ASSOCIATION OF LEG LENGTH WITH METABOLIC ABNORMALITIES UNDERLYING TYPE 2 DIABETES MELLITUS

Transcription

1 ASSOCIATION OF LEG LENGTH WITH METABOLIC ABNORMALITIES UNDERLYING TYPE 2 DIABETES MELLITUS by Luke W. Johnston A thesis submitted in conformity with the requirements for the degree of Master s of Science Graduate Department of Nutritional Sciences University of Toronto Copyright 2013 by Luke W. Johnston

2 Abstract Association of Leg Length with Metabolic Abnormalities Underlying Type 2 Diabetes Mellitus Luke W. Johnston Master s of Science Graduate Department of Nutritional Sciences University of Toronto 2013 The objective of this thesis was to determine the association of leg length (LL), a marker of early childhood conditions, with metabolic abnormalities underlying type 2 diabetes. Utilizing data from a population at-risk for diabetes, the associations of LL with i) insulin resistance (IR) and β-cell dysfunction and ii) a continuous metabolic syndrome risk score (MetScore) were analyzed. Results showed that shorter LL was associated with IR and β- cell dysfunction, and that the combination of short legs and large waist (a marker of adult obesogenic conditions) was associated with the greatest IR. Height, a marker of overall childhood conditions, was found to be inversely associated with the MetScore. Therefore, both adverse childhood conditions and early-late life mismatched conditions may increase the risk for diabetes through differing pathways. Improving childhood conditions (i.e. nutritionally or economically) may be an important strategy to prevent diabetes. ii

3 Acknowledgements A huge portion of my thanks goes to Anthony Hanley for being my supervisor and mentor, setting me loose on such an interesting field of research, and providing wise tidbits of information and pertinent sport analogies whenever necessary. His style of teaching is outstanding, giving me just enough information so that I know where to look to learn more but always able to go more in-depth when needed. My labmate, Sheena Kayaniyil, has also been incredibly instrumental in helping me with my research, such as with SAS code or wading through the PROMISE dataset, and who is always ready and willing to help out whenever she can. And of course, I want to thank my partner and wife, Veronica Diaz, for helping me get through the stresses of writing a thesis by supporting me and keeping me sane. My thanks likewise goes to Jill Hamilton and Thomas Wolever for their advice and suggestions. I would like to acknowledge my sources of funding (Ontario Graduate Scholarship) and conference travel award (Banting and Best Diabetes Centre). The research nurses also deserve special recognition for their expertise on running the PROMISE clinic visits, including Jan Neuman, Paula Van Nostrand, Stella Kink, and Annette Barnie of the Leadership Sinai Centre for Diabetes, Mount Sinai Hospital, and Sheila Porter and Mauricio Marin of the Centre for Studies in Family Medicine, University of Western Ontario. Finally, I would like to thank the PROMISE cohort participants for being involved in this study, without whom I would not be doing this fascinating area of research. iii

4 Contents 1 Introduction and Rationale 1 2 Literature Review Developmental Origins of Health and Disease Early postnatal conditions and growth in stature Type 2 Diabetes Mellitus Pathophysiology of insulin resistance and β-cell dysfunction Early life conditions and the development of diabetes Metabolic Syndrome Early growth and the clustering of metabolic risk factors Summary of Literature Review Objectives and Hypotheses 26 4 Research Design and Methods: The PROMISE Study 28 5 Short leg length, a marker of early childhood deprivation, is associated with metabolic disorders underlying type 2 diabetes mellitus: The PROMISE cohort study Introduction Research Design and Methods Results Conclusions Relationship of Continuous Metabolic Syndrome Scores with Insulin Resistance: A Comparison of Different Statistical Approaches Introduction Methods Results iv

5 6.4 Discussion Association of Stature and its Components with a Metabolic Syndrome Risk Score Introduction Methods Results Discussion Overall Discussion Overall Summary of Results Strengths Limitations Future Directions Conclusions Bibliography 78 v

6 List of Tables 2.1 Studies that have investigated the association of leg length with diabetes and insulin resistance (i.e. HOMA-IR) Studies that have investigated the association of leg length with components of Metabolic Syndrome (e.g. HDL or TAG) Metabolic, demographic, and anthropometric characteristics from 3-yr visit ( ) of non-diabetic participants from the PROMISE cohort Means of stature components according to categories of demographic variables from non-diabetic participants in the 3-yr visit ( ) of the PROMISE cohort Linear regression models showing associations of height, sitting height, leg length, and leg-to-height ratio with insulin sensitivity and β-cell function measures using the 3-yr visit data from non-diabetic PROMISE participants, adjusting for covariates Different continuous MetScore variables used within the published literature Area under the receiver operating characteristic curve and the odds ratio of dichotomized HOMA-IR (at the fourth quarter; Q4) and ISI (at the first quartile; Q1) for each MetScore Spearman correlation coefficients and R 2 obtained from linear regression between the different MetScores and insulin sensitivity Area under the receiver operating characteristic curve and the odds ratio of dichotomized HOMA-IR (at the fourth quartile; Q4) and ISI (at the first quartile; Q1) for the top MetScore using different blood pressure indices Spearman correlation coefficients and the R 2 obtained from linear regression for the top MetScore using different blood pressure indices and insulin sensitivity vi

