Written report Degree Project, 30 credits Medical study program, 2017 Are there any correlations between respiratory quotient and glucose homeostasis in obese children? By: Tutors: Anthony Bachtiar Anders Forslund, Roger Olsson
TABLE OF CONTENTS TABLE OF CONTENTS... 2 ABSTRACT... 4 Purpose... 4 Method... 4 Results... 4 Conclusion... 4 POPULÄRVETENSKAPLIG SAMMANFATTNING... 5 ABBREVIATIONS... 6 INTRODUCTION... 7 Paediatric Obesity... 7 Paediatric Type 2 Diabetes... 8 Obesity and Insulin... 9 Glucose Homeostasis and Oral Glucose Tolerance Test (OGTT)... 10 Indirect Respiratory Calorimetry and Respiratory Quotient... 11 Obesity and Metformin... 13 Respiratory Quotient and Insulin... 14 AIM OF STUDY... 14 METHODS... 15 Study Design... 15 Study Population... 15 Data Collection... 17 Anthropometric Measures... 17 Air Displacement Plethysmography (ADP) with Bod Pod... 18 Bioelectrical Impedance Analysis... 18 Caliper Skinfold Thickness Measurement... 19 Indirect Respiratory Calorimetry... 19 Oral Glucose Tolerance Test (OGTT)... 20 Statistics... 21 Ethical Approval... 21 2
RESULTS... 22 Demographic Baseline Data... 22 Dynamic Variables During Oral Glucose Tolerance Test (OGTT)... 22 Correlation Analysis... 23 RQ, BMR, Fasting Insulin Level and Insulin Resistance... 24 RQ-peak and RQ-mean, Anthropometry, Body Composition and Energy Expenditure.. 26 Fasting Insulin, HOMA-IR and Anthropometric Measurements... 27 RQ-peak, Glucose-peak and Insulin-peak During OGTT... 28 Impaired Glucose Tolerance and Difference in RQ and EE... 28 DISCUSSION... 30 RESULTS... 30 Principal Findings... 30 Comparison to Other Studies... 30 METHOD... 32 Selection of Study Setup and Method... 32 Data Analysis... 34 Interpretation of Blood Samples... 34 Conclusions... 35 ACKNOWLEDGEMENTS... 35 REFERENCES... 36 APPENDIX... 39 3
ABSTRACT Purpose The general purpose of this study is to examine if there are any correlations between indirect respiratory calorimetry, respiratory quotient (RQ) and glucose homeostasis in obese children and adolescents. Method In Uppsala, Sweden, data from indirect respiratory calorimetry and oral glucose tolerance tests (OGTT) on overweight children has been recorded and analysed regarding RQ, insulin resistance, glucose tolerance, energy expenditure and anthropometric measurements. The study population consisted of 203 children in the ages between 5 and 18. Results No significance in the correlation between RQ and glucose or insulin could be found. There was a weak negative relationship between RQ mean during OGTT, weight and waist circumference (P = 0.051, P = 0.07), but no significant correlation between RQ during OGTT and other anthropometric variables. No statistical significant difference could be found regarding RQ or energy expenditure during OGTT when comparing normal to impaired glucose tolerance. Conclusion The results from this study could not show any correlations between the respiratory quotient and glucose homeostasis in obese children. Indirect respiratory calorimetry may not be a reliable tool in this manner, but more studies are needed to get more knowledge about RQ and its role in obesity and diabetes. 4
POPULÄRVETENSKAPLIG SAMMANFATTNING Diabetes och övervikt är två av världens största hälsoproblem idag. Förändringen i den moderna människans livsstil under det senaste århundradet är den största orsaken till ökningen av fetma och diabetes, inte bara hos vuxna utan även hos barn. Vid överviktsenheten för barn och ungdomar och institutionen för klinisk nutrition och metabolism i Uppsala har man mätt och analyserat energiomsättning och så kallade sockerbelastningar på barn med övervikt eller fetma. Detta görs för att följa barnens energioch viktutveckling och se hur deras övervikt påverkar olika nivåer av socker, insulin och ämnesomsättning. Syftet med den här studien är att utvärdera om det finns något samband mellan kolhydrat- och fettförbränning och sockerbalans hos överviktiga barn och ungdomar. Totalt inkluderades 203 barn med olika grader av övervikt i åldrarna mellan 5 och 18 år. Studien är byggd på statistiska analyser mellan fett- och kolhydratförbränning, energiomsättning, insulin och sockernivåer, fetma och olika kroppsmått. I resultaten fann man inget signifikant samband mellan fett- och kolhydratförbränning och insulin eller sockernivåer. Inte heller något samband mellan dessa faktorer under en sockerbelastning, då barnet får dricka en sockerlösning på fastande mage. Men man kunde se ett litet samband mellan fett- och kolhydratförbränning under sockerbelastning och kroppsvikt samt midjemått. Man såg även en lite högre energiomsättning hos pojkar med normal sockertolerans under sockerbelastning, jämfört med pojkar med nedsatt sockertolerans. Få studier har gjorts inom detta forskningsområde och även om denna studie pekar på svaga samband, så är mätandet av energiomsättning och sockerbalans fortfarande intressant för att undersöka och identifiera riskfaktorer för viktuppgång och begynnande diabetes. Fler studier behövs i framtiden för att få ökad förståelse och slutsatser inom detta ämne. 5
ABBREVIATIONS Table 1: Description of various and essential terms and abbreviations ABBREVIATIONS DESCRIPTION ADP Beta-JUDO BIA BMI BMR BP EE FFM FM (Air-Displacement Plethysmography) A technology used in BodPod (technical instrument) to determine a patient s body composition. (Beta-cell Function in Juvenile Diabetes and Obesity) A science project involving obese or overweight children and adolescents within the EU. (Bioelectrical impedance analysis) A technology to determine a patient s body composition. (Body Mass Index) A measurement on the relationship between body mass (kg) and body height (m). Calculated with the formula weight / height 2. (Basal Metabolic Rate) The energy we consume in complete rest, minimum 12 hours after ingestion of food, i.e. the energy that is required to maintain our body s homeostasis and function. (BodPod) A technical instrument used to determine a patient s body composition by measuring body mass, body volume and body density. (Energy Expenditure) The amount of calories a subject consumes from either external work and/or internal heat. (Fat Free Mass) Also known as lean body mass, refers to all body components except fat, i.e. water, proteins, minerals etc. (Fat