Preventive Aspects of Early Nutrition Nestlé Nutrition Institute Workshop Series

Vol. 85 Preventive Aspects of Early Nutrition

Editors

Mary S. Fewtrell London, UK Ferdinand Haschke Salzburg, Austria Susan L. Prescott Perth, Australia Nestec Ltd., 55 Avenue Nestlé, CH–1800 Vevey (Switzerland) S. Karger AG, P.O. Box, CH–4009 Basel (Switzerland) www.karger.com

Library of Congress Cataloging-in-Publication Data

Names: Fewtrell, Mary S., editor. | Haschke, F., editor. | Prescott, Susan L., editor. Title: Preventive aspects of early nutrition / editors, Mary S. Fewtrell, Ferdinand Haschke, Susan L. Prescott. Description: Basel ; New York : Karger, [2016] | Series: Nestlé Nutrition Institute workshop series, ISSN 1664-2147 ; vol. 85 | Includes bibliographical references and index. Identifiers: LCCN 2015046033| ISBN 9783318056426 (hard cover : alk. paper) | ISBN 9783318056433 (electronic version) Subjects: | MESH: Child Nutritional Physiological Phenomena | Infant Nutritional Physiological Phenomena | Pediatric Obesity--prevention & control | Hypersensitivity--prevention & control | Congresses Classification: LCC RJ206 | NLM WS 130 | DDC 613.2083--dc23 LC record available at http://lccn.loc.gov/2015046033

The material contained in this volume was submitted as previously unpublished material, except in the instances in which credit has been given to the source from which some of the illustrative material was derived. Great care has been taken to maintain the accuracy of the information contained in the volume. However, neither Nestec Ltd. nor S. Karger AG can be held responsible for errors or for any consequences arising from the use of the information contained herein. © 2016 Nestec Ltd., Vevey (Switzerland) and S. Karger AG, Basel (Switzerland). All rights reserved. This book is protected by copyright. No part of it may be reproduced, stored in a retrieval system, or transmitted, in any form or by any means, electronic, mechanical, photocopying, or recording, or otherwise, without the written permission of the publisher.

Printed on acid-free and non-aging paper (ISO 9706) ISBN 978–3–318–05642–6 e-ISBN 978–3–318–05643–3 ISSN 1664–2147 e-ISSN 1664–2155

Basel · Freiburg · Paris · London · New York · Chennai · New Delhi · Bangkok · Beijing · Shanghai · Tokyo · Kuala Lumpur · Singapore · Sydney Contents

VII Preface X Foreword XIII Contributors

Allergy

1 Early Nutrition as a Major Determinant of ‘Immune Health’: Implications for Allergy, Obesity and Other Noncommunicable Diseases Prescott, S.L. (Australia/Western Australia) 19 The Future of Infant and Young Children’s Food: Food Supply/ Manufacturing and Human Health Challenges in the 21st Century Venter, C. (UK/USA); Maslin, K. (UK) 29 Infant Feeding: Foods, Nutrients and Dietary Strategies to Prevent Allergy Beyer, K. (Germany) 35 Using Food and Nutritional Strategies to Induce Tolerance in Food-Allergic Children Nowak-Węgrzyn, A. (USA) 55 Summary on Allergy Prescott, S.L. (Australia)

Obesity Prevention

59 Interrupting Intergenerational Cycles of Maternal Obesity Gillman, M.W. (USA) 71 Development, Epigenetics and Metabolic Programming Godfrey, K.M.; Costello, P.M.; Lillycrop, K.A. (UK)

V 81 Endocrine and Metabolic Biomarkers Predicting Early Childhood Obesity Risk Socha, P. (Poland); Hellmuth, C. (Germany); Gruszfeld, D. (Poland); Demmelmair, H.; Rzehak, P.; Grote, V.; Weber, M. (Germany); Escribano, J.; Closa-Monasterolo, R. (Spain); Dain, E.; Langhendries, J.-P. (Belgium); Riva, E.; Verduci, E. (Italy); Koletzko, B. (Germany) for the European Childhood Obesity Trial Study Group 89 Effects of Early Nutrition on the Infant Metabolome Hellmuth, C.; Uhl, O.; Kirchberg, F.F.; Grote, V.; Weber, M.; Rzehak, P. (Germany); Carlier, C. (Belgium); Ferre, N. (Spain); Verduci, E. (Italy); Gruszfeld, D.; Socha, P. (Poland); Koletzko, B. (Germany) for the European Childhood Obesity Trial Study Group 101 Postnatal High Protein Intake Can Contribute to Accelerated Weight Gain of Infants and Increased Obesity Risk Haschke, F. (Austria); Grathwohl, D.; Detzel, P.; Steenhout, P.; Wagemans, N.; Erdmann, P. (Switzerland) 111 Summary – Early Nutrition and Obesity Prevention Haschke, F. (Austria)

Complementary Feeding: Taste, Eating Behavior and Later Health

113 Can Optimal Complementary Feeding Improve Later Health and Development? Fewtrell, M.S. (UK) 125 Learning to Eat: Behavioral and Psychological Aspects Birch, L.L. (USA) 135 The Development of Flavor Perception and Acceptance: The Roles of Nature and Nurture Forestell, C.A. (USA) 145 Dietary Patterns during Complementary Feeding and Later Outcomes Emmett, P.M. (UK) 155 Nature and Nurture in Early Feeding Behavior Cooke, L.; Llewellyn, C. (UK) 167 Summary on Complementary Feeding: Taste, Eating Behavior and Later Health Fewtrell, M.S. (UK)

169 Subject Index

For more information on related publications, please consult the NNI website: www.nestlenutrition-institute.org

VI Contents Preface

Early-life nutrition is arguably the most critical determinant of future health. It provides the energy and critical building blocks for all development, determines our future tastes and eating behaviors, and supplies the crucial substrate for the establishment of a healthy microbiome, now recognized as one of the most important determinants of immune and metabolic health. Our diets are also an increasing source of exposure to adverse elements in the environment, including contaminants and pollutants in our food and water supply, which have greater potential to affect human health early in life than at any other stage. Optimizing early nutrition can have long-term benefits for long-term biological reserve and resilience, through effects on developing structure and physiological responses. Greater attention to this is key to maximizing human potential. Indeed, improved early-life nutrition has been a major element in dramatic improvements in life expectancy over the 20th century, particularly in high-income countries. However, access to healthy food remains an important factor in the wide global disparities in human health and longevity that still remain. Across the threshold of the 21st century, new nutritional challenges are emerging, with overnutrition and obesity now the most significant threats to the future of human health. Rising rates of inflammatory and metabolic diseases in children now underscore the importance of understanding the impact of the early environment on immune and metabolic health. For the first time in modern history, the current generation is expected to have a shorter life expectancy than their parents, simply because of obesity and the associated increased risk of noncommunicable diseases (NCDs), including diabetes, heart disease, mental ill health, some cancers, musculoskeletal disorders and immune diseases. In particular, the epidemic rise in very-early-onset NCDs, such as infant allergy, provide clear evidence of immune dysregulation and the rising early predisposition to inflammation. Diet-sensitive pathways are likely to be crucial in understanding how early- life conditions influence the finely balanced development of immune and meta-

VII bolic responses. In particular, diet is at the center of the emerging epigenetic paradigms that may underpin the rise in several NCDs. While epigenetic mechanisms provide a potential explanation of how nutritional exposures can affect fetal gene expression and subsequent disease risk, other diet-induced tissue compositional changes may also contribute directly to altered immune and metabolic function, e.g. through diet-induced changes in the microbiome. A better understanding of nutritional programming of immune health, nutritional epigenetics and the biological processes sensitive to nutritional exposure in early life may lead to dietary strategies that provide more optimal conditions during early programming, and reduce the burden of many metabolic and inflammatory diseases. While there is little doubt that changing dietary patterns are at the core of this modern NCD-driven health crisis, the importance of addressing this intervention in early life is still often underestimated. However, the increasing burden of childhood disease underscores that without early intervention there is very little hope of averting current trends. These concepts are supported by growing evidence that a ‘life course’ approach from the first moments of life will be more effective in reducing the long-term disease burden. With this philosophy, there is an increased focus on the next generation of parents and promotion of their health before conception. These efforts must be firmly grounded through promoting a far greater public awareness of the long-term implications of dietary choices in pregnancy, lactation and infancy. Advocacy for early-life nutrition needs to be matched with sound evidence and consistent advice for healthcare professionals, parents and the wider community. At the moment, there is still inconsistency around even basic advice such as the optimal timing of introducing complimentary foods to infants. This needs to be addressed relatively urgently. As we understand more about gene-environment interactions, it is also increasingly likely that we may need to tailor this advice according to the genetic background and environmental context. The first two sections of this workshop consider preventive aspects of early nutrition in relation to the development of allergy and obesity, including the role of the microbiome, the use of food and nutritional strategies to induce tolerance and reduce the risk of obesity, and genetic and epigenetic aspects of metabolic programming. In the third section of the workshop, the role of complementary feeding is discussed in relation to later health outcomes, including consideration of the determinants of flavor and food preferences and eating behaviors which may shape subsequent food habits and exposures. While the burden of modern diseases is complex and diverse, it is also clear that many NCDs share the same risk factors, with early life nutrition as a central

VIII Fewtrell · Haschke · Prescott common element. This underscores the need to take a more collaborative integrated cross-discipline, cross-sectoral approach to solving these problems. All of these issues need to be framed in the context of the wider environmental impact and sustainability of our food choices. This raises important questions around food diversity for nutrients, taste and deriving a balanced microbiome. It also raises the challenge of balancing the demands of modern life and the need of convenience food against the potential risks that this poses. As we investigate nutritional strategies as a critical avenue of improving health and life expectancy in the 21st century, we must consider ‘the future of our food’ in the wider context of the other social and economic challenges facing the world today. Mary S. Fewtrell Ferdinand Haschke Susan L. Prescott

Preface IX Foreword

The program was based on well-established and documented evidence that early-life events, including nutrition, play a powerful role in programming a person’s development, metabolism and health for the future. The implications of early nutrition programming are significant – particularly when it comes to the risk of suffering cardiovascular diseases, diabetes and obesity, as well as effects on immune function, allergy risk, and cognitive and behavioral outcomes later in life. It was concluded that by ensuring appropriate nutrition early in life, we have enormous potential to improve the health of future generations. The workshop program had three sessions. The first, led by Prof. Susan L. Prescott, addressed a fundamental topic on the role of nutrition in the determination of ‘immune’ health and implications for noncommunicable diseases. An excellent group of speakers took part in this session and explored the new opportunity to transform human health by diet, the role of gut microbiota in immune status, and strategies and interventions for the prevention and management of food allergies in children. Prof. Ferdinand Haschke chaired the session on obesity prevention. During these discussions, experts in epigenetics and metabolic programming shared scientific evidence on the use of biomarkers for predicting the risk of early obesity and discussed potential strategies for interrupting intergenerational cycles of obesity during pregnancy, early infancy and childhood. The final session, chaired by Prof. Mary S. Fewtrell, looked into complementary feeding, with a focus on the importance of optimal complementary feeding in short- and long-term health, and how its ‘programming’ can impact behavioral and psychological aspects, as well food preferences, in later life. We would like to thank the three Chairpersons Mary S. Fewtrell, Susan L. Prescott and Ferdinand Haschke for putting the excellent scientific program together.

X We would also like to thank all the speakers and scientific experts in the au- dience, who actively contributed to the workshop content and scientific discussions and enhanced the learning experience. Finally, we thank Mike Eddi, Liz Greenstreet and their teams in the UK for their valuable logistical support.

Dr. Jose M. Saavedra, MD Dr. Natalia Wagemans, MD, PhD Chairman Global Head Nestlé Nutrition Institute Nestlé Nutrition Institute Vevey, Switzerland Vevey, Switzerland

Foreword XI 85th Nestlé Nutrition Institute Workshop London, November 16–19, 2014 Contributors

Chairpersons & Speakers Prof. Kirsten Beyer Prof. Mary S. Fewtrell Augustenburger Platz 1 Childhood Nutrition Research Centre 13353 Berlin UCL Institute of Child Health Germany 30 Guilford Street E-Mail [email protected] London WC1N 1EH UK Prof. Leann L. Birch E-Mail [email protected] Department of Foods and Nutrition The University of Georgia Assoc. Prof. Catherine A. Forestell 176 Dawson Hall Department of Psychology Athens, GA 30602-3632 College of William & Mary USA Tyler Hall, Room 210A E-Mail [email protected] PO Box 8795 Williamsburg, VA 23185-8795 Dr. Lucy Cooke USA Health Behaviour Research Centre E-Mail [email protected] Department of Epidemiology and Public Health Prof. Matthew W. Gillman University College London Department of Population Medicine Gower Street Harvard Medical School/Harvard Pilgrim London WC1E 6BT Health Care Institute UK 133 Brookline Avenue, 6th floor E-Mail [email protected] Boston, MA 02215 USA Dr. Pauline M. Emmett E-Mail [email protected]. Centre for Child and Adolescent Health edu School of Social and Community Medicine Prof. Keith Godfrey University of Bristol MRC Lifecourse Epidemiology Unit Oakfield House, Oakfield Grove University of Southampton Bristol BS8 2BN Southampton General Hospital UK Mailpoint 95 E-Mail [email protected] Tremona Road Southampton SO16 6YD UK E-Mail [email protected]

XIII Prof. Ferdinand Haschke Prof. Piotr Socha Department of Pediatrics Department of Gastroenterology, Paracelsus Medical University Hepatology and Nutrition Disorders Müllner Hauptstrasse 48 Children’s Memorial Health Institute 5020 Salzburg Al. Dzieci Polskich 20 Austria 04-730 Warsaw E-Mail [email protected] Poland E-Mail [email protected] Dr. Christian Hellmuth Division of Metabolic and Nutritional Dr. Carina Venter Medicine Research Associate/Dietitian Dr. von Hauner Children’s Hospital Division of Allergy and Immunology University of Munich Medical Center Cincinnati Children’s Hospital Medical Lindwurmstrasse 4 Center 80337 Munich 3333 Burnet Avenue, MLC 7028 Germany Cincinnati, OH 45229 E-Mail christian.hellmuth@ USA med.uni-muenchen.de E-Mail [email protected] Assoc. Prof. Anna Nowak-Węgrzyn Department of Pediatrics Participants Icahn School of Medicine at Mount Sinai Jaffe Food Allergy Institute Margarita Thanhäuser/Austria One Gustave L. Levy Place, Box 1198 Martin Wald/Austria New York, NY 10029 Sergey Ukraintsev/Belarus USA Ana Paula Castro/Brazil E-Mail [email protected] Christiane Leite/Brazil Haiqi Li/China Prof. Susan L. Prescott Jie Mi/China School of Paediatrics and Child Health Xiaoyang Sheng/China The University of Western Australia Xiu Xu/China (M561) An an Yuan/China Princess Margaret Hospital Qianqian Zhu/China PO Box D184 Bertha Patricia Calderón Ortiz/Colombia Perth, WA 6001 Eliana Eugenia Cantillo Tinoco/Colombia Australia Silvana Dadán Muñoz/Colombia E-Mail [email protected] Wilson Daza Carreño/Colombia Janeth Jaramillo Cabal/Colombia Prof. Jose M. Saavedra Nicolás Ignacio Ramos Rodriguez/ Department of Pediatrics, Colombia Gastroenterology and Nutrition Alina Restrepo Velez/Colombia Johns Hopkins University School of José María Solano Suárez/Colombia Medicine Jiri Nevoral/Czech Republic Brady 320 Camilla Trab Damsgaard/Denmark 600 N. Wolfe Street Abdoul Gadirou Bah/France Baltimore, MD 21287-2631 Farid Boubred/France USA Laurence Foix l’Hélias/France E-Mail [email protected] Déo-gratias Rugemintwaza/France Georg Frey/Germany Mathilde Kersting/Germany Nina Ludwig/Germany Mike Possner/Germany Judit Cholnoky/Hungary Gárdos László/Hungary

XIV Contributors Bhaskar Raju Balagopal/India Laura Taylor/Switzerland Sridhar Ganapathy/India Sabine von Manteuffel/Switzerland Sanjeev Kumar Ganguly/India Maria Airainer/UK Sanjay Niranjan/India Michael Edde/UK Binayak Roy/India Sian Evans/UK Santosh Theodore Soans/India Gavin Fergie/UK Suresh Kumar Surapaneni/India Jackie Gaventa/UK Nishant Wadhwa/India Liz Greenstreet/UK Muhammad Azam/Ireland Ellie Grove/UK Nuala Collins/Ireland Kemi Ibinola/UK Dan O’Callaghan/Ireland Julie Lanigan/UK Pamela O’Connor/Ireland Anirban Maitra/UK Eleanor Power/Ireland Kate Maslin/UK Ita Saul/Ireland Caroline McCormack/UK Massimo Agosti/Italy Judy Moore/UK Francesco Tandoi/Italy Dilip Nathan/UK Adib Moukarzel/Lebanon Sue O’Neil/UK Anna Rybak/Poland Vanessa Shaw/UK Antonio Guerra/Portugal Anne Sidnell/UK Elena Kornienko/Russia Atul Singhal/UK Leyla Namazova-Baranova/Russia Sara Stanner/UK Xavier Dorca/Spain Charlotte Stirling-Reed/UK José Manuel Moreno/Spain Emma Sutton/UK Christian Braegger/Switzerland Paul Turner/UK Liya Denney/Switzerland David Tuthill/UK Irma Silva Zolezzi/Switzerland Leah Wood/UK Evelyn Spivey-Krobath/Switzerland Rachel Wood/UK Krisztina Suhajda/Switzerland Barbara MacFarland/USA

Contributors XV

Allergy

Fewtrell MS, Haschke F, Prescott SL (eds): Preventive Aspects of Early Nutrition. Nestlé Nutr Inst Workshop Ser, vol 85, pp 1–17, (DOI: 10.1159/000439477) Nestec Ltd., Vevey/S. Karger AG., Basel, © 2016

Early Nutrition as a Major Determinant of ‘Immune Health’: Implications for Allergy, Obesity and Other Noncommunicable Diseases

Susan L. Prescott School of Paediatrics and Child Health, University of Western Australia (UWA), Telethon Institute for Child Health Research (UWA), and Developmental Origins of Health and Disease (DOHaD) Consortium, Perth, WA , Australia; International Inflammation (in-FLAME) Network, World Universities Network (WUN), Perth, Western Australia

Abstract Early-life nutritional exposures are significant determinants of the development and future health of all organ systems. The dramatic rise in infant immune diseases, most notably allergy, indicates the specific vulnerability of the immune system to early environmental changes. Dietary changes are at the center of the emerging epigenetic paradigms that underpin the rise in many modern inflammatory and metabolic diseases. There is growing evidence that exposures in pregnancy and the early postnatal period can modify gene expression and disease susceptibility. Although modern dietary changes are complex and involve changing patterns of many nutrients, there is also interest in the developmental effects of specific nutrients. Oligosaccharides (soluble fiber), antioxidants, polyunsaturated fatty acids, folate and other vitamins have documented effects on immune function as well as metabolism. Some have also been implicated in modified risk of allergic diseases in observational studies. Intervention studies are largely limited to trials with polyunsaturated fatty acids and oligosaccharides, showing preliminary but yet unconfirmed benefits in allergy prevention. Understanding how environmental influences disrupt the finely balanced development of immune and metabolic programming is of critical importance. Diet-sensitive pathways are likely to be crucial in these processes. While an epigenetic mechanism provides a strong explanation of how nutritional exposures can affect fetal gene expression and subsequent disease risk, other diet-induced tissue compositional changes may also contribute directly to altered immune and metabolic function – including diet-induced changes in the microbiome. A better understanding of nutritional programming of immune health, nutritional epigenetics and the biological processes sensitive to nutritional exposures early in life may lead to dietary strategies that provide more tolerogenic conditions during early immune programming and reduce the burden of many inflammatory diseases – not just allergy. © 2016 Nestec Ltd., Vevey/S. Karger AG, Basel

Introduction

The current global health crisis and the pandemic of noncommunicable diseases (NCDs) are clearly rooted in complex modern societal and environmental changes, many of which have effects on developing immune and metabolic responses in early life. Early-life nutrition is one of the most significant determinants of the development, function and future health of all organ systems, and a major factor in both the risk and the prevention of NCDs. Dietary changes are at the center of the emerging epigenetic paradigms that underpin the rise in many modern inflammatory and metabolic diseases. There is growing evidence that exposures in pregnancy and the early postnatal period can modify gene expression and disease susceptibility. The low-grade inflammation that characterizes virtually all NCDs also suggests a central role of the immune system in the risk and pathogenesis of these conditions. In particular, the dramatic rise in infant immune diseases, most notably allergy, indicates the specific vulnerability of the immune system to early environmental changes. Understanding how environmental influences disrupt the finely balanced development of immune and metabolic programming is of critical importance. Diet-sensitive pathways are likely to be crucial in these processes. While an epigenetic mechanism provides a strong explanation of how nutritional exposures can affect fetal gene expression and subsequent disease risk, other diet-induced tissue compositional changes may also contribute directly to altered immune and metabolic function – including diet-induced changes in the microbiome. There is now evidence that the high-fat, low-fiber ‘western’ diet also adversely changes biodiversity of the gut microbiota, with recognized implications for immune development and metabolic function. Not only is this implicated in the increased risk of allergic disease, but associated effects on gut barrier function, systemic antigenic load and low-grade endotoxemia are also driving factors for insulin resistance, obesity and diabetes. This provides new perspectives on how modern dietary patterns may be increasing the risk of both immune and metabolic diseases. A better understanding of nutritional programming of immune health, nutritional epigenetics and the biological processes sensitive to nutritional exposures early in life may lead

2 Prescott to dietary strategies that provide more optimal conditions during early immune and metabolic programming, and reduce the burden of obesity and many inflammatory diseases – not just allergy.

Potent Effects of the Early Environment on Immune Health

As with other NCDs, genetic factors influence susceptibility but cannot explain the surge in allergic diseases in recent generations, and environmental factors must play a role. Other immune diseases have had similar increases during the same period [1]. Cleaner environments and declining exposure to infectious agents have been one of the leading explanations for this dual increase [1, 2] , but there are likely to be a number of other factors including modern diets and pollutants [3] . Allergic diseases are arguably the most common NCDs to appear in the 1st year of life leaving no doubt about the early vulnerability of the developing immune system. As a measure of immune health, they are, therefore, a useful early barometer of environmental impact, and an early measure of effectiveness of any intervention that we might try to prevent disease.

New Insights into Why Allergic Responses Evolved

Type 2 allergic responses are not inherently pathological and only cause disease when excessive or misdirected. Acute IgE-mediated responses appear to have evolved to protect against a broad and diverse range of environmental ir- ritants, toxins, venoms, parasites (such as mosquitoes and ticks) and noxious environmental chemicals (xenobiotics), including naturally occurring chemical compounds in plants (phytochemicals) [4] . This allows us to sense these threats at very low levels in the environment and react with rapid histamine reactions induced by IgE antibodies: sneezing, tearing, coughing, vomiting, diarrhea and itching, which serve to expel or remove these threats – helping our ancient ancestors survive. Moreover, these unpleasant symptoms also change behavior , conditioning avoidance of the noxious threat in the future [4] . This explains both the symptoms of the allergic response, the urgency of response and the types of allergen triggers. Most ‘modern’ allergens are chemi- cally related to the ancient environmental threats the type 2 responses evolved to react to, including plants (pollens and food allergens), mites, venoms, antibiotics (from molds), animals and fungi. This does not explain, however, the

Early Nutrition and Immune Health 3 dramatic increase in the tendency for misdirected allergic responses in the last few decades. The answer appears to lie with the lifestyle changes we have experienced in a relatively short period of time, and how our evolutionary training may lead to maladaptive responses under these modern conditions.

Evolutionary Adaptation to Overabundance – Close Links between Immunity, Body Weight and Inflammation

The mechanisms that have evolved to allow adaptation to variations in food and energy supply may also underpin the rising propensity for both immune and metabolic diseases in the modern context. Modern humans show all of the hall- marks of animals living in captivity. It is no coincidence that we are increasingly prone to both immune diseases and obesity-associated NCDs, and there is increasing evidence that these processes are linked. In times of scarcity, organisms live longer if they conserve energy and ‘tolerate’ nonlethal threats. But in more favorable conditions, there is greater benefit in mounting a strong immune response. We can see these dramatic shifts in metabolic and immune function in animals that undergo significant environmental changes. The Siberian hamster, like many seasonally breeding ro- dents, has evolved complex adaptations to maximize survival and reproductive success. To cope with winter’s scarcity and cold, these small animals reduce energy-demanding activities that are not essential for immediate survival [5– 7] . Mounting a strong immune response requires energy [8–10] . Under winter-like conditions, animals not only reduce their body mass and adipose tissue mass [7] , they also show significant changes in immune function, reducing their inflammatory responses [11] . In summer-like conditions, Siberian hamsters have significantly higher levels of circulating inflammatory cytokines such as interleukin (IL)-6 and produce higher fevers [11] . Antigen-presenting cells also show significantly increased activity in abundant conditions [5] . These findings are consistent with the immune suppression seen in human malnutrition [12, 13] and the increased inflammatory responses seen in obesity [14] . There are complex, intricate relationships between metabolism and immune function [14, 15]. The same hormones that regulate appetite, fat storage and metabolism also regulate immune function. Leptin is one of the main hormones involved in fat metabolism, and much higher levels (around five times) are seen in obesity [16] . Under conditions of food deprivation and reduced body fat mass, lower leptin levels reduce metabolic expenditure to conserve energy [15] .

4 Prescott In the Siberian hamster, lower leptin levels play a central role in suppressing inflammatory responses during ‘winter’ conditions – an effect that can be reversed artificially by infusing leptin [6] . Leptin enhances both innate and adaptive immune responses [15] , and is part of the IL-6 family of inflammatory cytokines. It increases the ability of innate immune cells to engulf and kill bacteria [17, 18]. It also stimulates adaptive immune responses by inducing proliferation and cytokine secretion by T cells [19, 20] . Leptin has been linked to an increased risk of both allergy [21] and autoimmunity [22, 23]. Higher leptin levels are associated with other ‘risk factors’, including psychological stress [24] and physical inactivity [25, 26] , independent of body weight. Insulin is another hormone important for both energy balance and immune function. It enhances uptake of nutrients, increasing cellular metabolism, energy requirements and protein synthesis. Under conditions of high blood sugar, increased insulin secretion promotes the production of inflammatory cytokines and T-cell activation [27] . Even in the early stages of overnutrition, these hormonal changes appear to contribute to changes in immune function and increased inflammation. As adipose tissue accumulates, it expresses and releases increasing levels of inflammatory cytokines such as IL-6, inducing low-grade systemic inflammation [28] , measurable as elevated baseline C-reactive protein (CRP). In early childhood, this is associated with allergic sensitization, in particular to foods [29] . Allergic IgE antibodies are higher among overweight children, although the exact nature of this relationship is still being explored [29] . This nexus between metabolism and the immune system clearly occurs at many levels: hormonal interactions, sensing of nutrients by immune cells and in the gut microbiota [14] .

Early Origins of Inflammation

There are significant differences in the baseline level of systemic inflammation (baseline CRP levels) in populations around the world. In developing regions of Asia and South America where lifestyle conditions are still quite traditional, CRP is actually significantly lower than in high-income regions such as the United States [30, 31]. This suggests that affluence is a risk for the kind of chronic ‘silent’ low-grade inflammation that leads to many NCDs. A number of studies have now shown that people with elevated CRP level at baseline are more likely to develop cardiovascular disease [32] and type 2 diabetes [33] , and have higher ‘all-cause’ mortality [34] than those with lower levels of this

Early Nutrition and Immune Health 5 inflammatory marker. Based on these studies, there has been some consensus that a baseline CRP >3 mg/l can identify individuals at high risk of cardiovascular disease [35] . About one third of adults in high-income regions of Europe and North America would fit into this category. Indeed, there is increasing evidence that our propensity for inflammation in adulthood (higher baseline CRP level) is determined by early-life conditions, including nutrition and more hygienic environmental conditions [36] . Many of these effects begin in pregnancy when immune responses also begin to develop [37, 38], and many are also implicated in later-onset inflammatory NCDs [39] . Low birth weight, as an indicator of a suboptimal prenatal nutrition, has been linked with a higher CRP level later in life in a number of studies [40–43] . Recent studies show that early poverty is also associated with higher CRP [44] , and exaggerated inflammatory responses in adolescence or adulthood [45, 46] . Finally, the protective effects of breastfeeding against obesity [47] and cardiovascular risk [48] may be the result of both metabolic benefits and anti-inflammatory effects. Thus, a range of early microbial, nutritional and toxic exposures appears to ‘program’ or ‘condition’ the level and patterns of inflammatory responses as we grow and age. Exactly how this manifests in each of us will depend on our ongoing environment and our genetic predisposition.

The Epidemics of Allergy and Obesity: A Two-Way Street?

While many NCDs are not manifest until later in life, allergic diseases appear within the first months of life [49] . This is a clear indication that the developing immune system is exquisitely sensitive to modern environmental pressures. It is also consistent with mounting evidence that the effects of environmental risk factors on immune development begin in utero [50] . These early effects could have additional longer-term implications for this new generation showing an increasing propensity to early-onset inflammatory diseases. Medicine is predisposed to operating in specialities and silos, and so the connections between allergic diseases and other NCDs have not been extensively recognized or explored until fairly recently. Several studies suggest that childhood obesity increases the risk of asthma [51, 52] and food allergy [29] . There has been interest in how the metabolic changes in obesity may contribute to the risk of allergy and inflammation in the airways. But, it must also be remembered that the effects of the obesity epidemic actually begin before birth.

6 Prescott Maternal obesity in pregnancy is a source of chronic low-grade inflammation for the developing fetus, with higher levels of inflammatory cytokines in the circulation [53] and in the placenta [54] . There is some preliminary evidence that children of overweight mothers have an increased risk of asthma and lung disease [55, 56], and that the levels of adipokines in cord blood may influence the risk of wheezing [57] . Again, more research is needed here. Once allergy has developed, obesity may exacerbate the symptoms of asthma and other allergic diseases. In sensitized animals, higher leptin levels, as seen in obesity, induce exaggerated allergic IgE antibody responses and increased airway inflammation [58]. Allergy is a systemic disease associated with systemic release of cytokines and chemokines, and with distal recruitment of inflammatory progenitors into the circulation from the bone marrow [59] . Low-grade systemic inflammation has now been so clearly linked with risk for metabolic dysregulation and vascular disease [60] . While there are some associations between cardiovascular and allergic diseases in later life [61, 62], the long-term multisystem implications of allergic inflammation in earlier life have not yet been determined, particularly in the current high-prevalence generation yet to reach maturity. Leptin levels are higher in allergic than nonallergic individuals, even after allowing for differences in body weight [21]. This suggests that allergic inflammation may have additional effects on leptin levels and, consequently, metabolism; in other words, the relationship between allergy and obesity might operate in both directions. Animal studies support this, showing serum leptin is increased during allergic reactions in the airways [58]. Allergen exposure not only induces inflammation in the airways, but also causes changes in the composition of adipose tissue and the levels of other adipokines in the blood- stream [63]. This is consistent with what is seen in human allergy. Leptin levels are very much higher in pollen-allergic individuals when exposed to pollen [64] . So, leptin may increase the systemic inflammatory response resulting from allergen exposure, which in turn causes a greater release of leptin from fat stores. Some of the prime suspects in the allergy epidemic are now emerging as likely culprits in the obesity epidemic. The ‘hygiene hypothesis’ has long been a major contender in the rise of allergy, with cleaner environments reducing the microbial diversity of our intestinal bacteria. Our modern (low-fiber, high-fat) dietary patterns add to this reduced biodiversity and allergy risk. For these reasons, prebiotic fiber and probiotic bacteria have both been used to prevent allergy in early life [65] . These potential treatments and prevention strategies are now being considered in preventing obesity to address the metabolic effects of the same risk factors.

Early Nutrition and Immune Health 7 Early-Life Nutritional Strategies to Promote Immune and Metabolic Health

Early life is an important opportunity to maximize future health. In particular, strategies that promote immune and metabolic health will have implications for both the development and future function of all organs and body systems. Pri- or to becoming pregnant, there should be greater attention on achieving and maintaining a healthy body; maternal obesity is a major risk factor for infertil- ity, adverse pregnancy outcomes and obesity and other NCDs in offspring in later life. Throughout pregnancy, optimal nutrition with a nutritionally balanced diet, coupled with daily physical activity, will help achieve a steady healthy pattern of weight gain. Gaining too much or too little weight in pregnancy may increase the risk of obesity, heart disease and diabetes, and their many complications in later life, especially when growth restriction in pregnancy is followed by rapid ‘catch-up’ and overnutrition after birth. Maternal diet also has specific effects on immune development and allergic risk [66–68] . In general, this includes ‘healthy’ dietary patterns such as the Med- iterranean diet, which, in addition to its cardiovascular benefits, appears to protect against wheezing in early childhood in some [69–71] but not all studies [72] . This may be due to the composite effects of higher intakes of antioxidant-rich fresh foods, dietary fiber and n-3 polyunsaturated fatty acids (PUFA). Individu- ally, many of these elements have demonstrated immune and metabolic benefits. Our own studies have shown that fish oil supplementation (n-3 PUFA) from 20 weeks gestation had effects on immune function, with reduced inflammation and oxidative stress in the newborns at risk of allergic diseases because of an immediate family history [73–75] . We also saw a reduction in egg sensitization and eczema severity at 12 months of age [73] . Postnatal fish oil supplementation appears to be less effective than improving n-3 PUFA levels in pregnancy [76] . Other nutrients of specific interest in immune and metabolic health include vitamin D, antioxidants (such as vitamins A, C and E) and prebiotics [68] . There are currently several randomized controlled trials of vitamin D in pregnant women and in young infants. As yet, the specific role in allergy prevention is still unclear. Optimizing folic acid status before conception and in the first trimester of pregnancy is critical for preventing neural tube defects. However, there are no clear benefits of continuing acid supplements beyond the first trimester. In fact, many women continue to consume five times the recommended daily intake, and there are emerging associations with high folate status and allergic disease in the offspring, including atopic dermatitis [77, 78] childhood wheeze [79] and asthma [80] .

8 Prescott After birth, breast milk is the most important source of nutrition for the developing infant. Breastfed infants generally show healthier growth patterns. Breastfeeding is also important in establishing infant colonization patterns, particularly bifidobacteria, that are important for establishing immune and metabolic homeostasis. Without the possibility of randomized controlled trials, it has been difficult to prove any specific allergy-protective effects of breast milk, and results from observational cohort studies are mixed. Observational studies also show protection from type 2 diabetes, with lower fasting insulin concentrations and insulin-like growth factor (IGF)-1 in breastfed infants than in theirformula-fed counterparts [81–83] , but again the data are mixed. Breast- feeding appears to also confer favorable effects on blood cholesterol, although the long-term effects on cardiovascular protection are less consistent [48, 84, 85] . Postnatal colonization is also affected by the delivery method, antibiotics, subsequent infant dietary patterns and a range of other environmental factors. One promising postnatal intervention has been the use of soluble ‘prebiotic’ fibers (oligosaccharides) that selectively stimulate the growth of beneficial gut microbiota, particularly bifidobacteria but also lactobacilli [86–88] . These bacteria ferment oligosaccharides to form short-chain fatty acids (such as acetate, butyrate and proprionate) that have direct anti-inflammatory effects [89]. This is very important for gut health, promoting the intestinal barrier function [90], and reduces systemic leakage of bacterial products such as endotoxin [91]. Short-chain fatty acids stimulate the development of regulatory immune responses. Clinical trials of prebiotics in humans show beneficial effects on the microbiome and immune function, with reduced systemic inflammation, plasma lipids and markers of the metabolic syndrome in overweight adults [92] . In young children, studies show beneficial effects on early colonization and a reduction in eczema with prebiotic supplementation [86, 93, 94] . There are several studies now underway to explore this in more detail. Probiotics have been far more extensively investigated in both infancy and late pregnancy for allergy prevention. Collectively, these studies suggest a protective effect on eczema [65, 95–97] , and a combination of pre- and postnatal probiotic supplementation appears most effective [98, 99]. However, there has been no consistent protection from other allergic outcomes [100]. So far, research has mostly focused on lactobacillus and bifidobacterial strains, and future studies are anticipated to determine if ‘next-generation probiotics’, such as stronger butyrate and proprionate producers and immunomodulatory Bacteroi- des strains, have more powerful effects [101] .

Early Nutrition and Immune Health 9 Infant Feeding: Is There a ‘Window of Tolerance’ for Reducing Allergy?

Complementary feeding before 3–4 months is associated with an increased risk of allergic disease, presumably because of immaturity of the gut and related immune system, and lack of an established microbiome [102] . For this reason, it is recommended that complimentary foods are not introduced before 17 weeks. On the other hand, there is mounting evidence that delayed complementary feeding (beyond 6 months) and allergen avoidance may actually increase allergy risk [102–104] . This has led to the speculation that there may be an ‘optimal window’ to introduce complementary foods somewhere between 4 and 6 months of age [102]. These concepts are based only on observational data, and clinical trials are currently underway to address this more definitively. Regular exposure to ubiquitous proteins may promote immune tolerance. This may be why allergen avoidance strategies (delayed exposure) may have actually contributed to an increased allergic risk. The most recent cohort studies show the delayed introduction of specific foods (oats, wheat, cow’s milk, fish and egg) is associated with increased allergic sensitization and allergic disease. This gives continued support for the current recommendations by expert bodies [105–107] that there is no evidence that delayed introduction of complementary foods beyond 4 months of age is beneficial. None of this negates recommendations to continue breastfeeding for as long as possible. Allergens secreted normally in mother’s milk appear to be an important early source of exposure [108, 109]. This may actually be important in initiating, maintaining and reinforcing normal tolerance to foods and even inhaled allergens [110, 111]. When breastfeeding is not possible, the use of hydrolyzed formulas appears to confer some protective effect compared with normal cow’s milk-based formulas, though more studies are needed, and the protective effects are not large. This is at odds with the World Health Organization’s recommendation of ‘exclusive’ breastfeeding for at least 6 months, which restricts intake other than breast milk, even water, but allows rehydration solutions, medicines, vitamins, minerals and other treatments if needed. While these global recommendations have been very much targeting infant mortality in the developing world, the obesity epidemic has more recently entered this agenda. There are concerns that early introduction of ‘carbohydrate-dense’ weaning foods could be promoting obesity in more affluent regions [112] through effects on both metabolic programming and taste preferences. Again, this has not been proven in humans, but animal studies suggest that increasing caloric intake in the immediate postnatal period increases the risk of higher body weight and overeating later in life, and that higher carbohydrate intake is a particular risk factor [113] .

10 Prescott However, the specific role of the duration of breastfeeding and the ‘timing’ of starting ‘solid’ foods in the risk of obesity is also unclear. One major study to address this followed 10,912 individuals from low-/middle-income countries over several decades [114]. They found no differences in the rates of obesity, overweight or diabetes between adults who were ‘ever’ breastfed compared with those who were never breastfed. There was also no relationship with the duration of breastfeeding. Earlier introduction of complementary food was associated with higher adult adiposity, but this effect was quite small. Almost all of the studies to examine the long-term effects of feeding choices are observational. The only situation in which babies have been randomly assigned to receive either breast milk (from a human milk bank) or a formula was a study on preterm infants [115] . When these children where followed to adolescence, there were no differences in body weight and obesity, but some markers of cardiovascular risk (blood lipid and inflammation profiles) were lower, suggesting some beneficial relationships. For ethical reasons, it is not possible to do similar clinical trials in healthy term infants. While breastfeeding should be promoted for its many benefits, it is difficult to claim with certainty that it specifically prevents either obesity or allergy. The optimal time to start complementary feeding needs more investigation. There are several randomized clinical trials of ‘early’ versus ‘later’ complementary feeding currently underway that are designed to assess the effects on allergic disease, and the effects on weight gain will also be of great interest. Finally, studies show that a significant proportion of infants already have food sensitization and clinical reactivity (including anaphylaxis) prior to the ‘first’ introduction of foods at 4–6 months [116] indicating that earlier preventive interventions will ultimately be required. This may not necessarily mean using allergens at all, but other immunomodulatory strategies. Again, if allergens are the ‘target’ and not the ‘cause’ of the allergy epidemic they may not hold the final solution. In summary, there is growing awareness that inflammation is an important part of the initiating events that lead to a range of disorders including cardiovascular disease, obesity and neuropsychiatric disorders. It is logical to wonder if the same environmental risk factors might be implicated in the rise in so many inflammatory diseases. Because eczema, food allergy and asthma appear so much earlier in life it is more obvious that early events must be important. In this context, the ‘allergy epidemic’ is the clearest evidence we have that the early immune system is exquisitely vulnerable to modern lifestyle changes [37] . This can also offer us vital clues about why we are also increasingly prone to other NCDs, which all generally feature inflammation and immune dysregulation. Given that diet, microbes, exercise, stress and modern pollutants all affect the

Early Nutrition and Immune Health 11 developing immune system, it is highly likely that these are also linked with the rise in so many other NCDs. These are among the many lifestyle factors that are associated with the significantly higher baseline levels of inflammatory cytokines that underpin many NCDs, and the logical targets in preventing disease. Finally, we can see how many of the ‘common risk factors’ for disease (diet, inactivity, smoking, pollutants, time indoors and declining microbial biodiversity) are inherently driven by wider social, cultural and economic factors, and the quality of our built and natural environment [117]. The WHO report on So- cial Determinants of Health: The Solid Facts summarizes the clear relationship between life expectancy and social standing – with an increasing burden of disease and premature death as we go down the social ladder [118] . In this context, it is clear that poor food quality and unhealthy eating patterns are much more likely in the socially disadvantaged [118]. As we have seen, all of these factors contribute to maternal health and family stress, to adversely influence the physical and psychological development of the next generation, and thus can also be our best chance to improve the future of human health.

Disclosure Statement

S.L.P. is supported by a Practitioner Fellowship from the National Health and Medical Research Council of Australia. She is on advisory boards of Nestlé Nutrition Institute and Danone Nutricia, and received speaker fees from these entities and from ALK Abello.

References

1 Bach JF: The effect of infections on suscepti- 6 Drazen DL, Demas GE, Nelson RJ: Leptin bility to autoimmune and allergic diseases. N effects on immune function and energy bal-

Engl J Med 2002; 347: 911–920. ance are photoperiod dependent in Siberian 2 Wills-Karp M, Santeliz J, Karp CL: The germ- hamsters (Phodopus sungorus) . Endocrinol-

less theory of allergic disease: revisiting the ogy 2001; 142: 2768–2775.

hygiene hypothesis. Nat Rev Immunol 2001; 7 Bartness TJ, Demas GE, Song CK: Seasonal

1: 69–75. changes in adiposity: the roles of the photo- 3 Martino D, Prescott SL: Epigenetics and pre- period, melatonin and other hormones, and natal influences on asthma and allergic air- sympathetic nervous system. Exp Biol Med

ways disease. Chest 2011; 139: 640–647. (Maywood) 2002; 227: 363–376. 4 Palm NW, Rosenstein RK, Medzhitov R: Al- 8 Demas GE, Chefer V, Talan MI, Nelson RJ:

lergic host defences. Nature 2012; 484: 465– Metabolic costs of mounting an antigen- 472. stimulated immune response in adult and

5 Yellon SM, Fagoaga OR, Nehlsen-Cannarella aged C57BL/6J mice. Am J Physiol 1997; SL: Influence of photoperiod on immune cell 273:R1631–R1637. functions in the male Siberian hamster. Am J

Physiol 1999; 276:R97–R102.

12 Prescott 9 Moret Y, Schmid-Hempel P: Survival for im- 23 Palmer G, Gabay C: A role for leptin in rheu-

munity: the price of immune system activa- matic diseases? Ann Rheum Dis 2003; 62:

tion for bumblebee workers. Science 2000; 913–915.

290: 1166–1168. 24 Otsuka R, Yatsuya H, Tamakoshi K, et al: 10 Svensson E, Raberg L, Koch C, Hasselquist D: Perceived psychological stress and serum Energetic stress, immunosuppression and the leptin concentrations in Japanese men. Obe-

costs of an antibody response. Funct Ecol sity (Silver Spring) 2006; 14: 1832–1838.

1998; 12: 912–919. 25 Hickey MS, Houmard JA, Considine RV, et 11 Bilbo SD, Drazen DL, Quan N, et al: Short al: Gender-dependent effects of exercise day lengths attenuate the symptoms of infec- training on serum leptin levels in humans.

tion in Siberian hamsters. Proc Biol Sci 2002; Am J Physiol 1997; 272:E562–E566.

269: 447–454. 26 de Salles BF, Simao R, Fleck SJ, et al: Effects 12 Polack E, Nahmod VE, Emeric-Sauval E, et of resistance training on cytokines. Int J

al: Low lymphocyte interferon-gamma pro- Sports Med 2010; 31: 441–450. duction and variable proliferative response in 27 Dandona P, Aljada A, Bandyopadhyay A: anorexia nervosa patients. J Clin Immunol Inflammation: the link between insulin resis-

1993; 13: 445–451. tance, obesity and diabetes. Trends Immunol

13 Cason J, Ainley CC, Wolstencroft RA, et al: 2004; 25: 4–7. Cell-mediated immunity in anorexia nervosa. 28 Visser M, Bouter LM, McQuillan GM, et al:

Clin Exp Immunol 1986; 64: 370–375. Elevated C-reactive protein levels in over-

14 Kau AL, Ahern PP, Griffin NW, et al: Human weight and obese adults. JAMA 1999; 282: nutrition, the gut microbiome and the im- 2131–2135.

mune system. Nature 2011; 474: 327–336. 29 Visness CM, London SJ, Daniels JL, et al: As- 15 Matarese G, La Cava A: The intricate inter- sociation of obesity with IgE levels and aller- face between immune system and metabo- gy symptoms in children and adolescents:

lism. Trends Immunol 2004; 25: 193–200. results from the National Health and Nutri- 16 Considine RV, Caro JF: Leptin and the regu- tion Examination Survey 2005–2006. J Al-

lation of body weight. Int J Biochem Cell Biol lergy Clin Immunol 2009; 123: 1163–1169,

1997; 29: 1255–1272. 1169.e1–1169.e4. 17 Moore SI, Huffnagle GB, Chen GH, et al: 30 McDade TW, Rutherford JN, Adair L, Kuza- Leptin modulates neutrophil phagocytosis of wa C: Population differences in associations

Klebsiella pneumoniae . Infect Immun 2003; between C-reactive protein concentration

71: 4182–4185. and adiposity: comparison of young adults in 18 Caldefie-Chezet F, Poulin A, Vasson MP: the Philippines and the United States. Am J

Leptin regulates functional capacities of poly- Clin Nutr 2009; 89: 1237–1245. morphonuclear neutrophils. Free Radic Res 31 McDade TW, Tallman PS, Madimenos FC, et

2003; 37: 809–814. al: Analysis of variability of high sensitivity 19 Farooqi IS, Matarese G, Lord GM, et al: Ben- C-reactive protein in lowland Ecuador re- eficial effects of leptin on obesity, T cell hypo- veals no evidence of chronic low-grade in-

responsiveness, and neuroendocrine/meta- flammation. Am J Hum Biol 2012; 24: 675– bolic dysfunction of human congenital leptin 681.

deficiency. J Clin Invest 2002; 110: 1093–1103. 32 Ridker PM, Buring JE, Cook NR, Rifai N: C- 20 Lord GM, Matarese G, Howard JK, et al: reactive protein, the metabolic syndrome, Leptin modulates the T-cell immune re- and risk of incident cardiovascular events: an sponse and reverses starvation-induced im- 8-year follow-up of 14,719 initially healthy

munosuppression. Nature 1998; 394: 897–901. American women. Circulation 2003; 107: 391– 21 Radon K, Schulze A, Schierl R, et al: Serum 397. leptin and adiponectin levels and their asso- 33 Pradhan AD, Manson JE, Rifai N, et al: C-re- ciation with allergic sensitization. Allergy active protein, interleukin 6, and risk of de-

2008; 63: 1448–1454. veloping type 2 diabetes mellitus. JAMA

22 Matarese G, Leiter EH, La Cava A: Leptin in 2001; 286: 327–334. autoimmunity: many questions, some an-

swers. Tissue Antigens 2007; 70: 87–95.