7 7.1 Metabolic, anthropometric, and demographic characteristics of participants without diabetes from the 3-yr visit ( ) of the PROMISE cohort Spearman partial correlation coefficients of stature components with continuous variables, adjusted for ethnicity, sex, and age, of participants without diabetes from the 3-yr visit ( ) of the PROMISE cohort Means of stature components according to categories of demographic variables from non-diabetic participants in the 3-yr visit ( ) of the PROMISE cohort Principal component loadings from the PCA with varimax rotation of both the traditional MetScore and the non-traditional MetScore Linear and logistic regression models of the associations of the stature components with the MetS and the MetScores using data from non-diabetic participants from the 3-yr ( ) visit of the PROMISE cohort, adjusted for covariates vii

8 List of Figures 4.1 CONSORT diagram of CANOE and PROMISE Interaction of leg length on the association of waist circumference with measures of insulin sensitivity viii

9 Acronyms ALT Alanine transaminase ARIC Atherosclerosis Risk in Communities AROC Area under the receiver operating characteristic curve AUC Area under the curve BMI Body Mass Index CANOE CAnadian Normoglycemia Outcomes Evaluation CRP C-reactive protein CVD Cardiovascular disease DBP Diastolic blood pressure DM Diabetes mellitus DOHAD Developmental origins of health and disease DXA Dual-energy x-ray absorptiometry EGFR Estimated glomerular filtration rate HDL High density lipoprotein HOMAIR Homeostasis Model Assessment of Insulin Resistance IFG Impaired fasting glucose IGIIR Insulinogenic Index over HOMA-IR IGT Impaired glucose tolerance ix

10 ISI Matsuda Insulin Sensitivity Index ISSI-2 Insulin Secretion Sensitivity Index 2 LDL Low density lipoprotein LHR Leg-to-height ratio MAP Mean arterial pressure MCR Microalbumin-to-creatinine ratio MetS Metabolic Syndrome MetScore Continuous Metabolic Syndrome Score NGT Normal glucose tolerance NHANES National Health And Nutrition Examination Survey OGTT Oral glucose tolerance test OR Odds Ratio PAR Predictive adaptive responses PC Principal component PCA Principal Component Analysis PROMISE PROspective Metabolism and ISlet Cell Evaluation SBP Systolic blood pressure SES Socio-economic status SH Sitting height TAG Triacylglyceride WHO World Health Organization x

11 Chapter 1 Introduction and Rationale Increasing scientific attention has been focusing on the developmental origins of health and disease (DOHaD) phenomenon, which posits that conditions during growth in gestation or early postnatal life play an etiological role in the pathogenesis of chronic diseases such as the metabolic syndrome (MetS) and type 2 diabetes [1, 2]. When developmental conditions are adverse, including poor nutrition, psychosocial stress, or low socioeconomic situations, there is a shift in growth processes to maximize immediate survival and emphasize development of higher priority organs, e.g. the brain. A consequence of this shift in growth is a propensity towards developing chronic diseases in later life, though the exact underlying mechanisms are unclear. While early life overnutrition leading to largefor-gestational age newborn babies or to childhood obesity similarly increases the risk for developing chronic diseases later in life [3], the focus of this thesis is on the implications of poor nutritional or environmental conditions on later disease risk. Weight and stature trajectories during childhood are a reflection of environmental conditions during growth (see Section 2.1.1). Interruptions in nutrient supply, arising from poor access to food or increased stress, may subtly alter the growth of muscles, bones, or fat tissue. Bone in particular is sensitive to chronic disruptions in nutrient supply [4]. Notably, each component of stature is influenced by different phases of growth. Leg length has been shown to be a sensitive marker of the conditions during the infant and early childhood growth phase (0 4 years of age) [5, 6], during which time other organs are also developing (e.g. the brain and pancreas) [7, 8]. Therefore, adult leg length can be used as a marker of early childhood conditions when investigating associations with chronic diseases. Several lines of research have implicated early childhood conditions in the etiology of type 2 diabetes (see Section 2.2.2). Adverse childhood economic or environmental conditions and rapid catch-up growth have each been found to be associated with a greater 1

12 CHAPTER 1. INTRODUCTION AND RATIONALE 2 risk for diabetes [9 12]. Research has also shown shorter leg length to be associated with diabetes and insulin resistance [13, 14], though there are some inconsistencies in the findings between studies [15]. In addition, there are a number of gaps and limitations in the literature, including the use of HOMA-IR, which is a fasting, hepatic measure of insulin resistance, and a lack of data regarding associations with β-cell function. The majority of studies of early life conditions and adult MetS (which is a constellation of interrelated risk factors for chronic diseases such as diabetes and cardiovascular disease) have used either childhood economic conditions as a marker of early childhood adversity or measures of weight during childhood to determine catch-up growth, and most found both were associated with either MetS or its individual components [16, 17]. A few studies found inconsistent associations of stature and its components with individual components of the metabolic syndrome [18 20], and fewer included a formal definition of the MetS as an outcome. No study has examined the association of leg length with a formal MetS definition, though a few have studied height and MetS (see Section 2.3.1). The primary gap in the literature is that no study has examined leg length with the MetS. The objectives for this thesis are to determine the association of leg length, which is a marker for early childhood conditions, with metabolic abnormalities underlying type 2 diabetes including insulin resistance, β-cell dysfunction (Chapter 5), and the MetS (Chapter 7).

