Nutrition and Disease Prevention: A Systems Approach to Metabolic Health

Resources

A systems epidemiology approach to nutrition, obesity, and diabetes

Brigham and Women's Hospital, Harvard Medical School. Channing Division of Network Medicine. The Nurses' Health Study.

Chan JM, Rimm EB, Colditz GA, et al. Obesity, fat distribution, and weight gain as risk factors for clinical diabetes in men. Diabetes Care. 1994;17(9):961-9.

Cheng S, Rhee EP, Larson MG, et al. Metabolite profiling identifies pathways associated with metabolic risk in humans. Circulation. 2012;125(18):2222-31.

Cornelis MC, Monda KL, Yu K, et al. Genome-wide meta-analysis identifies regions on 7p21 (AHR) and 15q24 (CYP1A2) as determinants of habitual caffeine consumption. PLoS Genet. 2011;7(4):e1002033.

Dettmer K, Aronov PA, Hammock BD. Mass spectrometry-based metabolomics. Mass Spectrom Rev. 2007;26(1):51-78.

Ebbeling CB, Feldman HA, Chomitz VR, et al. A randomized trial of sugar-sweetened beverages and adolescent body weight. N Engl J Med. 2012;367(15):1407-16.

Franks PW, Pearson E, Florez JC, et al. Gene-environment and gene-treatment interactions in type 2 diabetes: progress, pitfalls, and prospects. Diabetes Care. 2013;36(5):1413-21.

Ley SH, Hamdy O, Mohan V, Hu FB. Prevention and management of type 2 diabetes: dietary components and nutritional strategies. Lancet. 2014;383(9933):1999-2007.

Locke AE, Kahali B, Berndt SI, et al. Genetic studies of body mass index yield new insights for obesity biology. Nature. 2015;518(7538):197-206.

U.S. Department of Agriculture, U.S. Department of Health and Human Services. Office of Disease Prevention and Health Promotion. Dietary Guidelines for Americans, 2015. Updated May 9, 2015.

Qi Q, Chu AY, Kang JH, et al. Sugar-sweetened beverages and genetic risk of obesity. N Engl J Med. 2012;367(15):1387-96.

Ruyter de JC, Olthof MR, Seidell JC, Katan MB. A trial of sugar-free or sugar-sweetened beverages and body weight in children. N Engl J Med. 2012;367(15):1397-406.

Sun Q, Wedick NM, Pan A, et al. Gut microbiota metabolites of dietary lignans and risk of type 2 diabetes: a prospective investigation in two cohorts of U.S. women. Diabetes Care. 2014;37(5):1287-95.

Vinjé S, Stroes E, Nieuwdorp M, Hazen SL. The gut microbiome as novel cardio-metabolic target: the time has come! Eur Heart J. 2014;35(14):883-7.

Wang TJ, Larson MG, Vasan RS, et al. Metabolite profiles and the risk of developing diabetes. Nat Med. 2011;17(4):448-53.

Physiological changes associated with weight loss

Begg DP, Woods SC. The endocrinology of food intake. Nat Rev Endocrinol. 2013;9(10):584-97.

Davis EE, Katsanis N. The ciliopathies: a transitional model into systems biology of human genetic disease. Curr Opin Genet Dev. 2012;22(3):290-303.

Foster GD, Wyatt HR, Hill JO, et al. Weight and metabolic outcomes after 2 years on a low-carbohydrate versus low-fat diet: a randomized trial. Ann Intern Med. 2010;153(3):147-57.

Kaplan GD, Stifler LT. Very low-calorie diets for obesity. JAMA. 1994;271(1):24-5.

Kissileff HR, Thornton JC, Torres MI, et al. Leptin reverses declines in satiation in weight-reduced obese humans. Am J Clin Nutr. 2012;95(2):309-17.

Leibel RL, Rosenbaum M, Hirsch J. Changes in energy expenditure resulting from altered body weight. N Engl J Med. 1995;332(10):621-8.

Manning S, Pucci A, Finer N. Pharmacotherapy for obesity: novel agents and paradigms. Ther Adv Chronic Dis. 2014;5(3):135-48.

Pinto S, Roseberry AG, Liu H, et al. Rapid rewiring of arcuate nucleus feeding circuits by leptin. Science. 2004;304(5667):110-5.

Wang L, Meece K, Williams DJ, et al. Differentiation of hypothalamic-like neurons from human pluripotent stem cells. J Clin Invest. 2015;125(2):796-808.

Systems genetics to study heart disease

Bennett BJ, de Aguiar Vallim TQ, Want Z, et al. Trimethylamine-N-oxide, a metabolite associated with atherosclerosis, exhibits complex genetic and dietary regulation. Cell Metab. 2013;17(1):49-60.