Early Nutrition and Immune Health 13 34 Jenny NS, Yanez ND, Psaty BM, et al: Inflam- 45 Miller GE, Chen E, Fok AK, et al: Low early- mation biomarkers and near-term death in life social class leaves a biological residue

older men. Am J Epidemiol 2007; 165: 684– manifested by decreased glucocorticoid and 695. increased proinflammatory signaling. Proc

35 Pearson TA, Mensah GA, Alexander RW, et Natl Acad Sci U S A 2009; 106: 14716–14721. al: Markers of inflammation and cardiovas- 46 Miller GE, Chen E: Harsh family climate in cular disease: application to clinical and pub- early life presages the emergence of a proin- lic health practice: a statement for healthcare flammatory phenotype in adolescence. Psy-

professionals from the Centers for Disease chol Sci 2010; 21: 848–956. Control and Prevention and the American 47 Metzger MW, McDade TW: Breastfeeding as

Heart Association. Circulation 2003; 107: 499– obesity prevention in the United States: a sib-

511. ling difference model. Am J Hum Biol 2010;

36 McDade TW: Early environments and the 22: 291–296. ecology of inflammation. Proc Natl Acad Sci 48 Martin RM, Gunnell D, Smith GD: Breast-

U S A 2012; 109(suppl 2):17281–17288. feeding in infancy and blood pressure in later 37 Prescott SL: The Allergy Epidemic: A Mys- life: systematic review and meta-analysis. Am

tery of Modern Life. Perth, UWA Publishing; J Epidemiol 2005; 161: 15–26. 2011. 49 Prescott SL, Allen KJ: Food allergy: riding the 38 Prescott SL: Early origins of allergic disease: a second wave of the allergy epidemic. Pediatr

review of processes and influences during Allergy Immunol 2011; 22: 155–160. early immune development. Curr Opin Al- 50 Prescott SL, Saffery R: The role of epigenetic

lergy Clin Immunol 2003; 3: 125–132. dysregulation in the epidemic of allergic dis-

39 Prescott SL: Early-life environmental deter- ease. Clin Epigenet 2011; 2: 223–232. minants of allergic diseases and the wider 51 Gold DR, Damokosh AI, Dockery DW, Ber- pandemic of inflammatory noncommunica- key CS: Body-mass index as a predictor of

ble diseases. J Allergy Clin Immunol 2013; incident asthma in a prospective cohort of

131: 23–30. children. Pediatr Pulmonol 2003; 36: 514–521. 40 McDade TW, Rutherford J, Adair L, Kuzawa 52 Castro-Rodriguez JA, Holberg CJ, Morgan CW: Early origins of inflammation: microbial WJ, et al: Increased incidence of asthmalike exposures in infancy predict lower levels of symptoms in girls who become overweight or C-reactive protein in adulthood. Proc Biol Sci obese during the school years. Am J Respir

2010; 277: 1129–1137. Crit Care Med 2001; 163: 1344–1349. 41 Danese A, Pariante CM, Caspi A, et al: Child- 53 Ramsay JE, Ferrell WR, Crawford L, et al: hood maltreatment predicts adult inflamma- Maternal obesity is associated with dysregulation in a life-course study. Proc Natl Acad Sci tion of metabolic, vascular, and inflammatory

U S A 2007; 104: 1319–1324. pathways. J Clin Endocrinol Metab 2002; 87: 42 Tzoulaki I, Jarvelin MR, Hartikainen AL, et 4231–4237. al: Size at birth, weight gain over the life 54 Challier JC, Basu S, Bintein T, et al: Obesity course, and low-grade inflammation in in pregnancy stimulates macrophage accu- young adulthood: northern Finland 1966 mulation and inflammation in the placenta.

Birth Cohort study. Eur Heart J 2008; 29: Placenta 2008; 29: 274–281. 1049–1056. 55 Reichman NE, Nepomnyaschy L: Maternal 43 Sattar N, McConnachie A, O’Reilly D, et al: pre-pregnancy obesity and diagnosis of asth- Inverse association between birth weight and ma in offspring at age 3 years. Matern Child

C-reactive protein concentrations in the Health J 2008; 12: 725–733. MIDSPAN Family Study. Arterioscler 56 Haberg SE, Stigum H, London SJ, et al: Ma-

Thromb Vasc Biol 2004; 24: 583–587. ternal obesity in pregnancy and respiratory 44 Taylor SE, Lehman BJ, Kiefe CI, Seeman TE: health in early childhood. Paediatr Perinat

Relationship of early life stress and psycho- Epidemiol 2009; 23: 352–362. logical functioning to adult C-reactive pro- 57 Rothenbacher D, Weyermann M, Fantuzzi G, tein in the coronary artery risk development Brenner H: Adipokines in cord blood and

in young adults study. Biol Psychiatry 2006; risk of wheezing disorders within the first

60: 819–824. two years of life. Clin Exp Allergy 2007; 37: 1143–1149.

14 Prescott 58 Shore SA, Schwartzman IN, Mellema MS, et 71 Chatzi L, Torrent M, Romieu I, et al: Medi- al: Effect of leptin on allergic airway respons- terranean diet in pregnancy is protective for

es in mice. J Allergy Clin Immunol 2005; 115: wheeze and atopy in childhood. Thorax 2008;

103–109. 63: 507–513. 59 Holt PG, Sly PD: Interaction between adap- 72 Shaheen SO, Northstone K, Newson RB, et al: tive and innate immune pathways in the Dietary patterns in pregnancy and respira- pathogenesis of atopic asthma: operation of a tory and atopic outcomes in childhood. Tho-

lung/bone marrow axis. Chest 2011; 139: rax 2009; 64: 411–417. 1165–1171. 73 Dunstan J, Mori TA, Barden A, et al: Fish oil 60 Hotamisligil GS, Erbay E: Nutrient sensing supplementation in pregnancy modifies neo- and inflammation in metabolic diseases. Nat natal allergen-specific immune responses and

Rev Immunol 2008; 8: 923–934. clinical outcomes in infants at high risk of 61 Knoflach M, Kiechl S, Mayr A, et al: Allergic atopy: a randomised controlled trial. J Allergy

rhinitis, asthma, and atherosclerosis in the Clin Immunol 2003; 112: 1178–1184. Bruneck and ARMY studies. Arch Intern 74 Barden A, Mori TA, Dunstan JA, et al: Fish

Med 2005; 165: 2521–2526. oil supplementation in pregnancy lowers F2 62 Matheson EM, Player MS, Mainous AG 3rd, isoprostanes in neonates at high risk of atopy.

et al: The association between hay fever and Free Radic Res 2004; 38: 233–239. stroke in a cohort of middle aged and elderly 75 Prescott SL, Barden AE, Mori TA, Dunstan

adults. J Am Board Fam Med 2008; 21: 179– JA: Maternal fish oil supplementation in 183. pregnancy modifies neonatal leukotriene 63 Jung CC, Chang CC, Tsai YS, Su HJ: Allergen production by cord-blood-derived neutro-

exposure induces inflammation and affects phils. Clin Sci (Lond) 2007; 113: 409–416. adiponectin levels in adipose tissue. Toxicol 76 D’Vaz N, Meldrum SJ, Dunstan JA, et al:

Lett 2013; 223: 88–95. Postnatal fish oil supplementation in high- 64 Ciprandi G, De Amici M, Tosca MA, Mar- risk infants to prevent allergy: randomized

seglia G: Serum leptin levels depend on aller- controlled trial. Pediatrics 2012; 130: 674–682. gen exposure in patients with seasonal aller- 77 Kiefte-de Jong JC, Timmermans S, Jaddoe

gic rhinitis. Immunol Invest 2009; 38: VW, et al: High circulating folate and vitamin 681–689. B-12 concentrations in women during preg- 65 Pfefferle PI, Prescott SL, Kopp M: Microbial nancy are associated with increased preva- influence on tolerance and opportunities for lence of atopic dermatitis in their offspring. J

intervention with prebiotics/probiotics and Nutr 2012; 142: 731–738. bacterial lysates. J Allergy Clin Immunol 78 Dunstan JA, West C, McCarthy S, et al: The

2013; 131: 1453–1463; quiz 64. relationship between maternal folate status in 66 Shaheen SO: Prenatal nutrition and asthma: pregnancy, cord blood folate levels, and aller-

hope or hype? Thorax 2008; 63: 483–485. gic outcomes in early childhood. Allergy

67 Prescott SL, Clifton VL: Asthma and preg- 2012; 67: 50–57. nancy: emerging evidence of epigenetic inter- 79 Håberg SE, London SJ, Stigum H, et al: Folic actions in utero. Curr Opin Allergy Clin Im- acid supplements in pregnancy and early

munol 2009; 9: 417–426. childhood respiratory health. Arch Dis Child

68 West CE, D’Vaz N, Prescott SL: Dietary im- 2009; 94: 180–184. munomodulatory factors in the development 80 Whitrow MJ, Moore VM, Rumbold AR, Da- of immune tolerance. Curr Allergy Asthma vies MJ: Effect of supplemental folic acid in

Rep 2011; 11: 325–333. pregnancy on childhood asthma: a prospec-

69 Castro-Rodriguez JA, Garcia-Marcos L, Al- tive birth cohort study. Am J Epidemiol 2009;

fonseda Rojas JD, et al: Mediterranean diet as 170: 1486–1493. a protective factor for wheezing in preschool 81 Madsen AL, Larnkjaer A, Molgaard C, Mi-

children. J Pediatr 2008; 152: 823–828, 828. chaelsen KF: IGF-I and IGFBP-3 in healthy e1–828.e2. 9 month old infants from the SKOT cohort: 70 Chatzi L, Torrent M, Romieu I, et al: Diet, breastfeeding, diet, and later obesity. Growth

wheeze, and atopy in school children in Horm IGF Res 2011; 21: 199–204. Menorca, Spain. Pediatr Allergy Immunol

2007; 18: 480–485.

Early Nutrition and Immune Health 15 82 Madsen AL, Schack-Nielsen L, Larnkjaer A, 93 Arslanoglu S, Moro GE, Schmitt J, et al: Ear- et al: Determinants of blood glucose and in- ly dietary intervention with a mixture of sulin in healthy 9-month-old term Danish prebiotic oligosaccharides reduces the inci-

infants: the SKOT cohort. Diabet Med 2010; dence of allergic manifestations and infec-

27: 1350–1357. tions during the first two years of life. J Nutr

83 Owen CG, Martin RM, Whincup PH, et al: 2008; 138: 1091–1095. Does breastfeeding influence risk of type 2 94 Gruber C, van Stuijvenberg M, Mosca F, et diabetes in later life? A quantitative analysis al: Reduced occurrence of early atopic der-

of published evidence. Am J Clin Nutr 2006; matitis because of immunoactive prebiotics

84: 1043–1054. among low-atopy-risk infants. J Allergy Clin

84 Martin RM, Davey Smith G, Mangtani P, et Immunol 2010; 126: 791–797. al: Breastfeeding and cardiovascular mortal- 95 Pelucchi C, Chatenoud L, Turati F, et al: ity: the Boyd Orr cohort and a systematic re- Probiotics supplementation during preg-

view with meta-analysis. Eur Heart J 2004; 25: nancy or infancy for the prevention of atop- 778–786. ic dermatitis: a meta-analysis. Epidemiology

85 Rich-Edwards JW, Stampfer MJ, Manson JE, 2012; 23: 402–414. et al: Breastfeeding during infancy and the 96 Doege K, Grajecki D, Zyriax BC, et al: Im- risk of cardiovascular disease in adulthood. pact of maternal supplementation with pro-

Epidemiology 2004; 15: 550–556. biotics during pregnancy on atopic eczema 86 Moro G, Arslanoglu S, Stahl B, et al: A mix- in childhood – a meta-analysis. Br J Nutr

ture of prebiotic oligosaccharides reduces the 2012; 107: 1–6. incidence of atopic dermatitis during the first 97 Foolad N, Brezinski EA, Chase EP, Arm-

six months of age. Arch Dis Child 2006; 91: strong AW: Effect of nutrient supplementa- 814–819. tion on atopic dermatitis in children: a sys- 87 Boehm G, Lidestri M, Casetta P, et al: Supple- tematic review of probiotics, prebiotics, mentation of a bovine milk formula with an formula, and fatty acids. JAMA Dermatol

oligosaccharide mixture increases counts of 2013; 149: 350–355. faecal bifidobacteria in preterm infants. Arch 98 Fiocchi A, Burks W, Bahna SL, et al: Clinical

Dis Child Fetal Neonatal Ed 2002; 86:F178– use of probiotics in pediatric allergy F181. (CUPPA): a World Allergy Organization

88 Bouhnik Y, Vahedi K, Achour L, et al: Short- position paper. World Allergy Organ J 2012;

chain fructo-oligosaccharide administration 5: 148–167. dose-dependently increases fecal bifidobacte- 99 West CE, Prescott SL: Prebiotics and probi-

ria in healthy humans. J Nutr 1999; 129: 113– otics for prevention of allergic disease; in 116. Basow DE (ed): UpToDate. Waltham, Wolt- 89 Maslowski KM, Vieira AT, Ng A, et al: Regu- ers Kluwer Health, 2013, http://www.upto- lation of inflammatory responses by gut mi- date.com. crobiota and chemoattractant receptor 100 Azad MB, Coneys JG, Kozyrskyj AL, et al:

GPR43. Nature 2009; 461: 1282–1286. Probiotic supplementation during pregnan- 90 Wong JM, de Souza R, Kendall CW, et al: Co- cy or infancy for the prevention of asthma lonic health: fermentation and short chain fat- and wheeze: systematic review and meta-

ty acids. J Clin Gastroenterol 2006; 40: 235–243. analysis. BMJ 2013; 347:f6471. 91 Morita T, Tanabe H, Sugiyama K, et al: Di- 101 Neef A, Sanz Y: Future for probiotic science etary resistant starch alters the characteristics in functional food and dietary supplement of colonic mucosa and exerts a protective ef- development. Curr Opin Clin Nutr Metab

fect on trinitrobenzene sulfonic acid-induced Care 2013; 16: 679–687. colitis in rats. Biosci Biotechnol Biochem 102 Prescott SL, Smith P, Tang MLK, et al: The

2004; 68: 2155–2164. importance of early complementary feeding 92 Vulevic J, Juric A, Tzortzis G, Gibson GR: A in the development of oral tolerance: con- mixture of trans-galactooligosaccharides re- cerns and controversies. Pediatr Allergy Im-

duces markers of metabolic syndrome and munol 2008; 19: 375–380. modulates the fecal microbiota and immune

function of overweight adults. J Nutr 2013;

143: 324–331.

16 Prescott 103 Palmer DJ, Metcalfe J, Prescott SL: Prevent- 110 Verhasselt V, Milcent V, Cazareth J, et al: ing disease in the 21st century: the impor- Breast milk-mediated transfer of an antigen tance of maternal and early infant diet and induces tolerance and protection from aller-

nutrition. J Allergy Clin Immunol 2012; 130: gic asthma. Nat Med 2008; 14: 170–175. 733–734. 111 Yamamoto T, Tsubota Y, Kodama T, et al: 104 Palmer DJ, Prescott SL: Does early feeding Oral tolerance induced by transfer of food promote development of oral tolerance? antigens via breast milk of allergic mothers

Curr Allergy Asthma Rep 2012; 12: 321–331. prevents offspring from developing allergic 105 Greer FR, Sicherer SH, Burks AW: Effects of symptoms in a mouse food allergy model.

early nutritional interventions on the devel- Clin Dev Immunol 2012; 2012: 721085. opment of atopic disease in infants and chil- 112 Patel MS, Srinivasan M: Metabolic pro- dren: the role of maternal dietary restric- gramming in the immediate postnatal life.

tion, breastfeeding, timing of introduction Ann Nutr Metab 2011; 58(suppl 2):18–28. of complementary foods, and hydrolyzed 113 Patel MS, Srinivasan M, Laychock SG: Met-

formulas. Pediatrics 2008; 121: 183–191. abolic programming: role of nutrition in the 106 Host A, Halken S, Muraro A, et al: Dietary immediate postnatal life. J Inherit Metab

prevention of allergic diseases in infants and Dis 2009; 32: 218–228. small children. Pediatr Allergy Immunol 114 Fall CH, Borja JB, Osmond C, et al: Infant-

2008; 19: 1–4. feeding patterns and cardiovascular risk fac- 107 Agostoni C, Decsi T, Fewtrell M, et al: Com- tors in young adulthood: data from five co- plementary feeding: a commentary by the horts in low- and middle-income countries.

ESPGHAN Committee on Nutrition. J Pedi- Int J Epidemiol 2011; 40: 47–62.

atr Gastroenterol Nutr 2008; 46: 99–110. 115 Singhal A, Cole TJ, Fewtrell M, Lucas A: 108 Palmer DJ, Gold MS, Makrides M: Effect of Breastmilk feeding and lipoprotein profile cooked and raw egg consumption on oval- in adolescents born preterm: follow-up of a

bumin content of human milk: a random- prospective randomised study. Lancet 2004;

ized, double-blind, cross-over trial. Clin Exp 363: 1571–1578.

Allergy 2005; 35: 173–178. 116 Palmer DJ, Metcalfe J, Makrides M, et al: 109 Macchiaverni P, Rekima A, Turfkruyer M, Early regular egg exposure in infants with et al: Respiratory allergen from house dust eczema: a randomized controlled trial. J Al-

mite is present in human milk and primes lergy Clin Immunol 2013; 132: 387.e1–392. for allergic sensitization in a mouse model e1.

of asthma. Allergy 2014; 69: 395–398. 117 Marmot M: Social determinants of health

inequalities. Lancet 2005; 365: 1099–1104. 118 Wilkinson R, Marmot M: Social Determi- nants of Health: The Solid Facts, ed 2. Ge- neva, World Health Organization, 2003.

Early Nutrition and Immune Health 17

Allergy

Fewtrell MS, Haschke F, Prescott SL (eds): Preventive Aspects of Early Nutrition. Nestlé Nutr Inst Workshop Ser, vol 85, pp 19–27, (DOI: 10.1159/000439479) Nestec Ltd., Vevey/S. Karger AG., Basel, © 2016

The Future of Infant and Young Children’s Food: Food Supply/ Manufacturing and Human Health Challenges in the 21st Century

a, b a Carina Venter · Kate Maslin a b School of Health Sciences and Social Work, University of Portsmouth, Portsmouth , UK; Division of Allergy and Immunology, Cincinnati Children’s Hospital Medical Center, Cincinnati, OH, USA

Abstract Infant food and weaning practices are highly debated with lots of unanswered questions. It is becoming more apparent that early-life feeding may have an effect on the long-term health of humans, particularly for noncommunicable diseases such as obesity and allergic diseases. It is important to understand how environmental influences in early life can affect the development of the immune system and metabolic profiling. In terms of nutrition and diet, one should consider the role of the total/whole diet, as well as particular nutrients in the development of noncommunicable diseases. Providing the appropriate nutrition for infants during the weaning age needs to address factors such as the microbial load of the food, nutrient composition, presence/absence of allergens and appropriate textures. These factors are of importance irrespective of whether the food is homemade or produced commercially, and need to take environmental factors and food resources into account. © 2016 Nestec Ltd., Vevey/S. Karger AG, Basel

Health Challenges

Obesity There has been a marked increase in obesity rates over the past 20 years. Obe- sity and overweight affect more than 1.5 billion adults and account for 0.7– 2.8% of health care costs in both the developed and developing world [1, 2] . The cost of obesity and overweight to the UK economy was estimated at GBP 15.8 billion per year in 2007, including GBP 4.2 billion in costs to the NHS [3] . In 1993, 13% of men and 16% of women were obese compared to 24% of men and 26% of women in 2011 [4] . There has also been an increase in the incidence of childhood and adolescent obesity worldwide, an important predictor of adulthood obesity, morbidity and mortality [5–7] . In the USA, a third of children were classified as either overweight or obese in 2012. The average weight of a child has risen by more than 5 kg within three decades. This varies from an increase of 2 kg in children under 8 years of age to an increase of more than 8 kg in some adolescent age groups [8]. A survey in English schoolchildren found that 19.2 and 33.9% of 4-/5- and 10-/11-year- old children, respectively, are obese or overweight [9] . However, despite an increase in obesity-related hospital admissions in children, the obesity rates of reception age children in the UK have stayed stable, albeit at a high rate of about 22%. Maternal overweight [10–15] is associated with subsequent obesity in offspring, highlighting the importance of early-life factors in the development of obesity. In addition, increased birth weight has been associated with childhood and adolescent obesity. Data from the Isle of Wight have recently indicated [16] that an early persistent obesity was seen in obese infants but also in ‘normal- weight’ infants who went on to become obese teenagers/adults, i.e. some obese infants became obese adults but normal-weight infants could also become obese adults. Most importantly, this weight trajectory, also indicating the risk for developing asthma by 18 years of age, was set by 4 years. Maternal overweight before pregnancy and smoking during pregnancy were associated with increased risk in this group. The data, therefore, indicate that there are factors in early life that could lead to a ‘normal-weight’ infant developing into an overweight/obese infant that can persist into adulthood.

Early-Life Dietary Factors Affecting Obesity There is some evidence indicating the microbiota is affected in obese individuals, and most importantly that microbiota as early as 3 months of age could be related to obesity at 10 years of age. This has been summarized in a review by Clarke et al. [17] ( table 1 ). Diet, however, also affects the gut microbiota (e.g. a low fat diet is characterized by higher levels of Bacteroidetes and lower levels of Firmicutes). Weight reduction does not seem to affect gut bacteria [18] and it is therefore still unclear at this stage if the difference in gut bacteria is primarily related to body weight or diet. Early food exposure and introduction of solid food can also affect obesity outcomes, by affecting taste and food preferences to some extent.

20 Venter · Maslin Table 1. Gut bacteria in obesity (adapted from Clarke et al. [17])

Obese adults vs. lean adults (some conflicting data from papers) Lower or higher levels of Bacteroidetes Lower or higher levels of Firmicutes Lower levels of bifidobacteria (Some papers stating no difference) Obese children vs. lean children Higher levels of S. aureus Lower levels of bifidobacteria Lower levels of Faecalibacterium prausnitzii

Allergy Similar to the rise in obesity, we have seen a rise in allergic diseases. The WAO White Book on allergy states that ‘the prevalence of allergic diseases worldwide is rising dramatically in both developed and developing countries. These diseases include asthma; rhinitis; anaphylaxis; drug, food, and insect allergy; eczema; and urticaria (hives), and angioedema. This increase is especially problematic in children, who are bearing the greatest burden of the rising trend, which has occurred over the last two decades. In spite of this increase, even in the developed world, services for patients with allergic diseases are fragmented and far from ideal. Very few countries have comprehensive services in this field of medicine’ [19]. In the USA, the overall economic cost of food allergy is estimated at USD 24.8 billion annually (USD 4,184 per year per child). This includes direct medical costs such as clinician visits, costs borne by the family (lost labor productivity due to caregiver needing time off work) and the cost of specialized foods [20] . In the UK, allergy accounts for 6% of GP consultations and 10% of the prescribing budget [21] . The direct costs of allergy to the NHS budget is GBP 1 billion per annum. This further emphasizes the need for preventing allergic diseases.

Early-Life Dietary Factors Affecting Allergy Outcomes The role of the microbiota in the development of allergic disease has been re- searched for some time, with data indicating differences between the gut bacteria of allergic and nonallergic infants (table 2) [22] . This may be due to a decline in microbial exposure during early infancy, which in turn could be affected by environmental factors such as diet or early-life nutrition. Food diversity in early life may also affect allergy outcomes. In 2013, Nwaru et al. [23] indicated that by 12 months of age, less food diversity was associated with increased risk of any asthma, atopic asthma, wheeze and allergic rhinitis. Despite some controversy and debate, micronutrients such as vitamins A, E, C

Health Challenges through Nutrition 21 Table 2. Differences in bacterial load of infants with and without allergic disease [20]

Allergic vs. healthy infants Less enterococci in the 1st month Less bifidobacteria in the 1st year Higher counts of clostridia at 3 months Higher counts of S. aureus at 6 months Lower counts of Bacteroidetes at 12 months

and D, selenium and zinc may also affect allergy outcomes, as discussed in a review by Nurmatov et al. [24] in 2011. Finally, in 2013, Grimshaw et al. [25] indicated that a diet low in commercial baby foods was associated with less food allergy in the infant.

The Infant Diet

Infant foods and the weaning diet (also referred to as introduction of solid foods) have been at the forefront of dialogues in the scientific world and media over the past few decades. Answers to questions such as the most appropriate age of introduction of solid or allergenic foods, should breastfeeding continue alongside solid food introduction, what are the crucial times for introduction of different textures, and should organic or nonorganic foods be used have been sought. Central to all of these points is the use of homemade versus commercially available infant foods. The UK has seen an increase in baby food sales from GBP 303 million in 1995 to GBP 872 million in 2013 [26] . This is reflected in particular in the increase in sales of organic baby food. Worldwide, the baby food industry is worth USD 50 billion and is growing at a rate of 7% per year. These commercially prepared foods clearly constitute a large proportion of the infant diet, and should take into account factors that may play a role in the development of noncommunicable diseases. It is well known that association does not always equal causation. Black and Sharpe [27] published a paper looking at the association between fat intake and increase in allergic diseases in 1997. This association has never been proven in randomized controlled trials, and it will be interesting to see if there is any merit in the association seen between increase in sales of commercial baby foods and food allergy (using food-related anaphylaxis as a proxy; fig. 1 ). It is thought that gut microbiota has an effect on the development of allergic disease, and that certain foods and nutrients may also play a role in the prevention of allergic disease, e.g. fish/fish oil, vitamins A, E, C and

22 Venter · Maslin Consumption of baby food 8 Food anaphylaxis admissions 800 6 600 4 400 (GBP/child)

100,000 population 2 200 Hospital admissions per Consumption of baby food 0 0 20002005 2010

Fig. 1. Sales of commercial baby food in the UK versus admissions in food-related anaphylaxis [adapted from ref. 31 ].

D, selenium and zinc. Recent studies also indicate that a home-prepared [28] and more diverse diet leads to less allergic disease [29] . There are a number of ways in which the future of commercial baby foods can be improved to address the knowledge we have about possible preventative measures of allergic disease: the importance of microbes; nutrients that may play a role in the development of the immune system, and the importance of providing (some) homemade foods. The microbial load of commercially prepared baby food is negligible as food safety measures need to fulfill F 0 requirements. F0 is a term used in the canning industry to denote the minimum process required to destroy Clostridium botulinum spores, which are the most deadly of all bacteria, dependent on the material being processed. However, when home-cooked food is provided to infants, up to 65% of the daily microbial load may be provided by fresh fruits and vegetables. As only a finite number of foods are included in baby foods, an increase in consumption of commercial infant foods (with corresponding decrease in consumption of home-cooked foods) may affect the diversity of foods introduced during the weaning period and ultimately affect allergy outcomes. Finally, the importance of nutrients such as antioxidants and other nutrients also require consideration when discussing commercial baby food production. It is known that sterilization can reduce the vitamin C content by up to 50% and the vitamin B1 content by up to 30% [30] .

Focus Group Results Another very important factor to take into consideration is maternal/paternal experience of weaning and how they want the weaning message to be conveyed to them. This may not be important in terms of food production, but most baby food

Health Challenges through Nutrition 23 manufacturers also provide information to parents of young infants. We conducted 4 × 2.5 h focus group discussions with mothers of babies 4–7 months old. Babies had either commenced weaning solids already or were about to start weaning solids. The sample of mothers was split into groups by awareness/experience of allergies, social class and whether they were a first- or second-time mother. Three groups emerged from the analysis. These were ‘practical’, ‘balanced’ and ‘anxious’. Those in the ‘practical’ group were confident about weaning. They were often second-time mums who perceived prepared baby food as good (if not better) than homemade baby food. Mothers in the ‘balanced’ group had a balanced approach to weaning. They perceived homemade baby foods as ultimately the optimal choice, but conceded that it was not always practical to pre- pare and feed homemade foods. These mothers tended to examine food labels closely to determine the ingredient content. Those in the ‘anxious’ group tended to be either first-time mums with limited knowledge or second-time mums who had heard negative stories about weaning. As a result, this group often sought advice and guidance from experts. They also perceived homemade baby food as optimal, but some thought baby food has potential to be ‘safer’. Despite these three distinct typologies, all three groups had an underlying common perspective, viewing the goal of weaning as enjoyment of food and have development of a broad palate. They did not want the weaning message to be too directive, medical or pharmaceutical, and wanted the weaning message to center around ‘pure’, ‘simple’ and ‘healthy’. With regard to timing of weaning, there were two patterns. Second-time mothers tended to be ‘baby led’, using cues such as changes in sleep pattern, finishing milk quickly, being more irritable and watching others in the family eat. In contrast, ‘advice-led’ mothers sought advice from health care professionals, the internet, books, their own mothers or friends. Those in the higher social class were more likely to seek advice from health care professionals. Besides timing of weaning, the process of weaning (e.g. risk of choking) and what food to give were other sources of worry. For the majority of mothers these concerns were very short lived. Once weaning had started concerns over when and how were almost instantly overcome. However, concerns over what to feed and how quickly to introduce new flavors and foods into the diet were more variable between mothers. Generally, mothers commenced the weaning process with excitement and good intention of cooking homemade foods. However, even those who did a lot of home cooking reported that preparing fruit purees could be ‘hassle’. It was felt that prepared baby foods were composed of simple, safe ingredients. Indeed they were viewed by some to be superior to homemade foods especially if they were organic, prepared by better cooks and used better ingredients.

24 Venter · Maslin The choice of what prepared baby food to use was driven by three key factors: ‘taste’, ‘goodness’ and ‘the truth’. ‘Taste’ was characterized mainly by the description of the ingredients and recipes. ‘Goodness’ helped mothers decide whether the product was healthy. Finally, mothers were keen to know the ‘truth’ about prepared baby foods (i.e. what exactly was contained in the food and what was hid- den). Specifically, they were keen to discover whether the products contained milk, eggs, gluten and nuts, in addition to preservatives, coloring and salt. Inter- estingly, there was very little spontaneous mention of food allergies. When probed, mothers in the ‘balanced’ and ‘anxious’ groups showed some concerns, but the vast majority thought that food allergies are very individual and that exposure to a variety of foods early in the weaning process was important to identify any issues.

Conclusion

There is no doubt that the weaning diet will affect later health outcomes alongside maternal eating practices during pregnancy and breastfeeding as well as early milk (breast milk or formula) consumption. This poses the opportunity for families cooking homemade foods and industry producing commercial baby foods to provide infant foods in order to (perhaps) stem the tide of noncommunicable diseases such as allergy and obesity. Therefore, in the future, the introduction of food to infants should focus on three main factors: (1) Parental cooking skills to provide freshly cooked, homemade food (2) The possible bacterial content of commercial (sterilized vs. pasteurized) versus home-made foods (3) The particular nutrient content of baby foods and diversity of the infant diet. This will have to be provided with the backdrop of current dwindling world resources, focusing specifically on the availability and sustainable production of fish and meat, better food distribution and less food waste at home.

Disclosure Statement

Kate Maslin and Carina Venter have no conflict of interest regarding this chapter. Ca- rina Venter has acted as a consultant of provide lectures for Danone, Mead Johnson and Nestle in the past.

Health Challenges through Nutrition 25 References

1 Claessen H, Brenner H, Drath C, Arndt V: 16 Ziyab AH, Karmaus W, Kurukulaaratchy RJ, Repeated measures of body mass index and et al: Developmental trajectories of body risk of health related outcomes. Eur J Epide- mass index from infancy to 18 years of age:

miol 2012; 27: 215–224. prenatal determinants and health conse- 2 Withrow D, Alter DA: The economic burden quences. J Epidemiol Community Health

of obesity worldwide: a systematic review of 2014; 68: 934–941.

the direct costs of obesity. Obes Rev 2011; 12: 17 Clarke SF, Murphy EF, Nilaweera K, et al: 131–141. The gut microbiota and its relationship to 3 Public Health England: Obesity and Health. diet and obesity: new insights. Gut Microbes

2013, http://www.noo.org.uk/NOO_about_ 2012; 3: 186–202. obesity/obesity_and_health. 18 Duncan SH, Belenguer A, Holtrop G, et al: 4 Health and Social Care Information Centre. Reduced dietary intake of carbohydrates by Obesity. 2013, http://www.hscic.gov.uk/. obese subjects results in decreased concentra- 5 Adair LS: Child and adolescent obesity: epi- tions of butyrate and butyrate-producing demiology and developmental perspectives. bacteria in feces. Appl Environ Microbiol

Physiol Behav 2008; 94: 8–16. 2007; 73: 1073–1078. 6 Adair LS: Methods appropriate for studying 19 World Allergy Organization. WAO White the relationship of breast-feeding to obesity. J Book on Allergy. 2011, http://www.world

Nutr 2009; 139: 408S–411S. allergy.org/UserFiles/file/WAO-White-Book- 7 Park MH, Falconer C, Viner RM, Kinra S: on-Allergy_web.pdf. The impact of childhood obesity on morbid- 20 Gupta R, Holdford D, Bilaver L, et al: The ity and mortality in adulthood: a systematic economic impact of childhood food allergy in

review. Obes Rev 2012; 13: 985–1000. the United States. JAMA Pediatr 2003; 167: 8 Lobstein T, Jackson-Leach R, Moodie ML, et 1026–1031. al: Child and adolescent obesity: part of a big- 21 Royal College of Physicians: Allergy: The Un-

ger picture. Lancet 2015; 385: 2510–2520. met Need: A Blueprint for Better Patient 9 Chinthapalli K: A third of children finishing Care. A Report of the Royal College of Physi- primary school in England are overweight or cians Working Party on the Provision of Al-

obese. BMJ 2012; 345:e8488. lergy Services in the UK. London, Royal Col- 10 Weng SF, Redsell SA, Nathan D, et al: Esti- lege of Physicians, 2003, http://www.bsaci. mating overweight risk in childhood from org/pdf/allergy_the_unmet_need.pdf.

predictors during infancy. Pediatrics 2013; 22 Björkstén B, Sepp E, Julge K, et al: Allergy 132:e414–e421. development and the intestinal microflora 11 Weng SF, Redsell SA, Swift JA, et al: System- during the first year of life. J Allergy Clin Im-

atic review and meta-analyses of risk factors munol 2001; 108: 516–520. for childhood overweight identifiable during 23 Nwaru BI, Craig LC, Allan K, et al: Breast-

infancy. Arch Dis Child 2012; 97: 1019–1026. feeding and introduction of complementary 12 Brisbois TD, Farmer AP, McCargar LJ: Early foods during infancy in relation to the risk of markers of adult obesity: a review. Obes Rev asthma and atopic diseases up to 10 years.

2012; 13: 347–367. Clin Exp Allergy 2013; 43: 1263–1273. 13 Bogen DL, Hanusa BH, Whitaker RC: The 24 Nurmatov U, Devereux G, Sheikh A: Nutri- effect of breast-feeding with and without for- ents and foods for the primary prevention of mula use on the risk of obesity at 4 years of asthma and allergy: systematic review and

age. Obes Res 2004; 12: 1527–1535. meta-analysis. J Allergy Clin Immunol 2011;

14 Whitaker RC: Predicting preschooler obesity 127: 724–733. at birth: the role of maternal obesity in early 25 Grimshaw KE, Maskell J, Oliver EM, et al:

pregnancy. Pediatrics 2004; 114:e29–e36. Diet and food allergy development during 15 Linabery AM, Nahhas RW, Johnson W, et al: infancy: birth cohort study findings using Stronger influence of maternal than paternal prospective food diary data. J Allergy Clin

obesity on infant and early childhood body Immunol 2014; 133: 511–519. mass index: the Fels Longitudinal Study. Pe- 26 Mintel Reports. http://www.mintel.com/

diatr Obes 2013; 8: 159–169. mintel-reports.

26 Venter · Maslin 27 Black PN, Sharpe S: Dietary fat and asthma: is 30 Kon SK: The effect of commercial steriliza-

there a connection? Eur Respir J 1997; 10: 6– tion on the nutritive value of milk. VII. Con-

12. clusions. J Dairy Res 1938; 9: 207. 28 García AL, Raza S, Parrett A, et al: Nutrition- 31 Turner PJ, Gowland MH, Sharma V, et al: al content of infant commercial weaning Increase in anaphylaxis-related hospitaliza-

foods in the UK. Arch Dis Child 2013; 98: tions but no increase in fatalities: an analysis 793–797. of United Kingdom national anaphylaxis 29 Roduit C, Frei R, Loss G, et al: Development data, 1992–2012. J Allergy Clin Immunol

of atopic dermatitis according to age of onset 2015; 135: 956.e1–963.e1. and association with early-life exposures. J

Allergy Clin Immunol 2012; 130: 130.e5–136. e5.

Health Challenges through Nutrition 27

Allergy

Fewtrell MS, Haschke F, Prescott SL (eds): Preventive Aspects of Early Nutrition. Nestlé Nutr Inst Workshop Ser, vol 85, pp 29–33, (DOI: 10.1159/000439482) Nestec Ltd., Vevey/S. Karger AG., Basel, © 2016

Infant Feeding: Foods, Nutrients and Dietary Strategies to Prevent Allergy

Kirsten Beyer Department of Pediatric Pneumology and Immunology, Charité Universitätmedizin Berlin, Berlin , Germany

Abstract Food allergy is a common disease. In recent years, recommendations for the prevention of food allergy have been shifted from avoidance strategies to active oral tolerance induction. Due to evidence from observational studies, it has been suggested that sensitization occurs via the skin especially in children with atopic dermatitis due to skin barrier defects, whereas early oral introduction of the allergenic food(s) will promote tolerance. The current evidence does not justify recommendations about either withholding or encouraging exposure to potentially allergenic food(s) after 4 months once weaning has commenced, irrespective of atopic heredity. However, intervention studies are currently conducted to prove this hypothesis generated by observational studies. © 2016 Nestec Ltd., Vevey/S. Karger AG, Basel

Introduction

The pooled lifetime prevalence of self-reported food allergy is 17% in Europe [1] . There is some evidence that the prevalence may be increasing [1, 2]. To date, elimination of the offending food is the only treatment option. Although immu- notherapeutic strategies show promising results, they are currently not recommended outside clinical trials as their side effects and long-term efficacy are still under investigation. In parallel, patients at risk of anaphylactic reactions, or their caregivers, need to be prepared to treat themselves or their relatives in case of anaphylactic reactions due to accidental ingestions. Therefore, they need to carry emergency medication such as an epinephrine self-injector at all times. Look- ing at these therapeutic options, prevention methods seem to be very important.

The Route of Sensitization

For many years, it was recommended to avoid highly allergenic food for allergy prevention in children at risk of food allergy as it had been hypothesized that sensitization might occur due to early consumption of the allergenic foods. However, this view has been revised in the last 6 years after reevaluation of the evidence [3]. There is evidence from observational studies that sensitization might not only occur via the oral route but also via the skin in patients with atopic dermatitis who have a defective skin barrier function. There are several factors that point in this direction. First, food sensitization and food allergy is often found in children with atopic dermatitis [4] . Especially patients who develop atopic dermatitis within the first 6 months of life are at high risk of developing food sensitization, which increases with the severity of the eczema [5, 6] . In addition, a correlation between the application of peanut oil-containing oint- ments and the development of peanut allergy in patients with atopic dermatitis has been shown [7]. In parallel, environmental exposure to peanut proteins might be important. Peanut proteins have been shown in house and bed dust [8, 9] , and environmental peanut exposure increases with increasing household peanut consumption [10] . Moreover, peanut consumption in the eating area will result in an increase in peanut protein concentrations in the bed dust within 48 h [8] . It has been shown that children with peanut allergy had a 10-fold higher exposure to peanuts than atopic children without peanut allergy [11] . Finally, in a mouse model, it was demonstrated that disruption of the stratum corneum allows epicutaneous sensitization with ovalbumin or peanut proteins [12] .

Development of Oral Tolerance

Going even a step further, it is currently hypothesized that allergen exposure via the oral route at the right time point will result in tolerance development [6] . In a mouse model, a single oral application of high amounts of peanut, cow’s milk or hen’s egg resulted in oral tolerance and prevented sensitization to these allergens [13] . In humans, data from observational studies suggest a similar mechanism. The prevalence of peanut allergy in Jewish children in Israel is very low despite the fact that peanuts are introduced early into the infants’ diet [14, 15] .

30 Beyer In contrast, peanut allergy in Jewish children in the UK was 10 times higher despite avoidance of peanuts in infancy [14, 15]. Similar results came from other observational cohorts. Late introduction of wheat or hen’s egg increases the risk of developing wheat or hen’s egg allergy [16, 17]. In contrast, infants who were fed early and continuously cow’s milk formula or fish developed tolerance to these allergens [18, 19]. However, data from the UK part of the European birth cohort EuroPrevall showed that infants who were diagnosed with food allergy by the time they were 2 years of age were introduced to solids before 16 weeks of age and were less likely to be receiving breast milk when cow’s milk protein was first introduced into their diet [20] .

Current Recommendations for Allergy Prevention

For primary prevention of allergy, the current advice for all mothers includes a normal diet without restrictions during pregnancy and lactation [21] . For all infants, exclusive breastfeeding is recommended for at least the first 4–6 months of life [21] . If breastfeeding is insufficient or not possible, infants at high risk can be recommended a hypoallergenic formula with a documented preventive effect for the first 4 months [21] . In regard to prevention of atopic diseases, there is no need to avoid introducing complementary foods beyond 4 months of life [21] . Despite the fact that evidence is missing for allergy prevention through avoidance of allergenic food, there is also no strong evidence that early feeding of highly allergenic foods will induce tolerance. Currently, the evidence does not justify recommendations about either withholding or encouraging exposure to potentially allergenic foods after 4 months once weaning has commenced, irrespective of atopic heredity [21] .

Clues from Intervention Studies

Several intervention studies are currently ongoing to verify the hypothesis generated by observational cohorts that early introduction of allergenic food will result in the development of oral tolerance. Four of these studies focus on hen’s egg allergy, two on peanut allergy and one on several food allergens in parallel. The hen’s egg intervention studies aim at the general population or children at risk either by atopic heredity or the presence of atopic dermatitis. Both intervention studies with peanuts focus on children with atopic dermatitis [22] . The first results from the Australian hen’s egg intervention study in children with atopic dermatitis showed that about one third of the infants had allergic reactions upon

Nutritional Strategies to Prevent Allergy 31 introduction of the hen’s egg powder into their diet, with the majority reacting already after the first oral exposure [23]. The study was terminated early with a lower number of children than planned. Although infants who received hen’s egg powder had a lower rate of hen’s egg allergy at age 1 year than the children in the placebo group, the results for this primary end point were not statistically significant [23] . The fact that many infants react already after the first allergen exposure at 4–6 months of age suggests that this time point is already too late for tolerance induction. In this regard, the EAT (Enquiring about Tolerance) study in the UK introduces several allergenic foods in a controlled manner starting at 3 months of age. In contrast for peanut allergy it has been shown very recently in a randomized controlled trial in a country with high rates of peanut allergy that early feeding of peanut products at a median age of 8 months reduced the rate of peanut allergy by 80% in children with severe atopic dermatitis. How- ever, it is not known whether this finding can be extrapolated to lower-risk children or to countries with a lower prevalence of peanut allergy [24]. A common analysis of these intervention studies is planned within the EU-funded project iFAAM (Integrated Approaches to Food Allergen and Allergy Risk Manage- ment). With the results of these studies, a new paragraph in the guidelines of allergy prevention might be written.

Disclosure Statement

Kirsten Beyer has received consulting or speaker’s fees from Nestle, Danone, MedaPharma, ALK, Novartis, Unilever, Allergopharma, MedUpdate, HAL, Novartis, Bausch & Lomb, and funding from the European Union, German Research Foundation, ThermoFisher, Danone, DST and the Foundation for the Treatment of peanut allergy.

References

1 Nwaru BI, Hickstein L, Panesar SS, et al: The 3 Greer FR, Sicherer SH, Burks AW, et al: Ef- epidemiology of food allergy in Europe: a sys- fects of early nutritional interventions on the tematic review and meta-analysis. Allergy development of atopic disease in infants and

2014; 69: 62–75. children: the role of maternal dietary restric- 2 Sicherer SH, Sampson HA: Food allergy: epi- tion, breastfeeding, timing of introduction of demiology, pathogenesis, diagnosis, and complementary foods, and hydrolyzed for-

treatment. J Allergy Clin Immunol 2014; 133: mulas. Pediatrics 2008; 121: 183–191. 291–307; quiz 308. 4 Eigenmann PA, Sicherer SH, Borkowski TA, et al: Prevalence of IgE-mediated food allergy among children with atopic dermatitis. Pedi-

atrics 1998; 101:E8.