13 Chapter 2 Literature Review 2.1 Developmental Origins of Health and Disease Recent observations have demonstrated a link between environmental and nutritional conditions during fetal and early childhood ontogenesis and adult disease risk. These observations have led to the emergence of the developmental origins of health and disease (DO- HaD) paradigm, which manifests as ontogenetic changes to various metabolic processes that modulate susceptibility to chronic diseases. The following section briefly introduces the DOHaD hypotheses and provides an overview of the processes that may underlie the connection between early life and adult disease. Subsequent sections will review the current literature on the impact of early life conditions on adult chronic disease risk. DOHaD. Several hypotheses have been proposed to explain the phenomenon linking development, growth, and subsequent health status. Hales and Barker [1] put forward the thrifty phenotype hypothesis, the first hypothesis to gain widespread attention regarding the DOHaD phenomenon. The hypothesis proposed that environmental and nutritional factors during gestation were an important cause of type 2 diabetes mellitus. According to the hypothesis, undernourishment during gestation provides a nutritional and environmental forecast of the postnatal environment, prompting the development of a thrifty phenotype that is suited to poor postnatal conditions. The authors suggest underdevelopment of the pancreatic β-cells in both mass and function and insulin resistance in peripheral tissues to reserve glucose for critical organs necessary for fetal survival such as the brain and heart cause this thriftiness [1, 21]. The hypothesis was based on observations of greater risk of developing diabetes with decreasing birthweight among British males [1] and among individuals born during the Dutch Hunger Winter [22]. While this thrifty phenotype acquired by the fetus in response to an adverse in-utero 3

14 CHAPTER 2. LITERATURE REVIEW 4 environment may be suited to poor future environmental and nutritional conditions, it may be detrimental when the forecast does not match the actual postnatal environment. For instance, individuals born during the Dutch Hunger Winter (indicating poor gestational conditions) were more likely to develop diabetes in adulthood when nutritional and environmental conditions were improved [22], highlighting the impact of early-late life mismatch. In contrast, individuals born during the Leningrad siege in World War II did not have higher rates of diabetes, but the adulthood conditions were similarly deprived as the gestational conditions [23], suggestive of an early-late life match. However, there are several limitations to the thrifty phenotype hypothesis, including the lesser emphasis on the postnatal period and inadequate consideration on the role of the maternal environment, physiology, and behaviour during the offspring s ontogeny [2, 24 26]. In addition, the hypothesis makes little attempt to explain transgenerational influences. The more recently proposed predictive adaptive responses (PAR) hypothesis proposed by Gluckman et al. [2] states that mammals adjust their life history strategy using developmental plasticity in response to maternal and postnatal environmental and nutritional cues that predict later life conditions in an attempt to maximize survival and reproductive fitness (through to the age of sexual maturity) [2, 26]. If postnatal conditions improve compared to the intra-uterine conditions, the infant will compensate for the restricted growth as a fetus by growing rapidly. If both fetal and early postnatal predictions of later life conditions are incorrect, health may be compromised in later adulthood. For instance, populations undergoing rapid improvements in economic and nutritional conditions such as India and China are observing an increased prevalence of cardiometabolic diseases, as do populations which migrate from developing to developed countries [26]. The PAR expands on the thrifty phenotype hypothesis by including examples from other animal species to support the hypothesis and by including postnatal influences. However, there are also a number of criticisms of this theory, such as the fact that poor fetal and early childhood environmental conditions delay sexual maturity in contrast to what is observed, inadequate consideration of maternal influences on growth and development, and the insufficient explanation for transgenerational influences [24, 25]. Wells [24] extends aspects of the DOHaD phenomenon by focusing on the mother as the primary influence of fetal and infant growth, since the mother must balance her needs with those of all of her offspring. Rather than the fetus making predictions about the postnatal conditions, the mother provides these cues depending on her economic, social, nutritional, and familial demands by regulating placental or mammary transfer of substrates to her offspring necessary for growth. Her offspring reacts to her regulatory cues for growth, which consequentially develops the offspring s metabolic capability to tolerate variable postnatal environmental