Mass) Refers to the amount of fat in the body. HbA1c HOMA-IR (Haemoglobin A1c) Refers to glycated haemoglobin (A1c), which identifies average plasma glucose concentration. (The Homeostasis Model Assessment Insulin Resistance) A measurement of insulin resistance. Calculated by a formula, (G0 x I0)/22.5 = HOMA-IR, where G0 is the fasting glucose level and I0 is the fasting insulin level. IGT (Impaired glucose tolerance) Reduced ability to handle high glucose concentration. Definition: A 2-h plasma glucose level between 7.8-11.0 mmol/l after an OGTT. NIDDM Non-insulin Dependent Diabetes Mellitus OGTT PAEE (Oral Glucose Tolerance Test) A method to examine glucose tolerance and to establish the diagnosis of diabetes. (Physical Activation Energy Expenditure) The amount of calories consumed by physical activation. REDCap RMR (Research Electronic Data Capture) Safe, free and internet-based software that can design databases for science projects. Storing data and information on children and adolescents measured and examined in Uppsala and Salzburg. (Resting Metabolic Rate) The energy we consume at rest. Often used interchangeably with BMR but doesn t have the same requirement for rest and fast before examination. RQ (Respiratory Quotient) Examined in indirect respiratory calorimetry and calculated by VCO 2 / VO 2 = RQ. SAD (Sagittal Abdominal Diameter) An anthropometric measurement. Described in method. S-BMI SF T2DM TBW VCO 2 VO 2 (Smart Body Mass Index) A type of BMI that takes age and sex into account. (Subcutaneous Fat thickness) A method to determine a patient s body composition. Type 2 Diabetes Mellitus (Total Body Water) Total amount of water in the body. Volume of carbon dioxide Volume of oxygen 6
INTRODUCTION Paediatric Obesity Diabetes and obesity are two of our world s biggest health problems today. The change during the last century in the modern man s life style and behaviour is the main reason why there has been such an increase of diabetes and obesity not only among adults, but also among children and adolescents. Approximately 16.9 % of all US children is today obese, and up to 31.8 % is either obese or overweight (1). Those numbers have been quite consistent the last years in the US, but there is a significant difference in prevalence between boys and girls. The trend in studies from recent year has shown a significant increase in obesity prevalence among males but no significant change among females (1). However, estimates of childhood obesity tend to be higher in the US than in the rest of the world. In another study they compared the prevalence of obesity between the US, Canada and Mexico (2). A comparable prevalence of obesity among adolescents aged 12 to 19, was 11.7 % in Canada year 2004, and in Mexico 11.5 % year 2006. Compared to 18.4 % in the US year 2009-2010 in the same age group (17.8 % in year 2005-2006) (2). Childhood obesity as a serious problem is not only a truth in the developed countries, but in the developing world as well. In 2010, 43 million children were estimated to be overweight and obese, 35 million of these in developing countries (3). In Sweden, a study including fourth grade children from 1984/1985, 2000/2001 and 2004/2005 showed, by contrast to the American studies, that the prevalence of overweight and obesity in girls decreased from 2000/01 to 2004/05. In boys, there were no significant differences in prevalence between the different cohorts. Compare these with the data from 1984/85 were the prevalence of obesity plus overweight was significant lower among girls and boys (data presented in table 2) (4). 7
Table 2: Data from Swedish study on prevalence of overweight and obesity among children. (4) Academic Year 1984/1985 2000/2001 2004/2005 Girls n = 2088 (%) n = 2302 (%) n = 2134 (%) Overweight plus obesity 179 (8,6)* 450 (19,6)* 340 (15,9) Obese 17 (0,8)* 69 (3,0) 53 (2,5) Boys n = 2038 (%) n = 2381 (%) n = 2059 (%) Overweight plus obesity 147 (7,2)* 406 (17,1) 363 (17,6) Obese 14 (0,7)* 68 (2,9) 57 (2,8) *P < 0,01 Paediatric Type 2 Diabetes Extra astonishing is the increasing incidence of Type 2 diabetes (T2DM) among children and adolescents. In a consensus statement article from year 2000 by the American Diabetes Association, they found that before the 1990s, less than 4 % of all new-onset paediatric diabetes was T2DM. But in year 2000, different accounts showed between 8 % and up to 45 % of all new-onset cases of diabetes to be T2DM in American adolescents (5). It is not only in the US they found an increasing incidence of T2DM among adolescents. In another review article from 2005, compiled numbers from different published articles are discussed regarding the global spread of T2DM in children and adolescents (6). In Singapore, T2DM is accounting for around 10 % of all new cases of childhood diabetes. In Europe, Germany, in a large cohort study of children and adolescents with obesity, 1.5 % had T2DM, and 5.8 % had impaired glucose regulation. Another example, in a survey study from Auckland, New Zeeland, they found that T2DM accounted for 12.5 % of incident cases of diabetes in adolescents in the years 1997-1999. That number had increased to 35.7 % in years 2000-2001 (Add to that the development in prevalence of T2DM, from 1.8 % in 1996, to 11 % in 2002 of all childhood diabetes) (7). The mean body mass index (BMI) of type 2 diabetes subjects was 34.6 kg/m 2, and the mean age was 15. Even though substantial resources have been invested to change people s lifestyle and their environment, we must not forget that obesity is also a physiological condition and how energy metabolism is regulated widely across individuals. To identify and start interventions early to those who are metabolically prone to weight gain is a key to change the course of the obesity epidemic. 8
