/L [95% CI= -6.95 pmol/L, 3.77 pmol/L]). The overall quality of the evidence for the effect of low-dose fructose and allulose on HbA1c, fasting glucose and fasting insulin was graded as low-quality due to downgrades for serious indirectness and serious imprecision. The overall quality of evidence for the effect of tagatose was graded as moderate-quality due to downgrades for serious imprecision.

Conclusion: Pooled analyses indicated that small, ‘catalytic’ doses of fructose and tagatose may improve glycemic control over the longer term. There is a need for long- term randomized trials for all 4 sugars to improve our confidence in the effect estimates.

5.2 Introduction

Diabetes is a global health problem. Sugars have emerged as the dominant nutrient of concern in diabetes creating a need for caloric and low-caloric alternative sweeteners. The fructose moiety of sugar has been implicated as a potent driver of type 2 diabetes due to its unique set of biochemical, metabolic and endocrine responses3-5. This a result of undue emphasis on ecological observations, animal models of fructose overfeeding, and select human studies assessed in isolation. The important biological mechanisms by which small doses of fructose may assist in the hepatic handling of glucose have largely been ignored. Contrary to the concerns that fructose may have adverse metabolic effects, an emerging literature suggests that small (< 10g/meal) of fructose and its epimers (allulose, tagatose and sorbose) may improve the metabolic handling of glucose.

Fructose exists in foods either as free fructose (i.e. fruits, honey or high fructose corn syrup) or fructose bound to glucose (i.e. sucrose)34 . Allulose, the c-3 epimer of fructose, is a low-calorie sugar (0.4 kcal/g) found naturally in small amounts in dried fruits, brown sugar and maple syrup40. Tagatose, the c-4 epimer of fructose, is also a low-calorie sugar (1.5 kcal/g) found naturally in small amounts mainly in dairy products (certain kinds of milk, cheese and yogurt)46, 47. Sorbose, the c-5 epimer of fructose, is the least studied of the three epimers and its presence in foods is currently unknown.

Acute clinical evidence has demonstrated that small doses (<10g/meal) of fructose, allulose and tagatose decrease the postprandial glycemic responses to high glycemic index meals (oral glucose, maltodextrins, mashed potatoes and sandwiches) by ~4 – 30% in healthy participants and those with prediabetes or diabetes42, 49, 50, 88, 89, 92, 93. These acute effects have been shown to be sustainable over the long term in the case of fructose. A systematic review and meta-analysis of controlled feeding trials showed that small doses (defined as < 36g/day based on 3 meals at < 10g/meal and 2 snacks at < 3g/snack) of fructose in exchange for other carbohydrates (mainly starch) decreased 118 HbA1c by 0.4% . Whether the acute benefits of allulose and tagatose translate into meaningful reductions in long-term glycemic control remains unclear.

Our objective was to update and expand the previous systematic review and meta- analysis on small doses of fructose by synthesizing the current evidence of the effect of low-dose fructose and its epimers (allulose, tagatose and sorbose) on long-term markers of glycemic control.

5.3 Methods

This systematic review and meta-analysis was conducted according to the Cochrane Handbook for Systematic Reviews and Interventions119 and results were reported according to The Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines120. The study protocol was registered on ClinicalTrials.gov under the following identification number: NCT02776722.

5.3.1 Data Sources and Searches

MEDLINE, EMBASE and the Cochrane Central Register of Controlled Trials were searched through January 31, 2017. The search strategy is presented in Table 5.1. To limit the database searches to controlled trials, validated filters from McMaster University Healthy Information Research Unit were applied121.

Table 5.1. Search strategy for studies assessing the effect of fructose and its epimers (allulose, tagatose and sorbose) on markers of long-term glycemic control

Database Search Period Search Terms MEDLINE Through to January 1. exp Fructose/ 31, 2017 2. psicose.mp. 3. allulose.mp. 4. tagatose.mp. 5. sorbose.mp. 6. 1 or 2 or 3 or 4 or 5 7. exp Glucose/ 8. glycaemic.mp. 9. glycemic.mp. 10. glycaemia.mp. 11. glycemia.mp. 12. exp Insulin/ 13. exp Glucose Tolerance Test/ 14. OGTT.mp. 15. exp Hemoglobin A, Glycosylated/ 16. HbA1c.mp. 17. 7 or 8 or 9 or 10 or 11 or 12 or 13 or 14 or 15 or 16 18. 6 and 17 19. limit 18 to animals 20. 18 not 19 21. clinical trial.mp. 22. clinical trial.pt. 23. random:.mp. 24. tu.xs. 25. 21 or 22 or 23 or 24 26. 20 and 25 EMBASE Through to January 1. exp fructose/ 31, 2017 2. psicose.mp. 3. allulose.mp. 4. tagatose.mp. 5. sorbose.mp. 6. 1 or 2 or 3 or 4 or 5 7. exp glucose/ 8. glycaemic.mp. 9. glycemic.mp. 10. glycaemia.mp. 11. glycemia.mp. 12. exp insulin/ 13. exp oral glucose tolerance test/ 14. OGTT.mp. 15. exp hemoglobin A1c/ 16. HbA1c.mp. 17. 7 or 8 or 9 or 10 or 11 or 12 or 13 or 14 or 15 or 16 18. 6 and 17 19. limit 18 to animals 20. 18 not 19 21. random:.tw. 22. clinical trial:.mp. 23. exp health care quality/

24. 21 or 22 or 23 25. 20 and 24

Cochrane Through to January 1. Fructose/ Central 31, 2017 2. psicose.mp. Register of 3. tagatose.mp. Controlled 4. 1 or 2 or 3 Trials 5. Glucose/ 6. glycaemic.mp. 7. glycemic.mp. 8. glycaemia.mp. 9. glycemia.mp. 10. Insulin/ 11. exp Glucose Tolerance Test/ 12. OGTT.mp. 13. exp Hemoglobin A, Glycosylated/ 14. HbA1c.mp. 15. 5 or 6 or 7 or 8 or 9 or 10 or 11 or 12 or 13 or 14 16. 4 and 15

5.3.2 Study Selection

We included controlled feeding trials conducted in humans lasting >7 days investigating the effect of small doses (<50g/day or <10% energy/day) of fructose, allulose, tagatose and sorbose on long-term markers of glycemic control (HbA1c, fasting glucose or fasting insulin). The 50g/day or <10% energy/day dose threshold for chronic feeding allowed for the intake of fructose and its epimers as part of 3 main meals (<10g/meal) and 3 snacks (<5g/meal) per day and aligns with current guidelines not to exceed 10% energy from free or added sugars122, 123. Trials that lasted <7 days, administered fructose intravenously, lacked an adequate comparator or did not provide suitable endpoint data were excluded.

5.3.3 Data Extraction

Two reviewers (J.C.N and C.R.B) independently reviewed and extracted relevant data on trial characteristics and outcomes from each report. Any discrepancies were reconciled by consensus. Study authors were contacted for missing outcome data when it was indicated that an outcome was measured but not reported. Mean difference (MD) and standard error (SE) of the mean difference between the treatment arm (fructose, allulose, tagatose or sorbose) and comparator arm were extracted as the main endpoints for each outcome. Between treatment change-from-baseline differences (MD and SE) were preferred over end differences (MD and SE) as the primary endpoint. If trials did not report these values, we calculated them from available data using statistics or imputed them using standard formulas119. In the absence of numerical values for outcome measurements and inability to contact study authors, values were extracted from figures using Plot Digitizer where available124. When multiple comparators were present, starch and glucose were preferred to minimize the influence of hetereogeneity which would be present when combining arms. Risk of bias of included studies were assessed using the Cochrane Collaboration Risk of Bias Tool125.

5.3.4 Data Synthesis and Analysis

Primary pooled analyses were conducted on Review Manager (RevMan) version 5.3 (Copenhagen, Denmark) using the generic inverse variance method. Random-effects

65 models were used even in the absence of statistically significant heterogeneity, as they typically yield more conservative estimates. Fixed-effect models were only used when <5 trials were found for an outcome. Data were expressed as MD with 95% confidence intervals (CIs) for each outcome.

Inter-study heterogeneity was tested using Cochran’s Q statistic (significance at p<0.10) and quantified with the I2 statistic, where I2>50% was evidence of substantial heterogeneity. As a sensitivity analysis, we removed each study within a given outcome and recalculated the summary effect (the “leave one out” approach)126. If >10 studies were available for a given outcome and heterogeneity was substantial (I2>50% or

PQ<0.10), sources of heterogeneity were explored using a priori subgroup analyses by age, health status, comparator, dose, sugar form, design, duration, energy balance and risk of bias127-129. If >10 studies were found for a given outcome, we explored the possibility of publication bias by inspecting funnel plots and conducting Egger’s and Begg’s tests (each significant at p<0.10). If publication bias was suspected, results were shown without imputation and with “missing” studies imputed with Duval and Tweedie’s trim and fill method130. Subgroup and publication bias analyses were conducted on Stata (version 12, College Station, TX, USA).

5.3.5 Quality of Evidence Assessment

The overall quality of evidence was assessed using the grading of recommendations, assessment, development, and evaluation (GRADE) approach131. Evidence was graded as high, moderate, low or very low quality. Controlled feeding trials were graded as high-quality evidence by default and downgraded based on 5 pre-specified criteria: (1) risk of bias (assessed through the Cochrane Risk of Bias tool), (2) inconsistency (substantial unexplained inter-study heterogeneity, I2>50%, p<0.01), (3) indirectness (presence of factors that limited the generalizability of the results), (4) imprecision (the 95% CI for effect estimates were wide or crossed a minimally important difference for benefit or harm), and (5) publication bias (significant evidence of small-study effects).

5.4 Results

5.4.1 Search Results

Figure 5.1 illustrates the systematic search and selection of literature. 1764 reports were identified from database and manual searches, of which 1597 were excluded based on title and abstract. 167 reports were reviewed in full, of which an additional 153 reports were excluded. 14 reports including a total of 18 trials in 680 participants were included in the final analysis: 14 low-dose fructose trials132-143, 1 low-dose allulose trial92, 3 low-dose tagatose trials66, 86, 144 and 0 sorbose trials.

