Please note, this is recovered content from the former website of the New York Obesity Research Center website.


Examine eating behavior and body composition in children trough genetic and environmental glasses

Date: April 6th, 2006
Title: “Genetic and environmental contributions to eating behavior and body composition in children”
Speaker’s Name and Affiliation:

  1. Nicolas Stettler, Ph.D., Children’s Hospital of Philadelphia
  2. Julie Mennella, Ph.D., Monell Chemical Senses Center
  3. David Levitsky, Ph.D., Cornell University
  4. Streamson Chua, M.D., Ph.D., Albert Einstein College of Medicine
  5. Myles Faith, Ph.D., University of Pennsylvania Weight and Eating Disorders Program

Nicolas Stettler, “Association of fetal and infant growth with obesity.”

As the prevalence of obesity continues to increase among children and adolescents, obesity prevention is becoming a more urgent public health priority. A lifecourse approach to obesity prevention is promising, based on numerous epidemiological and mechanistic studies. High birth weight and rapid infancy weight gain are associated with an increased risk for obesity, but, in humans, small birth weight is not associated with obesity, in contrast to animal models. Early life programming or imprinting of the neurological and endocrine systems have been suggested as possible mechanisms for these associations.

Fetal and infancy growth patterns are associated with the long-term development of obesity both in epidemiological and mechanistic studies. Restricted fetal growth is associated with obesity in rats and with a central fat distribution in humans. Accelerated fetal growth and rapid weight gain are associated with increased risks for obesity. Due to inherent limitations of human and animal studies, closer collaborations between epidemiologists and experimental biologists will be necessary to improve our understanding of the lifecourse approach to obesity development and to design prevention strategies targeted to critical or sensitive periods of the lifecycle.


Q. Comment: I would be very careful about using BMI as a cut-off or sole indication of obesity. People of the same BMI can be at very different levels of risk. The distinction has not been widely recognized in this literature.
A. That is correct, BMI is a poor surrogate for level of overweight. However, body weight is also a poor surrogate.

Q. I noticed that in your data, Germany did not show the same relationships between rate of infant growth and adult obesity outcome. Have you looked at individual growth periods when children were 4 months/ 8 months?
A. Tim Cole in England is interested in this, but it is very challenging to do mathematically, because you have to correct for rate of infant growth. Comparing across age groups can be challenging.

Julie Mennella, “Flavor imprinting and the role of breastfeeding in the development of food preferences and habits.”

Dr. Mennella discussed the role of flavor programming, which begins in utero and continues during the first few years of life, on children’s food preferences and lifelong eating habits. Children in the U.S. continue to eat more than the recommended amounts of fat, sugar, and salt, and less than optimal amounts of fruits and vegetables. Sensory characteristics of foods dictate many of these intake patterns, and Dr. Mennella hypothesizes that many of the reactions children have to the foods they eat are imprinted from the time of conception.

Much like humans, young animals learn what to eat, and what not to eat by watching their mothers. Moreover, many flavors are transmitted through human milk, including garlic, ethanol, carrot, mint, and vanilla. Dr. Mennella presented data from an experiment to study the effects of exposure to carrot flavor during pregnancy or lactation on babies’ acceptance of carrots, and she found that those who were exposed to either carrot flavor or a greater variety of flavors during breastfeeding had higher acceptance of carrots than babies who were not exposed to carrot flavor. Thus, one of the greatest benefits to breastfeeding may be it’s ability to impact food preferences later in life by exposing children to a wider variety of flavors than formula feeding. In addition to her work on breastfeeding, Dr. Mennella has also found that babies who receive hydrolysate formulas from birth tend to have higher acceptances for bitter tasting foods such as broccoli later in life. Because of the sour and bitter flavor components present within hydrolysate formulas, infants learn from a very young age to accept these flavors.

Dr. Mennella briefly touched on some of the genetic variations that may impact on taste preferences, including the well-documented TAS2R38 gene that predicts variations in the ability to taste bitter thiourea compounds. There recent data suggests that variations at this genotype might impact not only on bitter taste perception, but also on preference for sweets. Interestingly, the effects between genotype and phenotype are more profound in children than they are in adults.


Q. If children are exposed to formulas early in life, how long do these effects (on food preferences) last?
A. The best evidence from this is from the PKU literature, and for these children, preferences can be affected by formula for a relatively long period of the child’s life, and perhaps into adulthood.

David Levitsky, “Teaching dietary restraint and weight monitoring to children: Is it good or bad?”

Dr. Levitsky presented some intriguing data collected from first year students at Cornell University that suggests that frequent weight monitoring can prevent the “freshman fifteen.” As background to these data, the speaker noted that while food intake surveys from the past 30 year suggest that food intake per person has either remained unchanged, or slightly decreased, if you examine food disappearance data from U.S.D.A, you can find an almost 35 kcal per day, per year increase in per capita energy intake. Despite this increase in calories, the amount of weight increase in the last few decades averages out to around 308 grams per day, a change that would be almost imperceptible on an ordinary bathroom scale.

In order to try to understand why so many freshman college students gain weight their first year of college, his lab has been conducting studies with freshman students at Cornell. In previous studies, reports indicated that these students gain an average of 2 kg during their first 12 weeks of college, and this weight gain has been attributed to increases in snacking and the frequency of eating in “all you can eat” dining halls (Levitsky et al., 2004). In the first study reported on, students were provided with scales and instructions on how to weigh themselves on a daily basis. After seven days of accumulating the data, students were sent a linear regression slope of their weight to let them know if they were gaining or losing weight that week. Results indicated that the experimental group who received feedback did not gain any weight during the first semester, while the untreated group gained 3.1 kg. Replications the following year further supported the initial results.