32 Beyer 5 Hill DJ, Hosking CS, de Benedictis FM, et al: 15 Toit Du G, Katz Y, Sasieni P, et al: Early con- Confirmation of the association between high sumption of peanuts in infancy is associated levels of immunoglobulin E food sensitiza- with a low prevalence of peanut allergy. J Al-

tion and eczema in infancy: an international lergy Clin Immunol 2008; 122: 984–991.

study. Clin Exp Allergy 2008; 38: 161–168. 16 Poole JA, Barriga K, Leung DYM, et al: Tim- 6 Lack G: Update on risk factors for food al- ing of initial exposure to cereal grains and the

lergy. J Allergy Clin Immunol 2012; 129: risk of wheat allergy. Pediatrics 2006; 117: 1187–1197. 2175–2182. 7 Lack G, Fox D, Northstone K, et al: Factors 17 Koplin JJ, Osborne NJ, Wake M, et al: Can associated with the development of peanut al- early introduction of egg prevent egg allergy

lergy in childhood. N Engl J Med 2003; 348: in infants? A population-based study. J Al-

977–985. lergy Clin Immunol 2010; 126: 807–813. 8 Trendelenburg V, Ahrens B, Wehrmann 18 Kull I, Bergström A, Lilja G, et al: Fish con- A-K, et al: Peanut allergen in house dust of sumption during the first year of life and de- eating area and bed – a risk factor for peanut velopment of allergic diseases during child-

sensitization? Allergy 2013; 68: 1460–1462. hood. Allergy 2006; 61: 1009–1015. 9 Brough HA, Makinson K, Penagos M, et al: 19 Katz Y, Rajuan N, Goldberg MR, et al: Early Distribution of peanut protein in the home exposure to cow’s milk protein is protective

environment. J Allergy Clin Immunol 2013; against IgE-mediated cow’s milk protein

132: 623–629. allergy. J Allergy Clin Immunol 2010; 126:

10 Brough HA, Simpson A, Makinson K, et al: 77. e1–82.e1. Peanut allergy: effect of environmental pea- 20 Grimshaw KEC, Maskell J, Oliver EM, et al: nut exposure in children with filaggrin loss- Introduction of complementary foods and of-function mutations. J Allergy Clin Immu- the relationship to food allergy. Pediatrics

nol 2014; 134: 867.e1–875.e1. 2013; 132:e1529–e1538. 11 Fox AT, Sasieni P, Toit Du G, et al: House- 21 Muraro A, Halken S, Arshad SH, et al: hold peanut consumption as a risk factor for EAACI Food Allergy and Anaphylaxis the development of peanut allergy. J Allergy Guidelines. Primary prevention of food al-

Clin Immunol 2009; 123: 417–423. lergy. Allergy 2014; 69: 590–601. 12 Strid J, Hourihane J, Kimber I, et al: Disrup- 22 Toit Du G, Roberts G, Sayre PH, et al: Identi- tion of the stratum corneum allows potent fying infants at high risk of peanut allergy: epicutaneous immunization with protein an- the Learning Early About Peanut Allergy tigens resulting in a dominant systemic Th2 (LEAP) screening study. J Allergy Clin Im-

response. Eur J Immunol 2004; 34: 2100–2109. munol 2013; 131: 135.e12–143.e12. 13 Strid J, Thomson M, Hourihane J, et al: A 23 Palmer DJ, Metcalfe J, Makrides M, et al: Ear- novel model of sensitization and oral toler- ly regular egg exposure in infants with ecze-

ance to peanut protein. Immunology 2004; ma: a randomized controlled trial. J Allergy

113: 293–303. Clin Immunol 2013; 132: 387.e1–392.e1. 14 Levy Y, Broides A, Segal N, Danon YL: Pea- 24 Du Toit G, Roberts G, Sayre PH, Bahnson nut and tree nut allergy in children: role of HT, Radulovic S, Santos AF, et al: Random-

peanut snacks in Israel? Allergy 2003; 58: ized trial of peanut consumption in infants at

1206–1207. risk for peanut allergy. N Engl J Med 2015;

372: 803–813.

Nutritional Strategies to Prevent Allergy 33

Allergy

Fewtrell MS, Haschke F, Prescott SL (eds): Preventive Aspects of Early Nutrition. Nestlé Nutr Inst Workshop Ser, vol 85, pp 35–53, (DOI: 10.1159/000439484 ) Nestec Ltd., Vevey/S. Karger AG., Basel, © 2016

Using Food and Nutritional Strategies to Induce Tolerance in Food-Allergic Children

Anna Nowak-Węgrzyn Jaffe Food Allergy Institute, Division of Pediatric Allergy, Mount Sinai School of Medicine, New York, NY , USA

Abstract Food allergy is an important and increasing public health problem worldwide, affecting predominantly infants and young children. There is an urgent need to develop effective treatment strategies to restore oral tolerance in food-allergic individuals. Among diverse research approaches, those involving native or heat-modified food proteins are most advanced and are currently being evaluated in clinical trials. Extensively heated (baked) milk and egg diets have already been adopted in clinical practice and benefit the majority of milk- and egg-allergic children. Oral, sublingual and epicutaneous immunotherapy with native foods remain in the sphere of clinical research with encouraging data suggesting that they may induce desensitization in a large proportion of treated patients and potentially permanent tolerance following an adequately long period of treatment. Syn- biotics appear to have the most beneficial role in the prevention of food allergy; Lactoba- cillus rhamnosus GG may promote the development of tolerance to milk in allergic infants. © 2016 Nestec Ltd., Vevey/S. Karger AG, Basel

Introduction

Food allergy is defined as an aberrant immunologic response towards an ingested food [1]. Food allergy can be mediated via IgE and non-IgE mechanisms. The prevalence of food allergy has increased worldwide, initially in the highly developed societies with so-called western lifestyle (US, UK, Canada, Australia and Western Europe), followed by Asia and Latin America [2] . With an estimated prevalence of 8% in US children, food allergy has become a serious public health problem. Currently, there is no cure for food allergy; dietary avoidance and management of accidental reactions remain the cornerstone of therapy. Consid- ering the risk of potentially fatal anaphylaxis, nutritional deficiencies, impaired growth, impaired quality of life and increased financial burden, there is an urgent need to develop effective treatment strategies for food allergy [1] .

Oral Tolerance

Food allergy is thought to result from the failure to develop tolerance or a breach in oral tolerance. Oral tolerance is defined as a state of selective systemic unresponsiveness to the ingested food antigens [3]. It is mediated by the gut-associated lymphoid tissue and is thought to depend on the generation of antigen- specific T-regulatory cells (Tregs). In some models, T-cell anergy and deletion are also associated with oral tolerance. It is hypothesized that food antigens are captured in the lamina propria and Peyer’s patches by CD103+ dendritic cells (DC) and are carried to the mesenteric lymph nodes. CD102+ DC induced Tregs (iTregs) by a mechanism dependent on TGF-β, retinoic acid and indole- amine-2,3-dioxygenase. DCs also induce gut-homing IgA-secreting plasma cells through a retinoic acid-dependent mechanism. Gut-homing iTregs are expand- ed in the lamina propria by IL-10-expressing CX3CR1+ macrophages. These iTregs are capable of suppressing systemic immune responses, including allergic sensitization, in an antigen-specific manner. Dietary factors (vitamins A and D) and microbial factors ( Clostridium spp. and Bacteroides fragilis polysaccharide A) promote the generation of iTregs. In contrast, bacterial adjuvants such as Staphylococcus aureus enterotoxin B and exposure to food allergens via an impaired skin barrier, e.g. in atopic dermatitis or filaggrin gene mutations, promote allergic sensitization.

Natural Resolution of Food Allergy

In the majority of children, allergy to cow’s milk, egg, soybean and wheat re- solves with age, a process that is frequently referred to as ‘outgrowing of food allergy’. The exact mechanism of the natural tolerance development in food- allergic children is not known. In general, food-specific IgE antibody levels and skin prick test (SPT) reactivity decrease, whereas casein and β-lactoglobulin-

IgG4 antibodies increase. There is an increase in peripheral blood and T cell- secreted IL-10 [4, 5] .

36 Nowak-Węgrzyn Allergen specific Allergen nonspecific

Native food proteins Modified food proteins Efficacy and safety data available Anti-IgE (in peanut allergy) Anti-IL-5 Efficacy and safety data available Efficacy and safety data available Probiotics Milk, egg and peanut OIT Extensively heated milk and egg diet Safety data available Milk, peanut and hazelnut SLIT Chinese herbs (FAHF-2) Milk and peanut EPIT

Safety data available Safety data available Pilot data available Milk OIT with anti-IgE Escherichia coli expressing Trichuris suis ova therapy Peanut OIT with anti-IgE recombinant modified Ara h 1, 2, 3 rectal vaccine Multiple-food OIT

Fig. 1. Diverse approaches to food allergy therapy tested in clinical trials. Reprinted with permission from Albin and Nowak-Węgrzyn [30] .

Potential Novel Therapies for Food Allergy

Therapies for food allergy aim at restoring oral tolerance. At this time, there is no conclusive evidence that permanent oral tolerance can be induced. The most advanced, diverse strategies for food immunotherapy have been extensively reviewed elsewhere ( fig. 1 ); here, we focus on the strategies for the treatment of IgE-mediated food allergy involving food proteins administered orally, sublin- gually or via skin patch, hypoallergenic formulas and pre-/probiotics [6] .

Baked Milk and Egg Diet It has been shown that high temperature changes protein conformation in cow’s milk and hen’s egg white; casein and ovomucoid are more heat-stable than whey proteins and ovalbumin [7] . A majority of children with milk and egg allergy tolerate extensively heated (baked) products containing milk and egg [8, 9]. Reactivity to baked milk is a marker of a more severe (higher risk of anaphylaxis) and more persistent milk allergy, whereas reactivity to baked egg is not associated with a more severe phenotype of egg allergy. Introduction of baked milk/egg into the diet is well tolerated and appears to accelerate acquisition of tolerance to unheated milk and egg, compared to children who are strict- ly avoiding milk/egg, according to the current standard of care [10, 11] . Baked milk diet is associated with decreasing levels of milk- and casein-specific IgE,

Strategies to Induce Tolerance 37 basophil reactivity and milk SPT wheal diameter, and increasing levels of casein-specific IgG4 . Tolerance to baked milk is associated with increased numbers of circulating FoxP3+ (forkhead box protein 3+) Tregs; these changes are similar to the immune changes occurring during oral immunotherapy (OIT) [12] .

Oral Immunotherapy OIT utilizes the pathways underlying oral tolerance [6] . The aim of food allergy therapy is first to achieve desensitization and then to reestablish permanent oral tolerance. Desensitization is a state of temporary antigen hyporesponsiveness that depends on the regular ingestion of the food. When dosing is interrupted or discontinued, the protective effect of desensitization is lost, and augmentation factors (e.g. viral infection, exercise, use of nonsteroidal anti-inflammatory drugs or menstruation) can trigger reactions to the previously tolerated maintenance dose. The immunologic mechanism of desensitization is not known, but it is associated with decreased reactivity of mast cells (measured by SPT reactivity) and basophils, and increased food-specific serum and salivary IgG4 and IgA antibodies, and initially increased but eventually decreased serum food-specific IgE antibodies. Similarly, the mechanism of permanent tolerance is not known, but it presumably involves the generation of Tregs followed by anergy and/or deletion of effector T cells. Specific Tregs increase and peak at around 12 months, with a subsequent decrease over time. Increased antigen-induced regulatory T- cell function is associated with hypomethylation of FoxP3. To date, no development of permanent tolerance (or even long-term desensitization) due to OIT as opposed to natural acquisition has been conclusively demonstrated. Individualized dosing and longer duration of OIT is associated with increased rates of sustained unresponsiveness [13, 14]. While permanent oral tolerance is the ultimate goal of OIT, desensitization is perceived as beneficial by many patients and parents. Results from two uncontrolled studies suggest that OIT may improve some aspects of life quality, such as dietary and social limitations, risk of accidental exposure and anxiety. The true impact of OIT on quality of life remains to be determined.

Dosing Schedule During OIT, food is mixed with a vehicle and eaten in gradually increasing doses. Dose escalations occur in a controlled setting. Most studies include an initial rapid dose escalation day that is followed by further dose escalation in the clinic on a biweekly schedule until the maintenance dose is achieved. Daily ingestion of tolerated doses during the buildup and maintenance phases occurs at home. The need for daily dosing raises concerns regarding patient

38 Nowak-Węgrzyn Tolerance food challenge

Buildup Maintenance Discontinuation- phase phase elimination diet

SPT wheal diameter, basophil activation

Specific IgE level

Regulatory T cells

Specific IgG4 level

Allergy Tolerance

Fig. 2. OIT time course and immunologic changes occurring during OIT. Reprinted with permission from Albin and Nowak-Węgrzyn [30] .

adherence over long periods of time, as well as unintentional dosing interrup- tions due to illness or travel. In a small clinical trial of milk OIT, twice weekly maintenance dosing was as safe and efficacious as daily maintenance dosing [15] .

Response to Food Oral Immunotherapy Figure 2 summarizes the different phases of OIT, as well as the general trends of immunologic changes seen in OIT. Desensitization failure may be associated with the most severe food allergy phenotype and high baseline food IgE levels, as opposed to desensitization success that may be associated with a mild- er, transient phenotype and higher chances of spontaneous resolution of food allergy.

Clinical Trials of Oral Immunotherapy for Food Allergy Many studies of food OIT have been published to date, but only a few had a rigorous design [16–18] . There is a significant heterogeneity in the study design, inclusion criteria, dosing schedule, duration of therapy and approach to evaluating the persistence of the protection following discontinuation of OIT. Selected relevant clinical trials are presented in detail in table 1 .

Strategies to Induce Tolerance 39 The dropout rate of this study was 35% (8 23 subjects); in 4 subjects, OIT was discontinued by the study physician because of adverse allergic reactions; 2 subjects dropped out because of compliance: 1 because of a severe infection, and 1 was partial responder who did not qualify for the final challenge Adverse reactions were frequent, particularly during the rush protocol; of 317 total doses OIT during rush protocol, 25 doses (7.9%) were associated with objective symptoms; of 6,137 doses OIT during the long-term buildup protocol, 160 doses (2.6%) were associated with objective symptoms Adverse reactions were common, but most were mild In the active arm, 9 of 19 subjects (47%) experienced adverse reactions during the initial day escalation During the buildup phase, adverse reactions occurred following 1.2% of 407 buildup doses None in the placebo group required treatment during initial day escalation or buildup dosing, but 3 (33%) were treated with epinephrine during the final peanut DBPCFC

4 levels 4 intermediate /FoxP3 high (p < 0.001) 4 increased (p < 0.001); after discontinuing OIT for 2 weeks, there was a small decrease in specific IgG The ratio of FoxP3 levels (p < 0.001) Peanut-specific IgE levels as well total IgE levels did not change with OIT CD4+ CD25+ T cells increased at the time of OFC (p = 0.04) in peanut OIT subjects After OIT, all subjects had reductions in IL-5, IL-4 and IL-2 (p < 0.001); after discontinuing OIT for 2 weeks, peanut-induced IL-5, IL-4 and IL-2 secretion was stable in most of the subjects After OIT, there was a small reduction in peanut SPT wheal diameter (p < 0.05), but after discontinuing OIT for 2 weeks, this reduction was no longer evident After OIT, peanut-specific IgG Compared to placebo, the peanut OIT group showed reductions in SPT size (p < 0.001), IL-5 (p = 0.01) and IL-13 (p = 0.02), and increases in peanut- specific IgG 22 of 23 (96%) subjects finished the initial rush protocol, but 17 of these subjects did not achieve the goal dose of 0.5 g peanut with rush therapy (they continued on individualized long-term buildup protocols to reach 0.5 g) Only 14 of 22 subjects (64%) reached a maintenance dose of at least 0.5 g peanut and completed OIT The median tolerated dose after completion of OIT was 1 g peanut During the initial day of rapid dose escalation, 26 of 28 (93%) subjects reached the maximum cumulative dose of 12 mg peanut protein or placebo Sixteen subjects received the full course of peanut OIT and all tolerated 5 g of protein following OIT 0.5 g peanut 4 g peanut Selected clinical trials in food OIT Table 1. Study/subjectsPeanut OIT Blumchen et al. [31], 2010 (n = 23; age 3–14 years) Subjects underwent a rush protocol for 1 week and then received OIT 8 weeks Success rateGoal maintenance dose: protein daily Immunologic changes Side effects/comments protein daily Varshney et al. [32], 2011 (n = 28; age 1–16 years) Subjects were randomized 19:9 to OIT or placebo for 48 weeks Goal maintenance dose:

40 Nowak-Węgrzyn This study is the first to demonstrate sustained unresponsiveness after peanut OIT During the first phase of study, 1 subject discontinued and 5 withdrew from the OIT arm; 4 subjects could not achieve target maintenance dose at 6 months in the OIT arm; control group, 4 subjects withdrew and 1 discontinued The number and nature of adverse events was similar in both groups after treatment, and most events were mild; the most common adverse event was oral itching, which occurred after 6.3% of all doses levels or FoxP3 CD4+ CD25+ T 4 cells between these groups In those who achieved sustained unresponsiveness, subjects had smaller skin test results, lower IgE levels specific for peanut, Ara h 1 and 2, and lower ratios of peanut-specific IgE/total IgE compared to those who did not achieve sustained unresponsiveness There were no differences in peanut IgG After OIT, there was a small reduction in peanut SPT wheal diameter After 24 weeks of OIT, there was an increase in peanut-specific IgE No significant within-patient differences were identified after treatment for basophil activation (although there was a reduction in mean fluorescence intensity and proportion of CD63 at lower peanut concentrations after OIT) Of the 39 subjects originally enrolled, 24 (62%) completed the protocol and had evaluable outcomes All subjects completing the study successfully ingested 5 g of peanut protein without symptoms during desensitization challenge Only 12 of 24 (50%) successfully ingested 5 g of peanut protein 1 month after stopping OIT and achieved sustained unresponsiveness Of those who failed the 5-gram challenge, the median amount of peanut protein ingested cumulatively before the development of symptoms was 3.75 g (range, 1.5–5 g) 84% of subjects receiving OIT were able to ingest 0.8 g of peanut daily for 26 weeks in the first phase, and 91% of subjects from crossover control group were able to ingest 0.8 g of peanut daily for 26 weeks during the second phase 24 of 39 subjects (62%) receiving OIT tolerated 1.4 g of peanut at the end first phase of the study, as compared to 0 46 control subjects (p < 0.001) the protocol 0.8 g peanut Continued Table 1. Study/subjectsVickery et al. [13], 2014 (n = 24; age 1–16 years) Subjects were treated for up to 5 years with peanut OIT, and then treatment was stopped for 1 month to test for sustained Success rateunresponsiveness Goal maintenance dose: was modified over time to permit dose increases to a maximum of 4 g peanut protein daily Immunologic changes Side effects/comments protein daily Anagnostou et al. [33], 2014 (n = 99; age 7–16 years) Phase 1: subjects were randomized 1:1 to OIT or control (standard of care food avoidance) for 26 weeks Phase 2: crossover from the control group to OIT for 26 weeks Goal maintenance dose:

Strategies to Induce Tolerance 41 All subjects in the OIT group had adverse events to some extent: 21 had mild symptoms and 4 moderate symptoms In the control group, 5 subjects had mild-to-moderate adverse events due to inadvertent ingestion of food allergens All 55 subjects completed the initial-day dose escalation, but 7 subjects withdrew before the maintenance phase (5 in the OIT group and 2 placebo group) There were no severe adverse events, and rates highest during the first 10 months of OIT Adverse events were associated with 25% of 11,860 doses of OIT and 3.9% 4,018 placebo The results of this clinical trial raise concerns about the long-term protection (sustained unresponsiveness) afforded by OIT; it remains to be determined whether a higher maintenance dose and longer duration of OIT might improve the rates sustained unresponsiveness antibody levels over time, 4 Compared with subjects who received placebo, those who received OIT had a decreased wheal size on SPT, reduced egg-induced basophil activation and increased egg-specific IgG Allergen-specific IgE levels decreased significantly both in subjects who developed natural tolerance during the elimination diet (p < 0.05) and in those treated with OIT (p < 0.001) whereas no change in egg-specific IgE antibody levels was noted At the 10-month challenge, 22 of 40 subjects (55%) receiving OIT tolerated 5 g of egg white powder, and 0 of 15 subjects (0%) the placebo group passed (p < 0.001) At the 22-month challenge, 30 of 40 subjects (75%) receiving OIT tolerated 10 g of egg white powder (were considered desensitized) At the 24-month challenge (6–8 weeks off OIT), 11 of 29 subjects (28%) tolerated 10 g egg white powder plus whole (were considered to have sustained unresponsiveness); according to the intention-to-treat analysis, 11 of the 40 subjects (28%) in the OIT group passed OFC at 24 months (p = 0.03 vs. placebo) 25 subjects were randomized to milk or egg OIT, and 14 underwent milk OIT 20 subjects were randomized to an elimination diet, 10 of which were avoiding milk In the OIT group, 16 of 25 subjects (64%) were able to integrate the food allergen into their diet; 9 of 25 subjects (36%) showed sustained tolerance after a secondary elimination diet, 3 of 25 (12%) showed tolerance with regular intake, and 4 of 25 (16%) were partial responders who never met the planned full maintenance dose In contrast, only 7 of 20 subjects (35%) showed tolerance (p = 0.05) doses were 2 g egg white Continued Table 1. Study/subjectsEgg OIT Burks et al. [34], 2012 (n = 55; age 5–11 years) Subjects were part of a multicenter double-blind placebo-controlled clinical trial Success rateGoal maintenance dose: powder daily Immunologic changes Side effects/comments Goal maintenance dose: increased (according to individual tolerance) to a maximum dose of 8.25 g cow’s milk protein or 2.8 hen’s egg protein; the maintenance phase mandated a minimum daily maintenance dose of 3.3 g cow’s milk protein and 1.6 g hen’s egg Milk OIT Staden et al. [35], 2007 (n = 45; age <1–12 years) Subjects were randomized to OIT or elimination diet; oral tolerance for all subjects was evaluated after a median 21 months (range 11–59 months)

42 Nowak-Węgrzyn Challenges were not performed in 2 subjects because of ongoing reactions with home dose escalations Adverse reactions were common and unpredictable, but overall rates of reaction decreased over time. In the 2,465 home doses recorded, there were 419 local reactions (17% of doses) Epinephrine was required for 6 (0.2%) doses in 4 subjects Of note, 1 subject developed symptoms consistent with eosinophilic esophagitis Three subjects and their families withdrew during maintenance because of personal problems not related to the study procedures Adverse events included asthma, oral itching, urticaria, rhinitis and abdominal pain, usually associated with concomitant illness or exercise There were 8 events in group A and 9 B (no significant difference between the groups: p = 0.08) Among 2,437 active OIT doses, there were 1,107 reactions (45.4%) Among 1,193 placebo doses, there were 134 reactions (11.2%) Local symptoms were most common and SPT 4 increase 4 levels increased over the follow- 4 up period from 3 to 17 months reactivity between groups A and B End-point titration milk SPT wheal decreased over time (p = 0.0091 vs. baseline) Milk-specific IgE levels did not change in either group Milk IgG levels increased significantly in the OIT group, with a predominant milk-IgG There were no significant differences in milk-specific IgE, IgG Milk-specific IgE levels decreased IgG 19 subjects completed treatment (12 receiving OIT and 7 placebo) After OIT, the median cumulative dose of milk inducing a reaction in the active group increased from 40 to 5,140 mg, but there was no change in the placebo group (p = 0.0003) 29 subjects completed the 12-month study: 15 randomized to daily milk and 14 randomized to twice weekly milk None of the subjects permanently discontinued their maintenance diet based on adverse events After 13–75 weeks of open-label dosing, challenges were conducted on 13 subjects; 6 tolerated 16 g with no reaction, and 7 reacted at 3–16 g 0.5 g cow’s Continued Table 1. Study/subjectsSkripak et al. [36], 2008 (n = 20; age 6–17 years) Subjects were randomized 2:1 to OIT or placebo for 13 weeks Goal maintenance dose: milk protein daily Success rate Immunologic changes Side effects/comments Pajno et al. [15], 2013 (n = 32; age 4–13 years) Subjects who were successfully desensitized with OIT were randomized to two maintenance regimens for 1 year: group A had to ingest 150–200 ml milk daily, and group B 150–200 ml milk twice weekly Narisety et al. [37], 2009 (n = 15; age 6–16 years) Open-label follow-up to Skripak study, which occurred over 13–75 weeks (median 17 weeks) 14 subjects were able to escalate daily doses, with maximum doses ranging from 1 to 16 g (median, 7 g)

Strategies to Induce Tolerance 43 This was a proof-of-concept phase I safety study, and the primary endpoint of study was occurrence allergic reactions Over the study period, there were 5 dropouts for reasons which included noncompliance with study medication and change of residence One subject in the multiple-food OIT group was unable to increase doses due eczema flares and was categorized a treatment failure

4 similar to the monotherapy group Peanut-specific IgE levels were stable after 1 year In the multiple-food group, there was an increase in peanut-specific IgG Rates of reaction per dose did not differ significantly between the two groups (median of 3.3 and 3.7% in multi- single-food OIT group, respectively; p = 0.31) In both groups, most reactions were mild but two 2 reactions requiring epinephrine occurred in each group Those on multiple-food OIT took longer to reach equivalent doses per food (median +4 months; p < 0.0001) 4 g of food Continued Table 1. Study/subjectsMultiple-food OIT Begin et al. [38], 2014 (n = 40; age 4–46 years) Pilot phase I study for multiple-food OIT (15 subjects on peanut OIT, 25 subjects on multiple-food OIT) Success rateGoal maintenance dose: protein per allergen daily Immunologic changes Side effects/comments OFC = Oral food challenge.

44 Nowak-Węgrzyn Oral Immunotherapy Combined with Anti-IgE Monoclonal Antibody Anti-IgE monoclonal antibody is a non-allergen-specific treatment that has been successfully tested as monotherapy for peanut allergy. In one study, subjects treated with the highest dose of anti-IgE showed an increased dose threshold response for peanut after 4 months [19] . In a logical next step, pretreatment with anti-IgE monoclonal antibody has also been shown to be effective and improve safety of food OIT, likely by lowering circulating IgE and downregulating expression of the IgE receptor on antigen-presenting cells (table 2 ). The preliminary data on pretreatment with omalizumab prior to OIT are encouraging, and large clinical trials are testing the efficacy of such approach.

Oral Immunotherapy with Multiple Foods Considering that approximately 30% of children undergoing OIT are allergic to multiple foods and that for aeroallergens, specific immunotherapy frequently combines multiple allergens, a safety trial of a multiple-food OIT was conducted in parallel to peanut OIT in subjects with peanut allergy (table 2 ). Pretreatment with omalizumab may also be an option for multiple-food OIT. These preliminary results suggest that multiple-food OIT may be a practical and safe option for patients with multiple food allergy, but these findings need validation by large and rigorous clinical trials.

Safety of Food Immunotherapy OIT is associated with acute adverse allergic reactions in virtually all patients, which are more common during dose escalation than during maintenance. Most of the reactions are mild and limited to the oropharynx. However, systemic reactions are seen and may occur at previously tolerated doses in the presence of augmentation factors. Outside of acute allergic reactions, many patients complain of gastrointestinal symptoms that may lead to study withdrawal. Gastro- intestinal side effects affect as many as 50% of patients, and those undergoing OIT may be at higher risk of developing eosinophilic esophagitis than the general allergic population [20] .

Sublingual Immunotherapy In sublingual immunotherapy (SLIT), a food allergen extract is kept in the mouth for 2–3 min and then spit out or swallowed. It is generally better tolerated and utilizes significantly lower doses as compared to OIT, but appears to have inferior clinical effects of desensitization [6] . Clinical trials of food SLIT have been reported for milk, peanut, hazelnut and peach (table 3 ). Some of these trials evaluated efficacy in patients with symptoms of pollen food syndrome rather than systemic food allergy. SLIT is generally better

Strategies to Induce Tolerance 45 OIT was more efficacious for desensitization than SLIT alone but was accompanied by more systemic side effects.There were symptoms with 1,802 (29%) of 6,246 SLIT doses and 2,402 (23%) of 10,645 OIT doses However, OIT had significantly more multisystem, upper/lower respiratory tract and gastrointestinal symptoms than SLIT This was a phase I study and the primary objectives were to examine the safety of this approach and to determine whether subjects could be dosed up to 2 g milk within 7–11 weeks of initiating desensitization

4 increased 15-fold 4 increased in all groups Milk-specific IgE and spontaneous histamine release decreased only in the OIT group Titrated milk SPT wheal diameter and basophil activity decreased in all groups Milk-specific IgG Within 1 week of treatment, the CD4+ T-cell response to milk was nearly eliminated Over the following 3 months, the CD4+ T-cell response returned, characterized by a shift from IL-4 to IFN-γ Milk IgE decreased Milk IgG 1 of 10 subjects in the SLIT group, 6 subjects in the SLIT/low-dose OIT group, and 8 of 10 subjects in the SLIT/high-dose OIT group passed the 8-gram milk protein challenge (p = 0.002, SLIT vs. OIT) After avoidance, 6 of 15 subjects (3 in each OIT group) regained reactivity, 2 after only 1 week off therapy The mean frequency for total reactions reported by week 24 was 1.6% (32 reactions of 2,199 doses in total for all 11 subjects) All subjects experienced some adverse events, though most reactions were defined as mild and needed no treatment 1 dose of epinephrine was given during rush desensitization, and 2 subjects received epinephrine at home during the maintenance phase 9 of the 11 subjects tolerated desensitization to a dose of 2 g milk within period of 7–11 weeks 2 g milk protein/day Clinical trials with combination food OIT Table 2. Study/subjectsOIT and SLIT safety efficacy study Keet et al. [22], 2012 (n = 30; age 6–17 years) Randomized clinical trial comparing milk OIT and SLIT with challenge performed after 12 and 60 weeks of maintenance Success rate Immunologic changes Side effects/comments Goal maintenance dose (milk protein/day) SLIT: 7 mg daily Low-dose OIT: 1 g High-dose OIT: 2 g OIT and anti-IgE-safety studies Nadeau et al. [39], 2011 (n = 11; age 7–17 years) Pilot phase I study using omalizumab in combination with oral milk desensitization (desensitization performed 9 weeks after the start of omalizumab treatment) Goal maintenance dose:

46 Nowak-Węgrzyn During initial escalation of rush multiple-food OIT, 13 participants (52%) experienced some symptoms However, subjects required minimal or no rescue therapy Subjects reported 401 reactions per 7,530 home doses (5.3%) with a median of 3.2 reactions per 100 doses 94% of reactions were mild, although there was one severe reaction requiring epinephrine the study (2.0% of 3,502 total peanut doses ingested) During the study, 6 of 13 subjects experienced mild or no allergic reactions, 5 subjects had grade 2 reactions, and subjects had grade 3 reactions, all of which responded rapidly to treatment antibody levels 4 showed median increases of 8.23 mgA/l (p < 0.0001). Peanut SPT decreased by a median of 8 mm (p < 0.0001) Not reported There were a total of 72 reactions during After 52 weeks of therapy, peanut-specific IgE did not change significantly, but peanut-specific IgG All 13 subjects (100%) reached the 0.5-gram peanut desensitization dose on the 1st day (cumulative dose, 992 mg), which was the primary outcome of the study, with minimal or no symptoms 12 of the 13 subjects (92%) reached 4-gram maintenance dose, which was a secondary outcome of the study, requiring a median time of 8 weeks 12 weeks after stopping omalizumab (week 32), 12 subjects underwent a DBPCFC (cumulative dose, 8 g peanut 11 of the subjects (85%) tolerated this challenge, and the 12th subject later passed an open challenge of 8 g peanut After pretreatment with omalizumab, 19 subjects tolerated all 6 steps of the initial escalation day (up to 1,250 mg of combined food proteins) The remaining 6 were started on their highest tolerated dose as their initial daily home doses Subjects reached their maintenance dose of 4 g of protein per allergen at a median 18 weeks, and all subjects reached maintenance dose by 9 months 4 g of food protein 4 g peanut protein/ Continued Table 2. Study/subjectsSchneider et al. [40], 2013 (n = 13; age 8–16 years) Pilot study using omalizumab in combination with oral peanut desensitization (desensitization performed 12 weeks after the start of omalizumab treatment) Success rate Immunologic changes Side effects/comments Goal maintenance dose: day Begin et al. [41], 2014 (n = 25; age 4–15) Phase 1 study using omalizumab with rush multiple-food OIT (omalizumab given 8 weeks prior to and 8 weeks following initiation of rush multiple-food OIT) Goal maintenance dose: per allergen daily

Strategies to Induce Tolerance 47 Table 3. Food SLIT

SLIT

Goal daily Milk: 1 ml or 7 mg maintenance Peanut: 165–2,000 μg, 3,696 μg in one crossover group dose; range from Hazelnut: 22 mg published studies Peach: 10 μg of Pru p 3 Dosing interval Milk, peanut, hazelnut: daily Peach: 3 times per week Allergen uptake Oral mucosal uptake by Langerhans cells

Safety/side effects Local, mild oropharyngeal

Desensitization Milk: 10–70% success rates Keet et al. [22], 2012 10% receiving SLIT were desensitized 60% receiving SLIT/low-dose OIT were desensitized 80% receiving SLIT/high-dose OIT were desensitized Peanut: 44–70% Kim et al. [10], 2011 After 12 months, subjects receiving SLIT safely ingested a median cumulative dose of 1,710 mg of peanut protein (equivalent to 6–7 peanuts), which was significant compared to the placebo group who safely ingested a median cumulative dose of 85 mg (<1 peanut) Hazelnut: Enrique et al. [42], 2005 45% ingested the highest planned dose (20 g) after 8–12 weeks of SLIT 71% ingested the highest planned dose after 4–12 months of SLIT Peach: Fernández-Rivas et al. [43], 2009 Actively treated patients tolerated at least 3 times (3–9 times) more peach in the DBPCFC after 6 months of SLIT Long-term Undetermined unresponsiveness Humoral Milk, peanut, hazelnut, peach: immunologic Variable changes in serum IgE changes Increased serum IgG4 Decreased SPT wheal diameters Basophil activation Milk: with food allergen No changes Peanut: Decreased T-cell responses Peanut: Decreased IL-5 production in PBMCs Decreased basophil activation No difference in Tregs Hazelnut: Increased IL-10 levels

PBMCs = Peripheral blood mononuclear cells.

48 Nowak-Węgrzyn tolerated with lower rates of systemic reactions as compared to OIT, which is partially due to 100- to 1,000-fold-lower starting doses in SLIT compared to OIT. In addition, the oral mucosa contains a limited number of proinflammatory cells, such as mast cells. However, the degree of desensitization appears to be inferior compared to OIT. A retrospective comparison of SLIT versus OIT for peanut showed that peanut OIT resulted in greater changes in peanut-specific IgE, IgG 4 and basophil activation as compared to peanut SLIT, and eliciting dose thresholds were lower during double-blind placebo- controlled food challenge (DBPCFC) at 12 months in patients who underwent SLIT [21] . One study compared these two routes of immunotherapy together in a rigorous, single-center clinical trial. Thirty milk-allergic children (aged 6–17 years) were randomized to SLIT or SLIT followed by low- or high-dose milk OIT [22] . Following therapy, 1 of 10 subjects in the SLIT group, 6 of 10 subjects in the SLIT/low-dose OIT group, and 8 of 10 subjects in the SLIT/high-dose OIT group passed the 8-gram desensitization challenge (p = 0.002, SLIT vs. OIT). Systemic reactions were more common during OIT escalation and maintenance (0.43% with high-dose and 0.08% with low-dose OIT) than during SLIT maintenance (0.02%). Milk OIT was more efficacious for desensitization than SLIT alone, but it was associated with more systemic side effects.

Epicutaneous Immunotherapy Skin has a high capacity for pleiotropic immune responses. The epidermis contains large numbers of antigen-presenting Langerhans cells and is not vascular- ized, reducing the risk of systemic side effects. The epicutaneous route for allergen immunotherapy has been introduced at the beginning of the 20th century, with allergen extracts being applied through scarified skin, with reported treatment success rates of 80% for co-seasonal treatment with aeroallergens. The interest in epicutaneous immunotherapy (EPIT) reemerged in the past decade, spurred by evidence that transcutaneous vaccination can be effective against infectious diseases ( Helicobacter pylori) and cancer. The initial randomized double-blind placebo-controlled study with grass pollen EPIT utilized patches with allergen that were applied weekly to tape-stripped skin [23]. Tape stripping enhances allergen penetration by removing stratum corneum and by activating keratinocytes to secrete IL-1, IL-6, IL-8, TNF-α and IFN-γ, which stimulate maturation and migration of DCs to the draining lymph nodes. Actively treated subjects had 70% improvement in allergic rhinitis symptoms compared with 20% in placebo-treated subjects. Local eczematous reactions under the tape were frequent, but no systemic reactions were observed. These results were confirmed in larger trials in grass-allergic adults and children [24, 25] .

Strategies to Induce Tolerance 49 Instead of tape stripping, penetration of the allergens through the epidermis can be enhanced by skin hydration, by application of a patch that leads to accu- mulation of sweat. This approach has been taken in EPIT studies in food allergy. In a clinical pilot study, an epicutaneous delivery system of milk (Viaskin® patches) was applied to children with milk allergy [26]. In a small pilot study, 18 children (mean age 3.8 years, range 10 months to 7.7 years) with cow’s milk

allergy were randomized 1: 1 to receive active EPIT or placebo. Children received three 48-hour applications (1 mg skimmed milk powder or 1 mg glucose as placebo) via the skin patch per week for 3 months. EPIT-treated children had a trend toward increased threshold doses at the follow-up oral milk challenge, from a mean of 1.8 ml at baseline to 23.6 ml at 3 months; there was no change in the placebo group. There were no severe systemic reactions; however, 1 child had repeated episodes of diarrhea following EPIT. Large clinical trials are currently underway for peanut and milk allergy. EPIT is applied by the patient at home, rotating the application site, and is generally better tolerated and more convenient than OIT.

Synbiotics and Hypoallergenic Formulas Synbiotics include pro- and prebiotics. A probiotic is defined as a live microor- ganism with potential health benefit, and a prebiotic is a nondigestible food ingredient that can stimulate the growth and metabolic activity of beneficial bacteria in the gut. Children with allergy have been shown to have different gut microflora as compared to healthy controls (e.g. higher levels of clostridia and lower levels of bifidobacteria), and some hypothesize that a beneficial gut microflora supports immunoregulation by promoting the development of Tregs. Pre- and probiotics have been studied extensively in the prevention of food allergy, with inconclusive results mostly due to a significant heterogeneity in the study design, including specific strains of bacteria used, and timing and duration of treatment [27] . In a mouse model of food allergy, treatment with a probiotic mixture was effective in the protection against anaphylaxis and caused a redirection of allergen-specific Th2-polarized immune responses towards Th1-Treg responses. Clostridia have been associated with protective immune responses. A first randomized, controlled trial of probiotic supplementation ( Lactobacillus casei CRL431 and Bifidobacterium lactis Bb-12) in the setting of cow’s milk allergy did not demonstrate accelerated tolerance in infants [28] . A recent nonrandomized clinical trial demonstrated that supplementation of an extensively hydrolyzed casein formula (eHCF) with a different probiotic (Lactobacillus GG) accelerated the development of tolerance to milk in infants with milk allergy as confirmed by DBPCFC. Berni Canani et al. [29] prospectively evaluated tolerance to milk

50 Nowak-Węgrzyn after 12 months of treatment with one of the following formulas: eHCF, eHCF with Lactobacillus rhamnosus GG (LGG), hydrolyzed rice formula (HRF), soy or amino acid. After 12 months of treatment, the rate of cow’s milk tolerance was significantly higher (p < 0.05) in the groups receiving eHCF (43.6%) or eHCF + LGG (78.9%) compared with the other groups, i.e. HRF (32.6%), soy formula (23.6%) and amino acid-based formula (18.2%). Subjects with persistent cow’s milk allergy were challenged again after 12 months of treatment, and 60% receiving eHCF alone compared to 45% receiving eHCF + LGG had a positive DBPCFC. It should be noted that there is a natural tendency to outgrow milk allergy over time, particularly in non-IgE-mediated milk allergy, which comprised the majority of subjects in each treatment arm. This study also suggested that if tolerated, formulas with more intact food protein (eHCF or HRF) may have a more beneficial effect on the development of tolerance. However, considering the nonrandomized design, there might have been bias, such as the infants with the most severe milk allergy phenotype were treated with the amino acid-based formula because they did not tolerate more allergenic formulas.

Conclusions

Food allergy is an important and increasing public health problem worldwide, affecting predominantly infants and young children. There is an urgent need to develop effective treatment strategies to restore oral tolerance in food-allergic individuals. Among diverse research approaches, those involving native or heat- modified food proteins are most advanced and are currently being evaluated in clinical trials. Extensively heated (baked) milk and egg diets have already been adopted in clinical practice and have been of benefit in the majority of milk- and egg-allergic children. OIT, SLIT and EPIT with native foods remain in the sphere of clinical research with encouraging data suggesting that they may induce desensitization in the large proportion of treated subjects and potentially permanent tolerance following an adequately long period of treatment. Synbiotics appear to have the most beneficial role in the prevention of food allergy; LGG may promote the development of milk tolerance in allergic infants.

Disclosure Statement

The author received grant support from NIH NIAID, FARE, Nestle, Nutricia, DBV and royalties from UpToDate. She is a member of the Merck and Nutricia advisory boards and speaker for Annenberg and Nestle.

Strategies to Induce Tolerance 51 References

1 Sicherer SH, Sampson HA: Food allergy: epi- 13 Vickery BP, Scurlock AM, Kulis M, et al: Sus- demiology, pathogenesis, diagnosis, and tained unresponsiveness to peanut in subjects

treatment. J Allergy Clin Immunol 2014; 133: who have completed peanut oral immuno-

291–307; quiz 308. therapy. J Allergy Clin Immunol 2014; 133: 2 Prescott SL, Pawankar R, Allen KJ, et al: A 468–475. global survey of changing patterns of food 14 Syed A, Garcia MA, Lyu SC, et al: Peanut oral allergy burden in children. World Allergy immunotherapy results in increased antigen-

Organ J 2013; 6: 21. induced regulatory T-cell function and hypo- 3 Berin MC, Sampson HA: Mucosal immunol- methylation of forkhead box protein 3

ogy of food allergy. Curr Biol 2013; 23:R389– (FOXP3). J Allergy Clin Immunol 2014; 133: R400. 500–510. 4 Sommanus S, Kerddonfak S, Kamchaisatian 15 Pajno GB, Caminiti L, Salzano G, et al: Com- W, et al: Cow’s milk protein allergy: immu- parison between two maintenance feeding nological response in children with cow’s regimens after successful cow’s milk oral de-

milk protein tolerance. Asian Pac J Allergy sensitization. Pediatr Allergy Immunol 2013;

Immunol 2014; 32: 171–177. 24: 376–381. 5 Fishbein AB, Qamar N, Erickson KA, et al: 16 Fisher HR, du Toit G, Lack G: Specific oral Cytokine responses to egg protein in previ- tolerance induction in food allergic children: ously allergic children who developed toler- is oral desensitisation more effective than al- ance naturally. Ann Allergy Asthma Immu- lergen avoidance? A meta-analysis of pub-

nol 2014; 113: 667.e4–670.e4. lished RCTs. Arch Dis Child 2011; 96: 259– 6 Nowak-Wegrzyn A, Albin S: Oral immuno- 264. therapy for food allergy: mechanisms and 17 Nurmatov U, Devereux G, Worth A, et al:

role in management. Clin Exp Allergy 2015; Effectiveness and safety of orally adminis-

45: 368–383. tered immunotherapy for food allergies: a 7 Bloom KA, Huang FR, Bencharitiwong R, et systematic review and meta-analysis. Br J

al: Effect of heat treatment on milk and egg Nutr 2014; 111: 12–22. proteins allergenicity. Pediatr Allergy Immu- 18 Nurmatov U, Venderbosch I, Devereux G, et

nol 2014; 25: 740–746. al Allergen-specific oral immunotherapy for 8 Nowak-Wegrzyn A, Bloom KA, Sicherer SH, peanut allergy. Cochrane Database Syst Rev

et al: Tolerance to extensively heated milk in 2012; 9:CD009014. children with cow’s milk allergy. J Allergy 19 Leung DY, Sampson HA, Yunginger JW, et al:

Clin Immunol 2008; 122: 342–347, 347.e1– Effect of anti-IgE therapy in patients with pea-

347.e2. nut allergy. N Engl J Med 2003; 348: 986–993. 9 Lemon-Mule H, Sampson HA, Sicherer SH, 20 Wood RA, Sampson HA: Oral immunothera- et al: Immunologic changes in children with py for the treatment of peanut allergy: is it egg allergy ingesting extensively heated egg. J ready for prime time? J Allergy Clin Immu-

Allergy Clin Immunol 2008; 122: 977.e1–983. nol Pract 2014; 2: 97–98. e1. 21 Chin SJ, Vickery BP, Kulis MD, et al: Sublin- 10 Kim JS, Nowak-Wegrzyn A, Sicherer SH, et gual versus oral immunotherapy for peanut- al: Dietary baked milk accelerates the resolu- allergic children: a retrospective comparison.

tion of cow’s milk allergy in children. J Al- J Allergy Clin Immunol 2013; 132: 476.e2–478.

lergy Clin Immunol 2011; 128: 125.e2–131.e2. e2. 11 Leonard SA, Sampson HA, Sicherer SH, et al: 22 Keet CA, Frischmeyer-Guerrerio PA, Thya- Dietary baked egg accelerates resolution of garajan A, et al: The safety and efficacy of egg allergy in children. J Allergy Clin Immu- sublingual and oral immunotherapy for milk

nol 2012; 130: 473.e1–480.e1. allergy. J Allergy Clin Immunol 2012; 129: 12 Leonard SA, Caubet JC, Kim JS, et al: Baked 448–455, 455.e1–455.e5. milk- and egg-containing diet in the manage- 23 Senti G, Graf N, Haug S, et al: Epicutaneous ment of milk and egg allergy. J Allergy Clin allergen administration as a novel method of

Immunol Pract 2015; 3: 13–23; quiz 24. allergen-specific immunotherapy. J Allergy

Clin Immunol 2009; 124: 997–1002.