Figure 5.1. Flowchart of literature search for the effect of small doses of fructose and its epimers (allulose, tagatose and sorbose) on glycemic endpoints (HbA1c, fasting glucose and fasting insulin). Electronic searches of MEDLINE, EMBASE and Cochrane Central Register of Controlled Trials were supplemented by manual searches of the references of included articles. FG, fasting glucose; FI, fasting insulin

5.4.2 Trial Characteristics

Characteristics of all trials included in the meta-analysis are shown in Table 5.2.

14 trials (n=339) assessed the effect of low-dose fructose on markers of long-term glycemic control. All trials were randomized (100%). Starch (8 comparisons) and glucose (6 comparisons) were the main comparators. Trial sizes tended to be small (median n=21), short (median duration=2.5 weeks, range 1 – 52 weeks) and were mainly conducted in an outpatient setting in Europe (7/14 trials) or North America (6/14 trials). Trials were performed in people with diabetes (6/14 trials), healthy subjects (3/14 trials) and those who were overweight or obese (3/14 trials). Fructose was provided in liquid or mixed forms at a median dose of 36g/d (or 7.2% daily energy intake) to participants in negative (6/14 trials), neutral (5/14 trials) or positive energy balance (3/14 trials).

Only 1 trial (n=17) assessed the effect of low-dose allulose on markers of long-term glycemic control. This was a randomized double-blind placebo-controlled parallel design trial where 17 healthy subjects consumed 5g allulose or 5g glucose with 3 meals for 12 weeks. This was an outpatient study conducted in Japan.

3 trials (n=376) assessed the effect of low-dose tagatose on markers of long-term glycemic control. Two of these trials were small (n=12 and n=8, respectively), were performed in healthy subjects and utilized glucose as the comparator. The third trial was a large (n=356), multi-center (USA & India) parallel-design trial where participants with type 2 diabetes consumed tagatose (15g x 3 per day) or Splenda (1.5g x 3 per day) dissolved in water for 40 weeks. Across all 3 trials, tagatose was either provided in solid, liquid or mixed form at a median dose of 45g/day.

As assessed by the Cochrane Risk of Bias Tool, majority of the low-dose fructose, allulose and tagatose trials were of low or unclear risk of bias with very few being at high risk across the 6 domains (Supplementary Figure 1).

Table 5.2 Characteristics of controlled feeding trials investigating the effect of low-dose fructose, allulose and tagatose on glycemic control

Mean age Feeding Dose Energy Follow-up Endpoints Funding Study Subjects (SD or Setting Design Randomized Form ‡ Comparator⁋ control* (g/d) † balance (wks) measured *** range) Fructose Turner et al. 4 HTG 48 (36-57) IP, USA C Met Yes 33-46 Liquid Starch Negative 2 FG & FI A, I 1979132 Turner et al. 1979 2 HTG+ 41 (40-42) IP, USA C Met Yes 33-46 Liquid Starch Negative 2 FG & FI A, I (DM)132 DM2

Rizkalla et al.1986 Glucose, HBA1c, FG & 23 OW/OB 22 (2) OP, France P Met Yes 36 Liquid Negative 2 A, I (E1)133, 134 Galactose FI

Rizkalla et al. Glucose, HBA1c, FG & 18 OW/OB 22 (2) OP, France P Met Yes 36 Liquid Negative 2 A, I 1986 (E2)133, 134 Galactose FI Paganus et al. 12.2 22 DM1 OP, Finland C Supp Yes 37 Mixed Starch Neutral 3 HbA1c I 1987 (guar)135 (8.9-15.9) Paganus et al. 12.3 8 DM1 OP, Finland C Supp Yes 37 Mixed Starch Neutral 3 HbA1c I 1987 (control)135 (10.7-14.8)

Grigoresco et al. HBA1c, FG & 8 DM2 40 (6.9) OP, France C Supp Yes 30 Liquid Starch Negative 8 A, I 1988136 FI Blayo et al. 14 DM1, Starch, 46.9 (13.1) OP, France P Supp Yes 20-30 Mixed Negative 52 HbA1c & FG A, I 1990137 6 DM2 Sucrose Sunehag et al. 36 H 12.4 (3.4) IP/OP, USA P Met Yes ~35.5 Mixed Starch Neutral 1 FG & FI A 2002138 Vaisman et al. 25 DM2 65.4 (10.7) OP, Israel P Supp Yes 22.5 - Starch Neutral 12 HbA1c - 2006139 Aeberli et al. 29 H 26.3 (6.6) OP, Switzerland C Supp Yes 40 Liquid Glucose Positive 3 FG A, I 2011140 Heden et al. 40 H+ 17.9 (1.9) OP, USA C Supp Yes net 35 Liquid Glucose Positive 2 FG A 2014141 OW/OB Lowndes et al. 95 H+ Glucose, 36.3 (11.0) OP, USA P Supp Yes 45 Liquid Neutral 10 FG & FI I 2015142 OW/OB Control Heden et al. 2015 7 OB 18 (1.1) OP, USA C Supp Yes net 35 Liquid Glucose Positive 2 FG & FI A (+/- exercise)143 Allulose

Hayashi et al. HBA1c, FG & 17 H 34 (3.7) OP, Japan P Supp Yes 15 Mixed Glucose - 12 I 201092 FI Tagatose Buemann et al. 8 H 26.2 (2.8) OP, Denmark C Supp Yes 30 Solid Sucrose Neutral 2 FG & FI A, I 199866 Boesch et al. 12 H (21-30) OP, Switzerland C Supp No 45 Mixed Sucrose - 4 FG & FI - 200186 Ensor et al. HBA1c, FG & 356 DM2 51.7 (10.4) OP, India & USA P Supp Yes 45 Liquid Splenda Neutral 40 A, I 2015144 FI

HTG, hypertriglyceridemia; DM1, diabetes mellitus type 1; DM2, diabetes mellitus type 2; H, healthy; OW, overweight; OB, obese; SD, standard deviation; IP, inpatient; OP, outpatient; C, crossover; P, parallel; Met, metabolic; Supp, supplemented; FG, fasting glucose; FI, fasting insulin; A, agency; I, industry * Met feeding control represents the provision of all meals, snacks and study supplements (test sugars and foods) during the study. Supp feeding control represents provision of study supplements † Doses preceded by ‘~’ represent average doses, where fructose was administered on % energy or g/kg body-weight basis. Doses preceded by ‘net’ represent the net difference between treatment (fructose) dose and comparator dose when treatment arms contained small amounts of the comparator, vice versa. ‡ Test sugar was provided in one of three forms: (1) a liquid form, where all or most of the test sugar was provided as beverages or crystalline powder to be added to beverages, or (2) in a mixed form, where all or most of the test sugar was provided as beverages, solid foods and/or crystalline fructose to be added to beverages and/or foods, or (3) a solid form, where the test sugar was administered in the form of a solid food (i.e. cake) ⁋ Comparator refers to the reference carbohydrate or control group (i.e. starch, sucrose, glucose or splenda) ** Values are for the ratio of carbohydrate: fat: protein *** Agency funding represents funding from government, university or not-for-profit health agency sources.

Figure 5.2 presents a super-plot of the summary of pooled effect estimates of low-dose fructose, allulose and tagatose on markers of long-term glycemic control. Forest plots for each individual sugar on HbA1c, fasting glucose and fasting insulin are presented in Supplementary Figures 2 – 6.

5.4.3 Effect of Low-dose Fructose & its Epimers on HbA1c

7 trials conducted in 105 participants with diabetes (type 1 and type 2) and those who were overweight/obese demonstrated a significant reduction in HbA1c when fructose was exchanged mainly for starch or glucose (MD= -0.38% [95% CI=-0.64%, -0.13%], p=0.003, I2=0%, heterogeneity p=0.44; moderate quality evidence). No significant effect was found with administration of low-dose allulose in 17 healthy subjects. Low-dose tagatose administration significantly reduced HbA1c compared to Splenda (MD= -0.20 [- 0.34, -0.06], p=0.004; moderate quality evidence) in 356 participants with type 2 diabetes. Sensitivity analyses through removal of individual trials did not change the overall significance or direction of the effect in any analyses. No subgroup analyses were conducted for any of the sugars since <10 trials were available for each analysis.

5.4.4 Effect of Low-dose Fructose & its Epimers on Fasting Glucose

In 12 trials involving 195 participants, low-dose fructose led to significant reductions in fasting glucose (MD= -0.14 mmol/L [95% CI= -0.25 mmol/L, -0.04 mmol/L], p=0.008, I2=31%, heterogeneity p=0.14; moderate quality evidence). No significant effect was found with administration of low-dose allulose in 17 healthy subjects. In 2 trials involving 368 participants with and without type 2 diabetes, low-dose tagatose administration led to significant reductions in fasting glucose (MD= -0.30 mmol/L [-0.57 mmol/L, -0.04 mmol/L], p=0.02, I2=0%, heterogeneity p=0.38; moderate quality evidence). This result was sensitive to the removal of the large, multi-center trial (MD = -0.12 mmol/L [95% CI= -0.61 mmol/L, 0.37 mmol/L], p=0.63). Subgroup analyses were not conducted for any of the sugars due to lack of significant heterogeneity (I2 <50%).

5.4.5 Effect of Low-dose Fructose & its Epimers on Fasting Insulin

In 10 trials involving 153 participants, low-dose fructose had no significant effect on fasting insulin. No significant effect was found with administration of low-dose allulose in 17 healthy subjects. In 3 trials involving 376 participants with and without diabetes, low- dose tagatose did not have a significant effect on fasting insulin. Sensitivity analyses through removal of individual trials did not change the overall significance or direction of the effect in any analyses. Subgroup analyses were not conducted for any of the sugars due to lack of significant heterogeneity (I2 <50%).

Figure 5.2. Summary of the pooled effect estimates from controlled feeding trials assessing the effect of low-dose (<50g/day) fructose, allulose and tagatose on markers of long-term glycemic control. To allow the summary estimates for each endpoint to be displayed on the same axis, mean differences (MDs) were transformed to standardized mean differences (SMDs) and pseudo-95% confidence intervals (CIs), which were derived directly from the original mean difference and 95% CI. N, number of participants.

5.4.6 Publication Bias

There was no evidence of publication bias through visual inspection of funnel plots, Egger’s tests or Begg’s tests for the effect of low-dose fructose on fasting glucose and fasting insulin where > 10 trials were available (Supplementary Figure 7).

5.4.7 GRADE Assessment

Table 5.3 shows the summary of the quality of evidence assessment for the effect of low-dose fructose, allulose and tagatose on long-term markers of glycemic control.