Dr. Levitsky concluded in his studies that frequent weight monitoring may be beneficial as a means of maintaining weight. Other studies support this, including individuals form the National Weight Registry (Klem et al., 1997), and data from Fujimoto et al., (1992). Some restriction of energy is necessary to sustain a weight loss, and it is only at the extremes of dietary restraint scores where these behaviors may indicate the presence of an eating disorder.


Q. In your data, the experimental group had higher body weights to start out with than the control group. Could that have affected your results?
A. Subjects were randomized into groups, but that doesn’t prevent regression to the mean from occurring. It is possible that the difference in baseline body weights might have had an effect.

Q. In the data that you presented on the number of kcal available per capita, does that take into account the fact that there are actually fewer children per adults today than there were a few decades ago?
A. Those figures are from USDA census data, and I don’t believe it takes into account the proportion of children and adults.

Q. Is it the actual behavior of weighing oneself repeatedly that is altering body weight or eating behavior, or is it something else? Have you tried to address this question?
A. We have started to get records from the weigh-ins to begin to answer this question of whether it is the behavior of weighing, or the outcome of this behavior that is benefiting the participants.

Streamson Chua, “Genetic models of human obesity in mice and humans”

Dr. Chua presented data from transgenic mouse models that indicate most of leptin’s actions on body composition, ingestive behavior, metabolism and reproduction are mediated by the central nervous system. This conclusion was based on the correction of the obesity phenotype of leptin receptor deficient mice with transgenes that express the B isoform of the leptin receptor in neurons. Dr. Chua then summarized studies in which chemically defined neurons were targeted to eliminate leptin receptor expression. Two arcuate nucleus neurons were studied due to their central roles in leptin biology – the arcuate POMC neuron and the arcuate AGRP/NPY neuron. Mice that were deficient for leptin receptor expression in either neuronal type develop mild obesity and increased fraction body fat. Mice that had deficient leptin receptor expression in both neuronal types exhibited a degree of obesity that was the sum of the individual lesions. However, these mice showed a transient hyperphagia immediately post-weaning that accounted for their increased mass along with preferential fat storage reflected by an increased respiratory quotient. Thus, leptin receptor deficiency in two arcuate nucleus cell types can account for some qualitative aspects of the phenotype due to complete loss of leptin signaling. However, the partial phenotype also shows that other leptin receptor bearing neurons, modulated by leptin’s actions, contribute in a significant manner to the regulation of ingestion and metabolism.


Q. In the transgenic mice, do you see a change in RQ?
A. We haven’t measured this yet. My prediction is that we’ll probably see a change in RQ.

Q. Did you look at leptin expression in the brain in your second study?
A. We haven’t looked at leptin expression in the brain specifically. There are probably many responsive sites to leptin.

Q. What about metabolic rate in the animals?
A. They were absolutely the same at three months of age, although I’m not sure why.

Q. Do any of these genetic manipulation have anything to do with hair density?
A. Noone has looked at that, although we do find that obese animals have more sparse hair.

Q. Is the fat mass accumulation brown or white fat?
A. It is mainly an accumulation of white fat.

Q. What happens if you restore LepR on taste cells?
A. We haven’t studied taste cells yet, so I’m not sure.

Myles Faith, “Behavior genetic contributions to child feeding and body composition”

Dr. Faith closed the seminar session with an overview of how traditional behavioral genetic studies can shed light on the study of child feeding and body composition. It has long been realized that there is an inherited component to body composition, and obese parents are more likely to produce obese children. Behavioral geneticists who study obesity are trying to determine how the “familial risk” for obesity impacts on a child’s eating behavior. A variety of “unmeasured genotype studies,” such as twin or heritability studies, have looked at the heritability of food intake, and typically the similarities in within-family food intake patterns are low, and often are not systematic. However, because most of these studies have been done using food intake questionnaires, there is certainly a need to validate these results in the laboratory.

With respect to laboratory studies in the heritability of food intake, Dr. Faith presented research from “Project Grow2gether,” a sibling study designed to determine the genetic and environmental components of child energy and macronutrient intake, and compensation ability. Results indicate that siblings eat similarly to one another, showing significant correlations for not only total energy, but macronutrient intake.

In the next portion of his talk, Dr. Faith introduced the notion of the classic twin model, which makes the assumption that phenotypes in each twin are influenced by an additive genetic effect, a common environmental effect, and a non-shared environmental effect. From twin studies, one can conclude that some food intake measures do show a significant heritable component, but the evidence is inconsistent. Detecting these genetic effects may require more refined eating phenotypes, or conducting experimental manipulations of the environment. A number of recent studies have introduced examples of potentially refined eating phenotypes, such as disinhibition, the reinforcing value of food, and response to environmental overfeeding. In his twin studies, Dr. Faith has seen evidence for a heritable component to eating related behaviors, such as “food requests” and the number of mouthfuls for minutes, so these may also serve as refined phenotypes for future investigations.

In conclusion, Dr. Faith suggested that greater collaborations between molecular genetic and behavioral genetics might elucidate some of the classic problems inherent within the field. Further, more contemporary models of the environment that include refined and specific eating phenotypes will also improve the ability of behavioral genetics studies.


Q. Are you moving away from heritability and into molecular genetics, as implied by your talk?
A. I believe that you can move forward by collaborating with molecular biologists, so I think both behaviorists and molecular biologists must work together. Examples of where this field might be moving are imagine studies, or environmental manipulations that are studied in the context of molecular genetic outcomes.

Q. What about taste, food choice, and in utero effects? These occur from such an early age.
A. I agree. Many behavioral genetic studies have not adequately addressed thse effects because they study only a single point in time.