Obesity and Insulin The level of glucose in our blood is carefully regulated and holds a stable level in healthy individuals despite big differences in food intake. Insulin s main task in the body is to reduce the level of glucose and together with glucagon, cortisol and adrenalin that increases the level of glucose, they maintain the glucose homeostasis. Insulin is a hormone produced by the beta cells in Langerhans islands in the pancreas. It has an anabolic effect and works through enabling glucose uptake in tissues sensitive to insulin, such as adipose, muscle and hepatic tissue. Insulin has a central role in the body s metabolism due to its biological effects that includes glucose uptake, glycogenolysis, gluconeogenesis, glycogenesis, protein synthesis, lipogenesis and lipolysis. There are many studies pointing to the relationship between obesity, Type 2 diabetes and insulin. In a review article by Goran et al (8) they assembled information from 131 different studies regarding obesity and different contributing factors in the increased risk of T2DM and cardiovascular disease in children and adolescents. The clearest factor contributing to increased risk was increased body fat. One of the studies using euglycemic and hyperglycemic clamp techniques, showed that obese adolescent girls had impaired glucose disposal and failed to suppress lipid oxidation and increase glucose oxidation in response to insulin infusion compared to lean control subjects (9). According to Goran et al in another study regarding the influence of total vs. visceral fat on insulin action, the results suggest that total fat mass is the major contributor to variance in insulin sensitivity in children (10). But on the other hand, it has been shown that the relationship between central fat and cardiovascular disease risk is due to visceral fat (10,11). One topic of growing interest is the ectopic fat storage syndrome. Most interesting in this area is that lipid accumulation in skeletal muscles may be important in the risk of diabetes because it is the primary site of insulin action. In a study done on rats where they examined the development of insulin resistance and the relationship to dietary fat and muscle triglyceride, it was found that triglyceride accumulation in muscle is correlated with impaired insulin action and insulin resistance (12). 9
It is well known today after multiple randomized studies in adults (13,14), that regular physical activity reduces insulin resistance, improves glucose tolerance and reduces the risk of T2DM and cardiovascular disease. This is explained through selective reduction in visceral fat, improved peripheral insulin sensitivity, or improved cardiovascular factors. But among children, the physical activity seems to affect insulin levels through its impact on body fat. Compared to adults where the physical activity seems to help per se, regardless of change in body composition (8). Better and earlier screening and identification of obesity-related complications are of great importance in the battle against paediatric obesity. Not least because the frequent asymptomatic nature of these conditions, which might go unnoticed for years and in worst case scenario lead to irreversible damage. Simultaneously, there is a current limitation of effective treatment and prevention programs that are specifically designed for the paediatric population and high-risk groups. Glucose Homeostasis and Oral Glucose Tolerance Test (OGTT) Glucose tolerance is determined by the effect of insulin as well as insulin-independent effects. Intake of glucose in our body gives a short hyperinsulinemia and hyperglycaemia, resulting in increased hepatic glycogen storage and decreased hepatic glucose production (15). The current definition of impaired glucose tolerance (IGT) is determined by the 2-h plasma glucose level after an OGTT, where the cut-point is 7.8 mmol/l (between 7.8 and 11 mmol/l is considered to be IGT) (16). In a study, it was showed that IGT is highly prevalent in obese children and adolescents, approximately 21-25 per cent. Impaired glucose tolerance was associated with insulin resistance even though beta-cell function still was relatively preserved (15). Obesity increases the risk of type 2 Diabetes through induction of impaired glucose tolerance and insulin resistance (17). There are various ways to examine insulin sensitivity and glucose tolerance in individuals. Oral glucose tolerance test (OGTT), euglycemic hyperinsulinemic clamp and frequent sampled intravenous glucose tolerance test (FSIGT) are most common clinically. But there are also formulas and mathematical indexes that are used to calculate 10
insulin sensitivity like homeostasis model assessment (HOMA) and quantitative insulin sensitivity check index (QUICKI) (18). The OGTT is a frequently used method to examine glucose tolerance and to establish the diagnosis of diabetes, mainly because it s simple to perform and at a low cost. But glucose measurements in the standard OGTT do not give adequate information about the dynamics of glucose and insulin action, insulin sensitivity or β-cell function, in contrast to euglycemic hyperinsulinemic clamp for example, which is more accurate in that manner (18). However, it is possible to predict an individual s insulin sensitivity and β-cell function from BMI and values in plasma obtained during an OGTT with acceptable accuracy (19). The limitations of the OGTT are partly because of too sparse taken blood samples, much since the response and release of insulin to glucose levels has a pulsatile character. Like in the standard OGTT, samples are taken before the intake of glucose, at 1 hour and 2 hours after ingestion (18). For this study, blood samples will be taken more frequently during the OGTT to get a more accurate image of the dynamics of insulin and glucose and also insulin sensitivity. Indirect Respiratory Calorimetry and Respiratory Quotient The assessment of energy expenditure in a subject has been carried out by direct and indirect calorimetry for more than a century, and development of new technologies has improved the ability to study energy turnover in a human being. The method of direct calorimetry measures the heat dissipated by the body, whereas indirect calorimetry measures the heat released by the oxidative processes (20). Today, indirect calorimetry is one of the most commonly used methods to measure energy expenditure (EE) in a subject. The regulation of energy metabolism varies widely among individuals and by measuring oxygen consumption and carbon dioxide production in a subject s breathing we get minute-by-minute information of various energy expenditure data (21). Such data includes resting and basal metabolic rate (RMR,BMR), thermic effect of food (TEF), energy cost of activity, which can be divided in every-day activity and physical activity energy expenditure (PAEE), and more importantly for this study, the respiratory quotient (RQ) and information on energy substrate utilization. The ratio between the volumes of carbon dioxide produced (VCO 2 ), and the oxygen consumed (VO 2 ), gives us the respiratory quotient (RQ). 11