The quality of the evidence for the effect of low-dose fructose and allulose on HbA1c, fasting glucose and fasting insulin were graded as low-quality owing to downgrades for serious imprecision and serious indirectness. Serious imprecision in the estimates for chronic fructose and allulose intake were due to 95% CIs of pooled effect estimates including the minimally important difference of 0.3%, 0.5 mmol/L and 5 pmol/L for

HbA1c, fasting glucose and fasting insulin, respectively. Serious indirectness in the estimates for chronic fructose intake were due to relative short follow-up duration. Only two trials had a follow-up duration of > 12-weeks. Serious directness in the estimates for chronic allulose intake were due to a limited number of trials and sample size (1 trial n= 17 healthy subjects).

Evidence for the effect of low-dose tagatose on HbA1c, fasting glucose and fasting insulin were graded as moderate-quality evidence owing to downgrades for serious imprecision (95% CIs of pooled effect estimates included the minimally important difference of 0.3%, 0.5 mmol/L and 5 pmol/L for HbA1c, fasting glucose and fasting insulin, respectively). The quality of evidence was not downgraded for serious indirectness even though 1 – 3 trials were available because 356 – 378 participants were included in the analysis. The large, multi-center trial (USA & India) in 356 participants with type 2 diabetes also had a sufficient follow-up duration (40 weeks).

Table 5.3. GRADE Quality of Evidence Assessment for the effect of low-dose fructose, allulose and tagatose on markers of long-term glycemic control

1 Serious imprecision for the effect of low-dose fructose, allulose and tagatose on HbA1c, fasting glucose and fasting insulin due to 95% CIs crossing the minimally important difference of 0.3% for HbA1c, 0.5 mmol/L for fasting glucose or 5 pmol/L for fasting insulin 2 For fructose trials, serious indirectness was due to the relatively short follow-up duration. Only 2 trials were >12 weeks. For allulose trials, serious indirectness was due to limited number of trials and sample size (1 trial, n = 17 healthy subjects). 3Bias cannot be excluded since we were unable to test for funnel plot asymmetry due to lack of power (<10 trials included in the analysis) 4 No serious indirectness for the effect of low-dose tagatose on HbA1c, fasting glucose and fasting insulin as 356 – 378 participants were included in the analysis even though only 1 – 3 trials were available. Trials were of sufficient length and assessed the effect of low-dose tagatose in a large population of interest. The one multi-center trial (USA & India) studying 356 participants with type 2 diabetes had a follow-up duration of 40 weeks.

5.5 Discussion

5.5.1 Summary of Findings

The results from our systematic review and meta-analysis of 14 controlled feeding trials involving 339 participants with and without diabetes over follow-up durations of 1-52 weeks showed that small ‘catalytic’ doses (22.5–46g/d) of fructose may improve glycemic control. We report that the available evidence suggested that ‘catalytic’ doses (15g/d) of allulose failed to improve markers of long-term glycemic control. These findings are limited to only one trial conducted in 17 healthy individuals. Further research is likely to have an important impact on our confidence in the effect estimate and is likely to change the estimate. We also report that results from 3 controlled feeding trials involving 376 participants with and without diabetes over follow-up durations of 2-40 weeks showed that small ‘catalytic’ doses (30-45 g/d) of tagatose may improve glycemic control. No trails were identified for sorbose.

5.5.2 Findings in Context of Previous Research

The results from our fructose analysis are consistent with a previous meta-analysis that showed a beneficial effect of ‘catalytic’ doses of fructose (22.5 – 36g/d) in isocaloric exchange for other carbohydrates on markers long-term glycemic control, without adversely affecting other cardiometabolic risk factors (body weight, fasting insulin, triglyceride and uric acid)118. These findings are also consistent with a recent meta- analysis that demonstrated a significant reduction in HbA1C and fasting glucose when fructose was substituted (mean 68 g/d, range 40 – 150 g/d) specifically with glucose or sucrose-sweetened foods145. Two previously conducted meta-analyses have also reported a beneficial effect of isocalorically exchanging fructose (median dose: 60 g/d) for other carbohydrates on glycated blood proteins in people with diabetes (~0.53% reduction in HbA1c), and without diabetes (<90g/d significantly improved HbA1c concentrations dependent on the dose, duration of study and severity of dysglycemia)146, 147.

To date, our systematic review and meta-analysis is the first to synthesize the evidence of small ‘catalytic’ doses of allulose, tagatose and sorbose on markers of long-term

76 glycemic control. Although several studies have documented the acute effect of allulose consumption on postprandial glycemic response42, 92, 107, 148, only one long-term trial was available for analysis. Tagatose was initially researched and developed as a full-bulk, low-calorie sweetener but after early data from animal and human studies suggested that it improved glycemia parameters, it was formulated as a potential drug for treating type 2 diabetes48. The phase 3 clinical trial for tagatose was included in our meta- analysis and indicated that tagatose was effective in treating markers of long-term 144 glycemic control such as, HbA1c and fasting glucose . Our findings partly align with results from previous non-placebo-controlled trials that have shown a tendency for glycemic improvement, but did not reach statistical significance. When 8 individuals with type 2 diabetes consumed 15g tagatose three times daily with food for 1 year, their

HbA1C concentrations fell from 10.6% to 9.6%, though results were non-significant (p=0.08)149. In an intention-to-treat analysis of a phase 2 clinical trial when 145 individuals with type 2 diabetes consumed 2.5g, 5g or 7.5g tagatose three times daily with food, reductions in HbA1c from baseline to six months were observed in the 5g (7.4% to 7.3%) and 7.5g (7.3% to 7.1%) tagatose group, though not statistically significant150.

5.5.3 Mechanism of Action

One potential mechanism to explain the improvements in glycemia observed is enhancement of hepatic glucokinase activity when small doses of fructose and tagatose are consumed with carbohydrates. It has been reported that hepatic glucokinase activity is decreased in individuals with type 2 diabetes10, 73. Glucokinase is inhibited by glucokinase regulatory protein (GKRP), and this action is enhanced in the presence of fructose-6-phosphate96, 108-112. Under fasting conditions, hepatic glucokinase is localized primarily in the nucleus, where it is bound to the glucokinase regulatory protein (GKRP) and fructose-6-phosphate95, 96, 113. In the postprandial state (presence of fructose or tagatose with glucose), fructose and tagatose are phosphorylated to fructose-1- phosphate and tagatose-1-phosphate by an enzyme called ketohexokinase85, 114. These metabolites compete with fructose-6-phosphate binding to GKRP, and in so doing, release glucokinase from GKRP151-154. This enables the liberated and activated

77 glucokinase to translocate from the nucleus to the cytosol where it can drive hepatic glucose uptake, promote glycogen synthesis, suppress hepatic glucose output and reduce plasma glucose levels. In support of this mechanism, ‘catalytic’ doses of fructose have shown a ~30% reduction in postprandial hepatic glucose output under hyperglycemic conditions in people with type 2 diabetes and ~3-fold increase in glycogen synthesis under euglycemic hyperinsulinemic conditions in people without diabetes78. Studies have also demonstrated hepatic glucose sensing impairment in individuals with impaired fasting glucose, which can be experimentally reversed by infusion of low-dose fructose80. It is important to note that support for the role of tagatose in this mechanism comes from in vitro and in animal models as human studies are lacking.

5.5.4 Clinical Implications

The reduction in HbA1c of 0.38% after chronic isocaloric substitution of fructose with other carbohydrates was clinically meaningful as it exceeded the threshold of >0.3% proposed by the U.S. Food and Drug Administration (FDA) for the development of new drugs for diabetes and lies at the lower limit of efficacy expected for oral hypoglycemic agents155, 156. Although many of these studies assessed the effect of exchanging one sweetener for another, use of fructose and tagatose as alternative sweeteners could be part of a broader strategy to decrease intake of excess calories from all sugars and refined starches while promoting the intake of more nutrient-dense foods that are high in whole grains, viscous fibers, fruit and vegetables, pulses, nuts and dairy or nondairy products. A benefit of ‘catalytic’ doses of fructose also has implications for consumption of low-GI fruit (i.e. apple has ~9-10g fructose)157. A secondary analysis of a randomized controlled trial in 152 participants with type 2 diabetes found that when comparing the highest with the lowest quartile of low GI fruit intake, the % change in HbA1c was reduced by -0.5%158. This reduction is similar to what we found in our analysis even though fructose was mainly consumed as crystalline fructose rather than in the form of low-GI fruit.

5.5.5 Consideration of a Dose Threshold for Harm

Although it appears that fructose in isocaloric exchange for other carbohydrates may benefit glycemia, a dose threshold for harm must also be considered. A meta-analysis in individuals with type 2 diabetes previously showed that fructose at doses >60g/d (in excesss of CDA recommendations) or >10% energy in isocaloric exchange for carbohydrate increased serum triglyceride levels159. Despite showing a tendency for improvement in HbA1c, Livesey and Taylor in their meta-analysis showed a consistent triglyceride-raising effect of fructose at doses >100 g/d (>95th percentile total US fructose intake) across different subject types147. Fructose intake in the U.S. population is also relatively high (mean, 54.7g/d; range, 38.4-72.8 g/d) accounting for 10.2% of total calorie intake raising the question of whether there is need for the addition of catalytic doses of fructose to the diet160. Similarly, a dose threshold for harm must also be considered in the case of tagatose. The presence of chronically elevated plasma uric acid levels (i.e. hyperuricemia) is a known risk factor for the development of gout. Ingestion of single high bolus doses (>30g) of tagatose has been associated with a mild, transient increase of plasma uric acid concentration in both healthy subjects and individuals with type 2 diabetes69, 85, 161. Gastrointestinal disturbances have also been reported with large single-doses (>25g) of fructose and tagatose intake in some individuals due to malabsorption of the sugars. In three separate studies, when the fructose dose was increased from 25g to 50g in 10% solution, the prevalence of fructose malabsorption increased from 0 to 37.5%, 11 to 58% and 50 to 80%, respectively51-53. A single-dose tolerance study in 73 young healthy male participants reported that consumption of 29 or 30g tagatose led to nausea and diarrahea in 15.1 and 31.5% of participants, respectively162. Another tolerance study in 8 healthy subjects and 8 individuals with type 2 diabetes reported reported diarrhea, nausea and/or flatulence in 100% of participants after consuming a single 75g dose of tagatose50.

