Childhood and adolescence represent critical times for bone growth and poor bone accumulation during this period may increase risk of osteoporosis later in life. Diet is a key modifiable factor that influences bone development. With this in mind, I am interested in using data from ALSPAC to explore the influence of three aspects of diet on bone density in children for my doctoral dissertation.

I: Breastfeeding and Bone Density

The first aspect of diet I would like to examine is breastfeeding. The composition of breast milk and the fact that it has been specifically designed for human infants provides reason to believe that breastfeeding may play an early, critical role in future bone health. Breast milk provides infants with the appropriate amount of fat, sugar, water, and protein that they need for healthy growth and development and is able to adapt to the needs of each infant. In addition, it is easier for infants to digest and allows for calcium to be absorbed more easily than formulas, which are often based on cow's milk that forms a hard curd in the infant's stomach that is not easily digested [1].

Few studies have been conducted on this topic and those that have provide conflicting evidence. Jones et al., for example, found that breastfeeding leads to increased bone mass at age 8 [2]. Children breastfed for 3 months or longer had significantly better total body, femoral neck and lumbar spine density [2]. Bishop et al. also found bone density of the radius to be higher in 5 year old children if they had been fed more human milk than infant formula [3]. However, Bishop et al. only studied children born preterm. Contrarily, a study by Harvey did not find any association between breastfeeding during the first year of life and bone density at 4 years of age [4]. In addition, several studies that have focused on breastfeeding and bone density in infancy seem to suggest that bone density is lower at this age in breastfed infants [5]. No studies were identified that examined the long-term effect of breastfeeding on bone density in children over 8 years of age.

Considering this, I am interested in further examining breastfeeding and bone density in children. In addition to examining whether a child has ever been breastfed in relationship to bone density, I would also like to examine breastfeeding duration and exclusivity and determine if the effect is different in preterm compared to fullterm infants.

II: Fruit and Vegetable Consumption and Bone Density

Current Western diets, which are abundant in protein rich foods, grains, and cereals, are considered to lead to increased acidity in the body due to their high content of components such as phosphorus, protein and chlorine [6]. To counteract this increased acidity, it has been theorized that the body draws on the basic salts of the bone (i.e. calcium) to serve as a buffer [7]. In children and adolescents, whose bones are still growing, this increased acid load may lead to achieving a peak bone mass below their genetic potential.

Vegetables and fruits, which contain high levels of basic elements such as potassium, calcium, magnesium and vitamin C [6], are gaining interest as important components of a bone healthy diet. Increased consumption of fruits and vegetables may counterbalance the effects of the more acidic elements of diet and, in turn, protect and promote bone growth. Other explanations for why fruits and vegetables are beneficial to bone include high levels of antioxidants such as beta-carotene and vitamin C, which may act to reduce oxidative stress [8], regulation of osteoblast differentiation and collagen formation by vitamin C or the role that vitamin K "plays in bone mineralization by acting as a co-factor in the carboxylation of the bone protein osteocalcin" [8].

To date, a handful of studies have been conducted on this topic in both adults and children and the findings are promising [6, 8-13]. In a study of 56 girls, for example, Tylavsky et al. reported that girls who consumed 3 or more servings of fruit and vegetables had 6.0% more whole body and 8.3% higher radius bone area than those who consumed fewer than 3 servings [10]. Vatanparast et al reported that higher intakes of vegetables and fruit had a significant effect on total body bone mineral content in boys [11]. They calculated that with the consumption of 10 servings of fruits and vegetables a day compared to one, total body bone mineral content would be 48.6 g higher in their male participants.