EE and BMR can be calculated by using the complete or abbreviated Weir equation, which builds on that the consumption of oxygen is proportional to the energy expenditure (22). Resting metabolic rate (RMR) is often used interchangeably with basal metabolic rate (BMR), but BMR has a stricter requirement with at least 12 hours after ingestion of food and complete rest before the measurement (23). By knowing the RQ, we know which energy substrate the subject is utilizing because different fuel source give different RQ. We can then compare the metabolism of different subjects and the relative proportion between fat and carbohydrate oxidation by simply comparing RQ (Figure 1) (23,24). Figure 1: Interpreting respiratory quotient When RQ is higher than 1, it means that under these circumstances carbohydrates are stored as fat in the body. Oxygen is produced when fat is synthesized, which means that the demand and consumption of oxygen decreases, which results in a RQ higher than 1. Therefore, an RQvalue over 1 indicates accumulation of fat in the body (23). Care must be taken when interpreting the RQ. There are several metabolic causes for an RQ less than 0.71, including untreated diabetes mellitus, oxidation of ethanol, lipolysis or underfeeding. There are also a few possible metabolic causes for RQ greater than 1.0 such as lipogenesis, overfeeding and high blood lactate level due to exercise for example (23,25). We have also methodological reasons for RQ greater than 1, for example errors in the calibration or in the instrument measuring the respiratory air. 12
But even though RQ varies in some degree depending on what environmental factors the subject has, RQ tend to vary more between individuals than within a single individual measured on separate days. Approximately 3-10 times more between individuals than observed within one single subject (24). It has long been suggested that reduced fat oxidation capacity favours positive fat balance and therefore predisposes to weight gain and obesity, especially in individuals that already faces a high fat diet (26). Obesity and Metformin It is known today that metformin has an effect on body fat, body weight and body composition in type 2 diabetic patients and non-insulin dependent diabetes mellitus (NIDDM). Apart from improving glycaemic control and decrease fasting hyperglycaemia, mainly by decreasing hepatic glucose output by the inhibition of gluconeogenesis, metformin has been proven to have other positive effects on obesity and type 2 Diabetes (27 29). Effects such as reduction of fasting insulin levels, improved sensitivity for insulin in muscles, increasing skeletal muscle uptake of glucose, and a delay of glucose absorption in the intestines was found in several double blind, placebo-controlled clinical studies (27,29,30). Regarding the beneficial effects of metformin on body weight, it has been seen as both an anorexigenic effect along with fat reduction due to an up-regulation of fat oxidation (27,29,30). The latter related to enzyme activity in the liver, UCP-1 in brown adipose tissue, and UCP-3 in skeletal muscle, both resulting in a significant reduction of visceral fat mass in both human subjects and in rats. Interestingly, in this study by Tokubuchi et al (29), the subjects receiving metformin had after their treatment a significantly decreased fasting respiratory quotient (RQ) and an increased post-prandial RQ, although their energy expenditure (EE) did not change. Suggesting that metformin accelerates fat oxidation during fast and has beneficial effects concerning visceral fat mass reduction and lipid metabolism (29,31). 13
Respiratory Quotient and Insulin In one study by Nakaya et al where they examined the RQ of patients with non-insulindependent diabetes, they analysed how the RQ change depending on whether the subjects received treatment with insulin or sulfonylurea or no treatment at all (32). The authors interestingly found that those subjects without treatment had higher glucose levels and a significantly lower RQ and a lesser tendency to weight gain, which suggested that the carbohydrate metabolism is impaired in uncontrolled diabetes. Treated patients showed a higher RQ, and 54 % of those with a RQ >1.0 (though a much lesser group) gain 3 kg, compared to only 16 % of those patients with a RQ less than 1.0. It was concluded that the improved blood glucose control under intense insulin therapy comes with a cost of increased risk of weight gain due to a higher proportion of fat since the metabolism is focused on carbohydrates (32). AIM OF STUDY Childhood obesity is today one of our biggest health problems globally and affects not only the individual, but also the society as a whole. The general purpose of this study is to examine if there could be any correlations between the respiratory quotient and glucose homeostasis in obese children and adolescents. By statistically analysing different factors including RQ, insulin resistance, glucose tolerance, obesity and anthropometric measurements in children, it could possibly give us a more reliable and accurate way that early on identifies those individuals that are metabolically prone to weight gain. The main hypothesis of this study is: - High fasting insulin and decreased insulin sensitivity correlates with high RQ - Those individuals with late glucose/insulin peak in OGTT have also a late RQ peak - Those individuals that have a higher RQ during the OGTT have higher anthropometric measurements, body fat percentage and lower energy expenditure. - Those individuals with impaired glucose tolerance have also a higher RQ 14