15 CHAPTER 2. LITERATURE REVIEW 5 conditions. Several criticisms have been raised regarding previous hypotheses. For instance, the PAR proponents criticise the thrifty phenotype hypothesis for its deterministic view that metabolism is programmed or set after birth/early postnatal period [2, 26]. Criticisms of the PAR hypothesis include its under-emphasis of the mother s role in her offspring s growth or the transgenerational role of the mother s ontogenetic environmental and nutritional conditions on her offspring s growth [24]. The predictive aspect of the ontogenetic adaptations to environmental conditions is particularly criticised as these adaptations could also be trade-offs for immediate survival given adverse conditions that have consequences on health outcomes in later life, without needing to invoke an ability of the offspring to predict later life environments [25]. Likely mechanisms which may mediate the DOHaD phenomenon include nutritional and environmental modifications to the epigenome. Epigenetics is the study of changes to gene expression that occur independent of changes in the DNA sequence and includes processes such as DNA methylation and histone modification [26 28]. During development, stressful or nutritionally deprived conditions can influence the methylation status of DNA strands, consequently affecting gene expression, and is supported by experimental evidence from mice [27, 29]. This thesis will not be focusing on the mechanisms underlying DOHaD, but rather on the influence of growth on later disease risk. Fetal growth. While both the thrifty phenotype and PAR hypotheses focus primarily on adaptations the fetus makes in response to the intrauterine nutritional environment, a mother s strategies to maximize her own and her offspring s fitness are not frequently considered [24]. From an evolutionary perspective, the mother and fetus are in constant conflict over nutrients and a balance must be maintained by the mother [24, 27]. It is in the best interest of the fetus to obtain as many nutrients from the mother as possible to maximize growth and development, and to support reproductive fitness later in life. The mother s best interests are to balance the amount of nutrients given to the fetus, to the mother s other offspring (i.e. through breastfeeding), and to still retain enough nutrients for herself to continue obtaining the resources necessary for survival of herself and her family. For instance, a potential mechanism through which the mother may modulate the growth of the fetus may occur through the placental interface whereby maternal hormones and substrates, or lack thereof, may influence fetal growth [24]. There are physical constraints that neither the fetus nor the mother may change such as maternal pelvic size and skeletal structure, which together dictate the size of the womb [2, 24, 26]. Other mechanisms which may arguably be outside of maternal control include her

16 CHAPTER 2. LITERATURE REVIEW 6 socioeconomic status (SES), regional economic circumstances, cultural behaviours (e.g. smoking or drinking during pregnancy, which were more prevalent prior to the era of strong public health campaigns urging against these practises), or her own metabolic or cardiovascular function (which may have been transgenerationally modulated by her own ontogenesis) [24, 26, 27, 30]. The mechanism behind the transgenerational modulation of the mother may occur through changes to the epigenome, which may then be passed on to her offspring [28]. Mechanisms within maternal control may include hormonal or nutrient synthesis (though unconsciously controlled), the quantity or quality of food consumed, and the level of energy expenditure. The fetus is able to modulate its growth either internally (e.g. hormonally) or externally via maternal nutrient supply. The main driver of growth during the fetal period is the availability of substrates (i.e. nutrients) [31], which may not necessarily be dependent on maternal dietary intake as detailed above. The fetus may respond to low substrate supply by reducing growth through several possible mechanisms: by redistributing blood and hence nutrients to higher priority organs (e.g. the brain); by decreasing peripheral insulin sensitivity; by reducing anabolic demand by decreasing insulin secretion and insulin-like growth factor-1; or potentially by self-inducing premature birth [4, 21, 26, 27]. Metabolic development during gestation may establish the capacity to tolerate variable postnatal metabolic environments, possibly by prenatally increasing cell numbers or mass of various organs. For instance, under-development and poor growth may lead to a reduced capacity to tolerate high metabolic demands as seen in obesogenic Western environments [27, 31]. As alluded to previously, changes in the metabolic function of an individual during the fetal stage may also transgenerationally influence the ontogenesis of an individual s offspring, potentially leading to a metabolic ghetto which could last many generations [24]. Fetal ontogeny can have significant impacts on the metabolic ability to withstand highly variable metabolic demands. However, postnatal conditions may modulate the influence of fetal conditions on metabolic development and capacity, suggesting an important role for infant and early childhood conditions. Postnatal growth. In contrast to the strong focus on the intra-uterine environment, there is less emphasis in the DOHaD literature on the role of infancy and early childhood growth on adult disease. While low birthweight increases the risk of chronic disease in obesogenic Western countries [1, 9, 21], the degree and timing of postnatal growth may also play an influential role in the pathogenesis of chronic diseases. Since growth in infancy is dependent on nutrition [31], growth restrictions during the fetal period may be alleviated if postnatal nutrition is improved, often resulting in catch-up growth i.e. moving from a low