Jackson Laboratory. Diversity outbred and collaborative cross mice to offer maximum allelic variation. JAX NOTES. 2009;(514).

O'Connor A, Quizon PM, Albright JE, et al. Responsiveness of cardiometabolic-related microbiota to diet is influenced by host genetics. Mamm Genome. 2014;25(11-12):583-99.

Olthof MR, Brink EJ, Katan MB, Verhoef P. Choline supplemented as phosphatidylcholine decreases fasting and postmethionine-loading plasma homocysteine concentrations in healthy men. Am J Clin Nutr. 2005;82(1):111-7.

Solberg Woods LC. QTL mapping in outbred populations: successes and challenges. Physiol Genomics. 2014;46(3):81-90.

Svenson KL, Gatti DM, Valdar W, et al. High-resolution genetic mapping using the Mouse Diversity outbred population. Genetics. 2012;190(2):437-47.

Wang Z, Klipfell E, Bennett BJ, et al. Gut flora metabolism of phosphatidylcholine promotes cardiovascular disease. Nature. 2011;472(7341):57-63.

Gut microbia niches and their relevance to metabolic health

Gao X, Liu X, Xu J, Xue C, et al. Dietary trimethylamine N-oxide exacerbates impaired glucose tolerance in mice fed a high fat diet. J Biosci Bioeng. 2014;118(4):476-81.

Hoyles L, McCartney AL, Neve H, et al. Characterization of virus-like particles associated with the human faecal and caecal microbiota. Res Microbiol. 2014;165(10):803-12.

Russell WR, Hoyles L, Flint HJ, Dumas ME. Colonic bacterial metabolites and human health. Curr Opin Microbiol. 2013;16(3):246-54.

Treberg JR, Driedzic WR. Elevated levels of trimethylamine oxide in deep-sea fish: evidence for synthesis and intertissue physiological importance. J Exp Zool. 2002;293(1):39-45.

Systems science for complex problems

Epstein JM. Why model? J Artif Soc Soc Simul. 2008;11(4):12.

Hall KD, Hammond RA, Rahmandad H. Dynamic interplay among homeostatic, hedonic, and cognitive feedback circuits regulating body weight. Am J Public Health. 2014;104(7):1169-75.

Hammond RA, Ornstein JT, Fellows LK, et al. A model of food reward learning with dynamic reward exposure. Front Comput Neurosci. 2012;6:82.

Homer J, Milstein B, Wile K, et al. Modeling the local dynamics of cardiovascular health: risk factors, context, and capacity. Prev Chronic Dis. 2008;5(2):A63.

Homer J, Milstein B, Wile K, et al. Simulating and evaluating local interventions to improve cardiovascular health. Prev Chronic Dis. 2010;7(1):A18.

Homer J, Wile K, Yarnoff B, et al. Using simulation to compare established and emerging interventions to reduce cardiovascular disease risk in the United States. Prev Chronic Dis. 2014;11:E195.

Honeycutt AA, Wile K, Dove C, et al. Strategic planning for chronic disease prevention in rural America: looking through a PRISM lens. J Public Health Manag Pract. Epub 2014.

Loyo HK, Batcher C, Wile K, et al. From model to action: using a system dynamics model of chronic disease risks to align community action. Health Promot Pract. 2013;14(1):53-61.

Meadows DM. Thinking in Systems: A Primer. White River Jct, Vermont: Chelsea Green Publishing; 2008.

National Institutes of Health. Division of Nutrition Research Coordination. A National Nutrition Research Roadmap. 2014.

National Institutes of Health. Office of Behavioral and Social Sciences Research. Systems Science Listserv.

National Institutes of Health. Office of Disease Prevention. Strategic Plan for Fiscal Years 2014–2018. Updated February 10, 2015.

Mabry PL, Bures RM. Systems science for obesity-related research questions: an introduction to the theme issue. Am J Public Health. 2014;104(7):1157-9.

Sterman JD. All models are wrong: reflections on becoming a systems scientist. System Dynamics Review. 2002;18(4): 501-31.

Sterman JD. Learning from evidence in a complex world. Am J Public Health. 2006;96(3):505-14.

Mathematical modeling of metabolism and body weight dynamics

Dhurandhar NV, Schoeller D, Brown AW, et al. Energy balance measurement: when something is not better than nothing. Int J Obes (Lond). 2014.

Guth E. JAMA patient page. Healthy weight loss. JAMA. 2014;312(9):974.

Hall KD, Butte NF, Swinburn BA, Chow CC. Dynamics of childhood growth and obesity: development and validation of a quantitative mathematical model. Lancet Diabetes Endocrinol. 2013;1(2):97-105.