METHODS Study Design This study was a quantitative cross sectional cohort study. The material was based on Swedish children included in the Beta-JUDO (Beta-cell function in Juvenile Diabetes and Obesity) program between 2012 and 2016. Via collaboration between the institute of overweight and obesity for children and adolescents and the institute of clinical nutrition and metabolism at Uppsala university hospital (UAS) in Sweden, data have been recorded and stored from 457 overweight children. These children have different degrees of complications from their obesity and different levels of disturbance of their glucose homeostasis, including T2DM. Study Population Requirement for including children in this study was that they should be either overweight or obese and had undergone basic anthropometric measures, indirect respiratory calorimetry and an oral glucose tolerance test (OGTT). The children were born between 1995 and 2010 and the mean age was 12.7 ± 3.4 in total, with a minimum of 4.9 and a maximum of 18.1 years. The mean s-bmi for both sexes was 37.8 ± 4.2 and ranging between 27.7 and 52.7. For boys, the mean s-bmi was 38.5 ± 4.3, and for girls the mean s-bmi was 37.0 ± 3.9. The final study population, with complete indirect respiratory calorimetry values ended on 203 children, 117 (57.6 %) boys and 86 (42.4 %) girls (selection of study population explained in figure 2). 15
Children who had undergone indirect respiratory calorimetry (n=331) Children with data transferred and compiled in the RedCap database (n=227) Declined participation, or didn t exist in our code list or in the RedCap database. (n=104) Excluded due to incomplete indirect respiratory calorimetry (n=15) Children with complete indirect respiratory calorimetry (n=203) Excluded due to incorrectly calibrated indirect respiratory calorimetry (n=9) Excluded due to blood samples difficult to interpret (n=5) Excluded due to missing blood samples (n=11) Children with complete indirect respiratory calorimetry and correct blood samples (n=187) Figure 2: Descriptive flowchart regarding the selection of study population 9 patients were excluded due to incorrectly calibrated RQ, i.e. a fasting value over 1.0. This value is considered to be metabolically implausible or unlikely because fasting values of RQ tend to be low, around 0.83. No second calibration was done to see if the first value was correct or incorrect calibrated, therefore these cases were excluded. 5 subjects were excluded due to blood values difficult to interpret. Those cases had a double peak during the OGTT, i.e. same value on two different measurements. Since the oxidation adapts after the substrate that is available in the cell we excluded those cases because we could not undoubtedly know which of these times glucose had entered the cell and which of their peaks we should correlate to or put into account. But for the other statistical analyses using fasting values and other variables than blood samples during the OGTT, we chose not to exclude those cases, since their dynamic change and blood sample values didn t affect those analyses. 16
Data Collection The collection and extraction of data was done in several stages. The raw data had already been collected before this study during patient visits at the paediatric laboratory at UAS. These raw data were collected during the morning when a patient visited after a 12-hour fast and included anthropometric measures, weight, height, calculated BMI, calculated S-BMI, body composition using Bod Pod, BIA and SF measurements, indirect respiratory calorimetry during an OGTT, measured and calculated RQ, BMR, EE and blood samples of insulin and glucose before and during OGTT. All this data is stored in file folders and in Microsoft Excel documents (version 14.7). The later stage in collection of data was to compile these raw data in an Internet based database called RedCap. This is a web application from the Beta-JUDO study and is used to manage and store data from different patient visits, examinations, measurements, blood samples and questionnaires. Last stage was to extract and create a statistical report using this data from RedCap. Anthropometric Measures On the same morning the patient was called to do the OGTT and indirect respiratory calorimetry, they were first examined and measured using standardized methods. Table 3: Anthropometric measurements and method Measurement Method Height (cm) Weight (kg) Standing patient. Measured twice by a stadiometer. Calculated mean height. Determined on standard, calibrated scale. Patient wearing only underwear or light clothes. BMI (body mass index) Calculated from values above. Weight kg / (Height m) 2 Waist circumference (cm) Hip circumference (cm) Waist/Hip ratio Standing patient. Using a measuring tape. Measured in the middle between the superior edge of iliac crest and the lowest rib. Standing patient. Measuring tape. Measured over the greatest circumference of the hip. Calculated from values above. Waist cm / Hip cm. 17
Abdominal girth (cm) Sagittal abdominal diameter (cm) Thigh circumference (cm) Standing patient. Measuring tape. Measured over the widest point around the abdomen, usually in level of the umbilicus. Lying patient. Measured with ruler from the bed to the level of the umbilicus while the patient lies with bended knees and the lumbar region touching the bed. Measuring tape. Measured over the greatest circumference of the thigh. Air Displacement Plethysmography (ADP) with Bod Pod Using a Bod Pod is a well-recognised way to determine a patient s body composition. This gives a clear image of the amount of fat mass (FM) and fat free mass (FFM), a so-called 2- component model where FFM consist of water, protein and minerals. Bod Pod uses a technology called air displacement plethysmography (ADP) and is built on the same principles as hydrostatic weighing (Archimedes principle), i.e. an object displaces its own volume of water, but a Bod Pod uses air instead of water. It uses the physical relationship between pressure and volume and the application of Boyle s law and Hagen- Poiseuille equation (Poiseuille s law) to calculate body volume (33,34). The patient sat inside a closed space and the Bod Pod measured body mass (kg) using an electronic scale and the volume of the patient s body using ADP. Body density could then be calculated by dividing body mass with body volume (kg/m 3 =density). Once the overall density of the body was determined, the relative proportions of body fat and lean body mass was calculated based on the knowledge that FM and FFM has different density (fat mass= 0.9 g/cm 3 and fat free mass = 1.1 g/cm 3 at 37 Celsius) (35). The software to the institute s Bod Pod was Body composition tracking system (COSMED). Bioelectrical Impedance Analysis The use of bioelectrical impedance analysis (BIA) is another well-known way to determine a patient s FFM and total body water (TBW). The patient stood on a BIA monitor while holding electrode handles in his or her hands. Then a low, safe electrical signal was sent from metal electrodes through the patient s feet, legs and abdomen and later reached the electrode handles. The electrical signal passes quickly through water that is present in hydrated muscle 18