5.5.6 Strengths & Limitations

Our systematic review and meta-analysis presented several strengths: 1) a rigorous search and selection process of available literature examining the effect of small ‘catalytic’ doses of fructose and its epimers (allulose, tagatose and sorbose) on glycemic control; 2) inclusion of controlled trials which give greater protection against bias; 3) the pooled synthesis of data from 18 trials involving 680 participants; and 4) an assessment of overall quality of evidence using the GRADE assessment tool. However, a number of limitations also complicate the interpretation of these pooled analyses: 1) serious indirectness for the effect of low-dose fructose on HbA1c, fasting glucose and fasting insulin – most of the trials were of relatively short duration, with only 2 trials having follow-up durations > 12 weeks. It is possible that the shorter trials may have underestimated the HbA1c reduction, given the evidence that HbA1c reduces at ~0.1%/day at a steady state with a half-life of 5 weeks163 (2) serious indirectness for the effect of low-dose allulose on HbA1c, fasting glucose and fasting insulin – only 1 trial with a sample size of 17 healthy individuals was available for analysis. The meta- regression may have been underpowered to detect a true difference; and 3) serious imprecision for the effect of low-dose fructose, allulose and tagatose on HbA1c, fasting glucose and fasting insulin as the 95% CI of the pooled effect estimate crossed the clinically meaningful threshold for benefit.

5.6 Conclusions In conclusion, our pooled analyses indicated that small, ‘catalytic’ doses of fructose and tagatose may improve glycemic control over the longer term. Sources of uncertainty remain with indirectness and imprescision in the estimates for chronic intake of fructose and its epimers. There is a need for more long-term (>6 months) randomized clinical trials to clarify whether fructose and its epimers can serve as effective alternative sweeteners in the management of diabetes.

Supplementary Tables & Figures

Supplementary Figure 5.1. Risk of bias summary of controlled feeding trials assessing the effect of low-dose fructose (top), allulose (middle) and tagatose (bottom) on markers of long-term glycemic control. Colored bars represent the proportion of studies assessed as low (green), unclear (yellow) or high (red) risk of bias for the 6 domains of bias above according to criteria set by the Cochrane Risk of Bias tool125

Supplementary Figure 5.2. Forest plot for the effect of low-dose fructose (< 50g/day) on HbA1c. Pooled effect estimates for the overall effect is represented by the diamond. Data are expressed as mean differences (MDs) with 95% confidence intervals (CIs), using the generic inverse variance method with random effects models. Paired analyses were applied to all crossover trials. Inter-study heterogeneity was tested by the Cochran Q-statistic at a significance level of p<0.10 and quantified by I2, levels < 50% represent moderate heterogeneity, > 50% representing substantial heterogeneity, and > 75% representing considerable heterogeneity.

Supplementary Figure 5.3. Forest plot for the effect of low-dose fructose (< 50g/day) on fasting glucose. Pooled effect estimates for the overall effect is represented by the diamond. Data are expressed as mean differences (MDs) with 95% confidence intervals (CIs), using the generic inverse variance method with random effects models. Paired analyses were applied to all crossover trials. Inter-study heterogeneity was tested by the Cochran Q-statistic at a significance level of p<0.10 and quantified by I2, levels < 50% represent moderate heterogeneity, > 50% representing substantial heterogeneity, and > 75% representing considerable heterogeneity.

Supplementary Figure 5.4. Forest plot for the effect of low-dose fructose (< 50g/day) on fasting insulin. Pooled effect estimates for the overall effect is represented by the diamond. Data are expressed as mean differences (MDs) with 95% confidence intervals (CIs), using the generic inverse variance method with random effects models. Paired analyses were applied to all crossover trials. Inter-study heterogeneity was tested by the Cochran Q-statistic at a significance level of p<0.10 and quantified by I2, levels <50% represent moderate heterogeneity, >50% representing substantial heterogeneity, and >75% representing considerable heterogeneity.

Supplementary Figure 5.5. Forest plot for the effect of low-dose allulose (< 50g/day) on HbA1c (top), fasting glucose (middle) and fasting insulin (bottom). Pooled effect estimate for the overall effect is represented by the diamond. Data are expressed as mean differences (MDs) with 95% confidence intervals (CIs), using the generic inverse variance method with fixed effects models. Paired analyses were applied to all crossover trials. Inter-study heterogeneity was tested by the Cochran Q-statistic at a significance level of p<0.10 and quantified by I2, levels <50% represent moderate heterogeneity, >50% representing substantial heterogeneity, and >75% representing considerable heterogeneity

Supplementary Figure 5.6. Forest plot for the effect of low-dose tagatose (< 50g/day) on HbA1c (top), fasting glucose (middle) and fasting insulin (bottom). Pooled effect estimate for the overall effect is represented by the diamond. Data are expressed as mean differences (MDs) with 95% confidence intervals (CIs), using the generic inverse variance method with fixed effects models. Paired analyses were applied to all crossover trials. Inter-study heterogeneity was tested by the Cochran Q-statistic at a significance level of p<0.10 and quantified by I2, levels <50% represent moderate heterogeneity, >50% representing substantial heterogeneity, and >75% representing considerable heterogeneity

Chapter 6 : General Discussion

6.1 Summary

Two studies were conducted to assess the effect of small doses of fructose and its epimers (allulose, tagatose and sorbose) on glycemic control. The first study was an acute analysis of small doses (5g, 10g) of fructose and its c-3 epimer, allulose, on postprandial glucose metabolism in 24 individuals with type 2 diabetes. In this randomized double-blind crossover trial, we reported that 10g allulose modestly reduced the plasma glucose iAUC response to a 75g oral glucose challenge. 5g allulose, 5g fructose and 10g fructose did not have a significant effect on plasma glucose iAUC responses to the oral glucose challenge. The second study was a long- term analysis of low-dose (< 50g/d) fructose and all its epimers on glycemic control. In this systematic review, we reported that chronic feeding of low-dose fructose and its c-4 epimer, tagatose, led to significant reductions in HbA1c and fasting glucose, with no effect on fasting insulin. The available evidence suggested that low-dose allulose, the c-

3 epimer of fructose, did not significantly effect HbA1c, fasting glucose and fasting insulin. To date, no trials have investigated the role of low-dose sorbose, the c-5 epimer of fructose, on markers of long-term glycemic control.

6.2 Clinical Implications

Fructose, allulose and tagatose taste similar to sucrose and have been shown to have a minimal impact on blood glucose and insulin levels when consumed alone. In our acute clinical trial, we demonstrated that small amounts of allulose modestly lower the glycemic response to high glycemic index foods (i.e. oral glucose and standard meal load) in individuals with type 2 diabetes. It is currently unclear whether these acute reductions translate into meaningful reductions in HbA1c, given that only one trial was identified in our systematic review and meta-analysis. Even though we were unable to replicate previous findings that small doses of fructose reduce the postprandial glycemic response to high glycemic index meals, our systematic review and meta-analysis suggested that chronic substitution of low-dose fructose with other carbohydrates lead to sustainable improvements in glycemic control over the long-term. Previous trials have reported that small doses of tagatose also reduce the postprandial glycemic response to

89 high glycemic index meals. In our systematic review and meta-analysis, we demonstrated that these acute reductions are sustainable over the long-term.

Given the current concern with sugars in the diabetes epidemic, clinical implications of our findings are that fructose, allulose and tagatose could serve as alternative sweeteners in the management of type 2 diabetes. Rather than replacing one type of caloric or low-caloric sweetener with another, the use of fructose and its epimers should be part of a broader strategy to decrease the intake of excess calories from all sugars and refined starches while promoting the intake of more nutrient-dense foods that are high in whole grains, viscous fibers, fruit and vegetables, pulses, nuts and dairy or nondairy products.

6.3 Strengths & Limitations

The strengths of the acute clinical trial were:

1. We conducted a randomized double-blind controlled trial which is considered the gold standard for a clinical trial.

2. The crossover design of the trial reduced the between-subject variation by allowing each subject to act as his/her own control

3. The control drink (75g-OGTT) was administered twice providing a more reliable estimate of postprandial glycemic and insulinemic responses

4. Based on a power analysis, the sample size of the trial was sufficient for equivalence testing of the effect of fructose and allulose on plasma glucose iAUC.

The strengths of the systematic review and meta-analysis were:

1. We carried out a rigorous search and selection process of available literature examining the effect of small ‘catalytic’ doses of fructose and its epimers (allulose, tagatose and sorbose) on glycemic control

2. Inclusion of controlled trials which gives greater protection against bias

3. The pooled synthesis of data from 18 trials involving 680 participants

4. An assessment of overall quality of evidence using the GRADE assessment tool

Limitations of the acute clinical trial were:

1. The trial was not designed to examine the mechanism(s) by which allulose reduced the postprandial glycemic response to an oral glucose load

2. The doses examined in this study may have been insufficient to detect a dose- response/threshold

3. The acute design of the study creates uncertainty as to whether reductions in the postprandial glycemic response will manifest as sustainable improvements in glycemic control over the long-term.

Limitations of the systematic review and meta-analysis were:

1. Lack of available trials assessing the effect of low-dose sorbose on markers of long-term glycemic control.

2. Serious indirectness for the effect of low-dose fructose on HbA1c, fasting glucose and fasting insulin – most of the trials were of relatively short duration, with only 2 trials having follow-up durations > 12 weeks. It is possible that the shorter trials

may have underestimated the HbA1c reduction, given the evidence that HbA1c reduces at ~0.1%/day at a steady state with a half-life of 5 weeks163

3. Serious indirectness for the effect of low-dose allulose on HbA1c, fasting glucose and fasting insulin – only 1 trial with a sample size of 17 healthy individuals was available for analysis. The meta-regression may have been underpowered to detect a true difference

4. Serious imprecision for the effect of low-dose fructose, allulose and tagatose on

HbA1c, fasting glucose and fasting insulin as the 95% CI of the pooled effect estimate crossed the clinically meaningful threshold for benefit.

6.4 Future Directions

Findings from our acute clinical trial can inform future long-term clinical trials investigating the role of allulose in long-term glycemic control. We showed that 10g

91 allulose significantly reduced the postprandial glycemic response to a 75g oral glucose challenge. A potential future study could compare the effect of 10g allulose three times daily with meals compared to Splenda (to control for calories) for 1 year in individuals with type 2 diabetes. HbA1c would be measured as the primary endpoint. Secondary endpoints would include: body weight, fasting glucose, fasting insulin, blood pressure, lipids and uric acid.