Hall KD, Chow CC. Estimating changes in free-living energy intake and its confidence interval. Am J Clin Nutr. 2011;94(1):66-74.

Hall KD, Guo J, Dore M, Chow CC. The progressive increase of food waste in America and its environmental impact. PLoS One. 2009;4(11):e7940.

Hall KD, Sacks G, Chandramohan D, et al. Quantification of the effect of energy imbalance on bodyweight. Lancet. 2011;378(9793):826-37.

Hall KD. Predicting metabolic adaptation, body weight change, and energy intake in humans. Am J Physiol Endocrinol Metab. 2010;298(3):E449-66.

Heshka S, Anderson JW, Atkinson RL, et al. Weight loss with self-help compared with a structured commercial program: a randomized trial. JAMA. 2003;289(14):1792-8.

Lin BH, Smith TA, Lee JY, Hall KD. Measuring weight outcomes for obesity intervention strategies: the case of a sugar-sweetened beverage tax. Econ Hum Biol. 2011;9(4):329-41.

Smith TA, Lin BH, Lee JY. U.S Department of Agriculture. Economic Research Service. Taxing caloric sweetened beverages: potential effects on beverage consumption, calorie intake, and obesity. Economic Research Report No. (ERR-100). 2010.

Winkler JT. The fundamental flaw in obesity research. Obes Rev. 2005;6(3):199-202.

Behavioral science to inform food policies

Block JP, Roberto CA. Potential benefits of calorie labeling in restaurants. JAMA. 2014;312(9):887-8.

Rick S, Loewenstein G. Intangibility in intertemporal choice. Philos Trans R Soc Lond B Biol Sci. 2008;363(1511):3813-24.

Roberto CA, Agnew H, Brownell KD. An observational study of consumers' accessing of nutrition information in chain restaurants. Am J Public Health. 2009;99(5):820-1.

Roberto CA, Larsen PD, Agnew H, et al. Evaluating the impact of menu labeling on food choices and intake. Am J Public Health. 2010;100(2):312-8.

Roberto CA, Schwartz MB, Brownell KD. Rationale and evidence for menu-labeling legislation. Am J Prev Med. 2009;37(6):546-51.

The Psychology of Eating and Consumer Health (PEACH) Lab.

Sponsors

Bronze Sponsor

Metagenics and Metagenics Healthcare Institute for Clinical Nutrition

Nutrition plays an important role in prevalent chronic diseases such as obesity, heart disease, diabetes, and cancer. Heart disease and diabetes were among the top causes of death in 2012, responsible for 7.4 million and 1.5 million fatalities, respectively, worldwide. Nutrition science aims to deepen scientists' understanding of these diseases and point to prevention strategies. Yet the study of nutrition and its role in disease is complex. Recent genome-wide association studies (GWAS) have uncovered nearly 100 genetic loci associated with obesity. But those sites explain only 3% of obesity risk. Genes, gene–diet interactions, gut microbiota, hormones, and neural networks are all involved in regulating hunger and metabolism.

The conference focused on a systems approach to nutrition and disease prevention to untangle complex questions involving fields from genetics to food policy. It also approached epidemiological questions about how food groups impact disease, and psychological questions about why people make the food choices they do.

A systems approach to biology explores how different types of data combine to explain physiology and disease states. However, the disciplines represented at the conference, such as microbiology and epidemiology, have different meanings for the term systems science, and different tools to study it. Some of the presentations looked at how best to define, study, and apply the approach.

In his keynote address, Frank Hu talked about a type of systems science called systems epidemiology, which, unlike traditional epidemiology, aims to explain the mechanisms by which an exposure leads to a disease. For example, whereas traditional epidemiology and large cohort studies demonstrate a link between caffeine and a reduced risk of type 2 diabetes, systems epidemiology and 'omics data (genomics, transcriptomics, metabolomics) uncover the genes and metabolites mediating the association.

Systems epidemiology aims to use 'omics data, such as genome or exposome data shown here, to explain how exposures affect disease risk. (Image courtesy of Frank Hu)

Brian Bennett discussed how systems genetics, and the study of intermediate phenotypes such as transcriptome profiles, can identify pathways through which genes contribute to atherosclerosis. Patricia Mabry described models that predict the effect policy interventions will have on disease outcomes. Kevin Hall talked about models that capture the relationship between energy intake and body mass, revealing the drivers of weight regain and the obesity epidemic.

Several speakers presented data on pathways in metabolic health and eating behavior. Rudolph Leibel explained that weight loss decreases leptin hormone levels, which in turn increases appetite and reduces energy expenditure. Anne McCartney outlined her studies of the microbiome in the cecum, the first part of the colon. The microbiota here appear to have different metabolic activities when compared with the fecal microbiota, typically the focus of gut microbiota/microbiome studies. In the last talk, Christina Roberto shared her research into how menu labeling affects eating behavior and food choice, with a focus on food policy.