52 Nowak-Węgrzyn 24 Senti G, von Moos S, Tay F, et al: Epicutane- 34 Burks AW, Jones SM, Wood RA, et al: Oral ous allergen-specific immunotherapy amelio- immunotherapy for treatment of egg allergy

rates grass pollen-induced rhinoconjunctivi- in children. N Engl J Med 2012; 367: 233–243. tis: a double-blind, placebo-controlled dose 35 Staden U, Rolinck-Werninghaus C, Brewe F, escalation study. J Allergy Clin Immunol et al: Specific oral tolerance induction in food

2012; 129: 128–135. allergy in children: efficacy and clinical pat-

25 Agostinis F, Forti S, Di Berardino F: Grass terns of reaction. Allergy 2007; 62: 1261–1269. transcutaneous immunotherapy in children 36 Skripak JM, Nash SD, Rowley H, et al: A ran- with seasonal rhinoconjunctivitis. Allergy domized, double-blind, placebo-controlled

2010; 65: 410–411. study of milk oral immunotherapy for cow's

26 Dupont C, Kalach N, Soulaines P, et al: Cow’s milk allergy. J Allergy Clin Immunol 2008;

milk epicutaneous immunotherapy in chil- 122: 1154–1160. dren: a pilot trial of safety, acceptability, and 37 Narisety SD, Skripak JM, Steele P, et al: impact on allergic reactivity. J Allergy Clin Open-label maintenance after milk oral im-

Immunol 2010; 125: 1165–1167. munotherapy for IgE-mediated cow's milk

27 Jensen MP, Meldrum S, Taylor AL, et al: Ear- allergy. J Allergy Clin Immunol 2009; 124: ly probiotic supplementation for allergy pre- 610–612. vention: long-term outcomes. J Allergy Clin 38 Bégin P, Winterroth LC, Dominguez T, et al:

Immunol 2012; 130: 1209.e5–1211.e5. Safety and feasibility of oral immunotherapy 28 Hol J, van Leer EH, Elink Schuurman BE, et to multiple allergens for food allergy. Allergy

al: The acquisition of tolerance toward cow’s Asthma Clin Immunol 2014; 10: 1. milk through probiotic supplementation: a 39 Nadeau KC, Schneider LC, Hoyte L, et al: randomized, controlled trial. J Allergy Clin Rapid oral desensitization in combination

Immunol 2008; 121: 1448–1454. with omalizumab therapy in patients with 29 Berni Canani R, Nocerino R, Terrin G, et al: cow's milk allergy. J Allergy Clin Immunol

Formula selection for management of chil- 2011; 127: 1622–1624. dren with cow’s milk allergy influences the 40 Schneider LC, Rachid R, LeBovidge J, et al: A rate of acquisition of tolerance: a prospective pilot study of omalizumab to facilitate rapid

multicenter study. J Pediatr 2013; 163: 771. e1– oral desensitization in high-risk peanut-aller-

777.e1. gic patients. J Allergy Clin Immunol 2013;

30 Albin S, Nowak-Węgrzyn A: Potential treat- 132: 1368–1374. ments for food allergy. Immunol Allergy Clin 41 Bégin P, Dominguez T, Wilson SP, et al:

North Am 2015; 35: 77–100. Phase 1 results of safety and tolerability in a 31 Blumchen K, Ulbricht H, Staden U, et al: rush oral immunotherapy protocol to mul- Oral peanut immunotherapy in children with tiple foods using omalizumab. Allergy Asth-

peanut anaphylaxis. J Allergy Clin Immunol ma Clin Immunol 2014; 10: 7.

2010; 126: 83.e1–91.e1. 42 Enrique E, Pineda F, Malek T, et al: Sublin- 32 Varshney P, Jones SM, Scurlock AM, et al: A gual immunotherapy for hazelnut food al- randomized controlled study of peanut oral lergy: a randomized, double-blind, placebo- immunotherapy: clinical desensitization and controlled study with a standardized hazelnut

modulation of the allergic response. J Allergy extract. J Allergy Clin Immunol 2005; 116:

Clin Immunol 2011; 127: 654–660. 1073–1079. 33 Anagnostou K, Islam S, King Y, et al: Assess- 43 Fernández-Rivas M, Garrido Fernández S, ing the efficacy of oral immunotherapy for Nadal JA, et al: Randomized double-blind, the desensitisation of peanut allergy in chil- placebo-controlled trial of sublingual immu- dren (STOP II): a phase 2 randomised con- notherapy with a Pru p 3 quantified peach

trolled trial. Lancet 2014; 383: 1297–1304. extract. Allergy 2009; 64: 876–883.

Strategies to Induce Tolerance 53

Allergy

Fewtrell MS, Haschke F, Prescott SL (eds): Preventive Aspects of Early Nutrition. Nestlé Nutr Inst Workshop Ser, vol 85, pp 55–57, (DOI: 10.1159/000441078) Nestec Ltd., Vevey/S. Karger AG., Basel, © 2016

Summary on Allergy

The goal of the opening session of the workshop was to emphasize the importance of early nutrition in all aspects of health, mediated in part through critical early effects on the developing immune system. There was a strong focus on the interconnections between the rising burden of so many modern diseases. In particular, dietary patterns are at the heart of the rise in both early- onset noncommunicable diseases (NCDs), such as allergy, autism, mental ill health and mor- bidities associated with childhood obesity, as well as many later- onset NCDs, such as heart disease, diabetes and dementia. Indeed, it is now well recognized that inflammation and immune dysregulation underpin many of these diseases. In this regard, allergy was ‘the canary in the coal mine’, i.e. a very early measure of the impact of the modern environmental changes on the developing immune system. There are obvious links between immune dysfunction and metabolic dysregulation – including emerging links between allergy and obesity. This highlights the role of nutrition as a common pathway for reducing the burden of many of these conditions. The effects of dietary patterns on the microbiome were a central theme of this session. In particular, the declining intake of dietary fibers in western societies may be equally important as cleaner environments in the ‘dysbiotic drift’ associated with many modern diseases. The associated changes in bacterial species abundance and overall diversity have been associated with changes in the microflora, gut mucosa permeability, immune maturation and metabolic regulation. Short-chain fatty acids are emerging as the ‘molecule of the moment’ based on their powerful anti-inflammatory properties both locally in the gut and system- ically in the modulation of appetite metabolism and neurological function. These microbial digestion products of dietary fibers have recently been shown to have epigenetic effects, including effects on regulatory T cells, which explained their anti-inflammatory properties. These molecules and their metabo- lite-sensing receptors provide novel targets for treatment and protection of many NCDs, including obesity, allergy, autoimmunity and even mental ill health. The importance of food diversity was another platform of discussion in this session. This includes diversity of nutrients, taste and the microbiome delivered with natural foods. There were concerns raised about the impact of commercial processing of foods (including baby foods) not only regarding the reduction of nutrient density and vitamin depletion but also the loss of the normal microbiota found in healthy foods. The importance of consumer engagement was highlighted, and the need to examine these questions in a wider context including our environment and sustainability. There are recognized challenges in balancing the demands of modern life and the needs for convenience foods with the unanticipated consequences. We also examined specific dietary strategies for allergy prevention, including both the role of allergens and the role of specific nutrients (such as probiotics, PUFAs and vitamin D). There are now a number of clinical trials exploring the hypothesis that early introduction of allergenic foods may indeed reduce the risk of allergy by promoting oral tolerance. Preliminary results suggest this may be the case. It is also clear that there is a central role of the skin and mucosal barrier in the pathway to allergic disease, and strategies to improve this (and reduce eczema) may also reduce the risk of food allergy in some children. While there is evidence that probiotics, PUFAs and vitamin D, all have important immunomodulatory properties, there is still insufficient evidence for specific recommendations for allergy prevention. However, these essential nutrients have important roles in development, as reflected by government recommendations for intake of these nutrients to optimize health. Finally, there was a focus on allergy treatment, in particular addressing the increasing burden of food allergy and anaphylaxis particularly in young children. In the last 10 years, there has been a shift towards using allergens in sensitized individuals to prevent anaphylaxis (desensitization – which is temporarily and depends on continued exposure) and ideally the induction of permanent long-term tolerance (which is lasting and independent of allergen). These re- gimes are still under investigation, including safer delivery methods and immunomodulatory factors that may improve safety and effectiveness. In summary, early-life nutrition is arguably the most important early-life exposure in determining future health and disease risk – it is a central determinant of both our lifespan and our ‘health span’. We need clear and consistent advice

56 Prescott for parents and health care professionals. It increasingly likely that both treatment and prevention will need to be ‘personalized’ according to the genetic risk, phenotypic profile and environmental (including geographical) context. Nutri- tional strategies play a central role in both prevention and treatment of all NCDs – from the first moments of life. Susan L. Prescott

Summary on Allergy 57

Obesity Prevention

Fewtrell MS, Haschke F, Prescott SL (eds): Preventive Aspects of Early Nutrition. Nestlé Nutr Inst Workshop Ser, vol 85, pp 59–69, (DOI: 10.1159/000439487) Nestec Ltd., Vevey/S. Karger AG., Basel, © 2016

Interrupting Intergenerational Cycles of Maternal Obesity

Matthew W. Gillman Obesity Prevention Program, Department of Population Medicine, Harvard Medical School/Harvard Pilgrim Health Care Institute, Boston, MA , USA

Abstract Factors operating in the preconception and prenatal periods, such as maternal obesity, excessive gestational weight gain and gestational diabetes, predict a substantial fraction of childhood obesity as well as lifelong adverse health consequences in the mother. These periods may lend themselves to successful intervention to reduce such risk factors because parents may be especially willing to change behavior if it confers health advantages to their children. If effective interventions started before or during pregnancy can be maintained after birth, they have the potential to lower the risk of both maternal obesity in the next pregnancy and obesity in the growing child, thus helping to interrupt maternal and child intergenerational vicious cycles of obesity, diabetes and related cardiometa- bolic health consequences. While this paradigm is appealing, challenges include determining the magnitude, causality and modifiability of these risk factors, and quantifying any adverse consequences of intervention. © 2016 Nestec Ltd., Vevey/S. Karger AG, Basel

Introduction

Women who are obese entering pregnancy, especially those who gain excessive weight during gestation, tend to retain more weight postpartum, putting them at higher risk of later diabetes and cardiovascular disease. If they become pregnant again, they enter pregnancy at a higher weight status, reigniting this cycle. Excessive gestational weight gain (GWG) and gestational diabetes mellitus

This study was funded in part by a grant from the US National Institutes of Health (R37 HD 034568). Pregnancy Postpartum Later life

Unhealthful environment, behaviors

Post- Type 2 Prepregnancy Excess partum + diabetes, obesity GWG weight CVD, etc. retention Mother GDM

Altered fetal Fetus growth and metabolism Adult obesity, Altered Child type 2 Child body obesity composition diabetes, CVD, etc.

Unhealthful environment, behaviors

Fig. 1. Intergenerational vicious cycles of maternal obesity. CVD = Cardiovascular diseases.

(GDM), both more common among obese women, are associated with obesity in the next generation; childhood obesity portends many adverse health outcomes. Moreover, if the obese child is female, she is likely to enter any future pregnancies obese, thus perpetuating intergenerational cycles of obesity and its consequences (fig. 1 ). One could consider interrupting the cycles either once pregnancy starts – on moderating excessive weight gain or preventing or treating GDM – or before pregnancy by having women enter pregnancy at a lower weight status. In this article, I review evidence for the existence of these vicious cycles, which may very well be driving increasing rates of obesity and noncommunicable disease around the globe, and for promising ways to interrupt these cycles.

Maternal Obesity

In almost all regions of the world, obesity prevalence has increased among women in the past 30 years [1] . In the United States, no nationally representative data exist on prepregnancy obesity. However, NHANES data show that rates of overweight (BMI 25–29.9) and obesity (BMI >30) among women aged 20–39 years,

60 Gillman Table 1. Institute of Medicine 2009 GWG recommendations

BMI Recommended total gain category kg/m2 lb kg

Low (<18.5) 28–40 12.5–18 Normal (18.5–24.9) 25–35 11.5–16 Overweight (25–29.9) 15–25 7–11.5 Obese (30+) 11–20 5–9

the prime reproductive period, approximately tripled from the 1970s to 2005 [2] . In 2009–2010, almost one third of women in this age group were obese, and more than half were overweight or obese. Among pregnant women in 20 states participating in the Pregnancy Risk and Monitoring System (PRAMS), prepregnancy obesity rates rose from 17.6 to 20.4% just from 2003 to 2009 [3]. In the CDC’s Pregnancy Nutrition Surveillance System of over 1 million low-income women, prepregnancy obesity prevalence rose from 34.6 to 53.7% from 1988 to 2011, mirroring the overall US data for reproductive-age women [4] . These trends are disturbing because overweight and obesity during pregnancy are associated with several adverse pregnancy outcomes, including GDM, gestational hypertension and preeclampsia, and cesarean section, whose rates increased over the past 2–3 decades [5–10] . GDM is associated with higher lifetime risk of developing type 2 diabetes in the mother, and preeclampsia is associated with later cardiovascular disease [11, 12]. Type 2 diabetes and cardiovascular diseases are leading causes of death in the developed world and have emerged in the developing world as well [13, 14]. These ‘noncommunicable diseases’ now represent the largest fraction of deaths in all regions of the world except those with the lowest income. By 2030, approximately half a billion people will have type 2 diabetes. Obesity in the mother is also strongly related to obesity in the child. Child- hood obesity not only presages adult obesity, diabetes and cardiovascular disease, but also gives rise to adverse consequences in the child [15, 16]. These consequences include psychosocial sequelae, asthma, musculoskeletal problems and even type 2 diabetes among a small proportion of adolescents.

Gestational Weight Gain

In 2009, the US Institute of Medicine promulgated new recommendations for weight gain during pregnancy, which for the first time included guidelines for obese women ( table 1 ). They recommend that lighter women gain more than

Interrupting Intergenerational Cycles of Maternal Obesity 61 0.6 Without adjustment for maternal BMI 0.4 0.2876 With adjustment for maternal BMI 0.19 0.2 0.18 0

–0.2 Child BMI z score –0.2511 –0.4

–0.6 <10 10–14 15–19 20–24 25–29 30–34 35–39 40–44 45 Maternal GWG (lb)

Fig. 2. Association of GWG with offspring BMI before and after adjusting for maternal BMI. Data from the Growing Up Today Study, with permission [19] .

heavier women. Gain below each recommended range is called inadequate, within the range adequate and above the range excessive. In the CDC’s Preg- nancy Nutrition Surveillance System, rates of excessive GWG rose from 37.4% in 1988 to 48.0% in 2011, and inadequate gain dropped from 33.0 to 21.0% [4] . Overweight and obese women are more likely than their leaner counterparts to gain excessive weight [17] . Excessive GWG is associated with childhood obesity. One of the first studies, in the US prebirth cohort study Project Viva, showed that 3-year-old children of mothers who gained excessive weight had a mean BMI z score of 0.52, compared with 0.17 for inadequate gainers [18] . This study was followed by an analysis from the Growing Up Today Study showing that once adjusted for maternal BMI, higher amounts of GWG were linearly associated with BMI in the offspring at the age of 9–14 years ( fig. 2 ) [19] . A recent meta-analysis estimated that excessive weight gain during pregnancy (defined somewhat differently across the 12 included studies) was associated with an odds ratio of 1.3 (95% CI 1.2–1.5) for obesity in children ranging in age from 2 to 18 years [20] . While these observational studies confirm strong associations, they cannot confirm whether excessive GWG actually causes offspring obesity. One of the best ways to judge causality, magnitude of effect and ability to change behavior in the real world is via randomized controlled trials (RCTs) of behavior change to modify GWG with child follow-up. In general, interventions to change lifestyle behaviors during pregnancy offer some advantages. Women may be especially willing to change behavior for the health of the growing fetus. Pregnancy is a few months long; weight control in nonpregnant adults is very often

62 Gillman Table 2. Two randomized trials of behavior changes to limit GWG, LIMIT [23] and Healthy Moms [24], had many similar features

Feature LIMIT Healthy Moms

Enrolment 10–20 weeks <20 weeks Eligibility BMI >25 BMI >30 Weight gain goal 0–5 kg ±3% Intervention led by dietitian Yes, then research assistant Yes, then group leader Diet: theory based, tailored, with calorie calculation Yes: low glycemic index Yes: DASH Clinician involvement Passive Passive Individual vs. group Individual Group Sample size 2,200 120

DASH = Dietary Approaches to Stop Hypertension.

successful for the first few months. Given that the prenatal period is one of frequent clinical visits, one can imagine delivery system changes to enhance intervention effects. In addition, interventions started during pregnancy could be continued after birth, for the benefit of both mother and child. Nevertheless, we must be careful not to pin blame on mothers if they are unable to change behaviors in our ubiquitous obesogenic environment [21] . It is important to consider interventions that at least take the social context of the mother into account if not change the environment itself. Unfortunately, for the case of GWG and offspring obesity, the RCT evidence base is slim to date. Before 2014, only 10 RCTs, comprising just over 1,000 total participants, had attempted to reduce GWG among overweight or obese pregnant women [22] . The pooled effect of these trials was a reduction of 2.21 (95% CI –2.86 to –1.57) kg, but none followed the offspring. 2014 saw the publication of 2 such RCTs that are funded to follow the children to at least 1 year of age, but to date have reported results only until shortly after delivery. These two studies, LIMIT in Australia and Healthy Moms in Oregon (USA), had many features in common ( table 2 ) [23, 24] . Among the goals of both was to achieve GWG that is actually lower than the Institute of Medicine recommends. The major differences were that LIMIT involved individual intervention, whereas in Healthy Moms the primary intervention modality was weekly group visits; in addition, LIMIT had a much larger sample size. Per- haps surprisingly, Healthy Moms demonstrated reductions in GWG and large-for-gestational age births, whereas LIMIT did not affect either. LIMIT did show a modest reduction in macrosomia ( tables 3 , 4 ). We await results from childhood follow-up from both of these studies as well as other RCTs that are ongoing.

Interrupting Intergenerational Cycles of Maternal Obesity 63 Table 3. Selected results from LIMIT [23]

Lifestyle advice Standard care Adjusted treatment effect (n = 1,080) (n = 1,073) β/relative risk (95% CI)1

Total GWG (mean ± SD) at 36 weeks of gestation, kg 9.39±5.74 9.44±5.77 −0.04 (−0.55 to 0.48) Excessive GWG, n 380/897 (42%) 368/871 (42%) 0.99 (0.89 to 1.10) Preterm birth, n 62 (6%) 83 (8%) 0.74 (0.54 to 1.02) Cesarean section, n 370 (34%) 389 (37%) 0.95 (0.85 to 1.06) Preeclampsia, n 56 (5%) 53 (5%) 1.03 (0.71 to 1.47) GDM, n 148 (14%) 120 (11%) 1.21 (0.96 to 1.52) Large for gestational age, n 203 (19%) 224 (21%) 0.90 (0.77 to 1.07) Macrosomia >4 kg, n 164 (15%) 201 (19%) 0.82 (0.68 to 0.99)

1 Relative risks are shown except for total GWG (β value).

Table 4. Selected results from Healthy Moms [24]

Intervention Control Adjusted treatment effect (n = 56) (n = 58) β/relative risk (95% CI)1

Total GWG (mean ± SD) 3 weeks postpartum, kg 1.2±5.6 –2.6±5.5 −3.8 (−5.9 to –1.7) Excessive GWG/week, n 24 (44%) 47 (82%) Not reported Preterm birth, n 4 (7%) 1 (2%) 4.4 (0.4 to 220) Cesarean section, n 21 (38%) 26 (45%) 0.74 (0.35 to 1.56) Preeclampsia/growth hormone, n 5 (9%) 6 (10%) 0.85 (0.24 to 2.96) GDM, n 6 (11%) 7 (12%) 0.87 (0.28 to 2.78) Large for gestational age, n 5 (9%) 15 (26%) 0.28 (0.09 to 0.84) Macrosomia >4 kg, n 6 (11%) 13 (22%) 0.42 (0.15 to 1.18) Small for gestational age, n 3 (5%) 4 (7%) 0.76 (0.11 to 4.76)

1 Relative risks are shown except for total GWG (β value).

Gestational Diabetes

In rodent experiments, inducing GDM causes macrosomia in the offspring, followed by adiposity in the growing animal. If the animal is female, when she gets pregnant she is more likely to get GDM, causing obesity and GDM in the next generation, ad infinitum [25] . Given their longer lifespan and genetic and environmental variability, proving that such cycles exist in humans is more difficult. Two sib-pair studies suggest a causal relationship between diabetes during pregnancy (either preexisting diabetes or GDM) and higher BMI in the offspring, at least from school age to young adulthood. One of them consisted of 58 sibs from 19 Pima (American) Indian families, a population with endemic obesity and

64 Gillman diabetes [26]. The other, albeit comprising 46,000 brothers, was limited to male Swedish conscripts whose obesity rates were low and whose mothers had low rates of obesity and diabetes [27]. The few observational studies in which the investigators adjusted for maternal BMI do not show an association. In a 2011 meta-analysis of three studies, the pooled estimate was only 0.07 (95% CI –0.15 to 0.28) increase in BMI z score among offspring of women with GDM versus no diabetes [28]. Another study that adjusted for maternal BMI, an analysis of the Growing Up Today Study cohort comparing mothers with GDM versus no diabetes, showed an odds ratio of only 1.2 (95% CI 0.8–1.7) for obesity among 14,000 9- to 14-year-old children [29]. Because the mothers are all registered nurses who might be especially attuned to adequate GDM treatment, this modest result raised the possibility that treatment of GDM in pregnancy moderates the otherwise higher risk of obesity in the offspring. The Australian Carbohydrate Intolerance Study in Pregnant Women (ACHOIS), an RCT of treatment of mild-to-moderate GDM, resulted in a reduction in macrosomia and large-for-gestational age, as did another large trial in the US [30, 31]. The hope was that a reduction in large babies would translate into a reduction in large children. In a clever merging of the ACHOIS trial data with data from a South Australian statewide surveillance system of heights and weights among 4- to 5-year-old children, however, BMI was not lower among children of mothers in the intervention versus control groups [32] . Several explanations for this null result could exist. One of the more intriguing is that GDM causes fatter newborns, but this excess fat tends to disappear during infancy [33] . Some data suggest that the GDM-offspring obesity connection only reappears during school age and later [34, 35], suggesting that a ‘second hit’ of some sort is needed to manifest this phenomenon. In addition to treatment of GDM, attention is also focused on its prevention. Some ongoing RCTs aim to modify the diet of pregnant women irrespective of its impact on weight gain [36] . Nevertheless, interventions once pregnancy starts may be a drop in the bucket of GDM prevention. In the Project Viva cohort, in which the investigators measured diet during the first trimester with a validated food frequency questionnaire augmented by interviews for vitamins and supplements, on the whole neither nutrients nor foods nor food groups nor dietary patterns nor even GWG predicted GDM [37, 38]. By far the largest predictor was higher prepregnancy BMI. Other studies suggest that GWG is not a prime determinant of GDM [39], and universally cohort studies confirm that prepregnancy BMI is a strong predictor [40] . In summary, interventions to curb excessive GWG and to prevent (or treat) GDM are promising strategies to lower the risk of maternal and childhood obesity and thus interrupt the intergenerational cycles, but they are not yet proven.

Interrupting Intergenerational Cycles of Maternal Obesity 65 Prepregnancy Obesity

Entering pregnancy at an ideal (vs. higher) body weight is associated with better pregnancy and childhood outcomes. Thus, it might seem to ‘just make sense’ to treat obesity in women who will soon get pregnant. However, it is not clear that maternal obesity causes childhood obesity via its programming effects during pregnancy, rather than through genetic inheritance or shared postnatal environment. The aforementioned Swedish sib-pair study suggests otherwise [27] , as does a previous paper using the FTO gene as an instrumental variable in Men- delian randomization [41] . In addition, treating preconception obesity could carry risks. Rapid short- term weight loss in an adult can have adverse effects [42, 43] ; effects on the con- ceptus are unknown. In the US, it is difficult to recruit women in the preconception period as half of pregnancies are unplanned. Unplanned pregnancies are more common among minorities and women of lower socioeconomic status, who both have more preconception risks including obesogenic risk factors and are less likely to agree to be involved in research studies [44–47] . Although it is attractive to address preconception issues in the interpregnancy period, such studies miss the large proportion of women who have not yet had their first pregnancies. To date, preconception cohort studies are rare, and no intervention studies of weight loss exist [48] . A need exists to study preconception determinants – from society to biology – of outcomes in early and late pregnancy and of longer-term maternal and offspring outcomes. Given recent animal experimental studies showing paternal effects on pregnancy outcomes [49] , preconception studies would do well to include biological as well as social influences of fathers. A range of intervention studies is needed to address benefits and risks in highly controlled studies (efficacy studies), how to achieve effectiveness in real-world settings, and once proven effective, how to scale them for maximum impact.

Conclusion

Interventions starting in pregnancy, and perhaps before, have the potential to reduce risk factors like prepregnancy obesity, excessive GWG and GDM, and thus interrupt intergenerational vicious cycles of maternal and child obesity. However, the implications for clinical and public health practice require additional evidence from observational and intervention studies that follow children and mothers after delivery.

66 Gillman Disclosure Statement

Dr. Gillman receives royalties from Cambridge University Press as co-editor of Maternal Obesity, and from UpToDate as author of the chapter on dietary fat.

References

1 Finucane MM, Stevens GA, Cowan MJ, et al: 10 American College of Obstetricians and Gyne- National, regional, and global trends in body- cologists: ACOG Committee opinion No. mass index since 1980: systematic analysis of 548: weight gain during pregnancy. Obstet

health examination surveys and epidemiolog- Gynecol 2013; 121: 210–212. ical studies with 960 country-years and 9.1 11 Kim C: Maternal outcomes and follow-up

million participants. Lancet 2011; 377: 557– after gestational diabetes mellitus. Diabet

567. Med 2014; 31: 292–301. 2 Rassmussen KM: Public health policies relat- 12 Rich-Edwards JW, Fraser A, Lawlor DA, Ca- ing to obesity in childbearing women; in tov JM: Pregnancy characteristics and wom- Gillman MW, Poston L (eds): Maternal Obe- en’s future cardiovascular health: an under- sity. New York, Cambridge University Press, used opportunity to improve women’s

2012, pp 237–244. health? Epidemiol Rev 2014; 36: 57–70. 3 Fisher SC, Kim SY, Sharma AJ, et al: Is obe- 13 King H, Aubert RE, Herman WH: Global sity still increasing among pregnant women? burden of diabetes, 1995–2025: prevalence, Prepregnancy obesity trends in 20 states, numerical estimates, and projections. Diabe-

2003–2009. Prev Med 2013; 56: 372–378. tes Care 1998; 21: 1414–1431. 4 Table 16D: 2011 Pregnancy Nutrition Sur- 14 Lozano R, Naghavi M, Foreman K, et al: veillance System: Summary of Trends in Ma- Global and regional mortality from 235 ternal Health Indicators. Centers for Disease causes of death for 20 age groups in 1990 and Control, http://www.cdc.gov/pednss/pnss_ta- 2010: a systematic analysis for the Global

bles/pdf/national_table16.pdf (accessed Burden of Disease Study 2010. Lancet 2012;

December 30, 2014). 380: 2095–2128. 5 American College of Obstetricians and Gyne- 15 Ebbeling CB, Pawlak DB, Ludwig DS: Child- cologists: ACOG Committee opinion No. hood obesity: public-health crisis, common

549: obesity in pregnancy. Obstet Gynecol sense cure. Lancet 2002; 360: 473–482.

2013; 121: 213–217. 16 Franks PW, Hanson RL, Knowler WC, et al: 6 Ananth CV, Keyes KM, Wapner RJ: Pre-ec- Childhood obesity, other cardiovascular risk lampsia rates in the United States, 1980– factors, and premature death. N Engl J Med

2010: age-period-cohort analysis. BMJ 2013; 2010; 362: 485–493. 347:f6564. 17 Hunt KJ, Alanis MC, Johnson ER, et al: Ma- 7 Baraban E, McCoy L, Simon P: Increasing ternal pre-pregnancy weight and gestational prevalence of gestational diabetes and preg- weight gain and their association with birth- nancy-related hypertension in Los Angeles weight with a focus on racial differences. Ma-

County, California, 1991–2003. Prev Chronic tern Child Health J 2013; 17: 85–94.

Dis 2008; 5:A77. 18 Oken E, Taveras EM, Kleinman KP, et al: 8 Correa A, Bardenheier B, Elixhauser A, et al: Gestational weight gain and child adiposity at

Trends in prevalence of diabetes among de- age 3 years. Am J Obstet Gynecol 2007; 196: livery hospitalizations, United States, 1993– 322.e1–322.e8.

2009. Matern Child Health J 2015; 19: 635– 19 Oken E, Rifas-Shiman SL, Field AE, et al: Ma- 642. ternal gestational weight gain and offspring

9 Osterman MJK, Martin JA: Changes in cesar- weight in adolescence. Obstet Gynecol 2008;

ean delivery rates by gestational age: United 112: 999–1006.

States, 1996–2011. NCHS Data Brief 2013;

124: 1–8.

Interrupting Intergenerational Cycles of Maternal Obesity 67 20 Tie HT, Xia YY, Zeng YS, et al: Risk of child- 32 Gillman MW, Oakey H, Baghurst PA, et al: hood overweight or obesity associated with Effect of treatment of gestational diabetes excessive weight gain during pregnancy: a mellitus on obesity in the next generation.

meta-analysis. Arch Gynecol Obstet 2014; Diabetes Care 2010; 33: 964–968.

289: 247–257. 33 Parker M, Rifas-Shiman SL, Belfort MB, et al: 21 Richardson SS, Daniels CR, Gillman MW, et Gestational glucose tolerance and cord blood al: Society: don’t blame the mothers. Nature leptin levels predict slower weight gain in

2014; 512: 131–132. early infancy. J Pediatr 2011; 158: 227–233. 22 Oteng-Ntim E, Varma R, Croker H, et al: 34 Regnault N, Gillman MW, Rifas-Shiman SL, Lifestyle interventions for overweight and et al: Sex-specific associations of gestational obese pregnant women to improve pregnan- glucose tolerance with childhood body com-

cy outcome: systematic review and meta- position. Diabetes Care 2013; 36: 3045–3053.

analysis. BMC Med 2012; 10: 47. 35 Silverman BL, Metzger BE, Cho NH, Loeb 23 Dodd JM, Turnbull D, McPhee AJ, et al: An- CA: Impaired glucose tolerance in adolescent tenatal lifestyle advice for women who are offspring of diabetic mothers. Relationship to

overweight or obese: LIMIT randomised trial. fetal hyperinsulinism. Diabetes Care 1995; 18:

BMJ 2014; 348:g1285. 611–617. 24 Vesco KK, Karanja N, King JC, et al: Efficacy 36 Briley AL, Barr S, Badger S, et al: A complex of a group-based dietary intervention for lim- intervention to improve pregnancy outcome iting gestational weight gain among obese in obese women; the UPBEAT randomised women: a randomized trial. Obesity (Silver controlled trial. BMC Pregnancy Childbirth

Spring) 2014; 22: 1989–1996. 2014; 14: 74. 25 Aerts L, Van Assche FA: Animal evidence for 37 Burris HH, Rifas-Shiman SL, Kleinman K, et the transgenerational development of diabe- al: Vitamin D deficiency in pregnancy and

tes mellitus. Int J Biochem Cell Biol 2006; 38: gestational diabetes mellitus. Am J Obstet

894–903. Gynecol 2012; 207: 182.e1–182.e8. 26 Dabelea D, Hanson RL, Lindsay RS, et al: In- 38 Radesky JS, Oken E, Rifas-Shiman SL, et al: trauterine exposure to diabetes conveys risks Diet during early pregnancy and develop- for type 2 diabetes and obesity: a study of dis- ment of gestational diabetes. Paediatr Perinat

cordant sibships. Diabetes 2000; 49: 2208– Epidemiol 2008; 22: 47–59. 2211. 39 Li N, Liu E, Guo J, et al: Maternal prepreg- 27 Lawlor DA, Lichtenstein P, Langstrom N: nancy body mass index and gestational Association of maternal diabetes mellitus in weight gain on pregnancy outcomes. PLoS

pregnancy with offspring adiposity into early One 2013; 8:e82310. adulthood: sibling study in a prospective co- 40 Torloni MR, Betran AP, Horta BL, et al: Pre- hort of 280,866 men from 248,293 families. pregnancy BMI and the risk of gestational

Circulation 2011; 123: 258–265. diabetes: a systematic review of the literature

28 Philipps LH, Santhakumaran S, Gale C, et al: with meta-analysis. Obes Rev 2009; 10: 194– The diabetic pregnancy and offspring BMI in 203. childhood: a systematic review and meta- 41 Lawlor DA, Timpson NJ, Harbord RM, et al:

analysis. Diabetologia 2011; 54: 1957–1966. Exploring the developmental overnutrition 29 Gillman MW, Rifas-Shiman S, Berkey CS, et hypothesis using parental-offspring associa- al: Maternal gestational diabetes, birth tions and FTO as an instrumental variable.

weight, and adolescent obesity. Pediatrics PLoS Med 2008; 5:e33.

2003; 111:e221–e226. 42 Karmon A, Sheiner E: Timing of gestation 30 Crowther CA, Hiller JE, Moss JR, et al: Effect after bariatric surgery: should women delay of treatment of gestational diabetes mellitus pregnancy for at least 1 postoperative year?

on pregnancy outcomes. N Engl J Med 2005; Am J Perinatol 2008; 25: 331–333.

352: 2477–2486. 43 Kjaer MM, Lauenborg J, Breum BM, Nilas L: 31 Landon MB, Spong CY, Thom E, et al: A The risk of adverse pregnancy outcome after multicenter, randomized trial of treatment bariatric surgery: a nationwide register-based for mild gestational diabetes. N Engl J Med matched cohort study. Am J Obstet Gynecol

2009; 361: 1339–1348. 2013; 208: 464.e1–464.e5.

68 Gillman 44 Denny CH, Floyd RL, Green PP, Hayes DK: 47 Hagiwara N, Berry-Bobovski L, Francis C, et Racial and ethnic disparities in preconception al: Unexpected findings in the exploration of risk factors and preconception care. J Wom- African American underrepresentation in

ens Health (Larchmt) 2012; 21: 720–729. biospecimen collection and biobanks. J Can-

45 Finer LB, Zolna MR: Shifts in intended and cer Educ 2014; 29: 580–587. unintended pregnancies in the United States, 48 Temel S, van Voorst SF, Jack BW, et al: Evi-

2001–2008. Am J Public Health 2014; dence-based preconceptional lifestyle inter-

104(suppl 1):S43–S48. ventions. Epidemiol Rev 2014; 36: 19–30. 46 George S, Duran N, Norris K: A systematic 49 Ng SF, Lin RC, Laybutt DR, et al: Chronic review of barriers and facilitators to minority high-fat diet in fathers programs beta-cell research participation among African dysfunction in female rat offspring. Nature

Americans, Latinos, Asian Americans, and 2010; 467: 963–966.

Pacific Islanders. Am J Public Health 2014; 104:e16–e31.

Interrupting Intergenerational Cycles of Maternal Obesity 69

Obesity Prevention

Fewtrell MS, Haschke F, Prescott SL (eds): Preventive Aspects of Early Nutrition. Nestlé Nutr Inst Workshop Ser, vol 85, pp 71–80, (DOI: 10.1159/000439488 ) Nestec Ltd., Vevey/S. Karger AG., Basel, © 2016

Development, Epigenetics and Metabolic Programming

a, b c b, d Keith M. Godfrey · Paula M. Costello · Karen A. Lillycrop a b MRC Lifecourse Epidemiology Unit and NIHR Southampton Biomedical Research Centre, University Hospital Southampton NHS Foundation Trust and University of Southampton, and c Institute of Developmental Sciences, Academic Unit of Human Development and Health, d University of Southampton and Centre for Biological Sciences, University of Southampton, Southampton , UK

Abstract It is now widely recognized that the environment in early life can have important effects on human growth and development, including the ‘programming’ of far-reaching effects on the risk of developing common metabolic and other noncommunicable diseases in later life. We have shown that greater childhood adiposity is associated with higher maternal adiposity, low maternal vitamin D status, excessive gestational weight gain and short duration of breast-feeding; maternal dietary patterns in pregnancy and vitamin D status have been linked with childhood bone mineral content and muscle function. Human studies have identified fetal liver blood flow adaptations and epigenetic changes as potential mechanisms that could link maternal influences with offspring body composition. In experimental studies, there is now substantial evidence that the environment during early life induces altered phenotypes through epigenetic mechanisms. Epigenetic processes, such as DNA methylation, covalent modifications of histones and non-coding RNAs, can induce changes in gene expression without a change in DNA base sequence. Such processes are involved in cell differentiation and genomic imprinting, as well as the phenomenon of developmental plasticity in response to environmental influences. Elucidation of such epigenetic processes may enable early intervention strategies to improve early development and growth. © 2016 Nestec Ltd., Vevey/S. Karger AG, Basel

Developmental Influences and Common Noncommunicable Disorders

Patterns of health, illness and disease are influenced at different stages of the life course by a combination of genetic, epigenetic and environmental factors. Sub- stantial research has demonstrated that during prenatal development, responses to a range of stimuli are likely to ‘program’ the risk of metabolic and other noncommunicable disorders (NCDs), as articulated by the ‘developmental origins of health and disease’ or ‘DOHaD’ paradigm [1] . Subsequent environmental exposures, including nutritional, social, psychological, physical, lifestyle and oc- cupational factors, during infancy, childhood and adult life can modify or condition this risk of disease. Research over many years has shown that impaired fetal development, indicated by low birth weight, is not only associated with adverse childhood outcomes, such as stunting and reduced cognitive function, but also with increased morbidity in adult life from type 2 diabetes mellitus, metabolic syndrome, os- teoporosis, sarcopenia and coronary heart disease [1] . These findings have been extensively replicated and are known to be independent of adult environmental risk factors for these disorders. It is now known that small body size at birth, in addition to reducing later functional capacity, also conditions later responses to the childhood and adult environment. For example, an increased risk of coronary heart disease, hypertension and type 2 diabetes mellitus is associated with slow growth in utero, coupled with accelerated weight gain during childhood. In contrast, boys and girls who were born short and then gained height poorly during childhood have been found to have an increased risk of hip fracture [2] . Figure 1 shows the conceptual framework for ongoing research. The risk of NCDs increases across the life course as a result of declining plasticity and ac- cumulative effects of inadequate responses to new challenges (bottom: trian- gles). The greatest increase occurs in adult life, but the trajectory is set much earlier, being influenced by factors such as the mother’s diet and body composition before and during pregnancy, and fetal, infant and childhood nutrition. In early life, timely interventions can have a large effect on later disease risk (middle right: Early – arrow), while later intervention can remain impactful for vulnerable groups (top right: Late – arrow). Intervention in childhood and adolescence increases biological capital and may have an important impact on the next generation.

Early Development in Relation to Childhood Adiposity and Body Composition

Early life may be a critical period when appetite and regulation of energy balance are programed, with lifelong consequences for the risk of excess adiposity gain. In the UK Southampton Women’s Survey (SWS), we have demonstrated associations of preconception, pregnancy and early postnatal factors with childhood body composition, determined using dual X-ray absorptiometry at birth

72 Godfrey · Costello · Lillycrop Intervention Risk of Old None chronic age NCDs

Adulthood Late intervention Child/adolescent impactful for effective point for vulnerable next generation groups

Mother/infant Early effective point to intervention change trajectory improves functional Preconception capacity and biological capital responses to new challenges Life course

Developmental epigenetic plasticity Inadequate response to new challenges

Fig. 1. Conceptual framework illustrating a life course approach to NCD prevention and treatment.

and at age 4 and 6–7 years. Using the Institute of Medicine gestational weight gain categorization [3] , excessive gain was associated with greater offspring fat mass from birth to age 6–7 years [4] . Higher maternal adiposity and short duration of breast-feeding were independently associated with greater childhood adiposity, and low maternal vitamin D status in pregnancy with greater postnatal adiposity gain [5] . In relation to other aspects of childhood body composition, we have shown that maternal dietary patterns in pregnancy and vitamin D status are linked with childhood bone mineral content and muscle function [6, 7] . Insight into the potential impact of modifying early-life risk factors on later adiposity and obesity can be gained by evaluating their combined effects. Ex- amining the three perinatal risk factors for childhood adiposity mentioned above (excess gestational weight gain, low maternal vitamin D status and short duration of breast-feeding), together with smoking during pregnancy and maternal obesity, we found that among SWS children 15% had no early-life risk factors, 33% had 1, 30% had 2, 16% had 3 and 6% had 4 or 5 risk factors [8] . At both 4 and 6 years, there were positive graded associations between the number of early-life risk factors and adiposity and obesity outcomes. After taking account of confounders, the relative risk of being overweight or obese for children

Development, Epigenetics and Metabolic Programming 73 who had 4 or 5 risk factors was 3.99 (95% CI 1.83–8.67) at 4 years and 4.65 (95% CI 2.29–9.43) at 6 years when compared with children who had none [8] . Oth- er aspects of body composition are also associated with perinatal risk factors; we have identified influences of maternal dietary patterns, fatty acid status, physical activity, smoking and vitamin D status on offspring bone mineral parameters [9] . The long-term effects of maternal influences raise the important question of how best to approach modifying such influences. Studies of health behaviors have shown that few women succeeded in complying with nutrition and lifestyle recommendations for planning a pregnancy, and that the diets of infants and toddlers are strongly associated with the mother’s diet before pregnancy [10, 11] . Less than 6% of SWS women took the recommended amount of folic acid before pregnancy, but both folate intake and red cell folate increased markedly from before pregnancy to 11 weeks of gestation [12] , suggesting that women take adequate amounts only when they know they are pregnant. Prepregnant diet quality, smoking patterns, alcohol consumption and physical activity levels were similar in women who became pregnant within 3 months of interview to those who did not [10] . In early pregnancy, although smoking, alcohol and caffeine consumption fell, dietary quality (assessed using a ‘prudent diet score’ derived from principal component analysis [10, 11]) and consumption of fruit and vegetables hard- ly changed. Notably, younger and more disadvantaged women were the least likely to modify their behaviors when they became pregnant. These observations are now leading to trials of complex interventions to modify maternal diet and lifestyle.

Postnatal Influences on Childhood Body Composition

Few studies have objectively measured physical activity in young children. Using Actiheart devices in collaboration with the MRC Epidemiology Unit, we showed that 4-year-old SWS children who spent more time doing vigorous physical activity had a lower percentage of body fat and fat mass index, but adiposity was not related to sedentary and low-to-moderate-intensity activity [13]. This suggests that activity in young children needs to be vigorous in order to impact on adiposity. Additionally, greater grip strength was found in children who spent fewer hours in sedentary activity each day [14]. These findings have led us to study influences on physical activity, lean mass and muscle (handgrip) strength. We found modest positive associations between maternal vitamin D and n-3 fatty acid status in pregnancy and lean mass in childhood [15, 16], but a stronger association between maternal vitamin D status and grip strength at age 4 years,

74 Godfrey · Costello · Lillycrop independent of the child’s height and level of physical activity [7] . Our analyses also suggest that the postnatal environment is important. For example, at 4 years of age, we found that lean mass was greater in children whose weaning diets had complied with infant feeding guidance (diet based on fruits, vegetables and home-prepared foods, and longer breast-feeding) [17] . The potential for early-life influences to have far-reaching effects on adult health are illustrated by an ongoing follow-up of the Helsinki Birth Cohort Study. A comparison of siblings discordant for duration of breast-feeding has, for example, shown that both short (<2 months) and long ( ≥ 8 months) durations were associated with increased body mass index and greater percent body fat in later life [18] and that higher rates of diabetes were seen in those born before 35 weeks of gestation, compared with term births, even adjusting for birth weight [19] .

Underlying Mechanisms

SWS studies have provided evidence that prenatal developmental adaptations play important roles in the human propensity to deposit fat [20] . Among primates, human neonates have the largest brains but also the highest proportion of body fat. If placental nutrient supply is limited, the fetus faces a dilemma: should resources be allocated to brain growth or to fat deposition for use as a potential postnatal energy reserve? We hypothesized that resolving this dilemma operates at the level of umbilical blood distribution entering the fetal liver. In uncomplicated third-trimester SWS pregnancies, we used ultrasound to measure blood flow perfusing the fetal liver, or bypassing it via the ductus ve- nosus to supply the brain and heart [20] . Across the range of fetal size and independent of the mother’s adiposity and parity, greater liver blood flow was associated with greater offspring fat mass measured by dual X-ray absorptiometry, both in the infant at birth and at age 4 years. In contrast, smaller placentas less able to meet fetal demand for essential nutrients were associated with a brain-sparing flow pattern. This led us to propose that humans evolved a developmental strategy to prioritize nutrient allocation for prenatal fat deposition when the supply of conditionally essential nutrients requiring hepatic intercon- version is limited, switching resource allocation to favor the brain if the supply of essential nutrients is limited. Facilitated placental transfer processes for glucose and other nutrients evolved in environments less affluent than those now prevalent in developed populations, and we propose that in circumstances of maternal adiposity and nutrient excess these processes now also lead to prenatal fat deposition [20] .

Development, Epigenetics and Metabolic Programming 75 Molecular Processes

It has been argued that the associations between fetal or infant growth and later adult disease could represent the multiple (pleiotropic) effects of genes transmitted from mother to child. However, the Early Growth Genetics consortium showed only a small genotypic contribution to birth weight [21] . Rather, it appears that maternally mediated environmental modulation of gene expression in offspring and gene-environment interactions may be more important than purely heritable genetic risk. There is also growing evidence that epigenetic mechanisms (DNA methylation, histone modification and non-coding RNAs) are responsible for tissue-specific gene expression during growth and development, and that these mechanisms underlie the processes of developmental plasticity. Such ‘tuning’ of phenotype has potential adaptive value and fitness advan- tage because it adjusts the phenotype to current circumstances and/or matches responses to the predicted later environment [22]. When the phenotype is mis- matched to the later environment, e.g. from inaccurate nutritional cues from the mother or placenta, or from rapid environmental changes through improved socioeconomic conditions, risk of NCDs increases. Using a transcriptomic approach to examine perinatal influences on gene expression patterns in the umbilical cord, clear differences were found in the transcriptomic pattern of umbilical cords at different gestational ages, even within the normal range [23] . Gestational age-dependent expression was en- riched for signal transduction pathways (e.g. hedgehog) and in genes with roles in cytokine signaling and angiogenesis. In contrast, birth weight, even at extremes, was not a major influence on transcriptomic patterns. Transcriptome changes were found to relate to DNA methylation levels, with possible implications for the risk for NCDs later in life [23] . We have shown that epigenetic gene promoter methylation at birth is associated with the child’s later adiposity and measures of bone health [24, 25] . In these studies, associations were also observed between levels of RXRA methylation and mothers’ carbohydrate intake [24] , supportive of the concept that nutritional conditions in early pregnancy can affect a child’s adiposity in later life. Discovery and validation of perinatal epigenetic biomarkers of metabolic programming are a complex undertaking, in which both replication in independent cohorts and in vitro validation are critical. Figure 2 shows an illustrative work flow for ongoing research that we are undertaking. Alongside systematic genome-wide approaches, insights can also be gained from a candidate approach; recently, we have shown that peroxisomal proliferator-activated receptor-γ coactivator-1α promoter methylation in blood at 5–7 years is stable across childhood and associated with adiposity from 9 to 14 years [26] .