We failed to demonstrate a reduction in the postprandial glycemic response to the 75g oral glucose challenge with small doses of fructose. A previous study reported similar results, but found a significant reduction when 10g fructose was consumed 60 or 30 min prior to an instant mashed potato meal load in healthy subjects90. To date, it is unclear whether timing of fructose administration has an effect on the postprandial glycemic response in individuals with type 2 diabetes. A potential future study could administer 10g fructose 60, 30 and 0 min prior to an oral glucose challenge to individuals with type 2 diabetes in a randomized crossover fashion. Finger-prick capillary blood glucose could be measured at -60, -30, 0, 30, 60, 90 and 120 min. The primary endpoint would be plasma glucose iAUC.

6.5 Conclusions

Objective 1: To assess the acute effects of small doses (5g, 10g) of fructose and its c-3 epimer, allulose compared with control (0g) on glucose and insulin response to a 75g- oral glucose tolerance test (75g-OGTT)

 Conclusion: 10g allulose modestly reduced the postprandial glucose response to the 75g-OGTT without having a significant effect on the insulin response in individuals with type 2 diabetes. 5g allulose, 5g fructose and 10g fructose did not have a significant effect on postprandial glucose and insulin responses.

Objective 2: To assess whether there is a dose response over the proposed dose range (0g, 5g, 10g) for the effects of fructose and allulose on glucose and insulin responses to a 75g-OGTT in individuals with type 2 diabetes

 Conclusion: There was no dose response over the proposed dose range (0g, 5g, 10g) for the effects of fructose and allulose on glucose and insulin response to a 75g-OGTT in individuals with type 2 diabetes

Objective 3: To assess whether the effects of allulose and fructose are equivalent on the incremental area under the curve (iAUC) for plasma glucose across the 2 dose levels (5g and 10g) compared with control (0g)

 Conclusion: The effects of allulose and fructose were equivalent on the primary endpoint of plasma glucose iAUC across the 2 dose levels compared with control

Objective 4: To synthesize the evidence of the long-term effect of low-dose fructose and its epimers (allulose, tagatose and sorbose) on HbA1c, fasting glucose and fasting insulin  Conclusion: The available evidence suggested that low-dose fructose and

tagatose significantly reduced HbA1c and fasting glucose, without having a significant effect on plasma insulin. Only 1 trial was available for allulose which

showed that low-dose allulose did not have a significant effect on HbA1c, fasting glucose and fasting insulin in 17 healthy subjects. No trials were identified for low-dose sorbose.

References

1. Global report on diabetes. World Health Organization, Geneva, 2016.

2. International Diabetes Federation. IDF Diabetes, 7 ed. Brussels, Belgium: International Diabetes Federation, 2015. http://www.diabetesatlas.org

3. DiNicolantonio JJ, O'Keefe JH and Lucan SC. Added fructose: a principal driver of type 2 diabetes mellitus and its consequences. Mayo Clin Proc. 2015;90:372-81.

4. Lustig RH. Sickeningly Sweet: Does Sugar Cause Type 2 Diabetes? Yes. Can J Diabetes. 2016;40:282-6.

5. Lustig RH, Schmidt LA and Brindis CD. Public health: The toxic truth about sugar. Nature. 2012;482:27-9.

6. Canadian Diabetes Association Clinical Practice Guidelines Expert C, Goldenberg R and Punthakee Z. Definition, classification and diagnosis of diabetes, prediabetes and metabolic syndrome. Can J Diabetes. 2013;37 Suppl 1:S8-11.

7. Haffner SM, Lehto S, Ronnemaa T, Pyorala K and Laakso M. Mortality from coronary heart disease in subjects with type 2 diabetes and in nondiabetic subjects with and without prior myocardial infarction. N Engl J Med. 1998;339:229-34.

8. Canadian Diabetes Association Clinical Practice Guidelines Expert C and Cheng AY. Canadian Diabetes Association 2013 clinical practice guidelines for the prevention and management of diabetes in Canada. Introduction. Can J Diabetes. 2013;37 Suppl 1:S1-3.

9. Ceriello A and Colagiuri S. International Diabetes Federation guideline for management of postmeal glucose: a review of recommendations. Diabet Med. 2008;25:1151-6.

10. Rizza RA. Pathogenesis of fasting and postprandial hyperglycemia in type 2 diabetes: implications for therapy. Diabetes. 2010;59:2697-707.

11. Group AC, Patel A, MacMahon S, Chalmers J, Neal B, Billot L, Woodward M, Marre M, Cooper M, Glasziou P, Grobbee D, Hamet P, Harrap S, Heller S, Liu L, Mancia G, Mogensen CE, Pan C, Poulter N, Rodgers A, Williams B, Bompoint S, de Galan BE, Joshi R and Travert F. Intensive blood glucose control and vascular outcomes in patients with type 2 diabetes. N Engl J Med. 2008;358:2560-72.

12. Ismail-Beigi F, Craven T, Banerji MA, Basile J, Calles J, Cohen RM, Cuddihy R, Cushman WC, Genuth S, Grimm RH, Jr., Hamilton BP, Hoogwerf B, Karl D, Katz L, Krikorian A, O'Connor P, Pop-Busui R, Schubart U, Simmons D, Taylor H, Thomas A, Weiss D, Hramiak I and group At. Effect of intensive treatment of hyperglycaemia on

94 microvascular outcomes in type 2 diabetes: an analysis of the ACCORD randomised trial. Lancet (London, England). 2010;376:419-30.

13. Action to Control Cardiovascular Risk in Diabetes Study G, Gerstein HC, Miller ME, Byington RP, Goff DC, Jr., Bigger JT, Buse JB, Cushman WC, Genuth S, Ismail- Beigi F, Grimm RH, Jr., Probstfield JL, Simons-Morton DG and Friedewald WT. Effects of intensive glucose lowering in type 2 diabetes. N Engl J Med. 2008;358:2545-59.

14. Moritz T, Duckworth W and Abraira C. Veterans Affairs diabetes trial-- corrections. N Engl J Med. 2009;361:1024-5.

15. Canadian Diabetes Association Clinical Practice Guidelines Expert C, Imran SA, Rabasa-Lhoret R and Ross S. Targets for glycemic control. Can J Diabetes. 2013;37 Suppl 1:S31-4.

16. Bunn HF, Haney DN, Kamin S, Gabbay KH and Gallop PM. The biosynthesis of human hemoglobin A1c. Slow glycosylation of hemoglobin in vivo. J Clin Invest. 1976;57:1652-9.

17. Monnier L, Lapinski H and Colette C. Contributions of fasting and postprandial plasma glucose increments to the overall diurnal hyperglycemia of type 2 diabetic patients: variations with increasing levels of HbA(1c). Diabetes Care. 2003;26:881-5.

18. Woerle HJ, Neumann C, Zschau S, Tenner S, Irsigler A, Schirra J, Gerich JE and Goke B. Impact of fasting and postprandial glycemia on overall glycemic control in type 2 diabetes Importance of postprandial glycemia to achieve target HbA1c levels. Diabetes Res Clin Pract. 2007;77:280-5.

19. DeFronzo RA, Tobin JD and Andres R. Glucose clamp technique: a method for quantifying insulin secretion and resistance. Am J Physiol. 1979;237:E214-23.

20. Hucking K, Watanabe RM, Stefanovski D and Bergman RN. OGTT-derived measures of insulin sensitivity are confounded by factors other than insulin sensitivity itself. Obesity (Silver Spring). 2008;16:1938-45.

21. Bergman RN. Lilly lecture 1989. Toward physiological understanding of glucose tolerance. Minimal-model approach. Diabetes. 1989;38:1512-27.

22. World Health Organization: WHO Expert Committee on Diabetes Mellitus. Second Report. Geneva, World Health Org. (Tech. Rep. Ser., no. 646-1).

23. Jenssen T and Hartmann A. Emerging treatments for post-transplantation diabetes mellitus. Nat Rev Nephrol. 2015;11:465-77.

24. Grodsky GM. A new phase of insulin secretion. How will it contribute to our understanding of beta-cell function? Diabetes. 1989;38:673-8.

25. Phillips DI, Clark PM, Hales CN and Osmond C. Understanding oral glucose tolerance: comparison of glucose or insulin measurements during the oral glucose tolerance test with specific measurements of insulin resistance and insulin secretion. Diabet Med. 1994;11:286-92.

26. Wareham NJ, Phillips DI, Byrne CD and Hales CN. The 30 minute insulin incremental response in an oral glucose tolerance test as a measure of insulin secretion. Diabet Med. 1995;12:931.

27. Kim JK. Hyperinsulinemic-euglycemic clamp to assess insulin sensitivity in vivo. Methods Mol Biol. 2009;560:221-38.

28. Matsuda M and DeFronzo RA. Insulin sensitivity indices obtained from oral glucose tolerance testing: comparison with the euglycemic insulin clamp. Diabetes Care. 1999;22:1462-70.

29. Kahn SE. The relative contributions of insulin resistance and beta-cell dysfunction to the pathophysiology of Type 2 diabetes. Diabetologia. 2003;46:3-19.

30. Bergman RN, Ader M, Huecking K and Van Citters G. Accurate assessment of beta-cell function: the hyperbolic correction. Diabetes. 2002;51 Suppl 1:S212-20.

31. Bergman RN, Phillips LS and Cobelli C. Physiologic evaluation of factors controlling glucose tolerance in man: measurement of insulin sensitivity and beta-cell glucose sensitivity from the response to intravenous glucose. J Clin Invest. 1981;68:1456-67.

32. Kahn SE, Prigeon RL, McCulloch DK, Boyko EJ, Bergman RN, Schwartz MW, Neifing JL, Ward WK, Beard JC, Palmer JP and et al. Quantification of the relationship between insulin sensitivity and beta-cell function in human subjects. Evidence for a hyperbolic function. Diabetes. 1993;42:1663-72.

33. Retnakaran R, Qi Y, Goran MI and Hamilton JK. Evaluation of proposed oral disposition index measures in relation to the actual disposition index. Diabet Med. 2009;26:1198-203.

34. Park YK and Yetley EA. Intakes and food sources of fructose in the United States. The American journal of clinical nutrition. 1993;58:737s-747s.

35. White JS. Sucrose, HFCS, and Fructose: History, Manufacture, Composition, Applications, and Production. In: J. M. Rippe, ed. Fructose, High Fructose Corn Syrup, Sucrose and Health New York, NY: Springer New York; 2014: 13-33.