tissue, but meets resistance when it hits fat tissue. Based on the relationship between impedance and geometry, a subject s body composition can be calculated by inserting the measured resistance through different tissues (known as impedance) into scientifically validated BIA equations (36). All patients didn t use the exact same BIA instrument, some used the BIA Tanita MC980 monitor, some used the BIA Inbody monitor and some didn t want to participate in a BIA analysis, but all the results from the patient who fulfilled the examination were carefully noted. Caliper Skinfold Thickness Measurement A subject s body composition can also be calculated using a skinfold caliper by measuring subcutaneous fat thickness to the nearest millimetre over four sites: at the biceps, triceps, subscapular and supra-iliac areas of a standing patient. A logarithm of skinfold measurement was used in order to achieve a linear relationship with body density, and then formulate linear regression equations to estimate body density and body fat mass from single skinfold measurements, or from the sums of two or more skinfolds. There are separate equations and logarithms for different ages and sex, but still the estimated accuracy of these equations differ depending on skin site and obesity level (37,38). Indirect Respiratory Calorimetry The patient arrived in the morning after a 12-hour fast. After being examined and measured by the methods explained above, she or he wore a mask that covered the whole face, which enabled us to measure the respiratory air. Minute by minute, a computer then registered the volumes of oxygen consumed (L/min) and carbon dioxide produced (L/min) and the RQ and energy expenditure (Kcal/min and Kcal/day) could be calculated from those values. Beside from those numbers we also got the volume of oxygen per kilo body mass (ml/kg/min) and body temperature and pressure, saturated (BTPS L/min). The data was then collocated in a Microsoft Excel document (version 14.7) where the average of each unit was calculated and noted from each period of 30 minutes during a 2 hour OGTT. 19
Oral Glucose Tolerance Test (OGTT) The indirect respiratory calorimetry was done before and continuous with an OGTT. This was done to examine the glucose and insulin homeostasis and the progression of EE and RQ in each patient during the OGTT. The results reflect the glucose absorption, intake and disposal rate, as well as the natural endocrine response on glucose intake. Normally, the glucose concentration reaches its maximum ( 11 mmol/l) after approximately 30 min, and is back at fasting level (<6,1mmol/L) after approximately 2 hours. At -5 min before the start of the OGTT, different fasting biochemical markers were taken from the patient, including fasting glucose, insulin and HbA1c. The patient was then given 1.75g/kg glucose, maximum 75 g, dissolved in 3 dl water to drink (maximum time 5 min). Then at 5 min, 10 min, 30 min, 60 min, 90 min and 120 min, blood samples were extracted for analysis of glucose and insulin. After each blood sample test, the respiratory mask was attached on the patient and the respiratory flow was registered for 15 min. After that, 15 min was spent on measuring the body composition or to check and go through various data before the next blood sample test. To determine a patient s insulin resistance, HOMA-IR was calculated by a formula, (G0 x I0)/22.5 = HOMA-IR, where G0 is the fasting glucose level and I0 is the fasting insulin level. 20
Statistics For statistical analysis, the program IBM SPSS Statistics (version 23) was used (39). Correlation analysis was done according to Spearman s rank analysis because it takes outliers into better account and since the data was widely distributed. Correlation coefficient and significance was calculated from each requested variable. Correlation analysis was done between: - Fasting RQ, fasting insulin and HOMA-IR (a measurement of insulin resistance) - Fasting RQ and BMR (adjusted after weight) - RQ peak and RQ mean, anthropometric measurements, body fat percentage and EE during OGTT. - RQ peak time, glucose peak time and insulin peak time. Independent sample T-test to evaluate significance in mean differences: - Differences in fasting RQ, RQ peak, RQ mean and BMR between subjects with normal vs. impaired glucose tolerance Ethical Approval Ethical approval for the study and collection of data is accepted and available with registration number ULSCO Dnr 2010/036, ULSCO 2 Dnr 2012/318. 21
RESULTS Demographic Baseline Data In table 3 below the descriptive data of the study population is presented and shows the age, anthropometric measurements and body composition. Notable in this data is that the boys had slightly significant higher BMI and anthropometric measurements, but that the girls had significant higher percentage of body fat in the Caliper measurement (SF). Some of the subjects missed data from body composition examinations, mainly due to lack of their consent to participate. Valid subject numbers can be observed in the number (N) column. Table 4: Descriptive statistics from patient s anthropometric measurements and body composition. Total Boys Girls N Min Max Mean Std. Dev. N Min Max Mean Std. Dev. N Min Max Mean Std. Dev. Age 203 4,88 18,05 12,64 3,35 117 4,88 18,05 13,05* 3,20 86 5,44 17,88 12,08* 3,49 BMI-sds 208 1,6 4,7 3,221 0,521 120 1,6 4,7 3,34* 0,521 88 1,6 4,2 3,06* 0,4763 s-bmi 208 27,7 52,7 37,84 4,201 120 27,7 52,7 38,47* 4,337 88 27,7 48,3 36,97* 3,8646 BMI 208 19,99 54,34 32,99 6,55 120 22,95 54,34 33,85* 6,57 88 19,99 50,87 31,81* 6,36 Waist-hip ratio 208 0,76 1,52 0,98 0,07 120 0,76 1,14 0,99* 0,05 88 0,77 1,52 0,97* 0,09 SAD (cm) 204 