17 CHAPTER 2. LITERATURE REVIEW 7 height-for-age to a higher or normal height-for-age) or weight rebound (similar to catchup growth, but with weight being the predominate catch-up response; sometimes refered to as adiposity rebound) [21, 26]. Historically, catch-up growth or weight rebound in babies born small was viewed positively as it was thought to improve the development of a number of organs and improve survival; however, this view is now changing [32]. While catch-up growth in length and weight) may be beneficial, too rapid, too late, or too much growth may place a high metabolic demand on infants born small or underdeveloped, with potential consequences on health in later life [4, 21, 32]. A later catch-up growth may reflect a delay in the transition from infancy to childhood (the main driver of growth for this stage is growth hormone dependent), the delay itself may play a role in adult health [31]. Childhood growth trajectories may also continually be influenced by the mother s capacity to provide adequate nutrition and attention, such as through breastfeeding or by dividing attention between several offspring [24]. Organ development may continue into infancy and childhood, and therefore deprived early childhood conditions may compromise the capacity of organs to adequately respond to the metabolic demands in adulthood, especially in conditions typical of Western countries. In this section, the basic principles of the DOHaD paradigm were outlined. Currently, a number of hypotheses have been proposed regarding the DOHaD phenomenon, though there are limitations and criticisms associated with each of them. Regardless, the evidence suggests that since in utero and early childhood growth are substrate dependent, deficiencies in substrate availability may have ontogenetic consequences that impact disease risk in later life. A significant volume of literature has already been published on the role that fetal conditions have on later life disease risk, although research on the impact of early childhood conditions has not been as well characterized. Therefore, the proceeding sections will review the relationship between early childhood conditions, growth, and disease Early postnatal conditions and growth in stature While weight change is sensitive to acute environmental conditions, particularly nutritional conditions, growth in stature is more responsive to chronic environmental conditions [4]. Growth in stature follows a triple-logistic pattern, occurring in infancy, childhood, and adolescence [4]. Each growth stage can be represented by a logistic curve early rapid growth slowly abating until a plateau is reached and the next stage commences. The first stage starts with a period of rapid growth at birth and continues until it reaches a plateau at 5 years. The next stage starts at 6 years and continues until 8 years, and the final stage begins near or at puberty (10 12 years) and continues until the cessation of growth

18 CHAPTER 2. LITERATURE REVIEW 8 and the attainment of final adult stature [4]. Children born at a lower birthweight often experience a faster rate of growth in the first two years of life while children at a higher birthweight usually experience a slower rate of growth [4]. While there is a hereditary component to final attained height, genetics do not have as profound an impact on growth velocity as the conditions the child may experience during growth, such as the quality of nutrients, psychosocial stress, or low socioeconomic situations [33]. Tanner [34] stated that growth is a mirror of the conditions of society. Throughout human evolution, stature has undergone secular increases as well as decreases. For example, adult stature decreased during the Industrial Revolution, a period of rapid economic and social change which resulted in conditions which were far from ideal [33]. Adverse conditions during the early postnatal period may lead to a redistribution of blood to the brain and higher priority organs [21]; subtle alterations in skeletal length may be a reflection of these environmental insults. Height differences have been observed between developed and developing countries, with improvements in conditions often resulting in increases in height [33]. The short timescales in which height improvements have been observed is one indication that genetics do not play a large role compared to population-wide economic improvements, as gene pools do not significantly change in such brief time frames. Childhood growth patterns. In 1951, Leitch [35] published observations that longer legged children suffered less bronchitis than shorter legged children. This paper was also one of the earliest to postulate that interruptions in growth during early childhood have an effect on the length of individual components of stature and on disease risk (both acute and chronic), and that leg length could be used as an indicator of early childhood conditions. Early postnatal linear growth has since been attributed almost entirely to increases in leg length, as the legs represent 33% of overall length at birth and 50% by puberty (or peak height velocity), after which time growth occurs equally in the legs and trunk [5, 6, 36 38]. Early postnatal growth in leg length has a significant impact on final attained height, demonstrated by the higher correlation between leg length and height (r = 0.89) compared to sitting height and height (r = 0.68) [38]. In a 1958 British birth cohort, prepubertal height was most strongly associated with leg length but not sitting height [6]. Following the logistic pattern, the most rapid period of growth in leg length occurs from birth to 4 years, and slows down until peak height velocity during puberty ( 13 years) [37, 39, 40]. Pubertal timing may also influence relative leg length by initiating the childhood stage where growth occurs equally in the legs and trunk. In female participants of NHANES