36. Crapo PA, Kolterman OG and Olefsky JM. Effects of oral fructose in normal, diabetic, and impaired glucose tolerance subjects. Diabetes Care. 1980;3:575-82.

37. Lee BM and Wolever TM. Effect of glucose, sucrose and fructose on plasma glucose and insulin responses in normal humans: comparison with white bread. Eur J Clin Nutr. 1998;52:924-8.

38. Bantle JP, Laine DC, Castle GW, Thomas JW, Hoogwerf BJ and Goetz FC. Postprandial glucose and insulin responses to meals containing different carbohydrates in normal and diabetic subjects. N Engl J Med. 1983;309:7-12.

39. Jenkins DJ, Wolever TM, Taylor RH, Barker H, Fielden H, Baldwin JM, Bowling AC, Newman HC, Jenkins AL and Goff DV. Glycemic index of foods: a physiological basis for carbohydrate exchange. The American journal of clinical nutrition. 1981;34:362-6.

40. Oshima H KI, Izumori K. D-allulose contents in various food products and its origin. Food Sci Technol Res. 2006;12:137-143.

41. Iida T, Hayashi N, Yamada T, Yoshikawa Y, Miyazato S, Kishimoto Y, Okuma K, Tokuda M and Izumori K. Failure of d-psicose absorbed in the small intestine to metabolize into energy and its low large intestinal fermentability in humans. Metabolism. 2010;59:206-14.

42. Iida T, Kishimoto Y, Yoshikawa Y, Hayashi N, Okuma K, Tohi M, Yagi K, Matsuo T and Izumori K. Acute D-psicose administration decreases the glycemic responses to an oral maltodextrin tolerance test in normal adults. J Nutr Sci Vitaminol (Tokyo). 2008;54:511-4.

43. Kendall C WT, Jenkins A et al. A Randomized, Controlled, Crossover Study to Assess the Effects of a Sweetener on Postprandial Glucose and Insulin Excursions in Healthy Subjects. Glycemic Index Laboratories Toronto, ON, Canada. May 2014.

44. Wolever T, Jenkins AJ. A Randomized, Controlled, Crossover Study to Assess the Effects of a Sweetener on Postprandial Glucose and Insulin Excursions in Healthy Subjects. 2015. Glycemic Index Labs. Toronto, ON, Canada.

45. Wolever T, Jenkins AJ. A Randomized, Controlled, Crossover Study to Assess the Effects of a Sweetener on a Postprandial Glucose and Insulin Excursions in Subjects with Type 2 Diabetes. 2015. Glycemic Index Labs. Toronto, ON, Canada.

46. Levin GV. Tagatose, the new GRAS sweetener and health product. J Med Food. 2002;5:23-36.

47. Levin GV, Zehner LR, Saunders JP and Beadle JR. Sugar substitutes: their energy values, bulk characteristics, and potential health benefits. The American journal of clinical nutrition. 1995;62:1161s-1168s.

48. Lu Y, Levin GV and Donner TW. Tagatose, a new antidiabetic and obesity control drug. Diabetes Obes Metab. 2008;10:109-34.

49. Madenokoji N IH, Shimizu T, Hayakawa J, Sakashita M. Blunting Effect of D- tagatose on Blood Glucose when Administered Orally with Glucose in Volunteer Donors of Boundary Glycemic Level. J Jap Soc Clin Nutr. 2003;25:21–8.

50. Donner TW, Wilber JF and Ostrowski D. D-tagatose, a novel hexose: acute effects on carbohydrate tolerance in subjects with and without type 2 diabetes. Diabetes Obes Metab. 1999;1:285-91.

51. Rumessen JJ and Gudmand-Hoyer E. Absorption capacity of fructose in healthy adults. Comparison with sucrose and its constituent monosaccharides. Gut. 1986;27:1161-8.

52. Ravich WJ, Bayless TM and Thomas M. Fructose: incomplete intestinal absorption in humans. Gastroenterology. 1983;84:26-9.

53. Truswell AS, Seach JM and Thorburn AW. Incomplete absorption of pure fructose in healthy subjects and the facilitating effect of glucose. The American journal of clinical nutrition. 1988;48:1424-30.

54. Skoog SM and Bharucha AE. Dietary fructose and gastrointestinal symptoms: a review. Am J Gastroenterol. 2004;99:2046-50.

55. Kneepkens CM, Vonk RJ and Fernandes J. Incomplete intestinal absorption of fructose. Archives of disease in childhood. 1984;59:735-8.

56. Helliwell PA, Richardson M, Affleck J and Kellett GL. Regulation of GLUT5, GLUT2 and intestinal brush-border fructose absorption by the extracellular signal- regulated kinase, p38 mitogen-activated kinase and phosphatidylinositol 3-kinase intracellular signalling pathways: implications for adaptation to diabetes. The Biochemical journal. 2000;350 Pt 1:163-9.

57. Helliwell PA, Richardson M, Affleck J and Kellett GL. Stimulation of fructose transport across the intestinal brush-border membrane by PMA is mediated by GLUT2 and dynamically regulated by protein kinase C. The Biochemical journal. 2000;350 Pt 1:149-54.

58. Kellett GL and Brot-Laroche E. Apical GLUT2: a major pathway of intestinal sugar absorption. Diabetes. 2005;54:3056-62.

59. Kellett GL, Brot-Laroche E, Mace OJ and Leturque A. Sugar absorption in the intestine: the role of GLUT2. Annu Rev Nutr. 2008;28:35-54.

60. Tappy L and Le KA. Metabolic effects of fructose and the worldwide increase in obesity. Physiol Rev. 2010;90:23-46.

61. Sun SZ and Empie MW. Fructose metabolism in humans - what isotopic tracer studies tell us. Nutrition & metabolism. 2012;9:89.

62. Whistler RL, Singh PP and Lake WC. D-Psicose metabolism in the rat. Carbohydr Res. 1974;34:200-2.

63. Tsukamoto I, Hossain A, Yamaguchi F, Hirata Y, Dong Y, Kamitori K, Sui L, Nonaka M, Ueno M, Nishimoto K, Suda H, Morimoto K, Shimonishi T, Saito M, Song T, Konishi R and Tokuda M. Intestinal absorption, organ distribution, and urinary excretion of the rare sugar D-psicose. Drug Des Devel Ther. 2014;8:1955-64.

64. Williamson P, Schunk T, Woodyer R, Chiuu D, Song Q, Atiee G and Unger S. A single-dose, microtracer study to determine the mass balance of orally administered, 14C-labeled sweetener in healthy adult men (LB450). The FASEB Journal. 2014;28.

65. Matsuo T, Tanaka T, Hashiguchi M, Izumori K and Suzuki H. Metabolic effects of D-psicose in rats: studies on faecal and urinary excretion and caecal fermentation. Asia Pac J Clin Nutr. 2003;12:225-31.

66. Buemann B, Toubro S and Astrup A. D-Tagatose, a stereoisomer of D-fructose, increases hydrogen production in humans without affecting 24-hour energy expenditure or respiratory exchange ratio. The Journal of nutrition. 1998;128:1481-6.

67. Saunders JP, Zehner LR and Levin GV. Disposition of D-[U-14C]tagatose in the rat. Regulatory toxicology and pharmacology : RTP. 1999;29:S46-56.

68. Laerke HN and Jensen BB. D-tagatose has low small intestinal digestibility but high large intestinal fermentability in pigs. The Journal of nutrition. 1999;129:1002-9.

69. Buemann B, Toubro S, Holst JJ, Rehfeld JF, Bibby BM and Astrup A. D- tagatose, a stereoisomer of D-fructose, increases blood uric acid concentration. Metabolism. 2000;49:969-76.

70. Espinosa I and Fogelfeld L. Tagatose: from a sweetener to a new diabetic medication? Expert Opin Investig Drugs. 2010;19:285-94.

71. Bertelsen H, Jensen BB and Buemann B. D-tagatose--a novel low-calorie bulk sweetener with prebiotic properties. World Rev Nutr Diet. 1999;85:98-109.

72. Venema K, Vermunt SHF and Brink EJ. D-Tagatose increases butyrate production by the colonic microbiota in healthy men and women. Microbial Ecology in Health and Disease. 2005;17:47-57.

73. Caro JF, Triester S, Patel VK, Tapscott EB, Frazier NL and Dohm GL. Liver glucokinase: decreased activity in patients with type II diabetes. Horm Metab Res. 1995;27:19-22.

74. Velho G, Petersen KF, Perseghin G, Hwang JH, Rothman DL, Pueyo ME, Cline GW, Froguel P and Shulman GI. Impaired hepatic glycogen synthesis in glucokinase- deficient (MODY-2) subjects. J Clin Invest. 1996;98:1755-61.

75. Shiota M, Galassetti P, Monohan M, Neal DW and Cherrington AD. Small amounts of fructose markedly augment net hepatic glucose uptake in the conscious dog. Diabetes. 1998;47:867-73.

76. Shiota M, Moore MC, Galassetti P, Monohan M, Neal DW, Shulman GI and Cherrington AD. Inclusion of low amounts of fructose with an intraduodenal glucose load markedly reduces postprandial hyperglycemia and hyperinsulinemia in the conscious dog. Diabetes. 2002;51:469-78.

77. Wolf BW, Humphrey PM, Hadley CW, Maharry KS, Garleb KA and Firkins JL. Supplemental fructose attenuates postprandial glycemia in Zucker fatty fa/fa rats. The Journal of nutrition. 2002;132:1219-23.

78. Petersen KF, Laurent D, Yu C, Cline GW and Shulman GI. Stimulating effects of low-dose fructose on insulin-stimulated hepatic glycogen synthesis in humans. Diabetes. 2001;50:1263-8.

79. Hawkins M, Gabriely I, Wozniak R, Vilcu C, Shamoon H and Rossetti L. Fructose improves the ability of hyperglycemia per se to regulate glucose production in type 2 diabetes. Diabetes. 2002;51:606-14.

80. Perreault L, Faerch K, Kerege AA, Bacon SD and Bergman BC. Hepatic glucose sensing is impaired, but can be normalized, in people with impaired fasting glucose. J Clin Endocrinol Metab. 2014;99:E1154-62.