76 Godfrey · Costello · Lillycrop Extract DNA Stratify samples into Enrich methylated DNA from perinatal samples in groups using the childhood and hybridize to genome- initial cohort phenotype of interest wide methylation array

Pathway and network analysis Bayesian analysis on DMROIs to identify genomic of array data to regions for more detailed analysis identify DMROIs

Validation of methylation of CpGs within DMROI by Replication of associations with CpGs of bisulfite pyrosequencing in a CpGs within the region interest in an independent larger number of subjects from functionally validated in group of subjects the initial cohort with childhood vitro, e.g. CpG role in phenotype measurements expression and knockdown of transcripts to examine Replication in an cellular effect independent cohort with information on related phenotypes

Fig. 2. Illustrative work flow for discovery and validation of perinatal epigenetic biomarkers of metabolic programming. DMROI = Differentially methylated regions of interest.

To examine the relative contributions of genetic and prenatal environmental influences to neonatal methylome variation, samples of umbilical cord DNA from 237 neonates in the Growing Up in Singapore Towards Healthy Outcomes cohort study were interrogated on both Illumina Omniexpress plus exome genotyping arrays and Illumina Infinium 450K methylation arrays; 958,178 single nucleotide polymorphisms (SNPs) and 411,107 cytosine-phosphoguanine (CpG) methylation sites were assayed. Our analysis algorithms identified 1,423 regions for which there was substantial interindividual variability in DNA methylation, which we termed variably methylated regions [27] . Principal component analysis of the genotypic data resulted in clear separa- tion of Indian neonates from Chinese and Malays on the first principal component, while Chinese and Malay neonates separated on the second principal component; using a similar analysis of the methylome variably methylated regions, the samples did not separate well by ethnicity on the first or second components. The absence of ethnicity driving the methylome in the same manner as observed for the genotype suggests that the genotype plays a subordinate role in specify- ing methylation levels. We chose nineteen parameters as surrogate measures of the prenatal environment, encompassing the mother’s nutrition, mental health and lifestyle, and analyzed the data using 39 competing statistical models of genetic polymorphism alone, prenatal environment alone and genetic differences interacting with the prenatal environment. The results showed that genetic differences alone best explained 25% of the epigenetic variation between neonates,

Development, Epigenetics and Metabolic Programming 77 with the remaining 75% best explained by the interaction of genetic differences and the prenatal environment [27]. Focusing on the effects of genotype, the strength of genotype and methylation associations was strongest for SNPs affecting CpGs where DNA methylation typically occurs, intermediate for pairs on the same chromosome (cis) and weakest for pairs on different chromosomes (trans); cis pairs tend towards short distances between the SNP and the CpG, with a mode of 0–10 base pairs, or 50–60 base pairs without the disrupting pairs [28] . The findings have fundamental implications for how epigenetic studies will be conducted in the future and for our understanding of how the mother’s nutrition and lifestyle have long-lasting effects on the health of the offspring.

Conclusion

NCDs, including obesity, diabetes, and cardiovascular, chronic lung, mental and neurological disorders, affect all countries and people of all ages, and the WHO has identified them as ‘the world’s biggest killers’. The 2011 High-Level Meeting of the United Nations General Assembly on the Prevention and Control of NCDs noted that maternal and child health is inextricably linked with NCDs and their risk factors. It stressed the importance of taking a life course approach to addressing NCDs [28] . Likewise, the UK Department of Health and other agencies now advocate a life course approach to disease prevention from preconception through pregnancy, infancy, early years, childhood, adolescence and teenage years, and through to adulthood and preparing for older age. Primary prevention of NCDs referred to above requires a deeper understanding of the underlying environmental and societal influences [29] and of the mechanisms underpinning the ‘memo- ry’ of early developmental exposures, alongside with the determination of whether specific reversal strategies can be achieved. Experimental evidence is accruing that endocrine or nutritional interventions during early postnatal life can reverse epigenetic and phenotypic changes induced, for example, by an unbalanced maternal diet during pregnancy [30] . Elucidation of these epigenetic processes may permit perinatal identification of individuals at risk of later NCD and facilitate a new generation of early intervention strategies to mitigate such risk.

Acknowledgments

K.M.G. is supported by the National Institute for Health Research through the NIHR Southampton Biomedical Research Centre and by the European Union’s Seventh Frame- work Program (FP7/2007-2013), project EarlyNutrition under grant agreement No. 289346.

78 Godfrey · Costello · Lillycrop Disclosure Statement

K.M.G. has received reimbursement for speaking at conferences sponsored by companies selling nutritional products, and is part of an academic consortium that has received research funding from Abbott Nutrition, Nestec and Danone.

References

1 Godfrey KM, Inskip HM, Hanson MA: The 11 Fisk CM, Crozier SR, Inskip HM, et al: Influ- long-term effects of prenatal development on ences on the quality of young children’s diets: growth and metabolism. Semin Reprod Med the importance of maternal food choices. Br J

2011; 29: 257–265. Nutr 2011; 105: 287–296. 2 Javaid MK, Eriksson JG, Kajantie E, et al: 12 Blunden CH, Inskip HM, Robinson SM, et al: Growth in childhood predicts hip fracture risk Postpartum depressive symptoms: the B-vita-

in later life. Osteoporos Int 2011; 22: 69–73. min link. Ment Health Fam Med 2012; 9: 5–13. 3 Institute of Medicine: Weight Gain during 13 Collings PJ, Brage S, Ridgway CL, et al: Phys- Pregnancy: Reexamining the Guidelines. ical activity intensity, sedentary time, and Washington, National Academies Press, body composition in preschoolers. Am J Clin

2009. Nutr 2013; 97: 1020–1028. 4 Crozier SR, Inskip HM, Godfrey KM, et al: 14 Inskip H, Macdonald-Wallis C, Kapasi T, et Weight gain in pregnancy and childhood al: Associations between grip strength of par- body composition: findings from the South- ents and their 4-year-old children: findings ampton Women’s Survey. Am J Clin Nutr from the Southampton Women’s Survey.

2010; 91: 1745–1751. Paediatr Perinat Epidemiol 2012; 26: 27–33. 5 Crozier SR, Harvey NC, Inskip HM, et al: 15 Moon RJ, Harvey NC, Robinson SM, et al: Maternal vitamin D status in pregnancy is Maternal plasma polyunsaturated fatty acid associated with adiposity in the offspring: status in late pregnancy is associated with findings from the Southampton Women’s offspring body composition in childhood. J

Survey. Am J Clin Nutr 2012; 96: 57–63. Clin Endocrinol Metab 2013; 98: 299–307. 6 Cole Z, Gale C, Javaid MK, et al: Maternal 16 Harvey NC, Moon RJ, Sayer AA, et al: Mater- dietary patterns during pregnancy and child- nal antenatal vitamin D status and offspring hood bone mass: a longitudinal study. J Bone muscle development: findings from the

Mineral Res 2009; 24: 663–668. Southampton Women’s Survey. J Clin Endo-

7 Harvey NC, Moon RJ, Sayer AA, et al: Mater- crinol Metab 2014; 99: 330–337. nal antenatal vitamin D status and offspring 17 Robinson SM, Marriott LD, Crozier SR, et al: muscle development: findings from the Variations in infant feeding practice are asso- Southampton Women’s Survey. J Clin Endo- ciated with body composition in childhood: a

crinol Metab 2014; 99: 330–337. prospective cohort study. J Clin Endocrinol

8 Robinson SM, Crozier SR, Harvey NC, et al: Metab 2009; 94: 2799–2805. Modifiable early-life risk factors for child- 18 O’Tierney PF, Barker DJ, Osmond C, et al: hood adiposity and overweight: an analysis of Duration of breast-feeding and adiposity in

their combined impact and potential for pre- adult life. J Nutr 2009; 139: 422S–425S.

vention. Am J Clin Nutr 2015; 101: 368–375. 19 Kajantie E, Osmond C, Barker DJ, Eriksson 9 Moon RJ, Harvey NC, Davies JH, Cooper C: JG: Preterm birth – a risk factor for type 2 Vitamin D and skeletal health in infancy and diabetes? The Helsinki Birth Cohort Study.

childhood. Osteoporos Int 2014; 25: 2673–2684. Diabetes Care 2010; 33: 2623–2625. 10 Inskip HM, Crozier SR, Godfrey KM, et al: 20 Godfrey KM, Haugen G, Kiserud T, et al: Fe- Women’s compliance with nutrition and life- tal liver blood flow distribution: role in hu- style recommendations before pregnancy: man developmental strategy to prioritize fat

general population cohort study. BMJ 2009; deposition versus brain development. PLoS

338:b481. One 2012; 7:e41759.

Development, Epigenetics and Metabolic Programming 79 21 Horikoshi M, Yaghootkar H, Mook-Kanamo- 26 Clarke-Harris R, Wilkin TJ, Hosking J, et al: ri DO, et al: New loci associated with birth PGC1α promoter methylation in blood at weight identify genetic links between intra- 5–7 years predicts adiposity from 9 to 14

uterine growth and adult height and metabo- years (EarlyBird 50). Diabetes 2014; 63: 2528–

lism. Nat Genet 2013; 45: 76–82. 2537. 22 Godfrey KM, Lillycrop KA, Burdge GC, et al: 27 Teh AL, Pan H, Chen L, et al: The effect of Non-imprinted epigenetics in fetal and post- genotype and in utero environment on inter- natal development and growth; in Gillman individual variation in neonate DNA methy-

MW, Gluckman PD, Rosenfeld RG (eds): Re- lomes. Genome Res 2014; 24: 1064–1074. cent Advances in Growth Research: Nutri- 28 United Nations: Political Declaration of the tional, Molecular and Endocrine Perspec- High-Level Meeting of the General Assembly tives. Nestle Nutr Inst Workshop Ser. Basel, on the Prevention and Control of Non-Com- Karger, 2013, vol 71, pp 57–63. municable Diseases. New York, United Na- 23 Stünkel W, Pan H, Chew SB, et al: Transcrip- tions, 2011. tome changes affecting Hedgehog and cyto- 29 Inskip H, Baird J, Barker M, et al: Influences kine signalling in the umbilical cord: implica- on adherence to diet and physical activity

tions for disease risk. PLoS One 2012; recommendations in women and children; 7:e39744. insights from six European studies. Ann Nutr

24 Godfrey KM, Sheppard A, Gluckman PD, et Metab 2014; 64: 332–339. al: Epigenetic gene promoter methylation at 30 Hanson MA, Godfrey KM, Lillycrop KA, et birth is associated with child’s later adiposity. al: Developmental plasticity and developmen-

Diabetes 2011; 60: 1528–1534. tal origins of non-communicable disease: the- 25 Harvey NC, Sheppard A, Godfrey KM, et al: oretical considerations and epigenetic mech-

Childhood bone mineral content is associated anisms. Prog Biophys Mol Biol 2011; 106: with methylation status of the RXRA pro- 272–280.

moter at birth. J Bone Miner Res 2014; 29: 600–607.

80 Godfrey · Costello · Lillycrop Obesity Prevention

Fewtrell MS, Haschke F, Prescott SL (eds): Preventive Aspects of Early Nutrition. Nestlé Nutr Inst Workshop Ser, vol 85, pp 81–88, (DOI: 10.1159/000439489) Nestec Ltd., Vevey/S. Karger AG., Basel, © 2016

Endocrine and Metabolic Biomarkers Predicting Early Childhood Obesity Risk

a b a Piotr Socha · Christian Hellmuth · Dariusz Gruszfeld · b b b Hans Demmelmair · Peter Rzehak · Veit Grote · b c Martina Weber · Joaquin Escribano · c d Ricardo Closa-Monasterolo · Elena Dain · e f f Jean-Paul Langhendries · Enrica Riva · Elvira Verduci · b Berthold Koletzko for the European Childhood Obesity Trial Study Group a b Children’s Memorial Health Institute, Warsaw , Poland; Dr. von Hauner Children’s Hospital, Ludwig c Maximilian University of Munich, Munich , Germany; Pediatric Research Unit, Universitat Rovira i d Virgili, Tarragona , Spain; Queen Fabiola Children’s University Hospital, Free University of Brussels, e Brussels , and Department of Pediatrics, Centre Hospitalier Chrétien St. Vincent, Liège-Rocourt , f Belgium; San Paolo Hospital, University of Milan, Milan , Italy

Abstract There is growing evidence of long-term effects of early dietary intervention in infancy on later obesity risk. Many studies showed reduced risk of obesity with breastfeeding in infancy, which could be related to the reduced protein intake with human milk compared to infant formula. In a randomized controlled trial (Childhood Obesity Project), we were able to show that infant formula with reduced protein content results in lower BMI both at 2 and 6 years. These effects seem to be mediated mainly by branched-chain amino acids which stimulate the insulin-like growth factor (IGF)-1 axis and insulin release. In this trial, we also showed an influence of high-protein diet on larger kidney size, which seems to be partly explained by a significant effect of free IGF-1 on kidney volume. The IGF-1 axis was shown to regulate early growth, adipose tissue differentiation and early adipogenesis in animals and in humans. Leptin and adiponectin can also be regarded as important endocrine regulators of obesity. These markers were tested in observational studies. Leptin seems to be closely correlated with BMI but changes in adiponectin require further exploration. Still, there is a lack of good data or some results are contradictory to indicate the role of either leptin or adiponectin in infancy for determining later obesity risk. © 2016 Nestec Ltd., Vevey/S. Karger AG, Basel Introduction

There is an increasing body of evidence that prenatal or early postnatal external factors, such as nutrition, modulate metabolic and epigenetic regulation of body development and function [1] . Many observational studies linked early nutrition with late health outcomes, but the processes by which disease risk is regulated remain to be elucidated. Part of this programming effect seems to be regulated by selected nutrients and hormones (fig. 1 ). Different organ systems have their own critical periods of development. Thus, short- and long- term consequences may differ depending on the time point of nutritional intervention and organ-specific responses [2] . Numerous studies associated breastfeeding with a decreased risk of obesity and other disorders in later life, as compared to infant formula feeding (FOF) [3, 4] . Several factors might be involved in such early programming effects of later obesity risk. We proposed high protein intake with FOF as a major causal factor [5] . Our hypothesis was built on earlier observations that dietary protein intake modulates blood concentrations of insulin-like growth factor (IGF)-1. The IGF-1 axis was shown to regulate early growth, adipose tissue differentiation and early adipogenesis in animals and in humans. The most sensitive window for programming effects is still uncertain and has been proposed to last from the first weeks of life [6] up to the first 2 years, since weight gain both in infancy and in the 2nd year of life have been demonstrated to impact on later obesity [1, 7] . We tested this hypothesis in the European Childhood Obesity Project (CHOP; see Appen- dix) trial, in which infants were randomized double blind to conventional FOF with a relatively high protein supply during the 1st year of life or to a

Metabolic programming Early nutrition

Metabolic Late health programming outcome

Which nutrients? Fig. 1. Metabolic program- Which hormones? ming: process linking early Which life period? nutrition with later health outcomes.

82 Socha et al. protein-reduced intervention formula with equal content of energy achieved by adjustment of total fat content and similar content of other nutrients. Feed- ing conventional formula in the 1st year induced faster weight gain and higher BMI up to 2 years without any difference in length growth compared to infants fed a reduced protein formula or breast-fed infants [8]. Data from the follow-up at 6 years of age confirmed the programming effect of nutritional intervention in the 1st year of life [9]. We investigated metabolic and endocrine mechanisms of obesity programming within CHOP and collected blood and urine samples in the participants of this study at the age of 6 months [10] . Several endocrine markers were studied to explore effects of early diet on later obesity risk, including insulin, IGF-1, leptin and adiponectin. These me- diators are of particular interest as they have been previously proposed to be related to obesity and metabolic syndrome [1] . Our hypothesis was built on earlier observations that diet and mainly protein intake modulate blood concentrations of IGF-1 [11] . The IGF-1 axis was shown to regulate early growth, adipose tissue differentiation and early adipogenesis in animals and humans [12–14] . IGF-1 has a strong structural homology to insulin, which is also reflected in the binding motifs of the IGF-binding proteins (IGFBPs). Amino acids, and in particular the branched-chained amino acids, were proven to stimulate insulin secretion [15] . In the CHOP study, we found consistent changes in biochemical and endocrine markers which fit the hypothesis of metabolic programming of obesity: increased plasma levels of nonessential amino acids, especially branched- chained amino acids, in infants fed the high-protein diet, which were accompanied by increased concentrations of total and free IGF-1, increased urinary C- peptide levels (reflecting increased insulin secretion) and lowered serum glucose levels [9] . Compared to formula-fed children, breastfed children had generally lower plasma amino-acid levels, a less active IGF-1 axis and lower insulin production (fig. 2 ) [9] . Many hormones are differently secreted by males and females, which is usually the case first during puberty but can be also observed in early life. Sexual dimorphism has been observed in many physiologic situations. We analyzed the IGF-1 axis response to high-protein feeding in regard to sex within the CHOP study. Total and free IGF-1 and IGFBP-3 concentrations were higher in girls than in boys. We observed a similar effect of the high- protein formula on the IGF-1 axis; still, the effects tended to be stronger in girls than in boys. The leptin concentration was higher in girls than in boys and was correlated to the IGF-1 axis parameters. We concluded that the endocrine response to a high-protein diet early in life may be modulated by gender [16] .

Biomarkers of Early Childhood Obesity 83 Branched-chain AA Essential AA Nonessential AA 1,000 2,000 3,000 *

800 2,500 1,500 * * 600 2,000

1,000 400 1,500 Serum concentration (μmol/l) concentration Serum 200 500 1,000

Total IGF-1 Free IGF-1

150 2.0 * Low protein 1.5 * 100 High protein 1.0 Breastfed (ng/ml) 50 0.5

Serum concentration Serum 0 0 b

Fig. 2. Serum amino-acid (AA) concentrations (a ) and IGF-1 serum levels ( b) in breastfed infants and infants receiving a low- or high-protein formula in the CHOP trial cohort at the age of 6 months (with permission from Socha et al. [10] ). Outside values are excluded. * p < 0.001 vs. high protein.

Within the CHOP trial, we also showed an influence of high-protein diet on larger kidney size which seems to be partly explained by a significant effect of free IGF-1 on kidney volume [17]. Thus, IGF-1 is involved in protein-induced kidney growth in healthy infants [18] . We analyzed also genetic regulation of IGF-1 secretion and were able to show that there is predominant nutritional regulation of the IGF-1 axis compared to the small influence of single nucleotide polymorphisms [19] . Insulin and IGF-1 were also investigated in other studies. Renault et al. [20] studied the EDEN mother-child cohort (n = 342 subsample) from pregnancy to 1 year of age. Maternal glycemia was correlated with birth weight of their children, and this relationship seemed to be mediated by fetal insulin and fetal

84 Socha et al. IGF-1. Moreover, high fetal insulin correlated inversely with growth during the 1st year of life in girls, which could be explained by partial insulin resistance in girls. Leptin was also investigated in infants whose obesity was assessed in later life. Savino et al. [21] investigated 237 healthy term infants in whom leptin levels were determined at 8 months of age and followed them up to the age of 8.8 years when BMI was measured. In this cohort, breastfed infants had significantly higher serum leptin levels than formula-fed ones. Children who were formula fed in infancy had a significantly higher BMI at the follow-up. Interestingly, the authors found 2.7 ng/ml leptin as cutoff value (median serum leptin level in breastfed infants) below which infants had a higher BMI in childhood, and they concluded that a higher leptin level in infancy may be inversely associated with BMI in childhood. As this is an observational study, it is difficult to infer causality. More- over, the contribution of leptin in breast milk to serum levels was not addressed in this study. Volber et al. [22] described leptin and adiponectin trajectories from birth to 9 years of age and concluded that there are developmental differences in leptin and adiponectin throughout childhood. Adiponectin showed weak correlations between birth and later values, with closer correlations observed between later ages. Similar results were obtained for leptin (significant correlations for children aged 2 and 5; 2 and 9, and 5 and 9 years). The authors identified several potential risk factors for altered childhood adipokine levels, including increased maternal sugar-sweetened beverage consumption during pregnancy and increased child birth weight. Leptin was closely and positively correlated with BMI, but changes in adiponectin require further exploration as the correlations were weaker. The positive correlation of BMI with leptin is not in line with what was described by Savino et al. [21] but the influence of other factors like breastfeeding or different time points of sampling can be considered. Interest- ingly, maternal leptin during pregnancy is also significantly related to infant birth weight. This effect was shown by Misra et al. [23] only in overweight and obese women, in whom an increase in the rate of change in maternal serum leptin in the second half of pregnancy was significantly associated with a decrease in infant birth weight, adjusted for gestational age at delivery. The authors found this effect to be distinct from that of maternal body weight. Further interesting observations come from the EPOCH study, which examined the association between cord blood leptin levels and BMI growth velocity from birth to 12 months of age among infants exposed and not exposed to overnutrition in utero. The authors found a negative correlation between cord blood leptin levels and the rate of change in BMI during the 1st year of life. They tried to explain this relationship by a feedback mechanism during the early postnatal period, with lower baseline leptin levels promoting increased BMI growth velocity [24] .

Biomarkers of Early Childhood Obesity 85 Interesting data were obtained from a randomized trial in small-for-gestational-age (SGA) infants, where a similar concept as in the CHOP trial was tested. The control population was a group of breastfed infants born appropriate for gestational age. SGA infants were breastfed or received FOF. FOF infants were randomized to receive either a standard (FOF1) or protein-rich formula (FOF2). Endocrine markers were assessed at birth and 4 months of age. At 4 months, circulating high-molecular-weight (HMW) adiponectin and IGF-1 seemed to be influenced by nutrition but not the gestational age at birth: the circulating levels of HMW adiponectin and IGF-1 in SGA breastfed infants were comparable with those in breastfed controls who were appropriate for gestational age, but they were elevated in SGA-FOF infants. Moreover, HMW adiponectin levels were higher in SGA-FOF1 than in SGA-FOF2 infants, whereas IGF-1 levels were higher in SGA-FOF2 than in SGA-FOF1 infants. Therefore, it seems that a high-protein diet has a different effect on adiponectin and IGF- 1 – it decreased HMW adiponectin and increased IGF-1 levels [25] . The results in SGA infants concerning IGF-1 correspond well with the results of the CHOP study. Currently, it is difficult to interpret results related to leptin and adiponectin as they come mainly from observational studies. The role of these hormones in predicting obesity is not clear. Breastfeeding effects may be mediated by leptin, but this could not be proven directly. Leptin seems to be closely correlated with BMI at different ages, but there are conflicting data in infancy, while such a relationship has not been shown for adiponectin. Whether leptin and adiponectin responses are mediated by specific nutrients in infancy is unclear and requires further studies. We conclude that evidence from clinical trials firmly demonstrates strong long-term effects of early diet in infancy on later obesity risk. These effects are related to altered plasma concentrations of nonessential amino acids, and secretion of IGF-1 and insulin. Better understanding of the underlying mechanisms of early-life nutritional programming on later health will open new opportunities for prevention.

Appendix

86 Socha et al. Tarragona, Spain); Marcello Giovannini, Enrica Riva, Carlo Agostoni, Silvia Scaglioni, Elvira Verduci, Fiammetta Vecchi and Alice Re Dionigi (University of Milan, Milan, Italy); Jerzy Socha, Anna Dobrzańska, Dariusz Gruszfeld, Piotr Socha, Anna Stolarczyk, Agnieszka Kowalik, Roman Janas and Ewa Pietraszek (Children’s Memorial Health In- stitute, Warsaw, Poland); Emmanuel Perrin (Danone Research Center for Specialized Nutrition, Schiphol, The Netherlands); Helfried Groebe, Anna Reith, and Renate Hof- mann (Klinikum Nurnberg Sued, Nurnberg, Germany); Berthold Koletzko, Veit Grote, Martina Weber, Sonia Schiess, Jeannette Beyer, Michaela Fritsch, Uschi Handel, Ingrid Pawellek, Sabine Verwied-Jorky, Iris Hannibal, Hans Demmelmair, Gudrun Haile, Wolfgang Peissner, Ulrike Harder, Franca F. Kirchberg, Melissa Theurich, Peter Rzehak, Christian Hellmuth and Olaf Uhl (Dr. von Hauner Children’s Hospital, Ludwig Maxi- milian University of Munich, Munich Germany), and Rüdiger von Kries (Institute for Social Pediatrics and Adolescent Medicine, University of Munich, Munich, Germany).

Acknowledgments

Work reported herein is carried out with partial financial support from the Commission of the European Communities, the 7th Framework Program, contract FP7-289346- EarlyNutrition and the European Research Council Advanced Grant ERC-2012-AdG – No. 322605 META-GROWTH. This paper does not necessarily reflect the views of the Commission and in no way anticipates the future policy in this area.

Disclosure Statement

For this specific topic the authors do not have any conflict of interest.

References

1 Koletzko B, Brands B, Chourdakis M, et al: 5 Koletzko B, von Kries R, Monasterolo RC, et The Power of Programming and the Early- al: Infant feeding and later obesity risk. Adv

Nutrition project: opportunities for health Exp Med Biol 2009; 646: 15–29. promotion by nutrition during the first thou- 6 Plagemann A, Harder T, Franke K, Kohlhoff sand days of life and beyond. Ann Nutr R: Long-term impact of neonatal breast-feed-

Metab 2014; 64: 141–150. ing on body weight and glucose tolerance in 2 Gruszfeld D, Socha P: Early nutrition and children of diabetic mothers. Diabetes Care

health: short- and long-term outcomes. 2002; 25: 16–22.

World Rev Nutr Diet 2013; 108: 32–39. 7 Ong KK, Loos RJF: Rapid infancy weight gain 3 Oddy WH, Mori TA, Huang RC, et al: Early and subsequent obesity: systematic reviews

infant feeding and adiposity risk: from infan- and hopeful suggestions. Acta Paediatr 2006;

cy to adulthood. Ann Nutr Metab 2014; 64: 95: 904–908. 215–223. 8 Koletzko B, von Kries R, Closa R, et al: Lower 4 Arenz S, Ruckerl R, Koletzko B, von Kries R: protein in infant formula is associated with Breast-feeding and childhood obesity – a sys- lower weight up to age 2 y: a randomized

tematic review. Int J Obes Relat Metab Dis- clinical trial. Am J Clin Nutr 2009; 89: 1836–

ord 2004; 28: 1247–1256. 1845.

Biomarkers of Early Childhood Obesity 87 9 Weber M, Grote V, Closa-Monasterolo R, et 18 Luque V, Escribano J, Grote V, et al: Does al: Lower protein content in infant formula insulin-like growth factor-1 mediate protein- reduces BMI and obesity risk at school age: induced kidney growth in infants? A second- follow-up of a randomized trial. Am J Clin ary analysis from a randomized controlled

Nutr 2014; 99: 1041–1051. trial. Pediatr Res 2013; 74: 223–229. 10 Socha P, Grote V, Gruszfeld D, et al: Milk 19 Rzehak P, Grote V, Lattka E, et al: Associa- protein intake, the metabolic-endocrine re- tions of IGF-1 gene variants and milk protein sponse, and growth in infancy: data from a intake with IGF-I concentrations in infants at randomized clinical trial. Am J Clin Nutr age 6 months – results from a randomized

2011; 94(6 suppl):1776S–1784S. clinical trial. Growth Horm IGF Res 2013; 23: 11 Ketelslegers JM, Maiter D, Maes M, et al: Nu- 149–158. tritional regulation of the growth hormone 20 Renault N, Botton J, Heude B, et al: Higher and insulin-like growth factor-binding pro- cord C-peptide concentrations are associated

teins. Horm Res 1996; 45: 252–257. with slower growth rate in the 1st year of life

12 Smith PJ, Wise LS, Berkowitz R, et al: Insu- in girls but not in boys. Diabetes 2011; 60: lin-like growth factor-I is an essential regula- 2152–2159. tor of the differentiation of 3T3-L1 adipo- 21 Savino F, Liguori SA, Benetti S, et al: High

cytes. J Biol Chem 1988; 263: 9402–9408. serum leptin levels in infancy can potentially 13 Grohmann M, Sabin M, Holly J, et al: Char- predict obesity in childhood, especially in

acterization of differentiated subcutaneous formula-fed infants. Acta Paediatr 2013; and visceral adipose tissue from children: the 102:e455–e459. influences of TNF-alpha and IGF-I. J Lipid 22 Volberg V, Heggeseth B, Harley K, et al: Adi-

Res 2005; 46: 93–103. ponectin and leptin trajectories in Mexican- 14 Nougues J, Reyne Y, Barenton B, et al: Differ- American children from birth to 9 years of

entiation of adipocyte precursors in a serum- age. PLoS One 2013; 8:e77964. free medium is influenced by glucocorticoids 23 Misra VK, Straughen JK, Trudeau S: Mater- and endogenously produced insulin-like nal serum leptin during pregnancy and infant growth factor-I. Int J Obes Relat Metab Dis- birth weight: the influence of maternal over-

ord 1993; 17: 159–167. weight and obesity. Obesity (Silver Spring)

15 Fajans SS, Quibrera R, Pek S, et al: Stimula- 2013; 21: 1064–1069. tion of insulin release in the dog by a nonme- 24 Kaar JL, Brinton JT, Crume T, et al: Leptin tabolizable amino acid. Comparison with leu- levels at birth and infant growth: the EPOCH

cine and arginine. J Clin Endocrinol Metab study. J Dev Orig Health Dis 2014; 5: 214–218.

1971; 33: 35–41. 25 de Zegher F, Sebastiani G, Diaz M, et al: Body 16 Closa-Monasterolo R, Ferre N, Luque V, et composition and circulating high-molecular- al: Sex differences in the endocrine system in weight adiponectin and IGF-I in infants born response to protein intake early in life. Am J small for gestational age: breast- versus for-

Clin Nutr 2011; 94(suppl):1920S–1927S. mula-feeding. Diabetes 2012; 61: 1969–1973. 17 Escribano J, Luque V, Ferre N, et al: In- creased protein intake augments kidney volume and function in healthy infants. Kidney

Int 2011; 79: 783–790.

88 Socha et al. Obesity Prevention

Fewtrell MS, Haschke F, Prescott SL (eds): Preventive Aspects of Early Nutrition. Nestlé Nutr Inst Workshop Ser, vol 85, pp 89–100, (DOI: 10.1159/000439491 ) Nestec Ltd., Vevey/S. Karger AG., Basel, © 2016

Effects of Early Nutrition on the Infant Metabolome

a a a Christian Hellmuth · Olaf Uhl · Franca F. Kirchberg · a a a b Veit Grote · Martina Weber · Peter Rzehak · Clotilde Carlier · c d e Natalia Ferre · Elvira Verduci · Dariusz Gruszfeld · e a Piotr Socha · Berthold Koletzko for the European Childhood Obesity Trial Study Group a Dr. von Hauner Children’s Hospital, Ludwig Maximilian University of Munich, Munich , Germany; b University Children’s Hospital Queen Fabiola, Université Libre de Bruxelles, Brussels , Belgium; c d Pediatric Research Unit, Universitat Rovira i Virgili, Tarragona , Spain; San Paolo Hospital, e University of Milan, Milan , Italy; Children’s Memorial Health Institute, Warsaw , Poland

Abstract Breastfeeding induces a different metabolic and endocrine response than feeding conventional infant formula, and it has also been associated with slower weight gain and reduced disease risk in later life. The underlying programming mechanisms remain to be explored. Breastfeeding has been reported to induce lower levels of insulin, insulin-like growth factor-1 and some amino acids (AAs) than formula feeding. In the Childhood Obe- sity Project (CHOP), infants fed a conventional protein-rich formula had a higher BMI at 2 and 6 years than those fed a protein-reduced formula. At 6 months, higher protein intakes induced increased plasma concentrations of branched-chain AAs (BCAAs) and their oxidation products, short-chain acylcarnitines. With increasing BCAA levels, these short-chain acylcarnitines increased proportionally only until a break point was reached, after which BCAAs seemed to escape their degradation. The resulting marked elevation in BCAA levels with high-protein (HP) intakes appears to contribute to increased insulin levels and to affect β-oxidation of fatty acids. The ratios of long-chain acylcarnitines to free carnitine decreased in infants who received a HP formula, which indicates a reduced initiation of β-oxidation. We conclude that HP intakes inducing high BCAA plasma levels may inhibit fat oxidation and thereby enhance body fat deposition and adiposity. © 2016 Nestec Ltd., Vevey/S. Karger AG, Basel Introduction

The infant’s metabolic response adapts to environmental and particularly dietary exposure, and appears to affect growth, body composition and later disease risks [1] . Compared to feeding a conventional infant formula, breastfeeding was shown to induce different metabolic and endocrine responses, and it has been associated with differences in growth, body composition and disease risk throughout childhood and in adult life [2]. Populations of breastfed (BF) children show a lower prevalence of overweight and obesity at school age. The ‘Ear- ly Protein Hypothesis’ links the amount of protein in infant feeding to weight gain in the first months of life, which is related to obesity risk in childhood and early adulthood [3]. The underlying programming mechanisms remain to be explored. Recent publications point to epigenetic modifications of genes affecting different proteins and hormones, like leptin, insulin-like growth factor (IGF)-1 and insulin [4]. Associations of endocrine and hormonal markers with obesity are discussed by Socha et al. [this vol., pp. 81–88].

Effect of the Infant Formula on the Metabolome

Regardless of the primary mechanism, changes will be induced in the metabolome of infants fed different diets, since metabolites (molecules <1,500 Da) are the downstream products of both genetic and epigenetic alterations as well as environmental factors, including diet. The metabolome is closely related to the phenotype. More important, metabolomics is capable to enhance the understanding of metabolic regulation in response to environmental influences [5] . It is expected that the ‘programming effects’ of infant formula with different contents of protein should be reflected in the metabolome and consequently in the regulation of metabolic pathways. Martin et al. [6] found differences in the metabolism of formula-fed (FF) babies from obese mothers compared to BF infants. Higher short-chain acylcarnitines C2, C3, C4 and amino acids (AAs) were found in stool of FF infants suggesting an increased breakdown of protein in the gut by bacteria. Differences in the urinary metabolome pointed to an increased protein metabolism in FF infants. Furthermore, β-oxidation and ketogenesis were affected by formula feeding. In 1998, Karlsland Akeson et al. [7] reported increased values of some essential AAs in blood plasma of healthy infants at the age of 6 months, who received exclusive breastfeeding until the age of 3 months and afterwards were randomly assigned to infant formulae with either 13, 15 or 18 g/l protein (F13, F15, F18 in table 1 ). Since the sample number was very small and the mothers were allowed to breastfeed as long as they wished and to intro-

90 Hellmuth et al. Table 1. Plasma concentrations of AAs (μM) in FF and BF infants in three different studies

Formula CHOP [11, 12], BeMIM [8], Karlsland Akeson et al. [7], 6 months 4 months 6 months LP HP BF LP HP BF F13 F15 F18 BF

Protein, g/100 kcal 1.8 2.9 – 1.9 2.2 – 1.9 2.2 2.7 – Sample size, n 260 262 158 82 82 92 10 7 8 26 Isoleucine 64 85* 58 69* 80* 61 57 70* 69* 52 Leucine 120 165* 106 127* 143* 121 99 120 117 94 Lysine 166 197* 145 194* 190* 175 149 179 164 152 Methionine 31 35* 24 31* 35* 25 25* 30* 33* 19 Phenylalanine 72* 84* 61 65* 60* 48 53* 65* 60* 44 Threonine 126 142* 119 140 184* 142 108 127* 135* 106 Tryptophan 56 67 60 71* 66 65 49 57 60 56 Valine 214* 304* 172 204* 232* 173 188 224* 225* 157 Alanine 440 420 430 361 382 362 332 350 333 334 Arginine 115 110 113 82 92 83 68 75 74 79 Asparagine 54 58 52 50* 56* 44 44 50 55 42 Aspartate 25 27 26 12 13* 11 10 12 12 10 Glutamine 605* 556* 664 561* 559* 620 594* 582* 603 662 Glutamate 122 115 130 142* 139* 171 89 74 76 80 Glycine 267* 230 220 178 177 170 199 243 213 200 Histidine 105* 107* 88 85 88 90 88 91 95 86 Serine 161* 159* 187 135* 133* 142 149 149 152 146 Tyrosine 83* 101* 66 92* 85 80 53* 65* 60* 44 Citrulline 32* 34* 27 21* 22* 15 22 26 28 20 Ornithine 116 116 121 80* 87 92 88 93 102 87 Proline 316 365* 319 213* 267 251 199 243 213 205

* p < 0.05 vs. BF. p values were obtained by a linear mixed model adjusted for study center and corrected for multiple testing (CHOP); by Kruskal-Wallis tests (BeMIM), and by nonparametric Kruskal-Wallis and Mann-Whit- ney tests and analyses of variance, using the post hoc test of Bonferroni/Dunn (Karlsland Akeson et al. [7]).

duce the assigned formulas gradually, the effect of the formula intake was van- ished. In the BeMIM study, infants of mothers who chose formula feeding received either a standard formula (2.2 g/100 kcal protein) or an intervention formula with lower protein (LP) content (1.89 g/100 kcal protein) and modified protein composition, which was introduced within 28 days after birth [8] . The intervention formula contained additional α-lactalbumin, free phenylalanine, free tryptophan and long-chain polyunsaturated fatty acids. A BF group was also followed for reference. Urea and AA levels, in particular nondispensable AA levels, were higher in the blood plasma of both groups of FF infants ( table 1 ). Nonessential or dispensable AAs were less affected by the different diets or even decreased in the plasma of FF infants, e.g. glutamine, glutamate or serine.

Early Nutrition and the Infant Metabolome 91 Childhood Obesity Project – Lower versus Higher Protein Intake

In a large, double-blind, randomized, clinical intervention trial, we studied the effect of infant and follow-on formula with conventionally higher protein (HP, 2.05 g/100 ml protein) or LP (1.25 g/100 ml protein) contents on infant growth and metabolic responses. In this trial, the HP formula led to higher BMI than the LP formula at 2 years of age [9] and at school age [10] . Total IGF-1 serum levels were increased in the HP group, whereas IGF-binding protein (IGFBP)-2 was lower and IGFBP-3 did not differ significantly between both formula groups at the age of 6 months [11] . At the age of 6 months, HP-fed children showed significantly increased plasma concentrations of nondispensable AAs (table 1 ), including the branched- chain amino acids (BCAAs; Ile, Leu and Val) [11] as well as increased levels of the oxidation products of BCAAs (short-chain acylcarnitines) compared to LP intake as well as to a reference group of BF infants (table 2 ) [12]. Also, urea increased significantly in both the LP and HP groups compared to the BF group.

Branched-Chain Amino-Acid Metabolism in Formula-Fed Infants

Given that elevated levels of BCAAs and their degradation products are associated with infant formula intake, the BCAA metabolism might be the potential key factor in the relation between formula feeding and later obesity development. BCAAs are less metabolized during the first pass in the liver compared to other AAs [13] . In general, dietary proteins are degraded in the intestine to peptides and free AAs, which are resorbed. After intestinal resorption and metabolism, the portal vein transports the AAs to the liver where they undergo first-pass metabolism. However, the key enzyme of BCAA oxidation, branched-chain α-keto acid dehydrogenase (BCKDH), is less present in the liver [13]. Thus, the BCAA output of the liver is enhanced compared to other AAs, and BCAAs are much more increased in the plasma of HP-fed children than other essential AAs. In the skeletal muscle, BCAAs are degraded for energy provision [14] . First, valine, leucine and isoleucine are transaminased by the branched-chain amino transferase to α-keto acids (fig. 1 ). These keto acids are subsequently reduced by BCKDH to short-chain fatty acids, which are bound to short-chain acylcarnitines C4 and C5 [15] . Further degradation products comprise the acylcarnitine

C3, C5-OH and C5: 1. The reduction step via BCKDH is the limiting factor in the degradation of BCAAs [13] . Leucine supplementation increases BCKDH activity [14] to ensure higher degradation of BCAAs in a state of high BCAA availability to keep BCAA levels in a physiological range. In infants in the Childhood

92 Hellmuth et al. Table 2. Means (SD) of plasma concentrations (μM) of short-chain acylcarnitines (Carn) in HP- and LP-fed infants participating in the CHOP trial

Metabolites LP (n = 260) HP (n = 262) p value

Free Carn 38 (7.05) 40 (7.32) <0.0001 Carn C2 5.4 (2.35) 4.8 (2.34) 0.14 Carn C3 313 × 10–3 (0.1) 479 × 10–3 (0.2) <0.0001 Carn C4-OH 72 × 10–3 (0.03) 67 × 10–3 (0.05) 1 Carn C3-OH 23 × 10–3 (0.004) 23 × 10–3 (0.004) 1 Carn C3:1 6.7 × 10–3 (0.002) 6.5 × 10–3 (0.002) 1 Carn C4 128 × 10–3 (0.05) 206 × 10–3 (0.09) <0.0001 Carn C4:1 12 × 10–3 (0.002) 12 × 10–3 (0.002) 1 Carn C5 95 × 10–3 (0.04) 154 × 10–3 (0.06) <0.0001 Carn C5-M-DC 41 × 10–3 (0.006) 39 × 10–3 (0.006) <0.0001 Carn C5-OH 39 × 10–3 (0.009) 45 × 10–3 (0.01) <0.0001 Carn C5:1 18 × 10–3 (0.007) 21 × 10–3 (0.008) <0.0001 Carn C5:1-DC 19 × 10–3 (0.009) 18 × 10–3 (0.01) 1 Carn C5-DC 25 × 10–3 (0.008) 20 × 10–3 (0.007) <0.0001

p values were obtained by a linear mixed model adjusted for study center and corrected for multiple testing (adapted from Kirchberg et al. [12]).

Valine Leucine Isoleucine

C5-oxo C6-oxo C6-oxo

BCKDH

C4 C5 C5

C4:1 C5:1 C5:1

C4-OH C6:1-DC C5-OH

C4-oxo C6:0-OH.DC C5-oxo NAD+

C3 NADH/H+ C3 C2

Fig. 1. Degradation pathway of BCAAs. Leucine activates the rate-limiting enzyme BCKDH. The occurring short-chain acyl chains are bound to free carnitine. oxo = Molecule contains a keto group; OH = molecule contains a hydroxyl group; DC = molecule contains two car- boxyl groups .

Early Nutrition and the Infant Metabolome 93 0.5

0.4

) 0.3 M

C5 (μ 0.2

0.1

0 50100 150 a Ile (μM)

0.5

0.4

) 0.3 M

C5 (μ 0.2

0.1

0 50100 150 200 250 300 b Leu (μM)

0.6

0.5

) 0.4 M 0.3 C4 (μ C4 Fig. 2. The relation between BCAAs (Ile, 0.2 Leu and Val) and their corresponding short-chain acylcarnitine in high-protein 0.1 fed infants indicates a concentration-de- 0 pendent saturation of BCAA catabolism in 200300 400 500 infants. Modified from Kirchberg et al. c Val (μM) [12] .

Obesity Project (CHOP), the European CHOP Trial Study Group (see Appen- dix) demonstrated a saturation of this pathway. Segmented regression models revealed that with increasing BCAA levels, the short-chain acylcarnitines only increased until a break point was reached (fig. 2 ) [12] . After this point, the corresponding short-chain acylcarnitines C4 and C5 did not longer increase with

94 Hellmuth et al. Table 3. Means (SD) of ratios of the long-chain acylcarnitines C14, C16 and C18 to free carnitine (Carn) in HP- and LP-fed infants participating in the CHOP trial

Ratios LP (n = 260) HP (n = 262) p value

C14/free Carn 1.2 × 10–3 (0.0003) 1.0 × 10–3 (0.0004) <0.0001 C16/free Carn 2.6 × 10–3 (0.0007) 2.2 × 10–3 (0.0008) <0.0001 C18/free Carn 0.8 × 10–3 (0.0002) 0.7 × 10–3 (0.0002) <0.0001

p values were obtained from a linear mixed model adjusted for study center and corrected for multiple testing (adapted from Kirchberg et al. [12]).

increasing BCAA concentration. Thus, BCAAs seemed to escape their degradation after a certain point of high plasma BCAA levels, which indicates a saturation of BCAA catabolism in infants. This was especially observed for the HP group, who reached higher plasma levels of BCAAs. To our knowledge, this is the first indication that above a certain plasma concentration, the BCAA degradation pathway becomes saturated. This could potentially be of major biological importance in infants fed a high amount of protein, and where a markedly increased risk of adverse effects mediated through BCAAs may result. Leucine, for instance, is a potent stimulator of insulin secretion [16] . Increased C-peptide/ creatinine ratios in HP fed infants were shown in the CHOP trial [6] . Furthermore, leucine depressed β-oxidation of fatty acids [17] . The ratio of long-chain acylcarnitines to free carnitine decreased in infants who received HP formula in the CHOP trial, which indicates a lower initial step of the β-oxidation ( table 3 ) [12] . Moreover, leucine deprivation resulted in reduced activity of fatty synthase genes [18] . This deregulation of fat metabolism may result in a lipid oversupply, which causes consequently lipotoxicity, insulin resistance and fat storage [19]. Thus, HP and BCAA intake may inhibit fat oxidation and thereby enhance body fat deposition and the risk of adiposity. This would explain the effects of HP feeding on increased weight gain during the first years of live. The absence of significant differences for leucine and isoleucine as well as acylcarnitines C4 and C5 between LP-fed and BF infants underline the influence of HP- feeding on the metabolism.

Other Indispensable Amino Acids

Infant protein supply also affects the metabolism of other AAs. Levels of aromatic AAs (AAAs), particularly phenylalanine, which promotes IGF-1 secretion, were also elevated in the plasma of conventional FF infants ( table 1 ). Levels

Early Nutrition and the Infant Metabolome 95 of the nonessential AAA tyrosine are also increased in the HP group due to transformation of phenylalanine to tyrosine [20]. The trials showed that HP diet resulted in higher levels of AAAs compared to LP diet ( table 1 ). BCAAs and AAAs compete for transportation in mammalian tissues [21] . Therefore, higher values of BCAAs result in a lower uptake of AAAs, e.g. in the brain, and higher AAA plasma levels [22] . Reduction in AAA levels in the brain lower the synthesis and the release of neurotransmitters like serotonin and catecholamines. This affects metabolic pathways. As BCAAs, AAAs are insulinogenic [16] and elevated levels are related to obesity and may predict future diabetes [23] . Regarding the effect of AAAs on IGF-1 levels, AAAs may represent the missing link between HP intake or leucine supplementation and elevated IGF-1 levels [24] . In conclusion, not only the amount of protein in formula, but also the composition and the kind of the protein used are related to the later adverse outcomes in FF infants. Nearly all other essential AAs were elevated in FF infants and were particularly strongly affected by HP diet (table 1 ). Higher concentrations of certain essential AAs, namely leucine, phenylalanine, tyrosine and lysine, are well known to contribute to an elevated insulin secretion [16]. In contrast, a contributory effect of elevated levels of essential AAs on growth hormone and IGF-1 levels may be assumed, but further investigations in human intervention studies are needed to get a more detailed picture of the underlying molecular mechanisms. However, elevated essential AAs and their effect on the secretion of growth factors may be an underlying mechanism of the Early Protein Hypothesis [3] .