The speakers offered insight into how systems science models are created and how data on molecular mechanisms, epidemiology, and food policies can be integrated. Key to systems-based approaches is gathering large data sets to capture a broad range of biological processes. Data gathering is often accomplished through high-throughput technologies, such as microarrays. As Hu emphasized, collaboration in data collection among research centers and disciplines is critical. Researchers also need to work with community representatives and leaders to inform food policy decisions.

Metabolic health is influenced by factors from the microbiological level to the societal. (Image courtesy of Patricia L. Mabry)

Keynote Speaker:
Frank Hu, Harvard T.H. Chan School of Public Health

Highlights

• Systems epidemiology uses 'omics data, such as metabolomics and exposomics, to explain traditional epidemiological associations.
• Metabolites of dietary components could serve as markers of disease risk and help explain mechanisms of disease development.
• Advancing systems epidemiology will require multidisciplinary collaborations and improved methods to measure exposures.

Keynote address: a systems epidemiology approach to nutrition, obesity, and diabetes

In his keynote address, Frank Hu of the Harvard T.H. Chan School of Public Health first compared systems epidemiology with traditional epidemiology, which links an exposure such as diet with a phenotype such as heart disease. Systems epidemiology aims to explain these connections through so-called 'omics data, including genomics, metabolomics, and exposomics; the last describes how exposures such as diet, lifestyle, and sleep contribute to disease.

Hu exemplified the potential uses of systems epidemiology by describing its applications in type 2 diabetes research. Traditional epidemiological studies, such as the Nurses' Health Study at Harvard University, associated consumption of coffee, whole grains, and green leafy vegetables with a decreased risk of type 2 diabetes, and consumption of sugar-sweetened beverages, fruit juices, and red meat with an increased risk. "The question is," Hu said, "What is the biological explanation underlying these associations?" Systems epidemiology has linked variants in caffeine metabolism genes with coffee consumption, and linked coffee consumption with downregulation of inflammatory genes and with certain metabolite signatures in urine.

Systems epidemiology can help explain the mechanisms by which foods are associated with increased or decreased risk of type 2 diabetes. (Image courtesy of Frank Hu)

Characterizing metabolite profiles in biological samples such as urine or blood is a hot area of nutrition research. Many of the small molecules are derived from diet; metabolite signatures could serve as both biomarkers of diet and predictors of disease risk. For example, blood levels of branched-chain amino acid (BCAA) metabolites are determined by dietary levels of these amino acids (leucine, isoleucine, and valine), and according to prospective cohort studies, higher blood metabolite levels may predict the risk of developing diabetes.

Urine metabolites of lignin, a polyphenol found in whole grains and lentils, could help explain the link between polyphenols in plant-based foods and decreased risks of diabetes and heart disease. Urine levels of these metabolites are inversely associated with development of type 2 diabetes, and the metabolites have been shown to have antioxidant activity and to bind to estrogen receptor alpha, which is associated with maintaining insulin sensitivity. These metabolites are produced by the gut microbiota, which because of their many influences on metabolism could potentially be a target for prevention and treatment of cardiometabolic diseases.

Measuring polyphenol metabolites could improve the assessment of dietary exposure to polyphenols. This assessment typically relies on food intake questionnaires, but differences in packaging and exposure to sunlight can cause polyphenol levels to vary within food groups.Systems epidemiology could also elucidate how genes and environmental factors interact to contribute to disease. Although large genome-wide association studies (GWAS) have uncovered 97 loci associated with obesity, genetics explains only 2.7% of BMI variation in the population. A trio of studies, including one by Hu and his colleagues, suggest that consumption of sugar-sweetened beverages can magnify genetic predisposition to obesity.

Despite strides in systems epidemiology and "near perfect" genomic measurements, Hu said, the field needs better ways to measure the exposome and better biostatistics and bioinformatics methods to integrate diverse data. Its multidisciplinary focus also requires more international collaborations and consortia like DIAGRAM and MAGIC, which are in progress.

Metabolites, characterized through metabolome studies, could potentially serve as biomarkers of nutrient intake or disease risk. (Image courtesy of Frank Hu)

Q&A

In the Q&A period Hu was asked about how to balance the need for more clinical research with the need to make recommendations for consumers and patients, especially considering ubiquitous misinformation online. He replied that incorrect nutritional information is a serious problem, but advocated working with the media to increase awareness about how diet impacts disease outcomes. He mentioned that an update to the Dietary Guidelines for Americans, to be released later in the year, will affect federally funded nutrition programs. (Hu served on the advisory committee for this Guidelines update.)