I would like to further examine the effect of actual fruit and vegetable (e.g. number of servings or grams of fruits and vegetables per day) consumption on bone density. I would also like to focus on the specific vitamins and minerals (potassium, magnesium, vitamin C, vitamin K, calcium, etc) that are found in fruits and vegetables and believed to influence bone density and examine their effects. Lastly, I would like to examine patterns of food consumption in the child and bone density. Cole et al. took this approach, but looked at patterns of maternal food consumption during pregnancy [14]. They found diets high in fruits, vegetables, and wholemeal bread, rice and pasta and low in processed foods to be associated with increased bone mineral content and areal BMD in children.

III: Protein and Bone Density

Lastly, while there is no doubt that protein, which makes up approximately one third of bone mass, is important for bone growth, there is conflicting evidence regarding how much is beneficial [15]. High protein intakes have been implicated as a potential contributors to an increased acid load, which may increase urinary calcium excretion and, in effect, bone resorption [16]. On the other hand, high protein may be beneficial due to the important role it plays in IGF-1 metabolism, which is important in calcium and phosphorus metabolism. The conflicting evidence, which has been the topic of a number of studies in adults [17-20], also emerges in several epidemiological studies in children. A study of 10 year old girls in China, for example, found higher protein intakes to be negatively associated with bone mineral content [21]. The authors noted that this group of girls also had low calcium intakes. Contrarily, Chevalley et al and Alexy et al report high protein to be beneficial [22, 23]. Alexy et al. reported that protein intake in 6-18 year old children was associated with higher bone mineral content as well as periosteal circumference and polar strength strain. Interestingly, these researchers also estimated dietary acid load using PRAL (potential renal acid load) and found higher PRAL to be associated with decreased bone mineral content and cortical area [23].

Considering the debate that remains regarding protein and bone health, I propose to examine the effects of protein on bone density using the ALSPAC data, paying particular attention to low and high protein consumption. I would also like to examine whether the relationship between protein and bone density is modified by calcium consumption. Lastly, I would like to examine the effect of acid load on bone density using an equation developed by Remer et al that uses protein, phosphorus, potassium, and magnesium [24] and by Frassetto et al that uses the ratio of protein to potassium [25] .

Analysis

The bone measure outcomes for this analysis will be derived from the DXA scans taken during the Focus at 9 clinic (mean age 9.9 years). In statistical modeling, we will use total body bone mineral content, areal bone mineral density and area-adjusted bone mineral content like Tobias et al [26].

For aim 1, we will define infant feeding as was done by Pontin et al (exclusive/predominant breastfeeding, complementary feeding and replacement feeding) [27]. In addition, children will also be categorized according to how long they were breastfed to examine the effects of duration of breastfeeding.

The main exposure data for aims 2 and 3 will come from diet diaries collected at the Focus at 10 clinic (mean age 10.6 years) and from the food frequency questionnaire (FFQ) administered as part of the Focus at 8 clinic (mean age 8.6 years). For aim 2, we will examine fruit and vegetable consumption (servings per day), specific vitamins and minerals (potassium, magnesium, vitamin C, vitamin K, calcium) and patterns of food consumption. Patterns of food consumption have been derived from the FFQ. For aim 3, the main exposure will be protein consumption, which will be examined in terms of grams per day. The ratio of protein to calcium as well as estimates of acid load (PRAL) will also be calculated. To reduce confounding, dietary data will be adjusted for total energy intake.

A number of potentially important confounders will be considered. These include height and weight, which are key to consider because DXA variables are size dependent and any influences of diet on bone could be due to an effect on overall body size. Family characteristics such as social class, housing tenure and maternal and paternal education, race, and occupation will also be included. Physical activity derived by self-report or accelerometer will be important to include. For aim 1, other pregnancy and birth related variables such as maternal age, gestational age, birthweight, and maternal smoking during pregnancy will be examined.

All analyses will be conducted using SAS 9.2. Linear regression will be used to examine the questions presented above. Analyses will most likely be done separately for boys and girls. In addition, a sensitivity analysis will be done to compare results in children with and without puberty information.

Depending on the results that are found, we may extend analyses to later time points, and to other measures such as hip DXA and tibial pQCT