Dispensable Amino Acids – The Decrease in Glutamine

Nonessential AAs are less affected by HP diet than indispensable AAs. The less- er influence of the diet on nonessential AAs appears to be due to regulating mechanisms. Since these AAs are endogenously synthesized, the human metabolism can downregulate the biosynthesis of these AAs in times of protein oversupply to keep levels in the tissues and the blood plasma constant. This regulation mainly appears in the intestine and the liver during the first pass, hence, in contrast to oral supplements, direct infusion may affect plasma levels. Not in accord with this hypothesis, glutamine levels are decreased in different studies investigating FF infants ( table 1 ). In the CHOP trial, glutamine was lower in HP-fed infants compared to LP-fed infants, and even the LP-fed infants had lower levels compared to BF infants. An alteration in the urea cycle is assumed, because urea is elevated in FF infants [8, 11, 25] . However, levels of other AAs involved in the urea cycle, namely glutamic acid, aspartic acid, arginine and or-

96 Hellmuth et al. nithine, showed no consistent picture or were not affected by formula diet ( table 1 ), whereas citrulline was elevated in the CHOP and the BeMIM trial. In contrast to glutamine, cells can recycle the other AAs during the urea or the aspartate cycle [26] . Glutamic acid can be recycled at the expense of glutamine. Thus, glutamine may be the only AA with decreased levels by an elevated urea cycle. Elevation in the urea cycle in the formula groups could result from enhanced protein intake and the subsequent higher protein metabolism. Another explanation of the lower glutamine levels in the HP group might be the contribution of glutamine to insulin secretion induced by leucine [27]. Leucine activates glutamate dehydrogenase in pancreatic islets resulting in consumption of glutamic acid. In pancreatic islets, glutamic acid is mainly provided by the intra- cellular conversion of glutamine to glutamic acid [28] . Hence, increased insulin release, enhanced by leucine levels, may decrease glutamine levels.

One Step Further – The Link to Early Weight Gain and Obesity Risk

In infants, metabolites that responded to HP supply have been previously reported as markers for obesity risk. BCAAs, nonesterified fatty acids, organic acids, acylcarnitines and phospholipids were identified as potential biomarkers for obesity in a recent review [29] . This indicates a relation of elevated BCAAs by HP diet to the obese state [21, 23] . Furthermore, a deregulation of the β-oxidation seems to be associated with the development of obesity and insulin resistance. Nevertheless, the underlying mechanisms and pathways require further exploration. The CHOP trial offers the possibility to analyze the onset of obesity and the change in metabolites over the period of obesity development longitudinally. For instance, it was

shown in the CHOP trial that lysophosphatidylcholine 14: 0 is strongly related to rapid weight gain in infancy in the first 6 months of life and to overweight/obesity at the age of 6 years [30]. However, unraveling the effects of infant formula on the metabolome remains challenging, and further trials will provide insights in the molecular mechanism and help to optimize infant formulas. Metabolites like keto acids, or intermediates of the citric acid cycle or from gluconeogenesis, should be analyzed in response to formula feeding and may give new insights in the future.

Conclusion

HP intake in excess of metabolic requirements increases BCAA concentrations in infant plasma to levels at which the normal catabolic capacity for BCAAs is exceeded. Thereby, high dietary protein supply to infants may stimulate mark-

Early Nutrition and the Infant Metabolome 97 edly enhanced secretion of the growth factors insulin and IGF-1, and induce signaling effects inducing excessive weight gain. Moreover, HP intake in infants appears to inhibit initiation of β-oxidation and thus may contribute to enhanced fat storage and increased adiposity, probably by enhanced BCAA levels. Elevat- ed levels of BCAAs and disturbed β-oxidation have been shown in previous observational studies to be associated with obesity and risk of cardiovascular disease. Thus, BCAA metabolism might present a mechanism linking infant formula feeding and obesity risk.

Appendix

The European CHOP Study Group Philippe Goyens, Clotilde Carlier, Pascale Poncelet, Elena Dain and Joana Hoyos (Free University of Brussels, Brussels, Belgium); Françoise Martin, Annick Xhonneux, Jean- Paul Langhendries and Jean-Noel Van Hees (Centre Hospitalier Chrétien St Vincent, Liège, Belgium); Ricardo Closa-Monasterolo, Joaquin Escribano, Veronica Luque, Georgina Mendez, Natalia Ferre and Marta Zaragoza-Jordana (Universitat Rovira I Vir- gili, Tarragona, Spain); Marcello Giovannini, Enrica Riva, Carlo Agostoni, Silvia Sca- glioni, Elvira Verduci, Fiammetta Vecchi and Alice Re Dionigi (University of Milan, Milan, Italy); Jerzy Socha, Anna Dobrzańska, Dariusz Gruszfeld, Piotr Socha, Anna Stolarczyk, Agnieszka Kowalik, Roman Janas and Ewa Pietraszek (Children’s Memo- rial Health Institute, Warsaw, Poland); Emmanuel Perrin (Danone Research Center for Specialized Nutrition, Schiphol, The Netherlands); Helfried Groebe, Anna Reith, and Renate Hofmann (Klinikum Nurnberg Sued, Nurnberg, Germany); Berthold Koletzko, Veit Grote, Martina Weber, Sonia Schiess, Jeannette Beyer, Michaela Fritsch, Uschi Handel, Ingrid Pawellek, Sabine Verwied-Jorky, Iris Hannibal, Hans Demmelmair, Gu- drun Haile, Wolfgang Peissner, Ulrike Harder, Franca F. Kirchberg, Melissa Theurich, Peter Rzehak, Christian Hellmuth and Olaf Uhl (Dr. von Hauner Children’s Hospital, Ludwig Maximilian University of Munich, Munich Germany), and Rüdiger von Kries (Institute for Social Pediatrics and Adolescent Medicine, University of Munich, Mu- nich, Germany).

Acknowledgment

This work was financially supported by the Commission of the European Communities, the seventh Framework Program, contract FP7-289346-EarlyNutrition, and the European Research Council Advanced Grant ERC-2012-AdG (No. 322605 META- GROWTH). This paper does not necessarily reflect the views of the Commission and in no way anticipates the future policy in this area. The funders had no role in study design, data collection and analysis, decision to publish or preparation of the paper. Funds to support the writing of the paper were provided by Nestlé Nutrition, Vevey, Switzerland.

98 Hellmuth et al. Disclosure Statement

C.H. received funds by Nestle Nutrition, Vevey, Switzerland to support the preparation of the paper. C.H. and P.S. presented the work at the corresponding Nestle Nutrition workshop. All other authors have nothing to disclose.

References

1 Koletzko B, Brands B, Chourdakis M, et al: 10 Weber M, Grote V, Closa-Monasterolo R, et The power of programming and the Early- al: Lower protein content in infant formula Nutrition project: opportunities for health reduces BMI and obesity risk at school age: promotion by nutrition during the first thou- follow-up of a randomized trial. Am J Clin

sand days of life and beyond. Ann Nutr Nutr 2014; 99: 1041–1051.

Metab 2014; 64: 187–196. 11 Socha P, Grote V, Gruszfeld D, et al: Milk 2 Arenz S, Ruckerl R, Koletzko B, et al: Breast- protein intake, the metabolic-endocrine re- feeding and childhood obesity – a systematic sponse, and growth in infancy: data from a

review. Int J Obes Relat Metab Disord 2004; randomized clinical trial. Am J Clin Nutr

28: 1247–1256. 2011; 94: 1776S–1784S. 3 Koletzko B, von Kries R, Closa R, et al: Can 12 Kirchberg FF, Harder U, Weber M, et al: Di- infant feeding choices modulate later obesity etary protein intake affects amino acid and

risk? Am J Clin Nutr 2009; 89: 1502S–1508S. acylcarnitine metabolism in infants aged

4 Ruchat S-M, Bouchard L, Hivert M-F: Early 6 months. J Clin Endocrinol Metab 2015; 100: infant nutrition and metabolic programming: 149–158. what are the potential molecular mecha- 13 Brosnan JT, Brosnan ME: Branched-chain

nisms? Curr Nutr Rep 2014; 3: 281–288. amino acids: enzyme and substrate regula-

5 Nicholson JK, Lindon JC: Systems biology: tion. J Nutr 2006; 136: 207S–211S.

metabonomics. Nature 2008; 455: 1054–1056. 14 Shimomura Y, Murakami T, Nakai N, et al: 6 Martin FP, Moco S, Montoliu I, et al: Impact Exercise promotes BCAA catabolism: effects of breast-feeding and high- and low-protein of BCAA supplementation on skeletal muscle

formula on the metabolism and growth of during exercise. J Nutr 2004; 134: 1583S– infants from overweight and obese mothers. 1587S.

Pediatr Res 2014; 75: 535–543. 15 Roe DS, Roe CR, Brivet M, et al: Evidence for 7 Karlsland Akeson PM, Axelsson IE, Raiha a short-chain carnitine-acylcarnitine translo- NC: Protein and amino acid metabolism in case in mitochondria specifically related to three- to twelve-month-old infants fed hu- the metabolism of branched-chain amino

man milk or formulas with varying protein acids. Mol Genet Metab 2000; 69: 69–75. concentrations. J Pediatr Gastroenterol Nutr 16 Kuhara T, Ikeda S, Ohneda A, et al: Effects of

1998; 26: 297–304. intravenous infusion of 17 amino acids on 8 Fleddermann M, Demmelmair H, Grote V, et the secretion of GH, glucagon, and insulin in

al: Infant formula composition affects ener- sheep. Am J Physiol 1991; 260:E21–E26. getic efficiency for growth: the BeMIM study, 17 Goichon A, Chan P, Lecleire S, et al: An en- a randomized controlled trial. Clin Nutr teral leucine supply modulates human duo-

2014; 33: 588–595. denal mucosal proteome and decreases the 9 Koletzko B, von Kries R, Closa R, et al: Lower expression of enzymes involved in fatty acid

protein in infant formula is associated with beta-oxidation. J Proteomics 2013; 78: 535– lower weight up to age 2 y: a randomized 544.

clinical trial. Am J Clin Nutr 2009; 89: 1836– 18 Cheng Y, Meng Q, Wang C, et al: Leucine 1845. deprivation decreases fat mass by stimulation of lipolysis in white adipose tissue and upreg- ulation of uncoupling protein 1 (UCP1) in

brown adipose tissue. Diabetes 2010; 59: 17– 25.

Early Nutrition and the Infant Metabolome 99 19 Samuel VT, Petersen KF, Shulman GI: Lipid- 25 Picone TA, Benson JD, Moro G, et al: induced insulin resistance: unravelling the Growth, serum biochemistries, and amino

mechanism. Lancet 2010; 375: 2267–2277. acids of term infants fed formulas with amino 20 Matthews DE: An overview of phenylalanine acid and protein concentrations similar to

and tyrosine kinetics in humans. J Nutr 2007; human milk. J Pediatr Gastroenterol Nutr

137: 1549S–1555S; discussion 1573S–1575S. 1989; 9: 351–360. 21 Newgard CB, An J, Bain JR, et al: A 26 Haberle J, Boddaert N, Burlina A, et al: Sug- branched-chain amino acid-related metabolic gested guidelines for the diagnosis and man- signature that differentiates obese and lean agement of urea cycle disorders. Orphanet J

humans and contributes to insulin resistance. Rare Dis 2012; 7: 32.

Cell Metab 2009; 9: 311–326. 27 Sener A, Malaisse WJ: L-leucine and a non- 22 Fernstrom JD: Branched-chain amino acids metabolized analogue activate pancreatic islet

and brain function. J Nutr 2005; 135: 1539S– glutamate dehydrogenase. Nature 1980; 288: 1546S. 187–189. 23 Adams SH: Emerging perspectives on essen- 28 Malaisse WJ, Sener A, Carpinelli AR, et al: tial amino acid metabolism in obesity and the The stimulus-secretion coupling of glucose-

insulin-resistant state. Adv Nutr 2011; 2: 445– induced insulin release. XLVI. Physiological 456. role of L-glutamine as a fuel for pancreatic

24 Dawson-Hughes B, Harris SS, Rasmussen islets. Mol Cell Endocrinol 1980; 20: 171–189. HM, et al: Comparative effects of oral aro- 29 Rauschert S, Uhl O, Koletzko B, et al: Metab- matic and branched-chain amino acids on olomic biomarkers for obesity in humans: a

urine calcium excretion in humans. Osteopo- short review. Ann Nutr Metab 2014; 64: 314–

ros Int 2007; 18: 955–961. 324. 30 Rzehak P, Hellmuth C, Uhl O, et al: Rapid growth and childhood obesity are strongly

associated with lysoPC(14: 0). Ann Nutr

Metab 2014; 64: 294–303.

100 Hellmuth et al. Obesity Prevention

Fewtrell MS, Haschke F, Prescott SL (eds): Preventive Aspects of Early Nutrition. Nestlé Nutr Inst Workshop Ser, vol 85, pp 101–109, (DOI: 10.1159/000439492 ) Nestec Ltd., Vevey/S. Karger AG., Basel, © 2016

Postnatal High Protein Intake Can Contribute to Accelerated Weight Gain of Infants and Increased Obesity Risk

a b b Ferdinand Haschke · Dominik Grathwohl · Patrick Detzel · c d d Philippe Steenhout · Natalia Wagemans · Peter Erdmann a b Department of Pediatrics, Paracelsus Medical University, Salzburg , Austria; Nestlé Research c d Center, Lausanne , and Nestlé R&D and Nestlé Nutrition Institute, Vevey , Switzerland

Abstract Worldwide, 38% of women are now overweight (BMI 25–30) or obese (BMI ≥ 30). There is increasing evidence that maternal obesity can result in unfavorable (epigenetic) pre- and postnatal programming of important genes of the offspring. Infants of overweight mothers show faster weight gain during infancy, which is associated with higher risk of obesity during childhood and adult life. This can have lifelong consequences such as increased risk of noncommunicable diseases. Many studies indicate that infants of obese and nonobese mothers who were fed traditional (high-protein) formulas gain more rapidly weight than breastfed infants. An updated meta-analysis (n = 1,150) indicates that infants from four continents who were fed a whey-based, low-protein (1.8 g/100 kcal) formula with an essential amino-acid profile closer to breast milk grow in accordance with the World Health Organization (WHO) growth standard (0–4 months). A new experimental low-protein (1.61–1.65 g protein/100 kcal) formula for infants between 3 and 12 months of age was recently tested in two randomized clinical trials. One trial in the general US population indicates lower weight between 4 and 12 months of age in infants fed the low-protein formula when compared to infants on the high-protein formula (p = 0.031). Weight gain was not inferior to the WHO growth standards. Longitudinal analysis of odds ratios from 4 to 12 months of age showed a lower incidence of infants with weight >85th percentile in the low-protein group compared with the high-protein group (p = 0.015). In the second trial, which was conducted in Chile and included infants of mothers with BMI >25, infants fed the low-protein formula gained less weight between 4 and 12 months (p = 0.022) and until 24 months (p = 0.031) than the high-protein group. Weight gain was similar to the breastfed reference group. In both trials, biomarkers of protein metabolism (insulin-like growth factor-1 and C-peptide) of the low-protein groups were closer to breastfed infants than the respective biomarkers of the high-protein groups. Health economic analyses indicate that feeding low-protein formulas to nonbreastfed infants would result in cost savings for both the individual and the society. Preventive measures against childhood and adult obesity should include promotion of breastfeeding for 6 months or longer, and use of low-protein formulas in nonbreastfed infants. © 2016 Nestec Ltd., Vevey/S. Karger AG, Basel

Protein Intake and Growth

In the early 1990s, a World Health Organization (WHO) working group documented lower weight gain and percent body fat between 5 and 12 months of age in infants who were exclusively breastfed for 4–6 months compared with formula-fed infants [1] . During that time, the protein content of most infant and follow-up formulas was greater than 2.25 and 2.6 g/100 kcal, respectively. Slower growth and leanness of breastfed infants were the reasons why the WHO created growth standards [2] that define ‘healthy growth’ of predominantly breastfed (>6 months) infants living in different continents and favorable environments. Among the conditions required for enrollment into the WHO growth study was a maternal BMI between 18 and 25 – i.e. mothers were not malnourished. The WHO standard now serves as a global growth standard for weight, length/height, BMI and head circumference, and is used in most countries of the world. There is increasing evidence that high protein intake in infancy and during the 2nd year of life can have negative effects on long-term health [3, 4]. Epidemiological and clinical evidence [5, 6] and more recently randomized controlled trials [7–9] have demonstrated that infants who are fed formulas with high protein content show faster weight gain than infants fed low-protein formulas. Accelerated growth during infancy is associated with higher obesity risk in childhood and adult years [10] . In a recent meta-analysis [11] , investigators found a twofold higher risk of childhood obesity and a 23% increased risk of adult obesity with each unit increase in weight z scores. In light of the above evidence that suggests that early protein intake may have long-term health implications, in this paper we review the protein needs of young children and their actual protein intake, clinical trials with high- and low-protein infant formulas, and long- term health economic aspects of low-protein infant formulas.

Protein Needs versus Actual Intakes

Protein needs for growth can be estimated from body composition studies in children [12] . Protein requirements are higher during the first months of life because daily gains in weight and body protein are higher than during later

102 Haschke · Grathwohl · Detzel · Steenhout · Wagemans · Erdmann infancy and childhood [12, 13]. Protein requirements and protein/energy ratios (PE, in %: kcal from protein/kcal from food) can be calculated for each age range by using the revised recommendations from the Joint WHO/FAO/UNU Expert Consultation [14] . Protein concentration in breast milk decreases from an average of 2.09 g/100 kcal during the 1st month of lactation to 1.28 g/100 kcal at 3–4 months and about 1.24 g/100 kcal by 9–12 months of lactation [15] . All recent studies show that protein concentration in breast milk steadily decreases over the course of lactation and is below 1.4 g/100 kcal during the months 6–12 [16, 17] . The exclusively breastfed infant receives a diet with a PE of about 5% between 3 and 6 months. A PE of 7.7% (+2 SD) has been calculated to be safe at 6 months of age [14] . Formula-fed infants can receive up to 2–3 times more protein than breastfed infants [17, 18], in particular after 6 months when follow-up formulas are offered. The upper limit of protein concentration in follow-up formulas in Europe is 3 g/100 kcal [19]. The Codex Alimentarius still proposes 3 g/100 kcal as the lower limit. A formula with 3 g of protein per 100 kcal and an energy content of 75 kcal/100 ml offers a PE of 16%. When compared to breastfed infants at 6 months, feeding of high-protein formulas resulted in higher concentrations of the branched-chained amino acids leucine, isoleucine and valine, as well as higher serum levels of blood urea nitrogen (BUN), insulin, C-peptide and insulin-like growth factor (IGF)-1 [8, 9] . Data on protein intake of children between 3 months (formula-fed only) and 5 years indicate that protein intakes continue to exceed 2–3 times the requirements. After 6 months, the high protein intakes from cow’s milk-based products and other high-protein complementary foods play dominant roles [20] .

Clinical Trials

When the first whey-based low-protein infant formula with 1.8 g of protein/100 ® kcal (casein:modified whey ratio of 40: 60; NAN OPTIPRO 1 ) was clinically tested in a randomized clinical trial, it turned out that growth (0–4 months) was similar to a breastfed reference group [21] . The whey-based formula [22] had a lower glyco-macro-peptide concentration, essential amino acids closer to those in mature breast milk between 21 to 58 days [23] and lower concentrations of branched-chain amino acids than in conventional formulas. Infants fed this formula during the first months had biomarkers (BUN, serum insulin and branched- chained amino acids) that were closer to breastfed infants than to infants who were fed formulas with higher protein content [24] . An initial meta-analysis that included 4 clinical trials of the formula with 1.8 g of protein/100 kcal indicated that both weight and length after 4 months of exclusive formula feeding were

High Protein Intake and Later Obesity 103 Italy, Palermo, 1998 (n = 53) Italy, Palermo, 1999 (n = 108) South Africa, Johannesburg, 2002 (n = 37) Australia, Adelaide, 2002 (n = 123) Italy, Palermo, 2003 (n = 107) France, multi center, 2003 (n = 224) China, multi center, 2003 (n = 369) France, Lyon, 2005 (n = 68) France, Nancy, 2005 (n = 61) Fixed-effect total (n = 1,150) Random-effect total (n = 1,150)

–1.0 –0.5 0 0.5 1.0 Weight-for-age z score (WHO)

Fig. 1. Meta-analysis: weight z scores at 4 months of infants fed a whey-based formula with 1.8 g protein/100 kcal vs. WHO standard [2] . close to the WHO standards [25] . A more comprehensive meta-analysis is presented in figure 1, which summarizes the results from 1,150 infants enrolled in 9 clinical trials. All studies included in this meta-analysis were randomized, double-blind trials, and had several parallel groups (higher-protein formulas) and breastfed infants who served as reference groups. All infants started with the 1.8 g/100 kcal formula between birth and 2 weeks of age. Weight measurements were done at enrollment, and at 1, 2, 3 and 4 months of age. For those infants who were enrolled within 2 weeks after birth, data on birth weight were taken from hospital records. The whey-based low-protein formula was fed exclusively until 4 months. In the trials, the formula was studied with and without additional ingredients such as long-chain polyunsaturated fatty acids, prebiotics and probiotics. Data from all study groups on the low-protein formula were compared with the WHO standards. This meta-analysis based on individual data from all trials (n = 1,150; mean dropout rate in all trials was 24% at 4 months) is more sensitive than a meta-analysis based on group summary statistics (e.g. odds ratios), and it is possible in this type of analysis to adjust for influencing covariates, such as sex and birth weight. The z scores at 4 months of age were analyzed by ANCOVA correcting for the z score at birth (baseline), sex and study. The meta- analysis with fixed effects did not correct for different standard deviations in the individual trials. The meta-analysis with random effects was based on a mixed model which corrected for baseline and sex (fixed effects) and study variability (random effect). The more sensitive random-effect model (fig. 1 ) indicates a very small mean difference in weight z scores of 0.091 (95% CI –0.164, +0.345, non- significant) at 4 months, which corresponds to a weight of 76 g (840 g = 1 SD)

104 Haschke · Grathwohl · Detzel · Steenhout · Wagemans · Erdmann Table 1. Infant and follow-up protein formulas in randomized controlled trials

Study Infant formulas, Follow-up formulas, Lower/upper limit g/100 kcal g/100 kcal EU/USA [19], g/100 kcal

European CHOP 1.77 vs. 2.2 2.9 vs. 4.4 1.8–3.0 multicenter [7, 27] (0–6 months) (6–12 months) US multicenter [9] 1.8 1.61 vs. 2.15 1.8–4.5 (0–3 months) (3–12 months) Chile [8] – infants of 1.8 1.65 vs. 2.7 overweight mothers (0–3 months) (3–12 months)

above the 0 z core of the WHO standard. One study, which was carried out in China, seemed to be an outlier because weight at 4 months was significantly higher (+0.91 z score; i.e. +764 g) than the WHO standard. The discussion of whether the WHO growth standard also applies to Chinese (Asian) infants is still ongoing. A longitudinal study in 30 Chinese cities indicated that exclusively breastfed infants had similar birth weights as the WHO standard, whereas weight at different ages during the 1st year was 80–510 g higher [26]. Without the Chinese study, the mean weight difference to the z score WHO standard would have been as small as 3 g. Our meta-analysis, therefore, indicates that infants who receive a whey-based formula with 1.8 g/100 kcal high-quality protein between 0 and 4 months of age grow according to the WHO standard [2] . Lower and upper limits of protein concentrations in formulas are presented in table 1 . Three recent randomized controlled trials with similar experimental designs compared growth and metabolic outcome of infants who received formulas with higher and lower protein contents until 12 months of age. The Childhood Obesity Project (CHOP), a multicenter European trial [27] (table 1 ), compared outcomes of feeding formulas with 2.9 and 4.4 g protein/100 kcal from 4 to 12 months of age. BMI was significantly higher between 6 and 12 months in the group that received the higher-protein formula. A follow-up examination at 6 years indicated a significantly lower BMI in the low-protein formula group (p = 0.009) and a lower prevalence of childhood obesity (RR –2.87; 95% CI –1.22, –6.75) [7] . BUN, insulin and IGF-1 in serum were lower at 6 months in the group receiving the lower-protein formula [28] . Two recent trials tested an experimental whey-based formula (3–12 months) with a protein content that was below the minimum protein limit of the EU and USA ( table 1 ). The whey fraction of the formula was further modified [29] to be close to the essential ( fig. 2 ) and branched-chained amino acid pattern in mature

High Protein Intake and Later Obesity 105 Threonine 250 Phenylalanine + Cystine tyrosine 200

150

100 Tryptophan Valine 50

Histidine Methionine

Lysine Isoleucine

Leucine

Fig. 2. Concentrations of essential amino acids (mg/100 kcal). Dark line = High-protein formula (2.7 g/100 kcal) [8] ; grey line = low-protein formula (1.65 g/100 kcal) [8] ; light-grey line = mature breast milk (59–135 days) [23] . breast milk between 59 and 135 days of lactation [23] . Protein contents in the low- and high-protein formulas >3 months were 37–38 and 39–51% lower than the respective CHOP [27] formulas. In a multicenter US study [9], weight gain (g/day) from 3 to 6 months (primary outcome) was similar in the low- and high-protein study groups. How- ever, longitudinal analysis indicated lower weight from 4 to 12 months (p = 0.031) and a lower percentage of infants with weight >85th percentile (p = 0.015) in the group fed the low-protein formula. Serum biochemical parameters in the low-protein group reflected the lower protein intake and were closer to biomarkers in breastfed infants than to those in the high-protein group. In the Chilean study [8] , we tested whether a formula with low protein content slows weight gain in infants of overweight mothers ( table 1 ). Infants fed the low-protein formula gained less weight (–1.77 g/day; p = 0.028) between 3 and 6 months than infants fed the high-protein formula. The weight of infants fed the low- protein formula remained below that of infants fed the high-protein formula until 2 years of age (p = 0.031) but was similar to the weight of breastfed infants. Biomarkers of protein nutrition of infants fed the low-protein formula were close to those of breastfed infants. Three randomized clinical trials therefore indicate that low-protein formulas slow rapid growth during the 1st year of life in

106 Haschke · Grathwohl · Detzel · Steenhout · Wagemans · Erdmann infants of normal-weight and overweight mothers while supporting normal growth. Biomarkers of protein nutrition reflect the lower protein intake and levels are more like those of breastfed infants.

Potential Health Economic Benefit of Low-Protein Formulas

Global, regional and national prevalence of overweight and obesity in children and adults has risen steadily between 1980 and 2013 [30] . Being overweight or obese is related to a wide range of diseases, most notably cardiovascular diseases, type 2 diabetes, osteoarthritis and certain types of cancer. The corresponding health risks of obesity are associated with significant health system costs, loss of health-related quality of life and economic costs, such as reduced productivity. Obesity is not only an individual and societal health issue, because of the strong comorbidities associated with this health state it is also a major burden to society. Obesity accounts for nearly 7% of total health care costs globally [31] . Low-protein formulas help to prevent accelerated weight gain in nonbreastfed infants, which can reduce the risk of adult obesity [11, 32]. Recently, a health economic modeling approach was presented [33] which showed the economic benefits of using low-protein formulas in nonbreastfed infants. The modeling approach demonstrated that the reduction in weight gain during infancy that is associated with the use of a low-protein formula instead of a standard high-protein formula leads to an improved BMI in adulthood. Over their lifetimes, infants are 10.5% less likely to be obese, which in turn results in a reduced risk of obesity-related diseases (probability of experiencing clinical events reduced by 2.2– 3.3%, depending on the disease), health care cost savings (reduction in direct medical costs of 3.9%), and improvements in productivity (reduction in indirect economic costs of 4.1%) and quality of life. Because obesity-related comorbidities materialize some 45–60 years after infancy, the discounting of the savings will play a large role and significantly reduce the overall value of these impacts. The total savings in a country like Mexico are estimated at USD 64 per infant [33] . If all nonbreastfed infants of overweight and obese mothers in Mexico were fed low-protein formula, Mexico would save an estimated USD 69.3 million per year.

Conclusions

It has been shown that formula-fed infants, in particular those who receive a traditional follow-up formula between 6 and 12 months of age, receive excess protein compared with breastfed infants or the revised WHO/FAO/UNU

High Protein Intake and Later Obesity 107 recommendations [14] . Three randomized clinical trials now indicate that excess protein intake can result in accelerated weight gain, which is associated with a higher risk of adult obesity. Feeding low-protein formulas to nonbreastfed infants slows rapid weight gain during infancy, which helps to reduce the adult obesity risk for the individual and provides health care cost savings for the society.

Disclosure Statement

F.H. is a board member of the Nestlé Nutrition Institute (NNI) and receives consultant fees from Nestec Ltd. and other Nutri-Health Companies. D.G., P.D., P.S., N.W. and P.E. are NESTEC Ltd. employees. Some clinical trials which are presented (refs. 8, 9, 24) were financially sponsored by Nestlé Nutrition Research & Development, Nestec Ltd.

References

1 Dewey KG, Heinig MJ, Nommsen LA, et al: 7 Weber M, Grote V, Closa-Monasterolo R, et Breast-fed infants are leaner than formula-fed al: Lower protein content in infant formula infants at 1 y of age: the DARLING study. reduces BMI and obesity risk at school age:

Am J Clin Nutr 1993; 57: 140–145. follow-up of a randomized trial. Am J Clin

2 WHO Multicentre Growth Reference Study Nutr 2014; 99: 1041–1051. Group: WHO Child Growth Standards: 8 Inostroza J, Haschke F, Steenhout P, et al: Length/Height-for-Age, Weight-for-Age, Low-protein formula slows weight gain in Weight-for-Length, Weight-for-Height and infants of overweight mothers. J Pediatr Gas-

Body Mass Index-for-Age: Methods and De- troenterol Nutr 2014; 59: 70–77. velopment. Geneva, World Health Organiza- 9 Ziegler EE, Fields DA, Chernausek SD, et al: tion, 2006, http://www.who.int/childgrowth/ Adequacy of infant formula with protein software/en/. content of 1.6 g/100 kcal for infants between 3 Koletzko B, von Kries R, Closa R, et al: Can 3 and 12 months: a randomized multicenter infant feeding choices modulate later obesity trial. J Pediatr Gastroenterol Nutr 2015, Epub

risk. Am J Clin Nutr 2009; 89: 1502S–1508S. ahead of print. 4 Michaelsen KF, Greer FR: Protein needs early 10 Li L, Kleinman K, Gillman MW: A compari- in life and long-term health. Am J Clin Nutr son of confounding adjustment methods with

2014; 99: 718S–722S. an application to early life determinants of 5 Harder T, Bergmann R, Kallischnigg G, et al: childhood obesity. J Dev Orig Health Dis

Duration of breastfeeding and risk of over- 2014; 5: 435–447. weight: a meta-analysis. Am J Epidemiol 11 Druet C, Stettler N, Sharp S, et al: Prediction

2005; 162: 397–403. of childhood obesity by infancy weight gain: 6 Rolland-Cachera MF, Deheeger M, Akrout an individual-level meta-analysis. Paediatr

M, et al: Influence of macronutrients on adi- Perinat Epidemiol 2012; 26: 19–26. posity development: a follow up study of nu- 12 Fomon SJ, Haschke F, Ziegler EE, Nelson SE: trition and growth from 10 months to 8 years Body composition of reference children from

of age. Int J Obes Relat Metab Disord 1995; birth to age 10 years. Am J Clin Nutr 1982;

19: 573–578. 35(5 suppl):1169–1175. 13 Dewey KG, Beaton G, Fjeld C, et al: Protein requirements of infants and children. Eur J

Clin Nutr 1996; 50(suppl 1):S119–S150.

108 Haschke · Grathwohl · Detzel · Steenhout · Wagemans · Erdmann 14 Joint WHO/FAO/UNU Expert Consultation: 25 Haschke F, Steenhout P, Grathwohl D, Protein and amino acid requirements in hu- Haschke-Becher E: Evaluation of growth and man nutrition. World Health Organ Tech early infant feeding: a challenge for scientists,

Rep Ser 2007; 935: 1–265. industry and regulatory bodies. World Rev

15 Fomon SJ: Requirements and recommended Nutr Diet 2013; 106: 33–38. dietary intakes of protein during infancy. Pe- 26 Wang HS; China Breastfeeding Growth and

diatr Res 1991; 30: 391–395. Development Study Group: A longitudinal 16 Dewey KG, Beaton G, Fjeld C, et al: Protein study of growth of breastfed infants in rural requirements of infants and children. Eur J areas of six economically better developed

Clin Nutr 1996; 50(suppl 1):S119–S150. provinces in China (in Chinese). Zhonghua

17 Haschke F, Ziegler EE, Grathwohl D: Fast Er Ke Za Zhi 2010; 48: 484–491. growth of infants of overweight mothers: can 27 Koletzko B, von Kries R, Closa R, et al: Lower

it be slowed down? Ann Nutr Metab 2014; protein in infant formula is associated with 64(suppl 1):19–24. lower weight up to age 2 y: a randomized

18 Singhal A, Kennedy K, Lanigan J, et al: Nutri- clinical trial. Am J Clin Nutr 2009; 89: 1836– tion in infancy and long-term risk of obesity: 1845. evidence from 2 randomized controlled trials. 28 Socha P, Grote V, Gruszfeld D, et al: Milk

Am J Clin Nutr 2010; 92: 1133–1144. protein intake, the metabolic-endocrine re- 19 EFSA Panel on Dietetic Products, Nutrition sponse, and growth in infancy: data from a and Allergies (NDA): Scientific opinion on randomized clinical trial. Am J Clin Nutr

the essential composition of infant and fol- 2011; 94(suppl):1776S–1784S.

low-on formulae. EFSA J 2014; 12: 3760. 29 Haschke F, Magliola C, Steenhout P: Nutri- 20 Butte NF, Fox MK, Briefel RR, et al: Nutrient tional composition. Patent WO 2008071667 intakes of US infants, toddlers, and pre- A1. Geneva, World Intellectual Property Or- schoolers meet or exceed dietary reference ganization, 2008.

intakes. J Am Diet Assoc 2010; 110:S27–S37. 30 Ng M, Fleming T, Robinson M, et al: Global, 21 Räihä NC, Fazzolari-Nesci A, Cajozzo C, et regional, and national prevalence of over- al: Whey predominant, whey modified infant weight and obesity in children and adults formula with protein/energy ratio of 1.8 during 1980–2013: a systematic analysis for g/100 kcal: adequate and safe for term infants the Global Burden of Disease Study 2013.

from birth to four months. J Pediatr Gastro- Lancet 2014; 384: 766–781.

enterol Nutr 2002; 35: 275–281. 31 WHO: Obesity and Overweight. http://www. 22 Ballevre O, Haschke F, Jost R, et al: Composi- who.int/dietphysicalactivity/media/en/gsfs_ tion comprising casein protein and whey obesity.pdf. protein. Patent EP 1220620 B1. Munich, 32 Nettleton JA, Jebb S, Risérus U, et al: Role of European Patent Office, 2005. dietary fats in the prevention and treatment 23 Lönnerdahl B, Zhang Z, Adelman A, et al: of the metabolic syndrome. Ann Nutr Metab

Amino acid profiles in term and preterm hu- 2014; 64: 167–178. man milk through lactation: a systematic re- 33 Marsh K, Orfanos P, Möller J, et al: The eco-

view. Nutrients 2013; 5: 4800–4821. nomic impact of low protein infant formula 24 Axelsson IE, Ivarsson SA, Räihä NC: Protein for the children of overweight and obese intake in early infancy: effects on plasma mothers. EAPS Barcelona, 2014, poster 311. amino acid concentrations, insulin metabo-

lism, and growth. Pediatr Res 1989; 26: 614– 617.

High Protein Intake and Later Obesity 109

Obesity Prevention

Fewtrell MS, Haschke F, Prescott SL (eds): Preventive Aspects of Early Nutrition. Nestlé Nutr Inst Workshop Ser, vol 85, pp 111–112, (DOI: 10.1159/000441079) Nestec Ltd., Vevey/S. Karger AG., Basel, © 2016

Summary – Early Nutrition and Obesity Prevention

Epidemiology indicates a dramatic increase in overweight and obesity rates both in young women and in children from developed and developing countries. Pre- ventive measures through nutrition should start before conception and continue during the infant and early childhood period (‘first 1,000 days’). Weight management in young women and lifestyle changes (physical activity) are important steps. After birth, accelerated weight gain during the first 12 months is a strong predictor of obesity later in life. Continuation of breastfeeding >6 months slows weight gain during infancy and therefore seems to have a preventive effect. Ran- domized clinical trials on growth of infants who were on low- and high-protein formulas indicate that protein concentrations >2.2 g/100 kcal result in faster weight gain during infancy. In one trial, follow-up until the age of 6 years shows higher BMI in children who had been on high-protein formulas. In most developed countries, children continue to have higher protein intake during the preschool period than recommended by the WHO, but the effects on weight gain are not well documented. Maternal obesity and inadequate nutrition such as high protein intake during the first 2 years might result in unfavorable changes in the function of key genes in the offspring. Whatever the primary mechanism of high protein intake during infancy might be, changes are visible in the metabolome of infants fed high- protein diets, since metabolites (molecules <1,500 Da) are the downstream products of both genetic and epigenetic determinants as well as environmental factors, including diet. It has been shown that the ‘programming effects’ of infant formulas with different protein concentrations are reflected in the metabolome and in metabolic pathways. Infants fed high-protein formulas show significantly increased plasma concentrations of the branched-chain amino acids (BCAA) Ile, Leu and Val. The elevation in BCAA levels contributes to increased insulin levels, and it seems to affect ß-oxidation of fatty acids. Thus, high protein and BCAA intakes may inhibit fat oxidation and thereby enhance body fat deposition and the risk of adiposity. High protein intake also increases blood concentrations of insulin-like growth factor (IGF)-1. In humans, the IGF axis has been shown to regulate early growth, adipose tissue differentiation and early adipogenesis. However, it must be mentioned that at present we have no single biomarker which predicts with adequate precision the risks of accelerated growth and obesity later in life. Ferdinand Haschke

112 Haschke Complementary Feeding: Taste, Eating Behavior and Later Health

Fewtrell MS, Haschke F, Prescott SL (eds): Preventive Aspects of Early Nutrition. Nestlé Nutr Inst Workshop Ser, vol 85, pp 113–123, (DOI: 10.1159/000439501) Nestec Ltd., Vevey/S. Karger AG., Basel, © 2016

Can Optimal Complementary Feeding Improve Later Health and Development?

Mary S. Fewtrell Childhood Nutrition Research Centre, UCL Institute of Child Health, London , UK

Abstract Nutrition and growth during early infancy influence later health and development, but most research has focused on the period of milk feeding, and the possibility that the timing, content or method of complementary feeding (CF) might have similar later effects has received less attention. Such effects are plausible, given that the CF period is one of rapid growth and development when infants are susceptible to nutrient deficiencies and excesses, and during which there are marked changes in diet with exposure to many new foods, tastes and feeding experiences. CF practices could influence later outcome by several potential mechanisms, including programming effects, but also direct effects on food preferences, appetite and eating behavior. Investigating these issues is challenging given the diversity and complexity of CF practices, which limit the feasibility and generalizability of randomized trials in this field. Available evidence relating CF practices to later health and development are currently limited in quantity and quality, but suggest that avoiding the introduction of solid foods before 4 months may reduce the risk of subsequent obesity and allergy. Whilst recommendations for different aspects of CF may be developed in the future based on broad principles, they will need to be tailored for different populations. © 2016 Nestec Ltd., Vevey/S. Karger AG, Basel Introduction

Increasing evidence supports the concept that nutrition and growth early in life not only affect short-term health but also ‘program’ later outcomes, including brain development and the risk of obesity and cardiovascular disease, with effect sizes likely to be of public health significance. Most data on nutritional programming come from studies examining milk feeding in the first months of life: typically the effects of breastfeeding versus formula feeding or modifications to infant formulas. The possibility that nutrition during the complementary feeding (CF) period could also influence later health outcomes has received less attention, and most research has, understandably, focused on meeting nutrient requirements and optimizing short-term health outcomes. However, it is entirely plausible that the second 6 months of life, when the infant’s diet is rapidly diversifying, also represents a critical period in terms of growth, brain development and the development of taste, food preferences and eating behavior, which might in turn influence later health and development. This paper will review evidence relating to the impact of CF practices for later health and development, discuss methodological issues in this area of research and consider potential mechanisms.

Definition and Current Recommendations

CF is defined by the World Health organization (WHO) as any food or liquid other than breast milk added to meet nutritional requirements, although other authorities have suggested that it is more pragmatic to consider CF as anything other than milk, whether breast milk or infant formula. CF is required during the second part of the 1st year of life for nutritional and developmental reasons. At this point, breast milk can no longer meet nutrient requirements, while from a developmental perspective infants also develop the ability to chew and start to show an interest in foods other than milk. The WHO recommends, as global policy, exclusive breastfeeding (EBF) for 6 months followed by the introduction of CF alongside continued breastfeeding based on considerations of the optimal duration of EBF in terms of nutritional adequacy and effects on short-term outcomes [1] . This recommendation has been extensively discussed, given the relatively poor evidence base particularly for infants in higher-income settings [2, 3], and the issue of timing of CF has largely overshadowed other important aspects, including the optimal nutritional content of the diet, foods and feeding practices. It is widely recognized that,

114 Fewtrell regardless of official recommendations, CF practices are strongly influenced by social and cultural factors, and vary markedly between and within countries depending on the availability, acceptability and affordability of foods.

Investigating Later Health Effects of Complementary Feeding

Methodological Issues Randomized controlled trials (RCTs) are accepted as the gold standard for evaluating the safety and efficacy of nutritional interventions, demonstrating causality and underpinning practice. Observational studies can demonstrate associations between nutritional interventions and later outcome but cannot establish causality. However, whilst RCTs have been increasingly applied in infant nutrition research (notably when testing modifications to breast milk substitutes), they are not always possible (e.g. comparing outcomes in breastfed vs. formula- fed infants), and relatively few trials have been conducted to investigate CF practices. The lack of RCTs in this area probably reflects two main problems. Firstly, mothers and caregivers often have strong opinions about CF practices and may not readily agree to be randomized to an alternative. Secondly, unlike the situation in early infancy where an infant receives a single food – milk – the CF period is complex, and it is more difficult to design effective and pragmatic interventions during this period when an infant’s diet is rapidly diversifying and behavioral and cultural aspects are becoming more important. Related to this is the problem of distinguishing between or investigating the effects of specific nutrients, foods or whole diets. These factors may also limit the generalizability of the findings of any intervention. Investigating the later effects of any intervention introduces additional problems such as the use of proxy measures for health outcomes that may not manifest for many years and cohort attrition over time, with statistical effects including the potential introduction of bias and loss of power [4] .

Potential Mechanisms CF practices could potentially influence subsequent health outcomes by a number of mechanisms, which may not be mutually exclusive and which may be difficult to separate in practice (table 1 ). Other papers in this Session focus in detail on taste, psychological and behavioral aspects of CF and the impact of dietary patterns on subsequent outcomes. Thus, the remainder of this paper will summarize the available data relating CF practices to later development and health outcomes.

Effects of Optimal Complementary Feeding 115 Table 1. Potential mechanisms for effects of CF practices on later outcomes

Effect Examples

1 Programming effects (feeding Inadequate iron intake might permanently alter brain practice results in a permanent structure and function; nutritional practices could change in structure or function permanently influence hormonal pathways which has later consequences) or gut microbiota with subsequent effects on later growth and metabolic processes 2 Altered taste or food preferences Exposure to salt or sugar might result in preference for which influence later dietary these tastes which persists intake, with positive or negative Exposure to vegetables during the critical period could effects on health outcomes facilitate later acceptance 3 Effects on eating behavior, The mode of feeding (e.g. breastfeeding vs. bottle-/spoon- appetite or satiety feeding vs. ’baby-led’ feeding), content of the diet or parenting style influence these variables, which then persist Effects on these variables could be programmed and emerge later

Timing of Introduction of Complementary Feeding and Later Health

Obesity Two recent systematic reviews examined the association between the timing of CF and the risk of obesity in infancy and childhood. The first [5] identified 24 studies, providing data from over 34,000 participants for interpretative analysis. No clear association between the age of introduction of solid foods and obesity was found. The second [6] identified 21 studies that looked at the timing of CF and childhood BMI, of which 5 found that early (<3, <4 and <5 months) introduction was associated with higher childhood BMI. Seven studies looked at the timing of CF and body composition, and 1 reported increased percent fat in those with CF introduced before 15 weeks. The authors concluded that there is some evidence suggesting that introduction of CF at or before 4 months may increase the risk of overweight. With the exception of 1 RCT, all of the studies included in these reviews were observational. Data published since these reviews or not meeting their inclusion criteria are broadly consistent with these conclusions. Thus, an observational analysis of the Belarus Promotion of Breastfeeding Intervention Trial (PROBIT) cohort [7] reported no effect of EBF for 6 versus 3 months on weight, height or BMI at the 6.5-year follow-up. Other more recent cohort studies have paid particular attention to the methodological issues recognized to be problematic in previous studies. Huh et al. [8] performed a prospective observational study in 847 preschool-

116 Fewtrell aged children. The primary outcome was obesity at age 3 years and the primary exposure was the time of introduction of solid foods, classified as <4, 4–5 and ≥ 6 months. The introduction of solid foods before 4 months was associated with a 6-fold increased risk of obesity in formula-fed infants, with no association in breastfed infants. The effect was not explained by weight gain between birth and 4 months suggesting that it did not reflect earlier introduction of solid foods to rapidly growing infants. Moss and Yeaton [9] also reported potentially protective effects of breastfeeding and delaying solid foods beyond 4 months, with an additive effect. However, in this nationally representative US birth cohort studied up to 4 years, the beneficial effect of delaying CF was seen in both breastfed and formula-fed infants. Most recently, a follow-up of the only randomized trial comparing 4 versus 6 months EBF conducted in a high-income country (Iceland) used routinely collected data from child health records and reported no significant difference in weight, height, BMI or the proportions of overweight or obese children between groups [10] .