81. Hossain MA, Kitagaki S, Nakano D, Nishiyama A, Funamoto Y, Matsunaga T, Tsukamoto I, Yamaguchi F, Kamitori K, Dong Y, Hirata Y, Murao K, Toyoda Y and Tokuda M. Rare sugar D-psicose improves insulin sensitivity and glucose tolerance in type 2 diabetes Otsuka Long-Evans Tokushima Fatty (OLETF) rats. Biochem Biophys Res Commun. 2011;405:7-12.

82. Shintani T, Yamada T, Hayashi N, Iida T, Nagata Y, Ozaki N and Toyoda Y. Rare Sugar Syrup Containing d-Allulose but Not High-Fructose Corn Syrup Maintains Glucose Tolerance and Insulin Sensitivity Partly via Hepatic Glucokinase Translocation in Wistar Rats. Journal of agricultural and food chemistry. 2017;65:2888-2894.

83. Kruger CL, Whittaker MH, Frankos VH and Trimmer GW. 90-Day oral toxicity study of D-tagatose in rats. Regulatory toxicology and pharmacology : RTP. 1999;29:S1-10.

84. Bar A, Lina BA, de Groot DM, de Bie B and Appel MJ. Effect of D-tagatose on liver weight and glycogen content of rats. Regulatory toxicology and pharmacology : RTP. 1999;29:S11-28.

85. Buemann B, Gesmar H, Astrup A and Quistorff B. Effects of oral D-tagatose, a stereoisomer of D-fructose, on liver metabolism in man as examined by 31P-magnetic resonance spectroscopy. Metabolism. 2000;49:1335-9.

100

86. Boesch C, Ith M, Jung B, Bruegger K, Erban S, Diamantis I, Kreis R and Bar A. Effect of oral D-tagatose on liver volume and hepatic glycogen accumulation in healthy male volunteers. Regulatory toxicology and pharmacology : RTP. 2001;33:257-67.

87. Oku T, Murata-Takenoshita Y, Yamazaki Y, Shimura F and Nakamura S. D- sorbose inhibits disaccharidase activity and demonstrates suppressive action on postprandial blood levels of glucose and insulin in the rat. Nutr Res. 2014;34:961-7.

88. Moore MC, Cherrington AD, Mann SL and Davis SN. Acute fructose administration decreases the glycemic response to an oral glucose tolerance test in normal adults. J Clin Endocrinol Metab. 2000;85:4515-9.

89. Moore MC, Davis SN, Mann SL and Cherrington AD. Acute fructose administration improves oral glucose tolerance in adults with type 2 diabetes. Diabetes Care. 2001;24:1882-7.

90. Heacock PM, Hertzler SR and Wolf BW. Fructose prefeeding reduces the glycemic response to a high-glycemic index, starchy food in humans. The Journal of nutrition. 2002;132:2601-4.

91. Wolf BW, Wolever TM, Lai CS, Bolognesi C, Radmard R, Maharry KS, Garleb KA, Hertzler SR and Firkins JL. Effects of a beverage containing an enzymatically induced-viscosity dietary fiber, with or without fructose, on the postprandial glycemic response to a high glycemic index food in humans. Eur J Clin Nutr. 2003;57:1120-7.

92. Hayashi N, Iida T, Yamada T, Okuma K, Takehara I, Yamamoto T, Yamada K and Tokuda M. Study on the postprandial blood glucose suppression effect of D-psicose in borderline diabetes and the safety of long-term ingestion by normal human subjects. Biosci Biotechnol Biochem. 2010;74:510-9.

93. Kwak JH, Kim MS, Lee JH, Yang YJ, Lee KH, Kim OY and Lee JH. Beneficial effect of tagatose consumption on postprandial hyperglycemia in Koreans: a double- blind crossover designed study. Food Funct. 2013;4:1223-8.

94. Agius L and Peak M. Intracellular binding of glucokinase in hepatocytes and translocation by glucose, fructose and insulin. The Biochemical journal. 1993;296 ( Pt 3):785-96.

95. Van Schaftingen E, Detheux M and Veiga da Cunha M. Short-term control of glucokinase activity: role of a regulatory protein. FASEB J. 1994;8:414-9.

96. Detheux M, Vandercammen A and Van Schaftingen E. Effectors of the regulatory protein acting on liver glucokinase: a kinetic investigation. Eur J Biochem. 1991;200:553-61.

97. Vandercammen A, Detheux M and Van Schaftingen E. Binding of sorbitol 6- phosphate and of fructose 1-phosphate to the regulatory protein of liver glucokinase. The Biochemical journal. 1992;286 ( Pt 1):253-6.

101

98. Bureau of Nutritional Sciences, Food Directorate, Health Products and Food Branch, Health Canada. 2013. Draft Guidance Document on Food Health Claims Related to the Reduction in Post-Prandial Glycaemic Response. Retrieved from: https://chfa.ca/images/uploads/2012/08/Post-Prandial-Glycaemic-Response-Draft- Guidance.pdf.

99. Sievenpiper JL, Jenkins DJ, Josse RG and Vuksan V. Dilution of the 75-g oral glucose tolerance test improves overall tolerability but not reproducibility in subjects with different body compositions. Diabetes Res Clin Pract. 2001;51:87-95.

100. World Health Organization Study Group. Diabetes Mellitus: Report of a WHO Study Group. Geneva, World Health Organization, 1985, p 99

101. Schmidt FH. [Enzymatic determination of glucose and fructose simultaneously]. Klinische Wochenschrift. 1961;39:1244-7.

102. Peterson JI and Young DS. Evaluation of the hexokinase-glucose-6-phosphate dehydrogenase method of determination of glucose in urine. Analytical biochemistry. 1968;23:301-16.

103. Liverey JH HSRHaDR. Effect of time, temperature and freezing on the stability of immunoreactive LH, FSH, TSH, growth hormone, prolactin and insulin in plasma. Clin Biochem 1980;13:151-155.

104. Wolever TM, Jenkins DJ, Jenkins AL and Josse RG. The glycemic index: methodology and clinical implications. The American journal of clinical nutrition. 1991;54:846-54.

105. Kramer CK, Choi H, Zinman B and Retnakaran R. Glycemic variability in patients with early type 2 diabetes: the impact of improvement in beta-cell function. Diabetes Care. 2014;37:1116-23.

106. Bureau of Nutritional Sciences, Food Directorate, Health Products and Food Branch, Health Canada. 2013. Draft Guidance Document on Food Health Claims Related to the Reduction in Post-Prandial Glycaemic Response. Retrieved from: https://chfa.ca/images/uploads/2012/08/Post-Prandial-Glycaemic-Response-Draft- Guidance.pdf

107. Braunstein CB NJ, Glenn A, Viguiliouk E, Noseworthy R, St.Clair J, Khan TA, Nishi S, Blanco Mejia S, Wolever TMS, Josse RG, Kendall CWC, Sievenpiper JL. An acute randomized dose-finding equivalence trial of small, catalytic doses of fructose and allulose on postprandial carbohydrate metabolism in healthy participants: The Fructose and Allulose Catalytic Effects (FACE) Study. 35th International Symposium on Diabetes and Nutrition. 2017.

102

108. Iynedjian PB. Molecular physiology of mammalian glucokinase. Cell Mol Life Sci. 2009;66:27-42.

109. Vandercammen A and Van Schaftingen E. The mechanism by which rat liver glucokinase is inhibited by the regulatory protein. Eur J Biochem. 1990;191:483-9.

110. Vandercammen A and Van Schaftingen E. Competitive inhibition of liver glucokinase by its regulatory protein. Eur J Biochem. 1991;200:545-51.

111. Veiga-da-Cunha M and Van Schaftingen E. Identification of fructose 6- phosphate- and fructose 1-phosphate-binding residues in the regulatory protein of glucokinase. J Biol Chem. 2002;277:8466-73.

112. Agius L. Glucokinase and molecular aspects of liver glycogen metabolism. The Biochemical journal. 2008;414:1-18.

113. Miwa I, Mitsuyama S, Toyoda Y, Nonogaki T, Aoki S and Okuda J. Evidence for the presence of rat liver glucokinase in the nucleus as well as in the cytoplasm. Biochem Int. 1990;22:759-67.

114. Raushel FM and Cleland WW. Bovine liver fructokinase: purification and kinetic properties. Biochemistry. 1977;16:2169-75.

115. Hossain A, Yamaguchi F, Matsuo T, Tsukamoto I, Toyoda Y, Ogawa M, Nagata Y and Tokuda M. Rare sugar D-allulose: Potential role and therapeutic monitoring in maintaining obesity and type 2 diabetes mellitus. Pharmacol Ther. 2015;155:49-59.

116. Toyoda Y, Mori, S., Umemura, N., Futamura, N., Inoue, H., Hata, T., Miwa, I., Murao, K., Nishiyama, A., Tokuda, M. Suppression of blood glucose levels by D-psicose in glucose tolerance test in diabetic rats. Jpn Pharmacol Ther. 2010;38:261−269.

117. Shimabukuro M, Higa N, Chinen I, Yamakawa K and Takasu N. Effects of a single administration of acarbose on postprandial glucose excursion and endothelial dysfunction in type 2 diabetic patients: a randomized crossover study. J Clin Endocrinol Metab. 2006;91:837-42.

118. Sievenpiper JL, Chiavaroli L, de Souza RJ, Mirrahimi A, Cozma AI, Ha V, Wang DD, Yu ME, Carleton AJ, Beyene J, Di Buono M, Jenkins AL, Leiter LA, Wolever TM, Kendall CW and Jenkins DJ. 'Catalytic' doses of fructose may benefit glycaemic control without harming cardiometabolic risk factors: a small meta-analysis of randomised controlled feeding trials. The British journal of nutrition. 2012;108:418-23.

119. Higgins JPT, Green S. Cochrane Handbook for Systematic Reviews of Interventions Version 5.1.0 [updated March 2011]. The Cochrane collaboration Available from www.cochrane-handbook.org. 2011.

103

120. Moher D, Liberati A, Tetzlaff J, Altman DG and Group P. Preferred reporting items for systematic reviews and meta-analyses: the PRISMA statement. PLoS Med. 2009;6:e1000097.

121. Wilczynski NL, Morgan D, Haynes RB and Hedges T. An overview of the design and methods for retrieving high-quality studies for clinical care. BMC Med Inform Decis Mak. 2005;5:20.