19 CHAPTER 2. LITERATURE REVIEW 9 III, an earlier age of menarche (the start of puberty for females) was associated with a shorter stature, almost entirely due to shorter leg length [41]. A Swedish study of males using growth charts from medical records found that a later age of peak height velocity was an independent predictor of greater adult height, leg length, and femur and tibia length (using DXA to measure the bone lengths), while greater BMI at 6 years was a predictor of longer sitting height [42]. A British birth cohort similarly demonstrated that greater weight at 4 years of age was more strongly associated with sitting height than leg length [5]. In addition, greater early childhood BMI in the Swedish males was associated with an earlier pubertal timing, which was associated with shorter stature and shorter leg length; this was also observed in the British birth cohort [5]. In the Niagara Region in Ontario, Canada Liu et al. [43], found that children who were obese or overweight at grade 5 and later examined in grade 8 had shorter legs than normal weight children, even after controlling for the children who remained overweight or obese in grade 8. A similar association of lower leg length with greater fat mass has been seen in adult participants of NHANES III [44]. As described above, growth in the legs relative to total stature occurs predominately in the early postnatal period, beginning at birth, peaking during the transition into childhood (4 5 years), and slowing at puberty after which growth occurs equally in sitting height and leg length. Pubertal timing may also influence relative leg length and may be triggered by adiposity in early childhood (i.e. greater adiposity may advance the age of puberty, resulting in shorter leg length). Economic conditions. Economic circumstances have been shown to be associated with stature and its components. Country-wide improvements in economic conditions in Britain were associated with a secular trend in increasing leg length with little impact on sitting height [37]. In a British birth cohort, markers of lower economic circumstances such as larger family sizes ( 5 children), overcrowded sleeping quarters ( 1.5 children/room), and living in social housing were more strongly associated with shortness in height and leg length than in sitting height [6], each explaining 3% of the variation in growth of the components of stature before the age of 7, and much less afterwards [45]. In another British birth cohort, non-manual social class was associated with shorter leg length more than with sitting height [5]. Among multiple European countries, lower attainment of education was associated with shortness in height [46]. Meanwhile, populations who were displaced because of war were shorter and had shorter legs, while sitting height was less affected [47]; each subsequent displacement was associated with a corresponding decrease in height and leg length.

20 CHAPTER 2. LITERATURE REVIEW 10 Between three European countries, parental education and the number of assets were associated with longer stature components, especially with height [48]. The European country with the lowest parental assets also had the shortest leg length but similar sitting heights. However, this study was retrospective and used imprecise measures of childhood conditions, which may explain the unexpected finding that height was more strongly associated with childhood conditions than leg length. In British females, leg length, sitting height, and height were associated with current and childhood SES [14], though childhood SES was broadly defined as whether the father s occupation was non-manual or manual. During the 1970s to 1990s, a large number of Guatemalan Mayans immigrated to the US. Bogin et al. [49] compared the components of stature of those Mayans living in the US, who had better socioeconomic conditions, to Mayans still currently living in Guatemala, who had worse socioeconomic conditions, and found that the US Mayans were taller, had longer sitting heights, and had higher legto-height ratios (LHR), though their stature was still shorter than the general US population. Children living in rural subsistence agricultural communities in Mexico with marginal nutrition had slightly lower sitting height, but significantly compromised leg growth compared to children in Philadelphia, US [50]. Meanwhile, transition economies are observing increasingly higher rates of children who are both stunted and overweight or obese, suggesting either i) a conflict between poor nutrition during gestation and early childhood that stunts growth and the easy access to fast or processed high-calorie foods available for consumption post-weaning that promote fat storage or ii) the children s nutrition may be obesogenic but not favourable for bone growth that may be creating this mismatch in phenotype [4, 51]. Regional or national economic conditions have an indirect role on growth via accessibility to a good quality diet, the influence on psychosocial stress, and the presence of hygienic (e.g. sewage and waste disposal) and/or healthcare (e.g. hospitals) infrastructure. In particular, economic and environmental conditions during early childhood affect growth of the legs. Family environment and diet. Family and dietary influences, related but not entirely dependent on economic situations, also affect growth. For instance, in a British birth cohort maternal smoking was associated with shortness in height in the offspring, almost entirely due to reduced leg length [6]. In another British birth cohort, being breastfed had a considerably stronger influence on increasing leg length than sitting height and a larger energy intake at the age of 7 years was also more strongly associated with increased leg length than with sitting height [5]. The positive influence of breastfeeding on stature and

21 CHAPTER 2. LITERATURE REVIEW 11 particularly the leg length has been confirmed in a number of other studies [38, 45, 52]. Serious illness and parental separation, such as divorce, were associated with reduced sitting height, suggesting sitting height may be more strongly influenced by long-term, chronic impacts on growth throughout childhood (i.e years) [5]. Transgenerational influence. The growth of a mother during the fetal stage of her life influences the growth of her own offspring during gestation, which subsequently influences the offspring s growth postnatally, suggesting transgenerational action on the growth of an individual [33], likely through epigenetic modification [28]. Growth during gestation, using birthweight as a marker for growth, has a positive influence on a child s overall stature [5]; birthweight explains 8% of the variation in growth [45] and affects both components of stature equally [53]. Subsequent generations may also experience improvements in socioeconomic position, which increases a mother s affluence and allows her to invest more time and resources into her children, impacting their growth and consequently also affecting the growth of her children s offspring [33]. In two British birth cohorts, parental height was associated with the child s final attained height and was more strongly associated with leg length than sitting height [5, 6]; maternal height explained the greatest amount of variation in growth before (12%) and after (20%) 7 years of age [45]. Maternal leg length was more strongly associated with her offspring s birthweight than sitting height [54], suggesting not only the impact of a mother s own fetal conditions, but also her own growth during infancy on her offspring s growth. Single generation improvements in economic conditions, using a greater number of parental possessions as a proxy measure, was associated with longer sitting height and height in Chinese participants [55]. Improvements in economic conditions which occurred over multiple generations, by using literacy in both parents (since historically literacy was very low in China and only wealthier families could afford for their children to learn to read) combined with a greater number of parental possessions as a proxy measure, was associated with longer adult leg length [55]. This may suggest that in countries with recent improvements in economic circumstances, such as China, sitting height may be a better indicator of economic conditions, while leg length may reflect consistent and multigenerational improvements in economic conditions [55]. While nutritional deprivation may not have been as extreme in most parts of Canada years ago as in some developing countries, it is important to note that leg length can be affected by multiple factors which operate on a spectrum, such as exposure to secondhand smoke, frequency, quantity, and/or quality of breastfeeding, quality of dietary intake, micronutrient composition, SES conditions, and stress (e.g. [5, 6]), of which there was