14 36,2 24,52 3,942 118 14 36,2 25,08* 4,11 86 17,2 33,5 23,76* 3,5843 Total Body Fat (BIA) 155 21,7 56,0 42,03 6,43 86 21,7 56,0 41,91 6,40 69 25,3 55,3 42,19 6,52 Total Body Fat (BP) 186 20,3 60,8 44,28 7,10 109 25,5 58,0 45,10 7,01 77 20,3 60,8 43,12 7,11 Total Body Fat (Caliper) 184 20,0 47,6 33,70 5,32 97 20,0 40,3 30,25* 3,52 87 25,4 47,6 37,53* 4,26 Valid N (listwise) 138 78 60 * Significant difference at the between 0,05 level (2-tailed) boys and girls at the 0,05 level (2-tailed). Dynamic Variables During Oral Glucose Tolerance Test (OGTT) During the OGTT we followed and documented different parameters and variables. The dynamic changes of those variables are presented in figures 3 and 4. The charts are presented with 95 % confidence interval (CI) and with separate lines for boys and girls to get a more detailed view of the data. In the charts we can see that there is no big difference between the two genders, only small differences in the RQ, EE and glucose levels, where girls tend to have a slighter higher RQ than boys, and boys tend to have a slighter higher glucose level and EE than girls. Interesting observation is that the peak for RQ during OGTT seems to be at approximately 90 min, whereas the peak for EE, glucose and insulin seems to be at 30 min. 22
Figure 3: Dynamic change of RQ and EE during the OGTT for boys and girls. 200 Level of Insulin during OGTT Gender Boys Girls9,0 Level of Glucose during OGTT Gender Boys Girls Average Level of Insulin (microiu) 150 100 50 Average Level of Glucose (mmol/l) 8,0 7,0 6,0 5,0 0-5 5 10 30 60 Time (min) 90 120 4,0-5 5 10 30 60 Time (min) 90 120 Error Bars: 95% CI Error Bars: 95% CI Figure 4: Level of insulin and glucose during the OGTT for boys and girls Correlation Analysis Correlation analysis has been done for the entire cohort population (n=203), but for those analysis including blood samples the cohort population was reduced (n=187) due to missing samples or blood samples difficult to interpret. 23
RQ, BMR, Fasting Insulin Level and Insulin Resistance We could not find any significant correlations between fasting RQ and BMR, fasting insulin or HOMA-IR. On the other hand, we could find both a statistical significant negative correlation between BMR (adjusted after body mass) and fasting insulin and HOMA-IR (P < 0.001). Table 4 shows a relatively strong negative correlation with a coefficient of -0.377 between BMR and fasting insulin and -0.398 between BMR and HOMA-IR. Table 5: Spearman s correlation analysis between RQ, BMR, fasting insulin and insulin resistance Spearman's rho Fasting RQ Correlation Coefficient BMR (kcal/min/kg) Fasting Insulin HOMA-IR Correlations Fasting RQ BMR Fasting Insulin HOMA- IR 1,000,108,099,090 Sig. (2-tailed).,135,172,214 N 192 192 192 191 Correlation Coefficient,108 1,000 -,377 ** -,398 ** Sig. (2-tailed),135.,000,000 N 192 192 192 191 Correlation Coefficient,099 -,377 ** 1,000,991 ** Sig. (2-tailed),172,000.,000 N 192 192 192 191 Correlation Coefficient **. Correlation is significant at the 0.01 level (2-tailed).,090 -,398 **,991 ** 1,000 Sig. (2-tailed),214,000,000. N 191 191 191 191 The results from table 5 are illustrated graphically in figure 5 and 6, showing the data from fasting RQ, BMR and fasting insulin and how they correlate. 24
Figure 5: Scatter plot showing the data and correlation between fasting RQ and fasting insulin Figure 6: Scatter plot showing the data and correlation between BMR and fasting insulin 25
RQ-peak and RQ-mean, Anthropometry, Body Composition and Energy Expenditure The data regarding different variables during the OGTT are presented in table 6 to 9. Overall, the correlation between RQ peak (highest value of RQ during OGTT), RQ mean (average value of RQ during OGTT) and the other variables regarding anthropometry, body composition and energy expenditure was low. But there was a tendency between RQ-mean and weight (P = 0.051), even though it was not statistical significant. Presented in table 6, we found a negative correlation with a coefficient of -0.135, showing that weight and RQ mean have a small covariance where weight increases, RQ mean decreases. Table 6: Spearman s correlation analysis between RQ-peak, RQ-mean and anthropometric measurements Spearman's rho RQ peak RQ mean Correlation Coefficient RQ peak RQ mean Correlations Weight Waist circ Hip circ SAD Waist/hip ratio Waist/height ratio SAD/height ratio 1,000,913 ** -,035 -,036,016 -,028 -,108,030,066 Sig. (2-tailed).,000,612,609,818,687,119,672,351 N 208 208 208 208 208 204 208 208 204 Correlation Coefficient **. Correlation is significant at the 0.01 level (2-tailed).,913 ** 1,000 -,135 -,126 -,095 -,107 -,102 -,029,049 Sig. (2-tailed),000.,051,070,173,127,143,677,489 N 208 208 208 208 208 204 208 208 204 Table 7: Spearman s correlation analysis between RQ-peak, RQ-mean and body composition Spearman's rho RQ peak Correlation Coefficient RQ mean Correlations RQ peak RQ mean Total body fat (BIA) Total body fat (BP) Total body fat (SF) 1,000,913 **,035,035,020 Sig. (2-tailed).,000,665,640,790 N 208 208 155 186 184 Correlation Coefficient **. Correlation is significant at the 0.01 level (2-tailed).,913 ** 1,000,005 -,034,027 Sig. (2-tailed),000.,950,646,714 N 208 208 155 186 184 26
Table 8: Spearman s correlation analysis between RQ-peak, RQ-mean and BMI Spearman's rho RQ peak Correlation Coefficient RQ mean Correlations RQ peak RQ mean BMI-sds s-bmi BMI 1,000,913 **,020,043,002 Sig. (2-tailed).,000,773,541,978 N 208 208 208 208 208 Correlation Coefficient **. Correlation is significant at the 0.01 level (2-tailed).,913 ** 1,000,015,037 -,089 Sig. (2-tailed),000.,830,597,199 N 208 208 208 208 208 Table 9: Analysis between RQ-peak, RQ-mean, fasting BMR and energy expenditure during OGTT Spearman's rho RQ peak RQ mean Correlation Coefficient RQ peak Correlations RQ mean BMR EE (30 min) EE (60 min) EE (90 min) EE (120 min) 1,000,913 ** -,034 -,035 -,085 -,048 -,090 Sig. (2-tailed).,000,631,612,221,493,196 N 208 208 208 208 208 208 208 Correlation Coefficient **. Correlation is significant at the 0.01 level (2-tailed).,913 ** 1,000,049,055,021,054 -,006 Sig. (2-tailed),000.,482,432,766,439,936 N 208 208 208 208 208 208 208 Fasting Insulin, HOMA-IR and Anthropometric Measurements Additional correlation analyses were performed regarding fasting insulin and its relationship to anthropometric measurements. Interestingly, we found a statistical significant correlation between fasting insulin, HOMA-IR and several anthropometric measurements. Strongest correlation was found between fasting insulin and SAD measurement (r = 0.52, P < 0.001) and HOMA-IR and weight (r = 0.52, P < 0.001). These results can be observed in more detail in table 10. Table 10: Correlations between fasting insulin, HOMA-IR and anthropometric measurements Spearman's rank Fasting Insulin Fasting Insulin * Correlation is significant at the 0,05 level (2-tailed). HOMA- IR BMI-sds s-bmi BMI Weight Waist Hip SAD Waist/hipratio Total Body Fat (BIA) Total Body Fat (BP) Total Body Fat (SF) Corr. Coefficient 1 0,991** 0,059 0,075 0,490** 0,494** 0,496** 0,460** 0,517** 0,103 0,164* 0,301** 0,252** Sig. (2-tailed). 