Responding to an inquiry about how consumption of added sugar correlates with the obesity trend, Hu pointed to a decrease in overall consumption (especially of sugar-sweetened beverages) that could have contributed to the leveling of obesity prevalence or the decreased rates of obesity in some age groups seen in the last several years. After a related question about artificially sweetened drinks, Hu mentioned an analysis that suggested that replacing sugar with artificial sweeteners could reduce body mass in the short term but found an inconsistent association with cardiometabolic diseases in the long term. However the studies, in mice and humans, had flaws, such as administering high doses of artificial sweeteners; thus Hu recommended more research and advised against making artificially sweetened drinks the main replacement for sugar-sweetened beverages, at least for children.

After a question about whether to measure postprandial or fasting metabolite profiles, Hu noted that studies tend to look at fasting levels, which generally show stable metabolites. However, studies using the oral glucose tolerance test have found that some metabolite levels differ largely after meals compared with after fasting. Mealtime data may be more clinically relevant but is difficult to capture in large epidemiological studies. It would, however, be possible to measure both profiles in the clinic.

Speakers:
Rudolph L. Leibel, Columbia University
Brian J. Bennett, University of North Carolina, Chapel Hill
Anne L. McCartney, University of Reading, UK
 
Moderator:
Andrew Swick, Metagenics

Highlights

• Leptin is a key player in the physiological changes associated with weight loss.
• Systems genetics can uncover genes and intermediate phenotypes, such as metabolite levels, associated with disease.
• The cecal microbiota, which is proximal to the fecal microbiota, could be important in metabolic disease.

Physiological changes associated with weight loss

Rudolph L. Leibel of Columbia University began by discussing a problem most dieters encounter—weight lost is almost always regained, despite initial success, unless strict food restrictions and exercise are maintained. One molecule at the crux of this paradox is leptin, a hormone produced by fat cells in proportion to their numbers and size. Leptin undermines weight loss, because when its levels decrease, as fat cells shrink and possibly decrease in number, an anabolic response results and the body ramps-up food intake and reduces energy expenditure. This response is seen in obese and lean individuals, but the threshold leptin concentrations at which energy use drops are higher in those who are obese. As Leibel explained, myriad factors determine the threshold, including genes, environment, brain development, and perinatal exposures.

In clinical studies Leibel and his colleagues found that injections of leptin counteract the threshold effect. Administration of leptin after weight loss was associated with a smaller decline in energy expenditure—and a lower spike in muscle efficiency, which contributes to the energy expenditure drop. It also improved satiation and restored brain activity to the pre-weight-loss state. "We're trying to trick the brain into thinking the fat has not been lost," he said.

Leibel described therapies that restore normal physiology after weight loss as more promising for long-term outcomes than those that promote weight loss. His team developed a technique to generate neurons from pluripotent stem cells derived from the skin cells of people with monogenic types of obesity, such as Bardet-Biedl syndrome. The neurons had defective cellular leptin signaling.

Weight loss is associated with physiological changes including a decrease in energy expenditure. (Image courtesy of Rudolph L. Leibel)

Systems genetics to study heart disease

Brian J. Bennett of the University of North Carolina, Chapel Hill, introduced systems genetics and how it can elucidate relationships between genes, diet, and disease. Unlike classic genetics, which strives to link genotypes with phenotypes such as disease state, this approach looks at intermediate phenotypes, including transcriptome, proteome, and metabolome profiles, usually using quantitative and high-throughput techniques.

Bennett's team is using systems genetics to study atherosclerosis, the formation of plaques inside arteries. The researchers are focusing on a metabolite called trimethylamine-N-oxide (TMAO), which has been associated with atherosclerosis. TMAO levels are probably influenced by both diet and genes; it is a metabolite of choline (often phosphatidylcholine), which is found in foods such as eggs and meat and converted to TMAO by gut microbiota and liver enzymes.

To identify genes involved in regulating TMAO levels, Bennett and his collaborators took advantage of a genetic resource called Diversity Outbred (DO) mice, which are highly heterozygous and capture about 90% of the diversity between inbred strains. By working with a large population of DO mice, the team identified loci on chromosomes 12 and 14 that are associated with differences in TMAO levels.

These chromosomal regions comprise 165 genes, too many to study individually, so the researchers used a systems-based approach to help determine which genes contribute to TMAO levels. They pieced together co-expression networks, or hubs, of genes that follow the same pattern of upregulation or downregulation between mice. By comparing mice on normal chow with those on a high-fat, high-cholesterol diet, Bennett's team found two such hubs were associated with atherosclerosis and TMAO. The researchers are now studying these genes. "These are relationships that we didn't know about before we looked at the data," Bennett said. "The data are doing the work for us here."