22 CHAPTER 2. LITERATURE REVIEW 12 likely wide variability in Canada at this time. Taken together, growth in stature during early childhood occurs predominately in the legs and is sensitive to early childhood conditions. Pubertal timing initiates the period of equal growth in both components of stature, thus early puberty (which is associated with obesity) results in shorter relative leg length. There also is a transgenerational influence on leg length such that conditions during parental ontogeny may dictate the ontogeny of the offspring. Given that adult leg length is not susceptible to age-related shrinkage in stature (which result from reductions in sitting height caused by compression of the spine via vertebral fractures or intervertebral disc degeneration [56]), adult leg length can be used as a sensitive marker of early childhood conditions in countries with established and developed economies. 2.2 Type 2 Diabetes Mellitus Diabetes mellitus is a chronic disease in which the ability to regulate blood glucose is impaired, resulting in elevated blood glucose concentrations, and commonly, the development of micro and macrovascular complications including kidney, eye, as well as cardiovascular disease [57]. Type 1 diabetes is an autoimmune disease which results in the destruction of the pancreatic insulin-secreting β-cells and is typically detected during childhood. Type 2 diabetes is characterized by insulin resistance and eventual reductions in blood insulin concentrations. This form of diabetes typically manifests in adulthood, although youth-onset type 2 diabetes has been documented in some populations. According to the International Diabetes Federation 1, there were an estimated 366 million people globally who had diabetes in 2011 (the majority of whom had type 2 diabetes) and the number is expected to continue to rise. Traditional risk factors for type 2 diabetes include a family history of diabetes, older age, physical inactivity, non-european ancestry, obesity (including high waist circumference and BMI), hypertension, dyslipidemia, and smoking [58, 59]. Females who had gestational diabetes or gave birth to a baby weighing more than 9 lbs also have a greater risk for developing diabetes [58]. Genome-wide association studies have identified multiple genes associated with an increased risk for diabetes, however, the contributions of individual genes to overall risk is modest [59]. 1 Taken from on June 6, 2013

23 CHAPTER 2. LITERATURE REVIEW Pathophysiology of insulin resistance and β-cell dysfunction Type 2 diabetes develops as a result of a progressive failure of the pancreatic β-cells to secrete adequate insulin to overcome the insulin resistance that characterizes this disorder [60]. Insulin resistance is defined as the diminished response of key tissues such as muscle, liver, and adipose tissue to the normal action of insulin [60]. Peripheral (i.e muscle) insulin sensitivity is an important trait for the postprandial action of insulin given its ability to lower blood glucose by storing it as glycogen for later use [61]. In the fasted state, the brain consumes the majority of glucose. Post meal, peripheral insulin action accounts for two-thirds of the glucose disposal, with hepatic insulin action accounting for the remainder [61]. Although, insulin resistance is a central pathophysiological disorder in the natural history of type 2 diabetes, it is not sufficient to cause diabetes. Rather, the failure of the pancreatic β-cells to meet the challenge of insulin resistance is also required. The endocrine pancreas is a highly complex tissue that maintains blood glucose concentrations in a narrow physiological range which is constantly fluctuating due to dietary energy intake [62]. A functional pancreatic β-cell produces sufficient amounts of proinsulin (the precursor to insulin) in response to glucose, efficiently converts proinsulin to insulin and releases it, manages excess co-secreted peptide hormones (i.e. islet amyloid polypeptide), and maintains adequate islet mass [60, 63]. Insulin sensitivity of target tissues will modulate the amount of insulin necessary to reduce blood glucose. The tight interrelationship between insulin secretion and sensitivity can be represented as a mathematical hyperbola [60, 62]. Metabolic dysfunction occurs when the β-cells are unable to adequately secrete insulin in sufficient amounts to reduce blood glucose. Typically, β-cell failure occurs in the presence of insulin resistance and is critical in the pathogenesis of diabetes [60]. In the early stages of dysfunction, β-cell mass may increase in an attempt to compensate for the increased secretory demands [61, 64]. However, eventually β-cell function and mass decrease and hyperglycemia ensues. Considering insulin resistance and β-cell dysfunction are primary underlying abnormalities in the natural history of diabetes and may be targets of ontogenetic changes in the DOHaD phenomenon, the following section will explore the evidence of early life conditions, the risk for diabetes, and its underlying disorders Early life conditions and the development of diabetes Following the widespread attention that the thrifty phenotype hypothesis [1] received in attempting to explain the observed associations between decreasing birthweight and increasing diabetes risk, considerable scientific interest has focused on the early life environ-