Allergy Many developed countries have rising rates of food allergy, despite increasing advice to restrict and delay exposure to potentially allergenic foods, including cow’s milk, eggs, fish, gluten, peanut and seeds (see also papers from Session I). It has been suggested that the development of immune tolerance to a food antigen may require repeated exposure, perhaps during a critical early window, and perhaps modulated by other dietary factors including breastfeeding. Two systematic reviews have concluded that [11, 12] the evidence that delaying the introduction of solids beyond 4 months reduces the risk of subsequent allergy in either normal or high-risk infants is weak. A number of cohort studies have reported that relatively earlier introduction of allergenic foods, or greater food diversity during CF, is associated with reduced risk of allergy later in childhood [13]. However, the problem of reverse causality in observational studies in this field was well demonstrated by a study in 256 Finnish infants at risk for the development of allergy in whom the introduction of cereals and fish after 7 months was associated with an apparent increased risk of atopic eczema, which was no longer significant when the parents’ suspicions that their child might be exhibiting symptoms of allergic disease were taken into account [14] . Randomized trials are clearly needed to underpin recommendations on this issue, and 2 such studies are currently underway in the UK. The Learning Early about Peanut Allergy (LEAP) study addresses the timing of exposure to peanuts in high-risk infants, and the Enquiring about Tolerance (EAT) study is

Effects of Optimal Complementary Feeding 117 investigating early (from 3 to 4 months) versus delayed (beyond 6 months) introduction of six potentially allergenic foods in a general population of infants not selected for risk of atopy. Data from both studies should be available in the next 12 months.

Gluten Introduction and Celiac Disease The relationship between the introduction of gluten-containing cereals and subsequent development of celiac disease (CD) has been much studied, particularly in Sweden where the incidence of early-onset CD increased following advice to delay introduction of gluten until age 6 months, and fell to previous levels after the recommendation reverted to 4 months. Results from subsequent analyses performed in this cohort and other observational studies led several advisory bodies to conclude that, pending further data, it is prudent to advise mothers to introduce gluten between 4 and 7 months, i.e. at a time when they are still predominantly breastfeeding [2, 3]. Recently, the results of 2 RCTs investigating the timing of introduction of gluten in at-risk infants have questioned this advice. In the PreventCD trial [15] , 944 high-risk infants (with an affected first-degree relative and HLA-DQ2 or HLA-DQ4 haplotypes) from 7 European countries plus Israel were randomized to receive 100 mg per day of immunologi- cally active gluten versus lactose between 16 and 24 weeks. The primary outcome, biopsy-proven CD at 3 years, was not significantly different between randomized groups (hazard ratio 1.23; 95% CI 0.79–1.91) and was not influenced by breastfeeding duration or presence during the introduction of gluten. The risk of CD was significantly higher in girls and in those with HLA-DQ2 homozygosity. The CELIPREV trial [16] randomized 832 Italian newborns with a first-degree relative with CD to the dietary introduction of gluten (in the form of pasta, semolina and biscuits) at 6 or 12 months. HLA genotyping was performed at 15 months, and 553 subjects with a standard or high-risk genotype were studied further. At 2 years, significantly higher proportions of subjects randomized to receive gluten from 6 months had CD autoimmunity (16 vs. 7%, p = 0.002) and overt CD (12 vs. 5%, p = 0.01), but by age 5 years the between-group differences were no longer significant (autoimmunity 21 vs. 20% and overt CD 16 vs. 16%). Breastfeeding was not associated with the development of CD. The authors concluded that neither the delayed introduction of gluten nor breastfeeding modifies the risk of CD amongst high-risk infants, although later introduction of gluten was associated with delayed onset of the disease. The findings of these trials illustrate the need for caution when basing feeding recommendations on epidemiological data, and suggest that there is currently no basis for making specific recommendations for breastfeeding or CF based on preventing CD.

118 Fewtrell Cognitive Outcome Few studies have examined later cognitive outcome in relation to CF. An observational analysis of data from the PROBIT study in Belarus suggested no difference in cognitive outcome at 6.5 years between children exclusively breastfed for 3–4 versus 6 months [7], whilst follow-up of 1 small RCT of 4 versus 6 months of EBF in Iceland found no difference in developmental outcome at preschool age using routinely collected data [17] . Interestingly, in a follow-up of the PROBIT cohort at 11.5 years, children from the experimental arm, who had substantially increased duration and exclusivity of breastfeeding in infancy, had a significant reduction in problematic eating attitudes compared to the control group [18] .

Content of the Complementary Feeding Diet and Later Health Outcomes

Macronutrients The Finnish Special Turku Coronary Risk Factor Intervention Project (STRIP), a large (n > 1,000) trial, in which the intervention group was given dietary advice to reduce the intake of saturated fats from infancy throughout childhood, reported beneficial effects on plasma lipid profile and endothelial function, particularly in boys, with no adverse effects on growth to age 14 years or on cognitive development at age 5 years. The intervention resulted primarily in changes in dietary fat quality rather than quantity, and also in the consumption of generally healthier diets, notably with higher fiber intake and lower intakes of sucrose and fructose later in childhood [19]. Since the dietary recommendations were applied throughout infancy and childhood, it is not possible to distinguish programmed versus ongoing effects of the diet on the measured outcomes. Given evidence linking more rapid infant growth to later cardiovascular risk and obesity, and randomized trials showing adverse effects of higher-protein infant formulas on later fatness and blood pressure [20] , there is interest in whether high-protein intake during CF may have similar adverse effects. This is of some concern in higher-income countries where protein intake increases markedly during the CF period, particularly if CF is consumed alongside cow’s milk or infant formulas. Protein intakes during this period are often 2–3 times higher than the theoretical requirements. However, the trials investigating the later effects of protein have so far focused on protein intake from milk, and no randomized study has specifically addressed the issue of the protein content in CF and whether different protein sources have different effects. Observational studies suggest that although higher intakes of protein during the CF period are

Effects of Optimal Complementary Feeding 119 associated with higher BMI or percent fat in early childhood, this particularly applies to intake of dairy protein rather than meat or vegetable protein, perhaps due to stimulatory effects of dairy protein on insulin-like growth factor-1 and hence on growth [21–23] .

Micronutrients Iron deficiency is the commonest nutritional deficiency worldwide, and infants are at particular risk during the CF period due to their high requirements, especially in relation to energy needs. Iron deficiency anemia during this period is associated with irreversible effects on brain development; hence, many studies have examined the efficacy of increasing the provision of iron during CF by supplementation, food fortification or using naturally iron-rich foods such as meat. However, these have generally been short-term studies and few have reported outcomes beyond infancy. One observational study in the UK showed a positive association between intake of meat during CF and psychomotor development scores at 22 months [24] . Two studies have also tested the effect of iron-fortified follow-on formulas on later cognitive outcomes, with conflicting results: one showing a positive effect at age 2 years in high-risk infants and the other report- ing a negative effect in previously iron-replete infants at age 10 years [25] .

Salt Two randomized trials have shown that higher intakes of salt in early infancy are associated with higher blood pressure at the time, and the follow-up of one study suggested the effect persisted during adolescence [26]. Furthermore, infants may become accustomed to a salty taste, which could affect their subsequent food preferences and intake.

Method of Feeding

Regardless of the timing and content of the CF diet, it is possible that the way in which liquids and solids are given to the infant might be important for later outcomes by influencing appetite, food preferences and/or eating behaviors. An observational cohort study in the US suggested that bottle-feeding was associated with more rapid weight gain during the 1st year of life, regardless of the type of milk in the bottle [27] . Frequent bottle emptying encouraged by mothers and/ or high bottle-feeding intensity during early infancy also increased the likelihood of mothers pressuring their child to eat at age 6 years and was associated with low satiety responsiveness [28]. However, there are clearly complex interactions between mode and content of the feeding and the behavior of the care-

120 Fewtrell giver, and, where feasible, these may best be investigated in the controlled setting of a randomized trial. For example, experimental studies have shown that the glutamate content of the milk influences satiety and subsequent milk intake, as does the style of parenting [29] . Free glutamate, which is abundant in breast milk and also higher in extensively hydrolyzed formulas than in standard cow’s milk formulas, is thought to promote satiety, although it is not currently known whether these effects persist. Traditionally, infants have been spoon-fed with first solid foods in the form of purees, with subsequent introduction of semisolid and finger foods. However, in recent years, there has been an increasing tendency to avoid the initial ‘puree’ stage and progress straight to finger foods. Proponents of this ‘baby-led weaning’ method suggest that it may encourage improved eating patterns and reduce the risk of overweight and obesity by providing the infant with greater control over his/her intake and encouraging more responsive parenting [30] . However, given the self-selected nature of parents and infants who currently follow this practice, it is not clear whether such associations are causal. Furthermore, data are lacking on whether infants who are weaned using this approach obtain suf- ficient nutrients, including energy and iron, or eat a more diverse range of foods. These important issues ideally need to be tested in an RCT.

Conclusion

It is plausible, indeed likely, that CF practices influence later health and developmental outcomes and that a number of mechanisms, both nutritional and psychosocial, may be involved. Investigating these effects is challenging given the diversity and complexity of CF practices, which may limit the feasibility and generalizability of randomized trials in this field. Available data are limited in quantity and quality, but suggest that avoiding the introduction of solid foods before 4 months may reduce the risk of later obesity or allergy. It is likely that, whilst recommendations for different aspects of CF may be developed in the future based on broad principles and including consideration of later outcomes, they will need to be tailored for different populations.

Disclosure Statement

M.F. received an honorarium and expenses for participating in this NNI workshop. She has also previously received research funding from manufacturers of infant feeding products, and has participated in educational events organised by infant food manufacturers for which she has received an honorarium.

Effects of Optimal Complementary Feeding 121 References

1 Kramer MS, Kakuma R: Optimal duration of 13 Nwaru BI, Takkinen HM, Niemelä O, et al: exclusive breastfeeding. Cochrane Database Timing of infant feeding in relation to child-

Syst Rev 2002; 1:CD003517. hood asthma and allergic diseases. J Allergy

2 EFSA Panel on Dietetic Products, Nutrition Clin Immunol 2013; 131: 78–86. and Allergies (NDA): Scientific opinion on 14 Niinivirta K, Isolauri E, Nermes M, Laitinen the appropriate age for introduction of com- K: Timing of complementary feeding and the

plementary feeding of infants. EFSA J 2009; 7: risk of atopic eczema. Acta Paediatr 2014;

1423, www.efsa.europa.eu. 103: 168–173. 3 Agostoni C, Decsi T, Fewtrell MS, et al: Com- 15 Vriezinga SL, Auricchio R, Bravi E, et al: plementary feeding: a commentary by the Randomized feeding intervention in infants ESPGHAN Committee on Nutrition. J Paedi- at high risk for celiac disease. N Engl J Med

atr Gastroenterol Nutr 2008; 46: 99–110. 2014; 371: 1304–1315. 4 Fewtrell MS, Kennedy K, Singhal A, et al: 16 Lionetti E, Castellaneta S, Francavilla R, et al: How much loss to follow-up is acceptable in Introduction of gluten, HLA status, and the long-term randomised trials and prospective risk of celiac disease in children. N Engl J

studies? Arch Dis Child 2008; 93: 458–461. Med 2014; 371: 1295–1303. 5 Moorcroft KE, Marshall JL, McCormick FM: 17 Jonsdottir OH, Thorsdottir I, Gunnlaugsson Association between timing of introducing G, et al: Exclusive breastfeeding and develop- solid foods and obesity in infancy and child- mental and behavioral status in early child-

hood: a systematic review. Matern Child Nutr hood. Nutrients 2013; 5: 4414–4428.

2011; 7: 3–26. 18 Skugarevsky O, Wade KH, Richmond RC, et 6 Pearce J, Taylor MA, Langley-Evans SC: Tim- al: Effects of promoting longer-term and ex- ing of the introduction of complementary clusive breastfeeding on childhood eating feeding and risk of childhood obesity: a sys- attitudes: a cluster-randomized trial. Int J Ep-

tematic review. Int J Obesity 2013; 37: 1295– idemiol 2014; 43: 1263–1271. 1306. 19 Magnussen CG, Niinikoski H, Juonala M, et 7 Kramer MS, Matush L, Bogdanovich N, et al: al: When and how to start prevention of ath- Health and development outcomes in 6.5-y- erosclerosis? Lessons from the Cardiovascu- old children breastfed exclusively for 3 or lar Risk in the Young Finns Study and the

6 mo. Am J Clin Nutr 2009; 90: 1070–1074. Special Turku Coronary Risk Factor Inter-

8 Huh SY, Rifas-Shiman SL, Taveras EM, et al: vention Project. Pediatr Nephrol 2012; 27: Timing of solid food introduction and risk of 1441–1452. obesity in preschool-aged children. Pediatrics 20 Weber M, Grote V, Closa-Monasterolo R, et

2011; 127:e544–e551. al: Lower protein content in infant formula 9 Moss BG, Yeaton WH: Early childhood reduces BMI and obesity risk at school age: healthy and obese weight status: potentially follow-up of a randomized trial. Am J Clin

protective benefits of breastfeeding and de- Nutr 2014; 99: 1041–1051. laying solid foods. Matern Child Health J 21 Thorisdottir B, Gunnarsdottir I, Thorisdottir

2014; 18: 1224–1232. AV, et al: Nutrient intake in infancy and 10 Jonsdottir OH, Kleinman RE, Wells JC, et al: body mass index at six years in two popula- Exclusive breastfeeding for 4 versus 6 months tion-based cohorts recruited before and after and growth in early childhood. Acta Paediatr revision of infant dietary recommendations.

2014; 103: 105–111. Ann Nutr Metab 2013; 63: 145–151. 11 Prescott SL, Smith P, Tang M, et al: The im- 22 Günther AL, Remer T, Kroke A, Buyken AE: portance of early complementary feeding in Early protein intake and later obesity risk: the development of oral tolerance: concerns which protein sources at which time points and controversies. Pediatr Allergy Immunol throughout infancy and childhood are im-

2008; 19: 375–380. portant for body mass index and body fat 12 de Silva D, Geromi M, Halken S, et al: Prima- percentage at 7 y of age? Am J Clin Nutr

ry prevention of food allergy in children and 2007; 86: 1765–1772.

adults: systematic review. Allergy 2014; 69: 581–589.

122 Fewtrell 23 Thorisdottir B, Gunnarsdottir I, Palsson GI, 27 Li R, Magadia J, Fein SB, Grummer-Strawn et al: Animal protein intake at 12 months is LM: Risk of bottle-feeding for rapid weight associated with growth factors at the age of gain during the first year of life. Arch Pediatr

six. Acta Paediatr 2014; 103: 512–517. Adolesc Med 2012; 166: 431–436. 24 Morgan J, Taylor A, Fewtrell T: Meat con- 28 Li R, Scanlon KS, May A, Rose C, Birch L: sumption is positively associated with psy- Bottle-feeding practices during early infancy chomotor outcome in children up to 24 and eating behaviors at 6 years of age. Pediat-

months of age. J Pediatr Gastroenterol Nutr rics 2014; 134:S70–S77.

2004; 39: 493–498. 29 Ventura AK, Inamdar LB, Mennella JA: Con- 25 Domellöf M, Braegger C, Campoy C, et al: sistency in infants’ behavioural signalling of Iron requirements of infants and toddlers. J satiation during bottle-feeding. Pediatr Obes

Pediatr Gastroenterol Nutr 2014; 58: 119–129. 2015; 10: 180–187. 26 Geleijnse JM, Hofman A, Witteman JCM, et 30 Cameron SL, Heath A-LM, Taylor RW: How al: Long-term effects of neonatal sodium re- feasible is baby-led weaning as an approach striction on blood pressure. Hypertension to infant feeding? A review of the evidence.

1997; 29: 913–917. Nutrients 2012; 4: 1575–1609.

Effects of Optimal Complementary Feeding 123

Complementary Feeding: Taste, Eating Behavior and Later Health

Fewtrell MS, Haschke F, Prescott SL (eds): Preventive Aspects of Early Nutrition. Nestlé Nutr Inst Workshop Ser, vol 85, pp 125–134, (DOI: 10.1159/000439503) Nestec Ltd., Vevey/S. Karger AG., Basel, © 2016

Learning to Eat: Behavioral and Psychological Aspects

Leann L. Birch Department of Foods and Nutrition, University of Georgia, Athens, GA , USA

Abstract Because infants are totally dependent upon parents (or other caregivers) for care and sustenance, parents’ feeding practices are a key feature of the family environments in which infants and young children learn about food and eating. Feeding practices include not only what the child is fed, but also the how, when, why and how much of feeding. Extensive evidence indicates that parenting behavior influences a variety of child outcomes, including cognitive and socioemotional development, as well as the development of self-regulatory skills. The focus of this chapter is on what is known about how parenting, particularly feeding practices, influences the early development of several aspects of children’s eating behavior, including the acquisition of food preferences, self-regulatory skills, children’s reactivity to food cues, satiety responsiveness and ‘picky eating’. It is argued that traditional feeding practices, which evolved to protect children from environmental threats and ensure adequate intake in the context of food scarcity, can be maladaptive in current environments. An evidence base is needed to inform public policy to reduce early obesity risk in current environments, where too much palatable food is a major threat to child health. Results of recent research provides evidence that promoting responsive feeding practices can alter the development of eating behavior, sleep patterns and early self-regulatory skills, as well as reduce early obesity risk. © 2016 Nestec Ltd., Vevey/S. Karger AG, Basel

Introduction

During the first years after birth, normally developing children reach important developmental milestones: they learn to sit, crawl, stand, walk, talk and sleep through the night. They also learn to eat, and by the end of the second year of life, their patterns of intake have been transformed from the exclusive milk diet of early infancy to a modified adult diet, quite similar to the diets of other family members [1] . Of course, the eating behaviors of other family members are an important aspect of the early environment, affecting foods available to the infant and providing many opportunities for observational learning. Because infants are totally dependent upon parents (or other caregivers) for care and sustenance, parents’ feeding practices are a key feature of the family environments in which young children learn about food and eating. Feeding practices include not only what the child is fed, but also the how, when, why and how much of feeding. Extensive evidence indicates that parenting behavior influences a variety of child outcomes, including cognitive and socioemotional development, as well as the development of self-regulatory skills [2] . The focus of this chapter is on what is known about how parenting, particularly feeding practices, influences the early development of several aspects of children’s eating behavior, including the acquisition of food preferences, self-regulatory skills, children’s reactivity to food cues, satiety responsiveness and ‘picky eating’. In this chapter, the evidence presented includes findings from both basic research and primary prevention trials focused on altering parenting and feeding practices hypothesized to affect the development of children’s eating behavior during the first years of life [3] . During the first years of life, plasticity is high, and children ‘come equipped’ to learn to like and eat the diet that is available to them. These genetic predispositions include unlearned biases to like and ingest sweet, salty and umami tastes, and to reject sour and bitter. Second, they have a conservative, potentially protective bias to be ‘neophobic’, initially rejecting unfamiliar flavors and bad tasting, potentially toxic sour or bitter foods. These predispositions evolved in a very different food environment, where food, and especially sweet- and salty-tasting food, was relatively scarce, and food safety risks relatively high. In the current obesogenic food environment, industry has designed foods to be so exquisitely tuned to these taste biases, that they are no longer adaptive. Unless feeding practices are in place that promote experience with new foods to reduce neophobia, the result is young children eat diets too high in added sugar, salt and fat, and too low in vegetables, which are dominated by bitter and more likely to be avoided. It is argued below that many traditional feeding practices, which also evolved to protect children from environmental threats and ensure adequate intake in the context of food scarcity, are also maladaptive in current environments. Ev- idence-based policy is needed to inform how these traditional practices can be redesigned and adopted, serving to reduce obesity risk in current environments, where too much palatable food is a major threat to children’s health [4, 5] .

126 Birch Parents, Parenting and Traditional Feeding Practices: A Context for Learning

Parents provide genes and also play a powerful role in shaping the environments for gene expression. Early postnatal environments differ across families in a variety of ways, including whether the infant is fed breast milk, formula or both, the timing of introducing solids and table foods, the foods made available to the child, and schedules and routines for eating. While the focus here is on eating behavior, it is important to note that during this early period, parents and other caregivers control children’s access to screen time, and opportunities for active play and sleep [6] , which also affect energy balance and health. Because about two thirds of adults are currently overweight or obese [7] , most infants and young children live in families in which at least one parent is overweight or obese. Having an obese parent increases a child’s obesity risk, and intergenerational transmission of obesity is well documented [8] . However, less attention has been given to systematic differences in the family environments in which parents are or are not obese [9, 10] . There are data from observational studies indicating that obese and normal-weight parents differ in their own eating and activity patterns, and that they tend to use different feeding practices [10–12] . Obese women are less likely to exclusively breastfeed for long durations, more likely to formula feed and are more likely to encourage their infant to finish the bottle [9] . Such feeding practices are associated with overfeeding and higher weight status, and may operate conjointly with genetic risk to produce an obese phenotype. Because early parenting and feeding strategies are potentially modifiable, they are obvious targets for primary prevention [3] . Not surprisingly, with the transition to table foods, children’s diets soon resemble those of their parents and of adults in general. These diets are too high in energy, saturated fat, sugar and salt, and too low in fruits, vegetables, fiber and complex carbohydrates. However, even in infancy, the poor nutritional quality of infants’ and young children’s diets [1, 13] is evident, and the prevalence of excessive weight gain and overweight among infants and toddlers [14] underscores the need to address this issue. Data from the 2008 Feeding Infants and Toddler Study (FITS) reveal that infants’ and toddlers’ intakes exceed estimated caloric needs. Unfortunately, FITS provided no data on feeding practices or weight status, so there is an absence of evidence regarding associations between feeding practices and children’s energy intakes and weight status. Recent NHANES data also indicate that children are consuming too many foods and drinks that are high in fat and sugar, including sugar-sweetened beverages, dairy and grain desserts, and pasta dishes, which are contributing to diets in which about 40% of energy is from added sugar and fat [13] .

Learning to Eat 127 Do Traditional Feeding Practices Increase Obesity Risk in Current Environments?

Worldwide, the nutrition transition that has occurred in recent decades has created dramatic changes in the food supply and increased the availability of palatable, energy-dense, inexpensive foods [15] . We have argued previously that in the current environment, where too much food, not too little food, is a major risk to child health, traditional child feeding practices can actually exacerbate the adverse effects of our obesogenic environment on early obesity risk, promoting excessive intake and weight gain among infants, toddlers and preschoolers [16] . Traditional parenting and feeding practices evolved to protect and nurture children in the context of food scarcity, which, until recently, constituted the major environmental threat to infants and young children’s healthy growth and development. Traditional feeding practices include offering food as the default response to infant crying, feeding frequently when food is available, providing large portions, and pressuring or forcing children [17] . Recent research reveals that higher levels of ‘feeding to soothe’ (FTS) a fussy infant are related to high infant BMI z scores, but only for infants described by their mothers as high in temperamental negativity [18] ; negative infants whose mothers did not report using higher levels of FTS did not have higher BMI z scores, suggesting the key role of feeding practices in early infant weight status. A recent review also reported that excessive weight gain and obesity reported among infants who are higher in negativity may result from the more frequent use of FTS [19] . Other traditional feeding practices, including pressuring children to ‘finish the bottle’ or ‘clean their plates’, can also foster both ‘picky eating’ and excessive energy intake in today’s food environment. In one experiment, pressuring preschool children to eat a ‘healthy’ food (puréed vegetable soup) increased dislike of that food [20] , making it less likely that the food would be consumed, especially if other, more palatable foods are readily available. Paren- tal use of pressure to eat ‘healthy’ foods has also been linked to greater consumption of energy-dense sweet and savory snacks in preschoolers [21] . Another traditional practice is serving large portions of palatable foods, and there is ample evidence that young children do eat more when given larger portions of palatable entrées [22]. A recent meta-analysis indicated that doubling the size of an entrée increased children’s intake by about 20%. In addition to increasing energy intake, offering 3- to 5-year-old children larger portions of a palatable entrée can actually reduce intake of other foods on the plate, reducing fruit and vegetable intake and dietary variety at the meal [23] . Although feeding practices are often used with the intent to promote healthy patterns of intake in children, some feeding practices may disrupt the

128 Birch development of self-regulation of intake, in which eating is initiated by hunger and terminated in response to satiety cues. The use of FTS, discussed above, provides one example of feeding that is initiated in response to nonhunger cues. While evidence on the effects of feeding practices on infant intake is limited, infant ‘bottle finishing’ and FTS are commonly used with infants, and there is some evidence that such practices can promote greater consumption [9] . In an ongoing clinical trial, we are currently evaluating whether it is possible to reduce the use of traditional feeding practices, including use of FTS, serving large portions and feeding beyond satiety, by teaching new mothers more responsive feeding practices, including how to recognize and use hunger and satiety cues in feeding their infants, and also providing mothers with alternatives to FTS for a crying infant [3] . Mothers also received guidance on introducing solid foods, particularly on the use of familiarization, in combination with information on age-appropriate portion sizes and foods to offer and avoid. At this point, we are still in the data collection phase of this project, but results of a previous trial were promising [24] .

Learned Likes and Dislikes: Familiarization in Obesogenic Environments

With experience, things in the environment become familiar to the infant. Fa- miliarization is a very simple form of learning. It is a process of acquiring familiarity with objects, people, actions and their consequences [25]. The distinction between the familiar and unfamiliar is important due to the fact that familiarity has a very strong evaluative component: what becomes familiar tends to become preferred, and the unfamiliar tends to be avoided and disliked [15] . Infants learn to prefer people, objects, activities and foods that become familiar. Milk, as the single first food for infants, also becomes familiar. When weaning begins, milk provides the standard against which all other new foods and flavors are evaluated. For formula-fed infants, only the flavor of formula is familiar, but because a variety of flavors from the mother’s diet are transferred to the mother’s milk, breastfed infants have already become familiar with a variety of food flavors in the maternal diet [26] . Research of Mennella and Trabulsi [27] has revealed that these familiar flavors provide a ‘flavor bridge’, easing the transition to the foods of the adult diet consumed by the mother. In one study, breastfed infants showed more rapid acceptance of puréed vegetables during weaning, and experience with specific flavors (e.g. carrot) in breast milk promoted acceptance of that same flavor during complementary feeding. Pro- viding early experience with a variety of flavors in puréed foods also promoted acceptance of other unfamiliar flavors [5] . Early familiarization influences the

Learning to Eat 129 infants’ reactions to foods introduced at weaning and shapes the development of likes and dislikes for table foods. Unfortunately, many caregivers are not aware that familiarization is necessary to increase liking and intake of new foods. Interventions can help parents to see the ‘neophobic’ rejection response as a normal reaction to new foods, not ‘picky eating’, and research has shown that providing instructions on how to use repeated exposure to promote acceptance of new foods can have positive effects on infants’ responses to novel foods [24] . As mentioned above, infants also come equipped with predispositions to prefer or reject the basic tastes [28] . These predispositions include unlearned positive responses to sweet, salty and umami tastes, and rejection of bitter and sour tastes [26], although these initial responses to basic tastes can be modified through subsequent experience with food [29–31] . Our current food environment is tuned to our unlearned predispositions and characterized by the ready availability of inexpensive, energy-dense foods that are high in sugar and salt. Infants and young children will accept these foods and beverages the first time they are offered, even without repeated experience. It is, therefore, relatively easy to establish unhealthy dietary patterns, consisting primarily or exclusively of foods high in sugar and salt, while it takes more effort by caregivers to promote acceptance of a variety of healthy foods that lack these tastes and will only be accepted with repeated experience. The effects of exposure on the development of food and flavor preferences may differ with age and be greatest as weaning begins [5], and preferences formed in this early period can affect preferences for foods later in childhood [32] . Relatively minimal exposure can promote flavor preferences during early infancy, which may prove to be a sensitive period for learning flavor preferences [33]. In a study from our laboratory, infants who were just beginning to be offered puréed foods increased their intake of new fruits and vegetables following only a single exposure, and the effects of exposure generalized to other, similar puréed foods [34] . Learned liking or disliking of foods can also occur through associative learning, which involves the association of the food or flavor (conditioned stimulus) with the emotional tone of the unconditioned stimulus. Extensive research has provided evidence that these associative processes affect liking and intake in animal models [35] , and there is evidence that associative learning processes also play an important role in the acquisition of food likes and dislikes in young children [36] . They learn to associate foods with the emotional tone (either positive or negative) of social interactions during feeding, which affects food likes and dislikes. For example, liking for snack foods was increased when the food was either given as a reward or paired with positive adult attention [36] . In another experiment, children were served two different flavors of puréed vegetable soups at lunch in a preschool setting [20] . In the treatment condition, children were pressured by the

130 Birch adult at the table to ‘finish their soup,’ while no pressure to eat was applied in the control condition. Relative to control, children consumed less soup and made more negative comments about the soup they had been pressured to eat [20] . Eating has positive postingestive consequences, including feelings of satiety, which can increase liking for the foods eaten. There is extensive evidence on flavor-postingestive consequence learning from research with animal models, revealing that flavors paired with ingestion of foods with higher energy density are preferred to those associated with lower energy density [see ref. 35 for a review]. There is some evidence for such conditioning in young children; when 2- to 5-year-old children repeatedly consumed two different novel-flavored yogurts as snacks on alternate days, which were either high or low in energy density, greater increases in liking were obtained for flavors associated with higher energy density than those paired with low energy density yogurts [37] . Finally, in flavor-flavor learning, the conditioned stimulus is an unfamiliar flavor and the unconditioned stimulus is a familiar, preferred flavor. Following repeated pairing of the two during a series of tasting trials, in which the two are consumed together, the unfamiliar flavor can become associated with the preferred flavor, increasing liking of the new flavor, even when it is subsequently consumed alone, without the preferred unconditioned stimulus flavor. A recent study compared the effects of ‘mere exposure’ familiarization and flavor-flavor conditioning on 2- to 5-year-old children’s vegetable liking and intake [38] . Flavor-flavor learning consisted of trials in which an initially disliked vegetable was tasted either alone or with a preferred sauce the child could use as a ‘dip’ for the vegetable. We hypothesized that flavor-flavor learning would produce greater increases in liking and intake compared to familiarization, but the findings indicated that both familiarization and flavor-flavor conditioning resulted in significant increases in liking and intake after the test [38]. A challenge in familiarizing children with new foods and flavors is that tasting the food is necessary to alter preference and intake [39] . However, inducing children who are reluctant to taste a novel, disliked food can be difficult [39] . One of the benefits of pairing disliked vegetables with familiar, preferred dips was that it increased the likelihood that children would actually taste the vegetable, so that familiarization and associative processes come into play [38] .

Observational Learning: Using Social Influence to Facilitate Tasting, Liking and Intake

Social influence provides another powerful tool for promoting tasting, liking and intake of foods. Children are more likely to taste unfamiliar foods if they observe adults eating them than if the food is merely offered to the child [40] .

Learning to Eat 131 Peer modeling can also be effective; observers who watched peer models eating a food the observer disliked increased the observer’s willingness to choose and eat that food subsequently [41, 42] . Social influence affects even very young children; toddlers (14–20 months old) were more likely to try a new food after seeing it consumed by a familiar adult. We assessed young children’s (2–5 years old) responses to novel foods when an adult model (a) was not eating the food; (b) was eating a food of a different color, or (c) was eating a food of the same color as that offered to the child [43]. Children accepted and ate more of the novel food in the ‘same’ color condition, providing evidence that in young children food acceptance is promoted by specific social influence [43]. More research is necessary to understand what infants and toddlers are learning about food and eating through observation.

Summary and Conclusions

The transition to the adult diet begins in infancy, and by their 2nd birthday, children are consuming diets not very different from those of their parents. Children are born with a bias to prefer sweet and salty tastes, and to reject new foods and flavors; predispositions evolved which could have served a protective function in times of food scarcity, but in today’s world can foster unhealthy diets that are too high in sweet and salty foods. Children also learn an enormous amount about food and eating early in life, and much of this learning occurs in the family environment. Although preferences and eating behaviors continue to develop throughout the lifespan, patterns learned early tend to persist to shape both concurrent and subsequent dietary patterns, health and weight status. Parents’ feeding practices are an important feature of early environments in which infants and young children are learning about food and eating. Tradi- tional feeding practices, developed in response to the threat of food scarcity, still tend to be the default, despite dramatic changes in the food environment. These traditional practices can be maladaptive by exacerbating the effects of obesogenic environments, characterized by large portions of palatable, inexpensive food. Changes in the food environment are essential to addressing the obesity epidemic, but evidence-based alternatives to traditional feeding practices are also needed. Responsive parenting that is contingent, prompt and developmentally appropriate and fosters self-regulatory skills is an attractive alternative. Emerging evidence suggests that components of responsive parenting can also promote self-regulatory skills in feeding, which are likely to be beneficial as children are learning to eat in our obesogenic environment.

132 Birch Disclosure Statement

The author declares that no financial or other conflict of interest exists in relation to the contents of the chapter.

References

1 Siega-Riz AM, Deming DM, Reidy KC, et al: 11 Krahnstoever Davison K, Francis LA, Birch Food consumption patterns of infants and LL: Reexamining obesigenic families: parents’ toddlers: where are we now? J Am Diet Assoc obesity-related behaviors predict girls’ change

2010; 110:S38–S51. in BMI. Obes Res 2005; 13: 1980–1990. 2 Eshel N, Daelmans B, de Mello MC, Martines 12 Fernandez JR, Klimentidis YC, Dulin-Keita J: Responsive parenting: interventions and A, Casazza K: Genetic influences in child-

outcomes. Bull World Health Organ 2006; 84: hood obesity: recent progress and recom- 991–999. mendations for experimental designs. Int J

3 Paul IM, Williams JS, Anzman-Frasca S, et al: Obes (Lond) 2012; 36: 479–484. The Intervention Nurses Start Infants Grow- 13 Reedy J, Krebs-Smith SM: Dietary sources of ing on Healthy Trajectories (INSIGHT) energy, solid fats, and added sugars among

study. BMC Pediatr 2014; 14: 184. children and adolescents in the United States.

4 Paul IM, Bartok CJ, Downs DS, et al: Oppor- J Am Diet Assoc 2010; 110: 1477–1484. tunities for the primary prevention of obesity 14 Ogden CL, Carroll MD, Kit BK, Flegal KM:

during infancy. Adv Pediatr 2009; 56: 107– Prevalence of obesity and trends in body 133. mass index among US children and adoles-

5 Mennella JA, Nicklaus S, Jagolino AL, Your- cents, 1999–2010. JAMA 2012; 307: 483–490. shaw LM: Variety is the spice of life: strate- 15 Birch LL, Anzman SL: Learning to eat in an gies for promoting fruit and vegetable accep- obesogenic environment: a developmental

tance during infancy. Physiol Behav 2008; 94: systems perspective on childhood obesity.

29–38. Child Dev Perspect 2010; 4: 138–143. 6 Birch LL, Parker L, Burns A (eds): Early 16 Birch LL, Anzman-Frasca S: Learning to pre- Childhood Obesity Prevention Policies. fer the familiar in obesogenic environments; Washington, The National Academies Press, in van Goudoever H, Guandalini S, Kleinman 2011. RE (eds): Early Nutrition: Impact on Short- 7 Flegal KM, Carroll MD, Kit BK, Ogden CL: and Long-Term Health. Nestlé Nutr Work- Prevalence of obesity and trends in the distri- shop Ser, Pediatr Program. Basel, Karger, bution of body mass index among US adults, 2011, vol 68, pp 187–196; discussion 96–99.

1999–2010. JAMA 2012; 307: 491–497. 17 LeVine RA: Human parental care: universal 8 Maes HH, Neale MC, Eaves LJ: Genetic and goals, cultural strategies, individual behavior.

environmental factors in relative body weight New Dir Child Adolesc Dev 1988; 1988: 3–12.

and human adiposity. Behav Genet 1997; 27: 18 Stifter CA, Anzman-Frasca S, Birch LL, 325–351. Voegtline K: Parent use of food to soothe in- 9 Thompson AL: Intergenerational impact of fant/toddler distress and child weight status.

maternal obesity and postnatal feeding prac- An exploratory study. Appetite 2011; 57: 693–

tices on pediatric obesity. Nutr Rev 2013; 699. 71(suppl 1):S55–S61. 19 Anzman-Frasca S, Stifter CA, Birch LL: Tem- 10 Francis LA, Ventura AK, Marini M, Birch LL: perament and childhood obesity risk: a re- Parent overweight predicts daughters’ in- view of the literature. J Dev Behav Pediatr

crease in BMI and disinhibited overeating 2012; 33: 732–745. from 5 to 13 years. Obesity (Silver Spring) 20 Galloway AT, Fiorito LM, Francis LA, Birch

2007; 15: 1544–1553. LL: ‘Finish your soup’: counterproductive effects of pressuring children to eat on intake

and affect. Appetite 2006; 46: 318–323.

Learning to Eat 133 21 Campbell KJ, Crawford DA, Ball K: Family 32 Beauchamp GK, Mennella JA: Early flavor food environment and dietary behaviors like- learning and its impact on later feeding be-

ly to promote fatness in 5–6 year-old chil- havior. J Pediatr Gastroenterol Nutr 2009;

dren. Int J Obes (Lond) 2006; 30: 1272–1280. 48(suppl 1):S25–S30. 22 Fisher JO, Rolls BJ, Birch LL: Children’s bite 33 Mennella JA, Lukasewycz LD, Castor SM, size and intake of an entrée are greater with Beauchamp GK: The timing and duration of large portions than with age-appropriate or a sensitive period in human flavor learning: a

self-selected portions. Am J Clin Nutr 2003; randomized trial. Am J Clin Nutr 2011; 93:

77: 1164–1170. 1019–1024. 23 Savage JS, Fisher JO, Marini M, Birch LL: 34 Birch LL, Gunder L, Grimm-Thomas K, La- Serving smaller age-appropriate entree por- ing DG: Infants’ consumption of a new food tions to children aged 3–5 y increases fruit enhances acceptance of similar foods. Appe-

and vegetable intake and reduces energy den- tite 1998; 30: 283–295. sity and energy intake at lunch. Am J Clin 35 Myers KP, Sclafani A: Development of

Nutr 2012; 95: 335–341. learned flavor preferences. Dev Psychobiol

24 Paul IM, Savage JS, Anzman SL, et al: Pre- 2006; 48: 380–388. venting obesity during infancy: a pilot study. 36 Birch LL: Development of food preferences.

Obesity (Silver Spring) 2011; 19: 353–361. Annu Rev Nutr 1999; 19: 41–62. 25 Rheingold HL: Development as the acquisi- 37 Johnson SL, McPhee L, Birch LL: Condi-

tion of familiarity. Annu Rev Psychol 1985; tioned preferences: young children prefer

36: 1–18. flavors associated with high dietary fat. Physi-

26 Mennella JA: Flavour programming during ol Behav 1991; 50: 1245–1251.

breast-feeding. Adv Exp Med Biol 2009; 639: 38 Anzman-Frasca S, Savage JS, Marini ME, et 113–120. al: Repeated exposure and associative condi- 27 Mennella JA, Trabulsi JC: Complementary tioning promote preschool children’s liking

foods and flavor experiences: setting the of vegetables. Appetite 2012; 58: 543–553.

foundation. Ann Nutr Metab 2012; 39 Birch LL, McPhee L, Shoba BC, et al: What

60( suppl 2): 40–50. kind of exposure reduces children’s food neo-

28 Schwartz C, Issanchou S, Nicklaus S: Devel- phobia? Looking vs. tasting. Appetite 1987; 9: opmental changes in the acceptance of the 171–178. five basic tastes in the first year of life. Br J 40 Harper LV, Sanders KM: The effect of adults’

Nutr 2009; 102: 1375–1385. eating on young children’s acceptance of un-

29 Beauchamp GK, Cowart BJ: Congenital and familiar foods. J Exp Child Psychol 1975; 20: experiential factors in the development of 206–214.

human flavor preferences. Appetite 1985; 6: 41 Birch LL: Effects of peer models’ food choices 357–372. and eating behaviors on preschoolers’ food

30 Beauchamp GK, Moran M: Dietary experi- preferences. Child Dev 1980; 51: 489–496. ence and sweet taste preference in human 42 Hendy HM: Effectiveness of trained peer

infants. Appetite 1982; 3: 139–152. models to encourage food acceptance in pre-

31 Sullivan SA, Birch LL: Pass the sugar, pass the school children. Appetite 2002; 39: 217–225. salt: experience dictates preference. Dev Psy- 43 Addessi E, Galloway AT, Visalberghi E, Birch

chol 1990; 26: 546–551. LL: Specific social influences on the acceptance of novel foods in 2-5-year-old children.

Appetite 2005; 45: 264–271.

134 Birch Complementary Feeding: Taste, Eating Behavior and Later Health

Fewtrell MS, Haschke F, Prescott SL (eds): Preventive Aspects of Early Nutrition. Nestlé Nutr Inst Workshop Ser, vol 85, pp 135–143, (DOI: 10.1159/000439504) Nestec Ltd., Vevey/S. Karger AG., Basel, © 2016

The Development of Flavor Perception and Acceptance: The Roles of Nature and Nurture

Catherine A. Forestell Department of Psychology, College of William & Mary, Williamsburg, VA , USA

Abstract Our ability to perceive the broad range of flavors imparted by foods involves the assimila- tion of multiple chemosensory sensations: primarily those of taste and olfaction. Due to their adaptive value, these chemosensory systems are functional before birth and continue to mature throughout childhood. As a result, children live in their own flavor world, preferring foods that are high in sugar and salt over those that are sour and bitter tasting, such as fruits and vegetables. Although these flavor preferences are not consistent with a healthful diet, they can be ‘fine tuned’ by sensory experiences beginning prenatally. Through exposure to the flavors of amniotic fluid and breast milk, which reflect the foods within the mother’s diet, infants become more accepting of foods within their culture. In contrast, exclusively formula-fed children, who do not benefit from the ever-changing flavor profile of breast milk, learn only about the flavor of their formula. Early learning about flavors continues at weaning, through repeated exposure to a variety of foods. Thus, mothers who consume an array of healthy foods themselves throughout pregnancy and lactation, and subsequently feed their children these foods at weaning, can promote healthful eating habits in their children and families. © 2016 Nestec Ltd., Vevey/S. Karger AG, Basel

Introduction

Over the past several decades, with the emergence of developmentally appropriate methodologies for testing taste and odor sensitivity and preferences in children, we have begun to gain important insights into the unique flavor world in which children live [for a review, see ref. 1 ]. Based on this work, we are beginning to understand some of the factors involved in children’s maladaptive food choices. In this chapter, after providing a brief overview of the basic biology of taste and smell, advances in our understanding of sensory capabilities and acceptance patterns of the human fetus and infant will be reviewed. We will then discuss how early sensory experiences during feeding interact with the plasticity of the chemosensory system to shape subsequent preferences for foods.

Taste, Smell and Flavor Perception

According to its precise definition, taste refers to the sensations that occur when chemicals dissolved in saliva come into contact with taste receptors that are ar- ranged in groups of 50–100 cells called taste buds throughout the oral cavity. These clusters of cells send messages to the brain via three cranial nerves, the facial (VIIth), glossopharyngeal (IXth) and vagal (Xth) nerves, which allow us to perceive a small number of primary taste qualities, namely, sweet, salty, bitter, sour and savory. Olfaction arises from receptors located in the epithelium of the nasal cavity, which are innervated by a single cranial nerve (I). Unlike the limited number of primary tastes, there are thousands of distinctive odors with separable sensations. As shown in figure 1 , odor stimuli reach olfactory receptors either through inhalation through the nostrils, which comprises the orthonasal route, or volatile chemicals from foods may travel along the retronasal route through the back of the nasopharynx towards the roof of the nasal cavity. The latter occurs during chewing and swallowing in adults and children and suckling in infants when the nasal passages are open. When the nasal passages become blocked, as sometimes occurs when suffering from a cold, the aroma of foods is prevented from travel- ing along the retronasal route. As a result, many cold sufferers complain that their food ‘tastes’ bland or lacks flavor. The latter description is more accurate given that flavor , which in everyday language is often used interchangeably with taste, is elicited by a combination of taste and olfactory sensations. Indeed, it is olfactory sensations that contribute many of the distinctive characteristics of foods, such as the sensations of vanilla, strawberry or garlic, which are often erroneously described as tastes.

Prenatal Development of Taste and Odor Sensitivity

Relative to other sensory capacities such as vision and audition, the sense of taste begins to emerge relatively early. Just 8 weeks after conception, taste buds begin to appear and by the 13th–14th week they begin to morphologically resemble

136 Forestell Olfactory bulb and nerves

Orthonasal olfaction

Retronasal olfaction

Tongue

Fig. 1. Arrows represent orthonasal and retronasal routes of olfaction. Reprinted from Lipchock et al. [32] with permission from Elsevier.

those of the adult. Taste pores, which provide tastants with access to taste bud receptor cells, are generally considered to be markers of functional maturity [2] . Such pores have been identified in fetal fungiform papillae before the end of the 4th month [3]. Soon after (i.e. between 18 and 24 weeks), sucking and swallowing behaviors emerge and by 35–40 weeks of gestation these actions become co- ordinated [4]. This behavior stimulates the taste buds, influencing their synaptic connections and representing a major route of amniotic fluid absorption. Be- havioral studies using a variety of techniques suggest that by the last trimester, taste buds are capable of detecting and communicating information to structures that are responsible for organizing and controlling affective behaviors within the central nervous system [5, 6] . Because the amniotic fluid is in constant flux throughout pregnancy, with the concentrations of sugars, sodium and po- tassium salts, and various acids constantly changing, the fetus is therefore exposed to a rich taste environment [5] . It has been estimated that the near-term human fetus swallows 500–1,000 ml of amniotic fluid per day [4] and actively inhales more than twice this volume. The olfactory bulbs and receptor cells, which attain adult-like morphology by

The Development of Flavor Perception and Acceptance 137 the 11th week of gestation, are functional by this time. As a result, they are capable of detecting the continually changing odor profile of the amniotic fluid. In addition to containing chemicals with distinct taste properties, amniotic fluid contains volatile chemicals transmitted from the maternal diet [7]. The fetus not only perceives these olfactory changes, but as will be discussed later, there is evidence that these experiences are encoded and remembered. Given the extensive prenatal development of the taste and smell systems, it is not surprising that the newborn is sensitive and responsive to odor and taste stimuli after birth. These chemosensory systems continue to develop and change throughout childhood as a result of the interplay between children’s basic biology and their sensory experiences. In combination, these sensory changes will ultimately contribute to dietary habits and preferences in adulthood.