122. Guideline: Sugars intake for adults and children. Geneva: World Health Organization; 2015.

123. Dietary Guidelines Advisory Committee. Advisory Report to the Secretary of Health and Human Services and the Secretary of Agriculture. USDA; 2015

124. Huwaldt, J.A., 2015. Plot digitizer. Free software distributed from http://plotdigitizer.sourceforge.net/.

125. Higgins JP, Altman DG, Gotzsche PC, Juni P, Moher D, Oxman AD, Savovic J, Schulz KF, Weeks L, Sterne JA, Cochrane Bias Methods G and Cochrane Statistical Methods G. The Cochrane Collaboration's tool for assessing risk of bias in randomised trials. BMJ. 2011;343:d5928.

126. Greenhouse JB, Iyengar S. Sensitivity analysis and diagnostics. In: Cooper HM, Hedges LV, Valentine JC, eds. The handbook of research synthesis and meta-analysis. 2nd ed. Russell Sage Foundation 2009.

127. Higgins JP, Thompson SG, Deeks JJ and Altman DG. Measuring inconsistency in meta-analyses. BMJ. 2003;327:557-60.

128. Borenstein M, Hedges LV, Higgins JP, H.R. R. Introduction to meta-analysis. Wiley J, 2008.

129. Thompson SG, Higgins JP. How should meta-regression analyses be undertaken and interpreted? Stat Med. 2002;21(11):1559-73.

130. Duval S and Tweedie R. Trim and fill: A simple funnel-plot-based method of testing and adjusting for publication bias in meta-analysis. Biometrics. 2000;56:455-63.

131. Guyatt G, Oxman AD, Akl EA, Kunz R, Vist G, Brozek J, Norris S, Falck-Ytter Y, Glasziou P, DeBeer H, Jaeschke R, Rind D, Meerpohl J, Dahm P and Schunemann HJ. GRADE guidelines: 1. Introduction-GRADE evidence profiles and summary of findings tables. J Clin Epidemiol. 2011;64:383-94.

132. Turner JL, Bierman EL, Brunzell JD and Chait A. Effect of dietary fructose on triglyceride transport and glucoregulatory hormones in hypertriglyceridemic men. The American journal of clinical nutrition. 1979;32:1043-50.

104

133. Rizkalla SW, Baigts F, Fumeron F, Rabillon B, Bayn P, Ktorza A, Spielmann D and Apfelbaum M. Comparative effects of several simple carbohydrates on erythrocyte insulin receptors in obese subjects. Pharmacology, biochemistry, and behavior. 1986;25:681-8.

134. Rizkalla SW, Baigts F, Fumeron F, Grigoresco C, Jean-Joseph-Halfon P, Apfelbaum M, Tchobroutsky G and Slama G. [Influence of fructose on erythrocyte insulin receptors in obese diabetics and non-diabetics]. Journ Annu Diabetol Hotel Dieu. 1986:207-17.

135. Paganus A, Maenpaa J, Akerblom HK, Stenman UH, Knip M and Simell O. Beneficial effects of palatable guar and guar plus fructose diets in diabetic children. Acta Paediatr Scand. 1987;76:76-81.

136. Grigoresco C, Rizkalla SW, Halfon P, Bornet F, Fontvieille AM, Bros M, Dauchy F, Tchobroutsky G and Slama G. Lack of detectable deleterious effects on metabolic control of daily fructose ingestion for 2 mo in NIDDM patients. Diabetes Care. 1988;11:546-50.

137. Blayo A FA, Rizkalla S, Bruzzo F, Slama G. Effets metaboliques de la consommation quotidienne pendant un an de saccharose ou de fructose par des diabetiques. Médecine et Nutrition. 1990;26:909–913.

138. Sunehag AL, Toffolo G, Treuth MS, Butte NF, Cobelli C, Bier DM and Haymond MW. Effects of dietary macronutrient content on glucose metabolism in children. J Clin Endocrinol Metab. 2002;87:5168-78.

139. Vaisman N, Niv E and Izkhakov Y. Catalytic amounts of fructose may improve glucose tolerance in subjects with uncontrolled non-insulin-dependent diabetes. Clinical nutrition (Edinburgh, Scotland). 2006;25:617-21.

140. Aeberli I, Gerber PA, Hochuli M, Kohler S, Haile SR, Gouni-Berthold I, Berthold HK, Spinas GA and Berneis K. Low to moderate sugar-sweetened beverage consumption impairs glucose and lipid metabolism and promotes inflammation in healthy young men: a randomized controlled trial. The American journal of clinical nutrition. 2011;94:479-85.

141. Heden TD, Liu Y, Park YM, Nyhoff LM, Winn NC and Kanaley JA. Moderate amounts of fructose- or glucose-sweetened beverages do not differentially alter metabolic health in male and female adolescents. The American journal of clinical nutrition. 2014;100:796-805.

142. Lowndes J, Sinnett SS and Rippe JM. No Effect of Added Sugar Consumed at Median American Intake Level on Glucose Tolerance or Insulin Resistance. Nutrients. 2015;7:8830-45.

105

143. Heden TD, Liu Y, Park YM, Winn NC and Kanaley JA. Walking Reduces Postprandial Insulin Secretion in Obese Adolescents Consuming a High-Fructose or High-Glucose Diet. J Phys Act Health. 2015;12:1153-61.

144. Ensor M, Banfield AB, Smith RR, Williams J and Lodder RA. Safety and Efficacy of D-Tagatose in Glycemic Control in Subjects with Type 2 Diabetes. J Endocrinol Diabetes Obes. 2015;3.

145. Evans RA, Frese M, Romero J, Cunningham JH and Mills KE. Chronic fructose substitution for glucose or sucrose in food or beverages has little effect on fasting blood glucose, insulin, or triglycerides: a systematic review and meta-analysis. The American journal of clinical nutrition. 2017;106:519-529.

146. Cozma AI, Sievenpiper JL, de Souza RJ, Chiavaroli L, Ha V, Wang DD, Mirrahimi A, Yu ME, Carleton AJ, Di Buono M, Jenkins AL, Leiter LA, Wolever TM, Beyene J, Kendall CW and Jenkins DJ. Effect of fructose on glycemic control in diabetes: a systematic review and meta-analysis of controlled feeding trials. Diabetes Care. 2012;35:1611-20.

147. Livesey G and Taylor R. Fructose consumption and consequences for glycation, plasma triacylglycerol, and body weight: meta-analyses and meta-regression models of intervention studies. The American journal of clinical nutrition. 2008;88:1419-37.

148. Noronha JC, Braunstein, C., Glenn, A., Viguiliouk, E., Noseworthy, R., St. Clair, J., Khan T.A, Au-Yeung, F., Nishi, S., Blanco Mejia, S., Kendall C.W.C., Wolever T.M.S., Leiter, LA., Sievenpiper, J.L. . Effect of Small Doses of Fructose and Allulose on Postprandial Carbohydrate Metabolism in Type 2 Diabetes: An Acute Randomized Double-blind Crossover Trial 35th International Symposium on Diabetes and Nutrition 2017.

149. Donner TW, Magder LS and Zarbalian K. Dietary supplementation with d- tagatose in subjects with type 2 diabetes leads to weight loss and raises high-density lipoprotein cholesterol. Nutr Res. 2010;30:801-6.

150. Ensor M, Williams J, Smith R, Banfield A and Lodder RA. Effects of Three Low- Doses of D-Tagatose on Glycemic Control Over Six Months in Subjects with Mild Type 2 Diabetes Mellitus Under Control with Diet and Exercise. J Endocrinol Diabetes Obes. 2014;2:1057.

151. Agius L. Control of glucokinase translocation in rat hepatocytes by sorbitol and the cytosolic redox state. The Biochemical journal. 1994;298 ( Pt 1):237-43.

152. Agius L. The physiological role of glucokinase binding and translocation in hepatocytes. Adv Enzyme Regul. 1998;38:303-31.

153. Niculescu L, Veiga-da-Cunha M and Van Schaftingen E. Investigation on the mechanism by which fructose, hexitols and other compounds regulate the translocation of glucokinase in rat hepatocytes. The Biochemical journal. 1997;321 ( Pt 1):239-46.

106

154. Agius L, Peak M, Newgard CB, Gomez-Foix AM and Guinovart JJ. Evidence for a role of glucose-induced translocation of glucokinase in the control of hepatic glycogen synthesis. J Biol Chem. 1996;271:30479-86.

155. U.S. Department of Health and Human Services. Guidance for Industry Diabetes Mellitus: Developing Drugs and Therapeutic Biologics for Treatment and Prevention. [Draft Guidance]. Food and Drug Administration Center for Drug Evaluation and Research (CDER) 2008:1-30.

156. Inzucchi SE. Oral antihyperglycemic therapy for type 2 diabetes: scientific review. JAMA. 2002;287:360-72.

157. Human Nutrition Information Service. (1987) Sugar Content of Selected Foods: Individual and Total Sugars. Home Economics Research Report no 48. USDA W, DC.

158. Jenkins DJ, Srichaikul K, Kendall CW, Sievenpiper JL, Abdulnour S, Mirrahimi A, Meneses C, Nishi S, He X, Lee S, So YT, Esfahani A, Mitchell S, Parker TL, Vidgen E, Josse RG and Leiter LA. The relation of low glycaemic index fruit consumption to glycaemic control and risk factors for coronary heart disease in type 2 diabetes. Diabetologia. 2011;54:271-9.

159. Sievenpiper JL, Carleton AJ, Chatha S, Jiang HY, de Souza RJ, Beyene J, Kendall CW and Jenkins DJ. Heterogeneous effects of fructose on blood lipids in individuals with type 2 diabetes: systematic review and meta-analysis of experimental trials in humans. Diabetes Care. 2009;32:1930-7.

160. Vos MB, Kimmons JE, Gillespie C, Welsh J and Blanck HM. Dietary Fructose Consumption Among US Children and Adults: The Third National Health and Nutrition Examination Survey. The Medscape Journal of Medicine. 2008;10:160-160.

161. Saunders JP, Donner TW, Sadler JH, Levin GV and Makris NG. Effects of acute and repeated oral doses of D-tagatose on plasma uric acid in normal and diabetic humans. Regulatory toxicology and pharmacology : RTP. 1999;29:S57-65.

162. Buemann B, Toubro S and Astrup A. Human gastrointestinal tolerance to D- tagatose. Regulatory toxicology and pharmacology : RTP. 1999;29:S71-7.

163. Rech ME. Observations on the decay of glycated hemoglobin HbA1c in diabetic patients. Exp Clin Endocrinol Diabetes. 1996;104:102-5.