0,000 0,407 0,290 0,000 0,000 0,000 0,000 0,000 0,148 0,042 0,000 0,001 N 199 198 199 199 199 199 199 199 195 199 154 180 180 HOMA-IR Corr. Coefficient 0,991** 1 0,069 0,082 0,512** 0,520** 0,516** 0,484** 0,529** 0,096 0,164* 0,319** 0,240** Sig. (2-tailed) 0,000. 0,334 0,253 0,000 0,000 0,000 0,000 0,000 0,179 0,043 0,000 0,001 N 198 198 198 198 198 198 198 198 194 198 153 179 179 ** Correlation is significant at the 0,01 level (2-tailed). 27
RQ-peak, Glucose-peak and Insulin-peak During OGTT The correlation analysis investigating if individuals with a late RQ-peak also had a late insulin and/or glucose peak during the OGTT is presented in table 11. The results show no significant correlation between RQ peak and insulin or glucose peak, but a significant positive correlation between insulin peak time and glucose peak time, which is expected. Table 11: Spearman s correlation analysis between RQ-peak time, glucose-peak time and insulin-peak time Correlations Spearman's rho RQ-peak time Correlation Coefficient Glucose-peak time Insulin-peak time RQ-peak time Glucose-peak time Insulin-peak time 1,000 -,032,087 Sig. (2-tailed).,655,231 N 192 192 192 Correlation Coefficient -,032 1,000,528 ** Sig. (2-tailed),655.,000 N 192 192 192 Correlation Coefficient **. Correlation is significant at the 0.01 level (2-tailed).,087,528 ** 1,000 Sig. (2-tailed),231,000. N 192 192 192 Impaired Glucose Tolerance and Difference in RQ and EE Below is the data divided between subjects with normal glucose tolerance and impaired glucose tolerance (IGT). Different lines for boys and girls are presented in figures 7 and 8 to get at more detailed view of the data and differences. No significant difference in means could be found between the two groups concerning fasting RQ, RQ mean, RQ peak and BMR. But you can see a change in the dynamics for RQ and EE during the OGTT. Most apparent was that the boys with IGT had a lower EE through the whole OGTT compared to the group with normal glucose tolerance. Even though this was an interesting difference, the change was not statistical significant (P = 0.057). There was especially a big difference in the last EE-value for 120 min. The girls on the other hand, moved in the opposite direction where the group with IGT had higher EE during the OGTT than the group with normal glucose tolerance, though no statistical significance (table 12). 28
Figure 7: Difference in RQ during OGTT between subjects with normal and impaired glucose tolerance Figure 8: Difference in EE during OGTT between subjects with normal and impaired glucose tolerance Table 12: Difference in numbers between subjects with normal and impaired glucose tolerance Total Boys Girls Glucose Tolerance N Mean* Std. Dev Sig. (2-tailed) N Mean* Std. Dev Sig. (2-tailed) N Mean* Std. Dev Sig. (2-tailed) Fasting RQ Normal 101 0,88571 0,0447 0,496 55 0,87742 0,0422 0,227 46 0,89561 0,0461 0,895 Impaired 94 0,89024 0,0480 57 0,88766 0,0468 37 0,89421 0,0503 RQ mean Normal 101 0,94437 0,0522 0,944 55 0,93846 0,0500 0,959 46 0,95144 0,0544 0,912 Impaired 94 0,94443 0,0549 57 0,93892 0,0456 37 0,95291 0,0666 RQ peak Normal 101 1,00452 0,0815 0,918 55 0,99984 0,0758 0,931 46 1,01012 0,0885 0,749 Impaired 94 1,00568 0,0755 57 0,99870 0,0642 37 1,01644 0,0901 BMR Normal 101 0,01585 0,0033 0,349 55 0,01615 0,0034 0,064 46 0,01548 0,0032 0,469 (kcal/min/kg) Impaired 94 0,01542 0,0030 57 0,01505 0,0028 37 0,01600 0,0032 EE mean Normal 101 0,01622 0,0034 0,176 55 0,01647 0,0036 0,057 46 0,01592 0,0032 0,851 Impaired 94 0,01559 0,0030 57 0,01529 0,0029 37 0,01605 0,0032 *No significant difference between means 29
DISCUSSION RESULTS Principal Findings The results can be summarized to: there was no significance in the relationship between fasting RQ, fasting insulin and insulin resistance (table 5). Neither was there any statistical significant correlation between RQ peak and mean value during OGTT and the variables concerning anthropometric measurements, body composition and energy expenditure (table 6-9). But there was a weak negative relationship between the RQ mean value during OGTT and weight (P = 0.051), also between RQ mean and waist circumference (P = 0.07). Regarding the correlation between time for the RQ peak during the OGTT and time for glucose and insulin peak, no significant correlation could be found (table 11). Except for the significant correlation between glucose peak time and insulin peak time, but this result is expected. No significant differences could be found between the subjects who had normal glucose tolerance vs. impaired glucose tolerance regarding fasting RQ, RQ peak and mean value (table 12). However, there was an interesting difference between groups in the energy expenditure for boys during the OGTT, where the EE mean was higher for those boys with normal glucose tolerance than those with IGT, though this difference was not statistically significant. No significant difference could be found in the girl population or the study population as a whole. Comparison to Other Studies It is known that RQ is affected by a subjects diet (21,23,40), but in one study by Zurlo et al, Pima Indians spent one or several days in a respiratory chamber with the same diet and measured 24h-RQ during this period. Even though they were given the same strict metabolic condition, same diet and exercise, some individuals gained more weight than others, and this had a relationship with high RQ (24). 30