Determining co-expression hubs of genes with similar expression patterns helps identify genes that are important in disease. (Image courtesy of Brian J. Bennett)

Exploring new microbiota

Anne L. McCartney of the University of Reading is exploring how different populations of microbiota in the digestive system interact with diet and affect human health. Most research in this area has focused on the fecal, or gut, microbiota because samples are easy to obtain through noninvasive methods. But as McCartney argued, this approach is "short-sighted" because physiology changes along the digestive tract, and the majority of metabolic function takes place more proximally, toward the small intestine. Therefore, she is investigating the microbiota in the cecum, the first part of the colon.

McCartney and her collaborators have characterized cecal samples from 24 patients who underwent routine colonoscopy at King's College London. The team profiled the operational taxonomic units (OTUs) in the samples by pyrosequencing, and could identify patients as having either healthy guts or inflammatory bowel disease (IBD) based on these profiles. Culturing the samples has revealed 43 species thus far—a mixture of typical oral and fecal bacteria, suggesting that the cecum represents a transitional population between these two niches. The team is also exploring bacteriophages in cecal and fecal niches.

McCartney also returned to the topic of TMAO, the metabolite associated with atherosclerosis that Bennett studies. While TMAO is derived from choline-containing foods, it is also a dietary component itself, present in fish. McCartney's team found that TMAO enhanced the growth of certain bacteria, such as enterobacteria, and increased bacterial production of metabolites, including acetate, lactate, and trimethylamine (TMA). There were differences in metabolite production depending on whether a bacterial strain was isolated from cecal or fecal samples, suggesting differences in metabolic activity between the two niches.

Panel discussion

In a panel discussion and Q&A session moderated by Andrew Swick of Metagenics, Leibel expanded on the topic of leptin in weight loss. Muscle becomes more efficient in weight-reduced obese individuals because its use of myosin heavy chain isotypes changes. One proposed explanation is that leptin levels accompanying weight loss affect the hypothalamus, altering thyroid levels and the autonomic nervous system, and leading to the switch to the more energy-efficient myosin heavy chain isotype.

Leibel also talked about weight loss achieved through gastric bypass surgery, which is associated in about two-thirds of human studies with reduced resting energy expenditure. In mice, weight loss after this procedure is not associated with lowered energy expenditure (as diet-induced weight loss is); but, Leibel noted, "the jury is out on responses in humans." He pointed out that human studies are difficult to conduct because most people, eager to undergo surgery, do not want to go through a period of weight monitoring beforehand.

He next discussed how aging affects weight loss physiology, predicting that aging would promote rather than counteract weight gain due to sarcopenia, or age-related loss of muscle mass. However, he noted that many weight-loss studies have not included people older than 40–45 years old, so this question has not been directly studied.

It is difficult to find a single drug that would shift the leptin threshold and help maintain weight loss. "The system is so defended and so over-protected that by taking one or another element out of it you generally can't have a profound effect on body weight," Leibel said. Genome-wide association studies have associated at least 50 genes with obesity, and subtle expression and sequence differences in those genes are likely to affect the leptin threshold. In addition, differences in metabolite levels and neuronal connections (resulting from perinatal diet and other developmental factors) could affect brain activity. And the relative roles of genetics and neurobiology in physiology could differ among people.

One question concerned how epigenetics relates to systems genetics. Bennett pointed to recent studies of child and maternal health and paternal imprinting, which he predicted will change the way researchers think about information flowing from DNA to a phenotype. However, he noted that there are still many questions about which epigenetic tags (such as methyl groups) are important and which enzymes should be investigated.

When asked how to advise patients for weight-loss maintenance, Leibel described studies on resistance-training exercise, which makes muscles less efficient and thus could reverse some physiological changes associated with weight loss. In contrast, aerobic exercise enhances these changes by making muscles more efficient.

The yo-yo effect, the increase in the percentage of body fat that purportedly occurs when body mass increases and then drops back down, is popularly referred to as a reason diets fail. Leibel noted that his experience with people who have gained and lost body mass indicates the yo-yo effect is minimal. However, if weight cycling occurs before the leptin threshold is set, which presumably happens around puberty, the yo-yo effect could be more substantial during these years, as studies of adolescent gymnasts suggest. Epidemiological studies support the idea that people who are non-obese through adolescence will probably remain so later in life.

McCartney talked about the potential for the microbiota from the cecum and other regions of the colon to be used for transplant, as fecal samples are currently being used. Although there are disadvantages to taking samples from these more proximal regions—samples are harder to obtain and produce lower yield—bacteria from these niches could have different metabolic functions. "I think in 5 to 10 years we might be seeing a completely different story in relation to microbial transplant," perhaps including mixtures of pure bacterial cultures isolated from different gut regions, McCartney said. Swick added that there is now a heavier focus on the functional aspects of microbial transplants, such as whether bacteria are colonizing particular areas and having the desired functions.