24 CHAPTER 2. LITERATURE REVIEW 14 ment and adult disease risk. Considering that β-cell neogenesis occurs prenatally, with a small amount of neogenesis neonatally, and that β-cell proliferation and increases in size can occur up to 2 years postnatally [7], there is a potential for early childhood conditions to strongly influence the development of diabetes. In order to study associations of early life with disease risk, several methods could be employed to characterize early life conditions. Socioeconomic conditions, growth, and factors influencing growth are all markers of early life conditions, which can be used to study the association of early life on the risk for diabetes. Economic circumstances. Socioeconomic conditions in childhood have been associated with diabetes and glucose intolerance. Data from ethnically diverse US populations have shown that individuals who during their childhood had parents with higher income and education (proxy measures of childhood SES) were less likely to have diabetes [15, 65]. Similarly, British females who had parents with higher educational attainment and income during childhood were more likely to have lower insulin resistance, as measured by HOMA-IR [9]. In the Framingham Offspring Study, female participants who had a lower life-course socioeconomic position (which included childhood SES, using father s education, and the participant s current SES, using their current education and occupation) had an increased incidence of diabetes, but adjusting for BMI attenuated this finding [10]. Females with SES trajectories (SES from childhood to adulthood) that either remained low or progressed from high to low had a significantly greater incidence of diabetes, but after adjusting for BMI, only the high-low trajectory remained significant [10]. Attenuation of significance for the SES and diabetes associations after adjusting for BMI may suggest that BMI is in the causal pathway of the association. SES itself may not directly contribute to the development of diabetes, but rather low SES may lead to greater BMI which may increase the risk for diabetes. A cross-sectional survey in Korea, where significant social change occurred throughout the mid-20th century, found that individuals born after the social change who had low or high-low SES trajectories had a greater prevalence of diabetes, but no associations were seen for those born before the changes [66]. The differences seen in the timing of birth may be due to the use of education as a marker of SES, as attaining a higher education may not have assisted those born before the social changes in obtaining a good job and higher pay given the poverty and war during this time period. However, there are significant difficulties in trying to determine either childhood or adulthood SES using any of the frequently used variables (i.e. education, occupation, or income), including errors in recall and the validity of these measures as markers of SES across different time periods [67].

25 CHAPTER 2. LITERATURE REVIEW 15 Weight rebound. Growth can be determined by measuring weight, stature, or birthweight. Lower birthweight has been shown to increase the risk for diabetes [9, 11, 12]. However, studies on growth have found that while low birthweight is a risk factor for diabetes, early rapid weight rebound (which is a greater-than-expected increase in weight for any particular age; assessed either by BMI or weight) following low birthweight has been associated with an increased risk for diabetes and other chronic diseases [4, 9, 11, 12, 68]. The risk for diabetes is increased further if the rapid weight rebound occurred between 2 5 years of age and the weight gain was above average [11]. It has been suggested that this rapid weight rebound, not low birthweight per se, is the factor most associated with the development of chronic diseases [69]. The increased risk for diabetes as a result of childhood growth is likely mediated through increased insulin resistance [11, 21] and reduced β-cell proliferation [7, 70]. Considering other postnatal influences on growth and diabetes risk, use of birthweight as a measure of early life conditions can not provide a complete understanding of how early life impacts later diabetes risk. And while weight rebound during childhood may increase the risk for diabetes, this measure requires weight and height to be measured throughout childhood to assess risk (an impractical, time-consuming, and expensive method) and may represent a different phenomenon from those seen with restrictions in the growth of stature and its components [4]. Components of stature. Stature and its components have also been used as a measure of growth when examining early life conditions and risk for diabetes. Height is typically measured along with weight to obtain BMI, which is a risk factor for diabetes; however, height alone has been inversely associated with diabetes [65, 71, 72], though no such associations were observed in US Filipina females [15] and British females [14]. Sitting height was found to be positively associated with diabetes in older Chinese adults, though significance was attenuated after adjusting for BMI [73]. Another study of elderly Chinese individuals also found that a longer sitting height and higher leg-to-trunk ratio, but not leg length, was associated with higher fasting glucose and with the presence of diabetes, though the study s primary objective was to investigate associations of stature with cardiovascular disease risk [18]. Since leg length is a sensitive marker of early childhood conditions (reviewed in Section above) and early life conditions are an etiological risk factor for diabetes, some studies have investigated whether an association exists between leg length and diabetes risk (see Table 2.1). There have been several studies which have found a significant association of shorter leg length with a greater risk for having diabetes [14, 65, 71], while