Finally, McCartney returned the discussion to TMAO metabolism, explaining that much of the work in this area has looked at how dietary choline is converted first by gut bacteria and then by liver enzymes into TMAO. Her focus is on what happens to TMAO consumed through diet. Although her research indicates dietary TMAO is broken down by gut bacteria, it could also, she suggested, be adding to the body's TMAO stores. Bennett noted the paradoxical finding that fish intake is linked to reduced risk of heart disease, yet fish is the main dietary source of TMAO, which is associated with atherosclerosis.

Speakers:
Patricia L. Mabry, National Institutes of Health
Kevin D. Hall, National Institutes of Health
Christina A. Roberto, Harvard T.H. Chan School of Public Health
 
Moderator:
Liana Lianov, American College of Lifestyle Medicine

Highlights

• Systems science involves building models to predict the impact of a policy on health outcomes.
• The common dietary advice that reducing energy intake by 3500 kcal per week results in loss of one pound of body mass overestimates weight loss.
• People relapse soon after starting a diet but continue to lose weight for a while.
• Failure to adhere to diet probably plays a bigger role than lower energy expenditure in weight regain.
• Strategic science aims to influence policy decisions through targeted research.

Models to understand policy and physiology

Patricia L. Mabry of the National Institutes of Health (NIH) described how systems science methods can advance the study of complex problems such as obesity. These methods often involve creating simulations including computational, mathematical, agent-based, and dynamic models. As Mabry outlined, modeling can help both to explain phenomena and to make predictions about policy outcomes and trade-offs.

She described the Prevention Impacts Simulation Model (PRISM), which the U.S. Centers for Disease Control and Prevention developed in collaboration with the NIH, including the Office of Behavioral and Social Sciences Research (OBSSR), where Mabry is acting deputy director. Communities use the PRISM simulator to test the effects of various interventions on specific problems. In Austin, Texas, where the model debuted, researchers applied PRISM to study cardiovascular events, death, and disability. They defined known risk factors, including smoking and physical inactivity, and estimated the effect size of each based on published studies. Then they plugged into the model an intervention, such as a tobacco tax, to predict the effect it would have on smoking prevalence, downstream drivers of disease, and eventual cardiovascular outcomes.

Mabry recommended resources for learning about systems science methods, including a July 2014 theme issue on the topic in the American Journal of Public Health and a book titled Thinking in Systems: A Primer. She also recommended a listserv that she maintains through the OBSSR, to which she posts information about resources, conferences, and funding for research using systems science methods.

Kevin D. Hall of the NIH described insights from mathematical modeling of energy metabolism and body mass dynamics. He began by debunking commonly held medical advice about weight loss—that reducing energy intake by 3500 kilocalories (kcals) per week results in the loss of one pound of body mass. A model that he and his colleagues developed to capture the relationship between energy intake and body mass demonstrated that the 3500-rule overestimates weight loss. The rule also misleads, as Hall emphasized, both individual dieters and policy makers deciding which interventions to implement.

The model, developed based on diet studies by Rudolph Leibel and others, has numerous applications. In one example, Hall's team plugged in data from a study that reported weight changes and diet self-reports from participants. The model suggested that the decrease in energy expenditure that occurs as people lose weight, which Leibel discussed in his talk, accounts for only 20% of weight regain dieters experience. Explaining the rest was a relapse among dieters, who began consuming more calories soon after starting a diet. However, because calorie restriction has a delayed effects, people continued losing weight for a period after this relapse, a delay that could encourage dieters to cheat.

The widely used "3500 kcal per pound" rule overestimates the effect food policies such as soda taxes could have on weight loss. (Image courtesy of Kevin D. Hall)

Hall and his colleagues have also applied their model to quantify the calorie reduction required to maintain at least some weight loss. The results suggest that reducing energy intake by about 100 kcal a day from initial levels over several months can result a plateau effect after weight loss of about 10 pounds. The team also estimated that an increase of about 250 kcal in daily consumption per person explains the societal rise in obesity prevalence since the 1970s. In his final example, Hall discussed his team's work to improve predictions about the relationship between energy intake and body mass in children.

Exploring food choices and eating behaviors

Calorie labeling on menus may be more effective if it includes a statement about the recommended daily energy intake. (Image courtesy of Christina A. Roberto)

Christina A. Roberto of the Harvard T.H. Chan School of Public Health takes an approach to research she calls strategic science. Her team aims to design studies that could impact public health policy. Strategic science involves communication between scientists and "change agents," such as regulators, policy makers, and advocacy groups. These relationships help scientists understand, when designing studies, which research questions are important for policy decisions. The relationships also facilitate wider distribution of their findings.

Roberto described her team's study on menu labels containing calorie information, which compared three menus: one without calorie information, one with calorie information about each item, and one with that information and a reference statement relating the recommended energy intake of 2000 kcal daily. Although, compared with the unlabeled menu, both labeled menus led people to order and eat fewer calories, those whose menu lacked a reference point made up the difference by consuming more at home later. In keeping with her strategic science goals, Roberto shared this information with stakeholders at the FDA, which requires that chain restaurants have menu labels, in compliance with the Affordable Care Act.

Panel discussion

To begin the second panel discussion, keynote speaker Frank Hu asked Roberto whether research like hers could support the passage of legislation, such as a bill proposed in California that would introduce a warning label stating that sugary drinks contribute to obesity, diabetes, and tooth decay. (The state's Assembly Health Committee rejected the bill in June 2014; there are currently no U.S. laws on warning labels for sweetened drinks.) Roberto replied that science is only one aspect of food policy making. "I don't think [a] study comes out and then warning labels happen," she said, "but I think it's a small part of the discussion." She added that part of the argument against the bill is that there is not enough evidence suggesting it would change behavior.

Roberto fielded questions about the design and interpretation of her research. Elaborating on the definition of a change agent, she said people in many settings, including institutions and community-based groups, could fall into the category. She discussed the inherent limitations of intent-to-purchase studies (including research on warning labels): people tend to be more willing to impose sacrifices on their future than on their present selves. To gauge whether participants' intentions are predictive of actual behavior, Roberto's team takes into account past purchasing habits.

In response to an inquiry about the long-term effects of warning labels—and whether people would become desensitized to them, as some are to such labels on cigarette packaging—Roberto described labels as one of several solutions to influence eating behavior. Labels could inform consumers who might not be aware that some beverages, such as sports drinks, are high in sugar; but labels could be less effective among people who are committed to the lifestyle choice, just as cigarette warning labels are less effective among long-term smokers. Research suggests that menu labeling might be less effective for low-socioeconomic and other groups, so it is important to study impact on various populations. In her studies on menu labeling, Roberto explained, there were too few people (300 participants) to ascertain whether obese and non-obese participants behaved differently.

Hall discussed experimental data and model predictions on weight loss and adherence to diet. He described studies using doubly labeled water (used to measure metabolic rate), which show that people start slipping on diets within the first month but continue to lose weight for a longer period. People are not consciously cheating on diets; changes in the brain affect emotions and hunger and ingrained eating behaviors contribute to poor diet adherence. "The adherence issue and the hunger and the food intake are primary, and the metabolism slowing down is secondary" in explaining weight regain, he said.

In addition, Hall argued that the benefits of resistance training do not fully counteract the decrease in energy expenditure that occurs as people lose weight. His team showed that muscle is not highly metabolically active, and that gaining 1 kilogram (about 2 pounds) of muscle, which is a lot, only increases energy expenditure by 13 kcal per day.

Hall also discussed the implications of his estimate that a 250 kcal per day increase in average energy intake explains the increase in average adult weight since the 1970s. It does not mean, he pointed out, that if everyone ate 250 fewer calories daily the obesity epidemic would be reversed. "There's this very long time horizon" across several decades, he said. However, preventing excess energy intake among younger people could help change the trend. It is nevertheless meaningless to advise cutting energy intake by 250 kcal per day, Hall explained, because it is difficult for people to accurately gauge their consumption.

It is challenging to build models of eating behavior and energy expenditure that include variables such as sleep and lifestyle habits. Measuring eating behavior is itself difficult. Hall's team is interested in how both the ability to form new eating habits and the neurobiological basis of habit formation differ among people. His team is gathering data to build models of these phenomena.

Are eating habits important in normal-weight people? Hu concluded by explaining that, even in this group, both the quality and the quantity of food are important. The foremost goal is to prevent weight gain. Consuming a low-quality diet puts even normal-weight people on a trajectory for weight gain over several years.

Could targeting leptin and other signaling molecules help people maintain normal physiology after weight loss?

How do the microbiota of the cecum and associated bacteriophages affect metabolic function?

What are the genetic factors that affect levels of TMAO, a metabolite associated with atherosclerosis?

Can metabolites in blood and urine be used as biomarkers to predict risk of chronic diseases?

How can methods be improved to measure the many factors that contribute to body mass, such as diet, lifestyle, and pollutants?

How does the timing and combination of nutrient intake during development affect metabolic health?

How do factors such as health and socioeconomic status influence the effectiveness of calorie labeling and other interventions?