Taste Sensitivity and Acceptance throughout Childhood: The Role of Biology

Studies have used a variety of techniques to reveal that newborn infants can distinguish and differentially respond to basic tastes by emitting a combination of consummatory and reflexive responses, such as facial expressions that reflect hedonic or distaste reactions [1] . These studies have shown that newborn infants can perceive and display hedonic responses, such as licking, facial relaxation and smiling, to umami in soup broth and sweet tastes [8, 9] . Although little is known about the developmental progression of children’s sensitivity to and preferences for umami taste, we know that within days after birth, infants are adept at detecting dilute sweet solutions, differentiating varying degrees of sweetness and different kinds of sugars [10] . Preference for sweet taste remains heightened throughout childhood and declines to that observed in the adult during late adolescence [11] . Newborn infants respond with indifference to salty tastes [8, 9, 12] . This ability to detect and respond to salt, occurring between 2 and 6 months of age, is likely a result of postnatal maturation of salt detection mechanisms [12] . Subse- quently, between 3 and 11 years of age, a preference for salt emerges that is greater than that of adults [13, 14] . In contrast, neonates generally respond to bitter and sour tastes with aversive responses. At birth, concentrated bitter solutions elicit strong orofacial responses, such as gaping and nose wrinkling [8, 9] . However, consumption tests indicate that rejection of low-to-moderate concentrations of bitter is not evident until the 2nd week of life [15] , suggesting a developmental change in bitter perception and/ or the ability to regulate the intake of bitter solutions. Although initial reactions

138 Forestell to sour taste are negative (i.e. eye squinting and lip pursing [8, 9] ), for some children, these responses transform into preference by 18 months of age [16] . That the aforementioned responses to the basic tastes are remarkably similar across cultures [9] and species [17], and can be elicited in newborns with limited feeding experience [8, 9, 17, 18] as well as in those with anencephaly [18] , suggests that they are innate and a function of children’s basic biology. From an evolutionary perspective, these responses are thought to enhance survival. Pref- erence for salty, sweet and savory tastes is thought to attract us to foods such as salty-tasting minerals, energy-producing sugars, and vitamin- and protein-rich foods (such as glutamate which imparts a savory taste) that are important for growth and development. This is supported by research suggesting that preferences for sweet-tasting foods are correlated with periods of high growth [19] . Rejection of bitter and sour tastes is thought to inhibit ingestion of potentially dangerous substances such as poisons, many of which are bitter. Thus, children’s innate taste preferences help to explain why they like to consume the foods and beverages they do. Although research indicates that age is a particularly good predictor of differences in taste preferences, additional factors, such as our genes, contribute to individual differences in sensory perception and acceptance. To date, the TAS2R38 gene, which is responsible for variance in bitter taste perception, has been the most extensively studied. This work has demonstrated that variation in the TAS2R38 gene causes individual differences in perception of a class of bitter- tasting compounds commonly found in cruciferous vegetables [20] . Depending on their genotype, some individuals have a high sensitivity threshold for this class of bitters (nontasters), and as a result typically enjoy eating cruciferous vegetables. While those who have a lower threshold perceive the bitterness in cruciferous vegetables as either moderately intense (medium tasters) or very intense (supertasters). Research has shown that for medium tasters, the strength of the phenotype-genotype relationship for bitter sensitivity varies with age. That is, these children are more sensitive than adolescents, who are in turn more sensitive than adults to these bitter compounds [21]. These findings suggest that bitter taste sensitivity can decline throughout childhood, allowing some children to become more accepting of foods that contain these compounds over time.

Early Flavor Experiences and Their Role in Food Acceptance

Amniotic Fluid and Breast Milk As discussed above, sensory capacities emerge during the fetal period that allow the baby to respond to and learn about stimuli within the environment.

The Development of Flavor Perception and Acceptance 139 These stimuli include a wide range of odor volatiles (e.g. alcohol, garlic, vanilla and carrot) ingested by the mother, which have been shown to be transmitted to amniotic fluid and breast milk [see ref. 22 for a review]. Early exposures to these flavors serve to enhance acceptance and preference for similarly flavored foods at weaning. This has been demonstrated in human studies in which mothers consumed foods or beverages containing a particular target flavor (e.g. carrot juice) during pregnancy or lactation, while mothers in a control group did not. Several weeks after birth, infants whose mothers consumed the carrot juice flavor demonstrated a preference for baby cereal when carrot was added as a fla- vorant [23] relative to infants in the control groups. Thus, repeated exposure to flavors within amniotic fluid or breast milk may be one of the first ways that children learn about the foods within their culture. Research has supported this finding by showing that breastfed infants are more accepting of fruits and vegetables than formula-fed infants, but only if their mothers regularly eat these foods during lactation [24] . Hence, it appears that breast milk functionally serves as a ‘bridge’ that extends and connects the olfactory experiences of the fetus to the infants’ flavor experiences at weaning.

Formula Feeding In the United States, by 3 months of age, approximately 40% of infants are exclusively breastfed [25] , while the remaining infants receive formula, either in combination with breast milk or exclusively. Because formula has a fixed flavor profile, infants who are formula fed miss out on the ever-changing array of flavors provided by breast milk. However, we now know that the flavors of certain formulas may also enhance children’s acceptance of healthful foods at weaning. Cow’s-milk-based formulas (CMF) account for the majority of formula sales, as it is formulated for healthy infants. Extensively protein hydrolyzed formulas (ePHF), which contain hydrolyzed proteins in the form of peptides and free amino acids, is available to infants who suffer from cow’s milk protein allergies or intolerance. Because many of the free amino acids in ePHF taste sour and bitter, these formulas have an extremely unpalatable taste that is accompanied by an offensive odor. This is in stark contrast to CMF, which is described as having low levels of sweet and sour tastes and cereal-like odors. Using these inherent differences between these formulas as a model system, studies have demonstrated clear differences in the acceptance of basic tastes as a function of the type of formula fed to infants [26] . In one study, infants were tested on six occasions to measure their acceptance of sweet, salty, bitter, savory, sour and plain cereals. Results demonstrated that ePHF infants ate significantly more savory-, bitter-, and sour-tasting and plain cereals and displayed fewer facial expressions of distaste while eating the bitter and savory cereals than did

140 Forestell infants who were breastfed or fed CMF. Other studies have shown that these learned preferences may extend past the weaning period. Four- to 5-year-old children who were fed ePHF during infancy were more likely to prefer a sour- tasting apple juice and were more likely to preferentially consume broccoli, which has similar flavor notes to ePHF compared to children fed CMF [27] . In combination, these studies revealed that the aromas and tastes to which infants are exposed during formula feedings will depend on the type and brand of formula they are fed, which will in turn affect infants’ acceptance of foods at weaning.

Introduction of Complimentary Foods Complimentary foods, defined as all liquid, semisolid and solid foods that are fed in addition to breast milk or formula, are typically introduced to infants during weaning, which according to World Health Organization (WHO) guidelines generally should not occur before 6 months of age [28] . In addition to defining the optimal age for introducing complimentary foods, the WHO guidelines include recommendations about their nutritional content [28] . For example, drinks that are high in sugar content, such as soda, and excessive juice consumption should be avoided because they decrease the child’s appetite for more nutritive foods [28] . Such early experiences with sweet-tasting beverages may also have long-term effects; babies routinely fed sweetened water during the first months of life later exhibit a greater preference for sweetened water at 2 years [29] and at 10 years of age [30] compared to those who had little or no experience with sweetened water. Like experience with sweet, experiences with salt provide the infant with opportunities to learn about the level of saltiness to be expected in foods. Six-month-old infants’ acceptance of moderate and strong salt solutions [12] or salted cereals is enhanced by early experiences with foods that contain sodium. Because many foods and beverages manufactured for pediatric populations are high in refined sugar and salt content, children are learning to prefer and overconsume sweet and salty foods. This is of particular concern given that these dietary behaviors have been associated with obesity and high blood pressure, a leading risk factor for later health issues, such as heart disease and stroke. Early food environments should instead acquaint children a variety of healthful foods that meet nutritional needs for optimal growth and development [28] . Through repeated exposure to a variety of these foods, children learn to like and prefer them. Experimental studies have demonstrated that it may take as many as 8–10 exposures before infants learn to like new foods [24] . Moreover, exposure to an array of foods that vary in flavor and texture both within and between meals additionally serves to enhance children’s acceptance of novel foods [31] . However, many children’s diets are limited because parents often hesitate to continue feeding foods (such as sour-tasting fruits or bitter green vegetables) that initially

The Development of Flavor Perception and Acceptance 141 induce aversive facial responses. As a result of their limited diet, these children may become less willing to accept novel foods. Our research suggests that although infants may continue to show facial expressions of distaste (e.g. grimace) upon repeated exposure to a new food, their intake will increase with repeated exposures [25] . These findings suggest that caregivers should focus on the infants’ willingness to consume a food and not solely on their aversive facial expressions.

Closing Remarks

The convergence of findings from basic research has revealed that children live in different sensory worlds than adults, preferring sweeter, saltier and in some cases more sour tastes, while more intensely disliking bitter tastes when compared to adults. This is especially true for medium tasters for whom sensitivity to certain classes of bitter tastes declines throughout development. Although children have very clear likes and dislikes, recurrent exposures to a variety of taste and flavor stimuli associated with healthful eating, beginning before birth and continuing throughout infancy, can override these biological responses to some extent. To be sure, these experiences will not generally lead children to prefer bitter green vegetables over candy. However, their preferences will shift from sweet and salty foods toward a healthier array of flavors. By building an array of healthful foods into the family’s daily diet, parents will help their children develop healthful preferences that are maintained into adulthood.

Disclosure Statement

The author has no financial interests to disclose.

References

1 Forestell CA, Mennella JA: The Ontogeny of 3 Witt M, Reutter K: Embryonic and early fetal taste perception and preference throughout development of human taste buds: a trans- childhood; in Doty RL (ed): Handbook of mission electron microscopical study. Anat

Olfaction and Gustation, ed 3. New York, Rec 1996; 246: 507–523. Dekker, 2015, pp 797–829. 4 Ross MG, Nijland MJ: Development of inges-

2 Mistretta CM: Topographical and histologi- tive behavior. Am J Physiol 1998; 43:R879– cal study of the developing rat tongue, palate R893. and taste buds; in Bosma JF (ed): Third Sym- 5 Liley AW: Disorders of amniotic fluid; in As- posium on Oral Sensation and Perception: sali NS (ed): Pathophysiology of Gestation: The Mouth of the Infant. Springfield, Thom- Fetal Placental Disorders. New York, Aca- as, 1972, pp 163–186. demic Press, 1972, pp 157–206.

142 Forestell 6 Maone TR, Mattes RD, Bernbaum JC, Beau- 20 Wooding S, Gunn H, Ramos P, et al: Genetics champ GK: A new method for delivering a and bitter taste responses to goitrin, a plant

taste without fluids to preterm and term in- toxin found in vegetables. Chem Senses 2010;

fants. Dev Psychobiol 1990; 13: 179–191. 35: 685–692. 7 Mennella JA, Johnson A, Beauchamp GK: 21 Mennella JA, Pepino MY, Duke FF, Reed DR: Garlic ingestion by pregnant women alters Age modifies the genotype-phenotype rela- the odor of amniotic fluid. Chem Senses tionship for the bitter receptor TAS2R38.

1995; 20: 207–209. BMC Genet 2010; 11: 60. 8 Steiner JE: What the neonate can tell us about 22 Mennella JA: The chemical senses and the umami; in Kawamura Y, Kare MR (eds): development of flavor preferences in hu- Umami: A Basic Taste. New York, Dekker, mans; in Hale TW, Hartmann PE (eds): Text- 1987, pp 97–103. book on Human Lactation. Amarillo, Hale 9 Rosenstein D, Oster H: Differential facial re- Publishing, 2007, pp 403–414. sponses to four basic tastes in newborns. 23 Mennella JA, Jagnow CP, Beauchamp GK:

Child Dev 1988; 59: 1555–1568. Prenatal and postnatal flavor learning by hu-

10 Desor J, Maller O, Turner RE: Taste in accep- man infants. Pediatrics 2001; 107:E88. tance of sugars by human infants. J Comp 24 Forestell CA, Mennella JA: Early determi-

Physiol Psychol 1973; 84: 496–501. nants of fruit and vegetable acceptance. Pedi-

11 Desor JA, Beauchamp GK: Longitudinal atrics 2007; 120: 1247–1254. changes in sweet preferences in humans. 25 Grummer-Strawn LM, Scanlon KS, Fein

Physiol Behav 1987; 39: 639–641. SB: Infant feeding and feeding transitions

12 Stein LJ, Cowart BJ, Beauchamp GK: The de- during the first year of life. Pediatrics 2008;

velopment of salty taste acceptance is related to 122(suppl 2): S36–S42. dietary experience in human infants: a prospec- 26 Mennella JA, Forestell CA, Morgan LK,

tive study. Am J Clin Nutr 2012; 95: 123–129. Beauchamp GK: Early milk feeding influenc- 13 Beauchamp GK, Cowart BJ: Preference for es taste acceptance and liking during infancy.

high salt concentrations among children. Dev Am J Clin Nutr 2009; 90: 780S–788S.

Psychol 1990; 26: 539–545. 27 Mennella JA, Beauchamp GK: Flavor experi- 14 Desor JA, Greene LS, Maller O: Preferences ences during formula feeding are related to for sweet and salty in 9- to 15-year-old and preferences during childhood. Early Hum

adult humans. Science 1975; 190: 686–687. Dev 2002; 68: 71–82. 15 Kajiura H, Cowart BJ, Beauchamp GK: Early 28 World Health Organization: Guiding Prin- developmental change in bitter taste respons- ciples for Complementary Feeding of the

es in human infants. Dev Psychobiol 1992; 25: Breastfed Child. Geneva, World Health Or- 375–386. ganization, 2003, http://www.who.int/ 16 Blossfeld I, Collins A, Boland S, et al: Rela- nutrition/publications/guiding_principles_ tionships between acceptance of sour taste compfeeding_breastfed.pdf (accessed and fruit intakes in 18-month-old infants. Br October 10, 2014).

J Nutr 2007; 98: 1084–1091. 29 Beauchamp GK, Moran M: Acceptance of 17 Steiner JE, Glaser D, Hawilo ME, Berridge sweet and salty tastes in 2-year-old children.

KC: Comparative expression of hedonic im- Appetite 1984; 5: 291–305. pact: affective reactions to taste by human 30 Pepino MY, Mennella JA: Factors contribut- infants and other primates. Neurosci Biobe- ing to individual differences in sucrose pref-

hav Rev 2001; 25: 53–74. erence. Chem Senses 2005; 30(suppl 1):i319– 18 Steiner JE: The gustofacial response: observa- i320. tion on normal and anencephalic newborn 31 Mennella JA, Nicklaus S, Jagolino AL, Your- infants; in Bosma JF (ed): Fourth Symposium shaw LM: Variety is the spice of life: strate- on Oral Sensation and Perception. Bethesda, gies for promoting fruit and vegetable accep-

US Department of Health, Education and tance during infancy. Physiol Behav 2008; 94: Welfare, 1973, pp 254–278. 29–38. 19 Coldwell SE, Oswald TK, Reed DR: A marker 32 Lipchock SV, Reed DR, Mennella JA: The of growth differs between adolescents with gustatory and olfactory systems during infan- high vs. low sugar preference. Physiol Behav cy: implications for development of feeding

2009; 96: 574–580. behaviors in the high-risk neonate. Clin Peri-

natol 2011; 38: 627–641.

The Development of Flavor Perception and Acceptance 143

Complementary Feeding: Taste, Eating Behavior and Later Health

Fewtrell MS, Haschke F, Prescott SL (eds): Preventive Aspects of Early Nutrition. Nestlé Nutr Inst Workshop Ser, vol 85, pp 145–154, (DOI: 10.1159/000439505) Nestec Ltd., Vevey/S. Karger AG., Basel, © 2016

Dietary Patterns during Complementary Feeding and Later Outcomes

Pauline M. Emmett Centre for Child and Adolescent Health, School of Social and Community Medicine, University of Bristol, Bristol , UK

Abstract Guidelines for healthy infant feeding provide advice on breastfeeding and complementary feeding. The Avon Longitudinal Study of Parents and Children (ALSPAC) derived dietary patterns in comparison to infant feeding guidelines and by using principal components analysis (PCA). The ALSPAC cohort was recruited during pregnancy. Parent-completed questionnaires assessed diet at age 6 and 15 months. Children were weighed and measured at 7 years of age and IQ was assessed at 8 years. A complementary feeding utility index was calculated in relation to 14 feeding recommendations. High scores on the index were due to longer breastfeeding, and feeding more fruit and vegetables and less ready-prepared baby foods. The index scores were positively related to IQ and ‘healthy’ dietary patterns in childhood. In infancy four dietary patterns were derived from PCA at each age. Three occurred at both ages: ‘HM traditional’ (home-made meat, vegetables and desserts), ‘discretionary’ (processed adult foods) and ‘RM baby foods’ (commercial ready-made baby foods). A ‘breastfeeding’ pattern was derived at 6 months, with fruit and vegetables included. At 15 months, a ‘HM contemporary’ pattern included cheese, fish, nuts, legumes, fruit and vegetables. The ‘discretionary’ and ‘RM baby foods’ patterns at both ages were negatively associated, while the ‘breastfeeding’ and ‘HM contemporary’ patterns were positively associated with IQ. These results suggest that infant diet influences cognitive development in children and may set a trend for later eating patterns. © 2016 Nestec Ltd., Vevey/S. Karger AG, Basel

Introduction

Complementary feeding is the time when foods are first introduced into the infant’s diet and the aim is to move the child gradually from a milk-based to a family food-based diet. Many countries have recommended guidelines on how this should be achieved with information on the timing and type of foods to introduce. These recommendations are usually based on WHO guidelines [1] . It is important to bear in mind that the family diet should also be based on healthy eating principles such as those set out in the dietary guidelines for Americans [http://www.health.gov/dietaryguidelines/], for example. To evaluate the pattern of eating for groups of individuals, the foods consumed can be assessed in relation to guidelines for healthy eating such as those for complementary feeding. Each subject is allocated a score in relation to their closeness to the guidelines. Alternatively dietary patterns can be derived using data-driven methods such as principal component analysis (PCA). This method utilizes correlations between different foods and identifies foods that are often consumed together. Each subject has a score for each dietary pattern derived from the data. The number of dietary patterns derived depends on the complexity of the diets consumed by the subjects. Although dietary patterns have been derived to describe the diet of adults and children, there are very few published studies assessing dietary patterns during infancy. One UK study derived dietary patterns in infants at 6 and 12 months using PCA and found two similar patterns at each age [2] . These they named ‘infant guidelines’ and ‘family foods’ because of the types of foods which were strongly associated with the patterns. Further investigation showed that scores on the ‘infant guidelines’ pattern were positively associated with the verbal IQ of the children at 4 years of age [3] . However, there was no association with BMI at the same age [4] . Furthermore, data collected during infancy by the Avon Longitudinal Study of Parents and Children (ALSPAC) have been used to derive dietary patterns both in relation to complementary feeding guidelines and using PCA. These patterns have been related to social background and later childhood outcomes, and will be described in this chapter.

Subjects and Methods

ALSPAC is a birth cohort study which recruited pregnant women resident in the county of Avon (UK) with an expected delivery date between April 1991 and December 1992 (14,541 pregnancies) [5]. Ethical approval was obtained from the ALSPAC Law and Eth- ics Committee, and the Local Research Ethics Committees. The cohort was population based and broadly representative, at recruitment, of the population of women with children under 1 year in Avon [5]. The children (14,062 at birth, 13,988 alive at 1 year) have been followed by parental-completion questionnaires, educational records and by hands-on assessment at dedicated research clinics [5] . Details of the socioeconomic background of the family were collected by questionnaires. These include: highest maternal educational attainment, maternal age, smoking status and BMI before pregnancy. Mothers who completed the questionnaires over time have higher educational attain-

146 Emmett Table 1. Food intake and feeding behaviors for the lowest and highest quintiles scored on the CFUI in ALSPAC infants [adapted from ref. 8]

Index score Lowest quintile Highest quintile (n = 1,852) (n = 1,854)

Breastfeeding duration, months 1 (0–2) 8 (5–12)* Age starting solids, months 3 (2–3) 4 (3–4) Age starting cow’s milk, months 9 (6–11) 12 (9–12)* Age starting lumpy foods, months 7 (6–9) 7 (6–8) Protein foods Meat intake, times/week 5 (2–7) 4 (2–6)* Fish intake, times/week 0 (0–1) 1 (0–2) Egg intake, times/week 0 (0) 0 (0) Vegetable intake, times/day 1 (0.4–1.7) 1.7 (1.3–2.1)* Fruit intake, times/day 0.7 (0.3–1.0) 1.1 (0.9–1.6)* Ready-prepared baby food intake, times/week 14 (7–18) 8 (3–14)* Exposure to sugary drinks, n 1 (1–2) 1 (0–1) Exposure to energy-dense, nutrient-poor foods, n 1 (0–2) 0 (0–1)* Solid meals, n/day 3 (3–3) 3 (3–3) Always fed on demand, % 22 56* Fed iron-fortified baby cereal, % 77 100* No exposure to tea, % 77 100*

* p < 0.005, vs. the lowest quintile of the CFUI score. Medians (interquartile ranges) are shown except for feeding behaviors (%).

ment, are older and have more favorable health indicators than mothers who did not [5] . Infant diet was assessed by questionnaires at 6 and 15 months (n >11,000 at each age) [6] . These questionnaires covered milk feeding and a range of foods normally consumed by infants in the UK as well as the timing of introduction of foods and feeding behaviors. Food frequency questionnaires covering maternal and childhood diet were completed by parents at various times during the study. From 7 years of age, the children were invited annually to research clinics where they were weighed and measured using standardized procedures, and at 8 years their IQ was assessed using a validated age-adjusted shortened version of the Wechsler Intelligence Scale for Children [7] . At 7 years, a blood sample was taken for blood lipid measurements, and blood pressure was measured by a standardized procedure. The ALSPAC website contains details of all the data that are available through a fully searchable data dictionary (http://www.bris.ac.uk/alspac/researchers/data-access/data-dictionary/).

Infant Feeding Guidelines Score The dietary data at 6 and 15 months were assessed in relation to infant feeding guidelines (duration of breastfeeding, timing of solid introduction, feeding on demand and various recommendations about the types of foods and drinks to be introduced) compiled from international sources [8] . The scoring system was made up from 14 components (listed in table 1 ), each given the same weight with a total combined score calculated. It assessed the degree of adherence to the infant feeding guidelines and ranged from 0 to 1.0.

Dietary Patterns during Complementary Feeding 147 Data-Driven Dietary Patterns Using the questionnaire data collected at 6 and 15 months, PCA was performed on the standardized food items [9]. PCA forms linear combinations of the food variables by grouping together correlated variables, thus identifying underlying patterns within the data. The coefficients defining these patterns are called ‘factor loadings’ and the number patterns are chosen on the basis of a scree plot [10] and the interpretability of the factor loadings. A score (mean 0, standard deviation 1) was created for each child for each of the dietary patterns identified. Foods with factor loadings above 0.3 on each pattern were considered to have a strong association with that pattern and were used to help to describe and label the pattern. Dietary patterns were also derived using PCA from food frequency questionnaires assessing the child’s diet at age 3 and 7 years [11, 12] .

Results

For the comparison with the infant feeding guidelines, complete data were available for 6,065 infants. The difference between quintiles of the score derived from the guidelines (table 1 ) was most marked for breastfeeding duration, feeding on demand, age at introduction of cows’ milk, number of portions of vegetables and fruit, and the number of servings of commercial ready-prepared baby food [8] . The mean score on the index was 0.48 ± 0.10 (range 0.18–0.86). Better adherence to the guidelines was more likely in older mothers (p < 0.001) and those with higher education (p < 0.001). Mothers who were overweight or obese before pregnancy had lower scores on the guideline index than those of normal weight (p < 0.001). In longitudinal analysis ( table 2 ), better adherence to the guidelines adjusted for social background was independently associated with increasing scores on ‘health-conscious’ dietary patterns derived from PCA at both 3 and 7 years as well as decreasing scores on ‘processed’ dietary patterns at both ages [8, 13] . Bet- ter adherence to the guidelines was independently associated with higher verbal and total IQ [13] . There was no association with BMI at age 7 years after full adjustment; however, there were very weak negative associations with waist circumference and diastolic blood pressure and a marginal negative association with systolic blood pressure ( table 2 ). No associations were found with blood lipid profile at age 7 years (data not shown). The PCA carried out at 6 months (n = 7,052) and 15 months (n = 5,610) found four underlying patterns at each age [14]. At 6 months, the foods associated with the patterns were ‘home-made/prepared meat, vegetables and desserts’ (labeled: HM traditional); ‘biscuits (cookies), sweets, crisps (potato chips), fizzy drinks (soda) and tea’ (labeled: discretionary); ‘commercial ready-made baby foods’ (labeled: RM baby foods), and ‘breast milk, raw fruit and vegetables’ (labeled: breastfeeding). Three similar patterns were obtained at 15 months;

148 Emmett Table 2. Regression models of CFUI score (per 0.1 change in CFUI score) for infants as predictor of dietary patterns, IQ, anthropometry and blood pressure at later ages in childhood [adapted from ref. 8, 13]

Independent variables CFUI score β (95% CI)1 p value

Dietary pattern z-scores at 3 years2 (n = 6,065) Processed –0.23 (–0.26 to –0.21) <0.001 Healthy 0.18 (0.15 to 0.21) <0.001 Traditional 0.013 (–0.018 to 0.045) 0.40 Snacks 0.017 (–0.013 to 0.047) 0.25 Dietary pattern z-scores at 7 years2 (n = 4,326) Processed –0.16 (–0.20 to –0.13) <0.001 Traditional 0.03 (–0.01 to 0.06) 0.15 Health conscious 0.18 (0.14 to 0.21) <0.001 IQ points3 (n = 4,429) Total 1.92 (1.38 to 2.47) <0.001 Verbal 1.92 (1.37 to 2.48) <0.001 Performance 1.33 (0.74 to 1.92) <0.001 Anthropometry BMI, kg/m2 (n = 4,801) 0.05 (–0.12 to 0.01) 0.13 Waist circumference, cm (n = 4,798) –0.15 (–0.31 to –0.002) 0.046 Blood pressure (n = 4,685) Systolic, mm Hg –0.29 (–0.60 to 0.02) 0.06 Diastolic, mm Hg –0.24 (–0.47 to –0.01) 0.043

β coefficients (95% CI) are shown. 1 Multiple regression fully adjusted models include adjustment for ethnicity, maternal education, social class, marital status, maternal smoking, age and weight status at the time of pregnancy, other children in the family, gestational age, birth weight and twin pregnancy. 2 Additionally adjusted for other dietary patterns at that age. 3 Additionally adjusted for a measure of stimulation in the home environment from items in parentally completed questionnaires.

namely HM traditional, discretionary and RM baby foods. The fourth pattern at 15 months had high correlations with ‘cheese, fish, nuts, legumes, raw fruit and vegetables’ (labeled: HM contemporary). Each child received a score on each pattern and the level of their score was associated with maternal characteristics [14] . At both ages, scores on the ‘HM traditional’ pattern were positively associated with maternal education but not independently related to maternal age. The mother being obese before pregnancy was associated with a lower score on this pattern at 6 but not at 15 months. The ‘RM baby food’ pattern was not associated with maternal education at either age; scores were lower at 15 months if there were older siblings in the family. The ‘discretionary’ pattern was strongly negatively related to maternal age and education at both ages; high scores were more likely if the mother was

Dietary Patterns during Complementary Feeding 149 Table 3. Associations between dietary pattern scores at 6 and 15 months and full-scale IQ at age 8 years (n = 7,097) [adapted from ref. 15]

Pattern1 Full-scale IQ points β (95% CI) p value

Six months Discretionary –1.15 (–1.80 to –0.50) 0.001 RM baby food –0.63 (–1.06 to –0.19) 0.005 Breastfeeding 0.97 (0.49 to 1.45) <0.001 HM traditional 0.69 (0.18 to 1.21) 0.009 Fifteen months Discretionary –0.86 (–1.52 to –0.20) 0.012 RM baby foods –1.11 (–1.71 to –0.50) 0.001 HM contemporary 0.67 (0.07 to 1.26) 0.029 HM traditional –0.30 (–0.87 to 0.27) 0.301

β coefficients (95% CI) are shown. 1 Data were analyzed by multivariate linear regression adjusted for sex, gestational age, birth weight, ethnicity, twin pregnancy, parity, maternal education, social class, marital status, family income, smoking during pregnancy and a measure of stimulation in the home environment from items in parentally completed questionnaires. Additionally adjusted for all concurrent dietary patterns and at 15 months for the dietary patterns at 6 months as well.

young or less educated. Having a mother who was obese before pregnancy or having older siblings was also associated with higher scores on this pattern. The ‘breastfeeding’ pattern at 6 months showed a strong positive gradient with maternal education [14] . High scores were more likely with older siblings present or if the mother was older, but lower scores were likely if the mothers had been overweight or obese before pregnancy. The ‘HM contemporary’ pattern at 15 months also showed a strong positive gradient with maternal education and age, but there was no relationship with the presence of older siblings or maternal obesity [14] . Whether there was evidence of a relationship between the dietary patterns in infancy and IQ at age 8 years (n = 7,097) was investigated [15]. There were biases between those with or without data on IQ, and therefore imputation of missing data by statistical methods was employed to minimize any effects. Ad- justment was made for potential confounders including child’s sex and gestational age, maternal age, education and social class, and the other dietary patterns. At both ages, increases in the ‘discretionary’ pattern scores were associated with decreases in full-scale IQ points ( table 3 ); however, adjustments had attenuated the effect sizes considerably [15]. Again, at both ages, there were independent negative associations of IQ with the ‘RM baby food’ patterns. How- ever, IQ showed independent positive associations with the ‘breastfeeding’

150 Emmett pattern at 6 months and the ‘HM contemporary’ pattern at 15 months. The associations with the ‘HM traditional’ patterns were not consistent (table 3 ) [15] . To establish if these relationships are likely to be robust, comparisons were made between data collected in ALSPAC and similar data collected in a Brazilian cohort in Pelotas [16] . In the Pelotas cohort, at 11 years of age, blood pressure was available for 1,083 and BMI for 1,085 children. IQ had been measured at 4 years of age and was available for 506 children. The comparable measure of infant feeding between the two cohorts was duration of breastfeeding. In ALSPAC there were strong associations between prevalence of breastfeeding and social background, but this did not occur in Pelotas. In both cohorts, a more favorable social background was associated with lower blood pressure and higher IQ, whereas association with BMI differed by cohort, being lower in ALSPAC and higher in Pelotas. In ALSPAC, longer breastfeeding duration was associated with lower systolic and diastolic blood pressure and lower BMI, and with higher IQ. Whereas in Pelotas there were no associations of breastfeeding duration with blood pressure or BMI, but a similar positive association with IQ was found. It is likely, therefore, that early infant diet does have a direct influence on IQ but not on blood pressure or BMI.

Discussion

These results suggest that infant diet is likely to influence cognitive development in children and may set a trend for later eating patterns. The dietary patterns either in relation to infant feeding guidelines or derived directly from the data showed that longer breastfeeding occurred in conjunction with the inclusion of other recommended foods, such as fruits and vegetables, and behaviors such as feeding on demand. It may be the breastfeeding itself that influences better cognitive development as suggested by the Pelotas comparison [16] ; however, these other favorable eating patterns which occur alongside the breastfeeding are reflected in the later dietary patterns. It is likely that, in the long term, dietary patterns characterized by the consumption of plenty of fruits and vegetables and very few processed foods will be associated with better health outcomes at later ages [17] . The association of better adherence to the infant feeding guidelines in ALSPAC with higher childhood IQ is in agreement with the findings in another UK cohort, the Southampton Women’s Survey (SWS) [3]. As in ALSPAC, the association was greatly attenuated by adjustment for social background but re- tained significance. The SWS used PCA to derive dietary patterns from a food frequency questionnaire at 6 and 12 months, and in contrast to ALSPAC derived

Dietary Patterns during Complementary Feeding 151 three patterns at 6 months but only two at 12 months [2] . The SWS ‘infant guidelines’ patterns were characterized by fruit, vegetables and home-prepared foods at both ages and were comparable to ALSPAC ‘HM traditional’ and ‘breastfeeding/HM contemporary’ patterns combined. The SWS ‘family food’ patterns were characterized by white bread, biscuits, chocolate, savory snacks and chips, and were more like ALSPAC ‘discretionary’ patterns. At 6 months only, there was a ‘baby jar food’ pattern in SWS which was very similar to the ALSPAC ‘RM baby food’ pattern. The SWS infants were born up to 10 years later than ALSPAC infants, and this, as well as the different questionnaires used, may account for the slight differences in patterns found. However, the associations with social background in SWS were in line with ALSPAC findings. In SWS, the infant dietary patterns were shown to be associated with maternal dietary patterns prior to pregnancy [2]. In ALSPAC, maternal fruit and vegetable intake has been shown to be associated with childhood fruit and vegetable intake [18], but associations between maternal and infant dietary patterns have not been investigated. The findings that higher scores on the Complementary Feeding Utility Index (CFUI) were associated with higher scores on the ‘health-conscious’ dietary patterns at 3 and 7 years and lower scores on the ‘processed’ dietary patterns are important because these patterns track over time [19] and are therefore likely to be associated with adult diet in this cohort. A very important characteristic of the CFUI and these childhood patterns is in relation to fruit and vegetable consumption; high intakes of fruit and vegetables are associated with high scores on the CFUI and the ‘health-conscious’ patterns and low scores on the ‘processed’ patterns. In this respect, the distinction between feeding home-prepared foods and ready-made baby foods was investigated further in ALSPAC [20] . Children who were fed home-prepared fruits and vegetables at 6 months ate all types of fruits and vegetables more frequently at age 7 years than children who were fed ready-prepared baby food versions only. This finding is in line with those infants scoring in the top quintiles of the CFUI consuming more fruit and vegetables and less baby food than those scoring in the bottom quintile (table 1 ). Diet high in fruits and vegetables may be protective against cancer [21], obesity [19] and heart disease [22] . The lack of an association of the CFUI score with childhood BMI is in line with the SWS finding at 4 years of age [4]. However, SWS also measured fat and lean mass at this age, and although there was no association with fat mass there was a positive relationship between scores on the infant guidelines patterns and lean mass [4] . This association was independent of breastfeeding duration and social background. This is an intriguing finding which has not yet been explored in ALSPAC.

152 Emmett Observational cohort studies cannot show causation, but comparisons with findings from cohort studies with different confounding structures can go some way towards providing confidence that associations are robust and important. There is inevitably loss to follow-up in cohort studies; however, com- putational methods have been used in many of these analyses to minimize the effects of consequent biases. In the regression analyses on adjustment for social background, the effect size in the relationships between dietary pattern scores and outcomes was greatly reduced suggesting that there was substantial confounding and that some factors that have not been controlled for adequately still remain. For example, the associations were not adjusted for diet at the time of the outcome assessment. The findings from both SWS and ALSPAC are from geographically proscribed areas of one country, so may not be generalizable. Comparing infant diets with infant feeding guidelines was informative in relation to food intake and feeding behaviors. Higher scores on this index were associated with higher childhood IQ. Consistent dietary patterns were found at two time points in infancy, and these were associated with childhood IQ. Com- parison of relationships between breastfeeding duration and childhood IQ in two cohorts with different confounding structures confirmed that breastfeeding is an important influence on cognitive development. The infant dietary patterns showed that differences in many other foods and behaviors are associated with breastfeeding. Infant dietary patterns were associated with childhood dietary patterns and so may be important in setting children on the path to healthy eating.

Acknowledgments

We are extremely grateful to all the families who took part in this study, the midwives for their help in recruiting them and the whole ALSPAC team, which includes interview- ers, computer and laboratory technicians, clerical workers, research scientists, volun- teers, managers, receptionists and nurses. The UK Medical Research Council and the Wellcome Trust (grant ref. 102215/2/13/2) and the University of Bristol provide core support for ALSPAC. This article was commissioned by Nestec Ltd but was carried out independently. This publication is the work of the author who will serve as guarantor for the contents of this paper.

Disclosure Statement

P.M.E. has from time to time received research and consultancy funding from Wyeth Nutrition, Pfizer Nutrition Ltd., Plum Baby and Danone Baby Nutrition (Nutricia Ltd.).

Dietary Patterns during Complementary Feeding 153 References

1 WHO: Global Strategy for Infant and Young 13 Golley RK, Smithers LG, Mittinty MN, et al: Child Feeding. Geneva, WHO, 2001. Diet quality of UK infants is associated with 2 Robinson S, Marriott L, Poole J, et al: Dietary dietary, adiposity, cardiovascular, and cogni- patterns in infancy: the importance of mater- tive outcomes measured at 7–8 years of age. J

nal and family influences on feeding practice. Nutr 2013; 143: 1611–1617.

Br J Nutr 2007; 98: 1029–1037. 14 Smithers LG, Brazionis L, Golley RK, et al: 3 Gale CR, Martyn CN, Marriott LD, et al: Di- Associations between dietary patterns at 6 etary patterns in infancy and cognitive and and 15 months of age and sociodemographic

neuropsychological function in childhood. J factors. Eur J Clin Nutr 2012; 66: 658–666.

Child Psychol Psychiatry 2009; 50: 816–823. 15 Smithers LG, Golley RK, Mittinty MN, et al: 4 Robinson SM, Marriott LD, Crozier SR, et al: Dietary patterns at 6, 15 and 24 months of Variations in infant feeding practice are asso- age are associated with IQ at 8 years of age.

ciated with body composition in childhood: a Eur J Epidemiol 2012; 27: 525–535. prospective cohort study. J Clin Endocrinol 16 Brion MJ, Lawlor DA, Matijasevich A, et al:

Metab 2009; 94: 2799–2805. What are the causal effects of breastfeeding 5 Boyd A, Golding J, Macleod J, et al: Cohort on IQ, obesity and blood pressure? Evidence profile: the ‘Children of the 90s’ – the index from comparing high-income with middle-

offspring of the Avon Longitudinal Study of income cohorts. Int J Epidemiol 2011; 40:

Parents and Children. Int J Epidemiol 2013; 670–680.

42: 111–127. 17 Ambrosini GL, Emmett PM, Northstone K, 6 Emmett P: Dietary assessment in the Avon et al: Identification of a dietary pattern pro- Longitudinal Study of Parents and Children. spectively associated with increased adiposity

Eur J Clin Nutr 2009; 63:S38–S44. during childhood and adolescence. Int J Obes

7 Wechsler D, Golombok S, Rust J: WISC-III- (Lond) 2012; 36: 1299–1305. UK: Wechsler Intelligence Scale for Children. 18 Jones LR, Steer CD, Rogers IS, Emmett PM: Sidcup, Psychological Corporation, 1992. Influences on child fruit and vegetable intake: 8 Golley RK, Smithers LG, Mittinty MN, et al: sociodemographic, parental and child factors An index measuring adherence to comple- in a longitudinal cohort study. Public Health

mentary feeding guidelines has convergent Nutr 2010; 13: 1122–1130. validity as a measure of infant diet quality. J 19 Northstone K, Emmett PM: Are dietary pat-

Nutr 2012; 142: 901–908. terns stable throughout early and mid-child-

9 Gorsuch RL: Factor Analysis. Philadelphia, hood? A birth cohort study. Br J Nutr 2008;

Saunders, 1974. 100: 1069–1076. 10 Cattell RB: The scree test for the number of 20 Coulthard H, Harris G, Emmett P: Long-

factors. Multivariate Behav Res 1966; 1: 629– term consequences of early fruit and vegeta- 663. ble feeding practices in the United Kingdom.

11 North K, Emmett P; ALSPAC Study Team: Public Health Nutr 2010; 13: 2044–2051. Multivariate analysis of diet among three- 21 Potter JD, Steinmetz KA: Vegetables, fruit year-old children and associations with so- and cancer prevention: a review. J Am Diet

cio-demographic characteristics. Eur J Clin Assoc 1996; 96: 1027–1039.

Nutr 2000; 54: 73–80. 22 Ness AR, Powles JW: Fruit and vegetables, 12 Northstone K, Emmett P; ALSPAC Study and cardiovascular disease: a review. Int J

Team: Multivariate analysis of diet in chil- Epidemiol 1997; 26: 1–13. dren at four and seven years of age and associations with socio-demographic characteris-

tics. Eur J Clin Nutr 2005; 59: 751–760.

154 Emmett Complementary Feeding: Taste, Eating Behavior and Later Health

Fewtrell MS, Haschke F, Prescott SL (eds): Preventive Aspects of Early Nutrition. Nestlé Nutr Inst Workshop Ser, vol 85, pp 155–165, (DOI: 10.1159/000439507 ) Nestec Ltd., Vevey/S. Karger AG., Basel, © 2016

Nature and Nurture in Early Feeding Behavior

Lucy Cooke · Clare Llewellyn Health Behaviour Research Centre, Department of Epidemiology and Public Health, University College London, London , UK

Abstract Obesity has reached epidemic proportions and research into its prevention is increasingly focusing on the earliest stages of life. Avidity of appetite has been linked to a higher risk of obesity, but studies in infancy were scarce. The Gemini twin cohort was established to investigate genetic and environmental determinants of weight trajectories in early childhood with a focus on appetite and the home environment. Gemini families have been supplying questionnaire data at regular intervals, starting when the twins were 8 months old. Analyses of data on infant appetite and weight have provided a number of important findings. Firstly, a prospective study found that appetite in infancy drives weight gain more strongly than weight drives appetite, although the two processes do coexist. A further study using a subsample of twins discordant for appetite ruled out the possibility of familial confounding, suggesting a causal role for appetite in weight. Heritability estimates for appetitive traits were moderate to high (53–84%). Finally, multivariate analyses indicated that roughly one third of the genes related to weight are also related to appetite and vice versa. Environmental factors affecting appetite in infancy are understudied, but some potential strategies for minimizing over- or underconsumption by at-risk individuals are suggested. © 2016 Nestec Ltd., Vevey/S. Karger AG, Basel

Introduction

Obesity is one of the world’s great health challenges and is increasingly seen in childhood. Obesity is complex because there is strong evidence for both environmental and genetic influences. Environmental changes are widely agreed to account for the rapid increases in weight over the last 40 years. But within- population variation in weight is as high – maybe higher – as ever, and twin and family studies show that weight is highly heritable (50–90%) [1]. Since the completion of the human genome project in 2003, studies have started to identify common genetic variants associated with weight variation. The fat mass- and obesity-associated gene FTO was the first of these to be discovered and has the largest effect size, but there are now 32 variants robustly associated with adult and child body mass index [2]. These can be combined to create a polygenic obesity risk score that shows a quantitative association with BMI. One model put forward to resolve the apparent paradox of high genetic and environmental influences is the behavioral susceptibility theory (BST) [3] . BST proposes that genetic risk operates through appetitive traits (responsiveness to external food cues and internal satiety cues) that confer susceptibility to the environment. As the environment becomes more ‘permissive’ so genetic expression increases, implying gene-environment interplay in the development of obesity. At the heart of this theory is the hypothesis that genes influence adiposity through appetitive mechanisms, and that this process begins very early in life.

Evidence for an Appetitive Model of Obesity In support of an appetitive model of obesity, avidity of appetite has been associated with obesity risk in adults [4] and children. Obese children appear to be less satiety responsive (fill up less easily) [5] , to eat faster [6] , to value food more [7] and to be more food responsive (wanting to eat when seeing or smelling palatable food) [8] than their leaner peers. There also appears to be a graded association between appetite and weight in population samples including normal weight and overweight/obese individuals. A number of studies have found that higher food responsiveness (FR) is positively associated with weight – and satiety responsiveness (SR) negatively – in a linear fashion [3, 9], although as these studies were cross-sectional it was not possible to reject the possibility that the association worked in the opposite direction – higher weight resulting in increased appetite. Prospective studies addressed this question, finding that large differences in appetite in infancy (indexed by sucking rate) were linked to subsequent growth. A faster sucking rate predicted adiposity at ages 1, 2 and 3 years [10, 11] although the effect was no longer observed at 6 years of age. Likewise, Wright et al. [12] reported that appetite, rated at 6 weeks on a 5-point scale from ‘very poor’ to ‘very good’, predicted weight gain to 12 months, but not at 7–8 years. Finally, infants who frequently emptied their bottle without parental encouragement were 69% more likely than those who rarely did so to have excess weight (weight for age z score >1) at 12 months [13]. The suggestion is that appetite drives weight gain, but any reverse effect had yet to be explored.

156 Cooke · Llewellyn 1.0

0.9 0.81 0.77 0.8

0.7 0.54 0.6 0.53 0.47 0.5

0.4 0.34

0.3

0.2

0.1

Within-pair correlations (and 95% confidence intervals) correlations Within-pair 0 between twins for age- and sex-adjusted weight measures between twins for age- and sex-adjusted MZ DZ MZ DZ MZ DZ Birth weight SDS 6-m Weight change SDS weight SDS (0–6 m)

Fig. 1. Twin correlations for adiposity: birth to 6 months. m = Month; SDS = standard deviation scores.

Evidence for a Genetic Basis to Appetite The heritability of weight is well established in adults [1] children [14] and infants [15] . In the latter, heritability of birth weight was quite low at a moderate 38%, but by 6 months it had risen to 62%; moreover, 57% of variation in weight change (the speed at which infants grew between birth and 6 months) was at- tributable to genetic factors, suggesting that the processes through which genes influence growth begin soon after birth. Twin correlations for adiposity in monozygotic (MZ) and dizygotic (DZ) twins are shown in figure 1 . Recent studies in adults [16] and children [3, 6, 17] have indicated that appetitive traits are also highly heritable. However, studies of heritability of appetite in infancy were lacking, and, given the growing problem of childhood obesity, investigations of the genetic and environmental determinants of appetitive traits and their possible role in mediating genetic influences on weight were overdue.

The Gemini Study

The observation that a large proportion of the genetic effect on body weight has already been expressed by the time children are 4 years old [18] led Prof. Jane Wardle to establish Gemini – the Health and Development in Twins study.

Nature and Nurture in Early Feeding Behavior 157 Gemini is a longitudinal cohort study of UK families with young twins conducted by the Department of Epidemiology and Public Health at University College London. It is the largest twin study set up specifically to advance understanding of genetic and environmental influences on growth trajectories in early childhood, and has a focus on behavioral mechanisms in weight gain. Data have been collected on weight every 3 months from birth, and at multiple subsequent time points on the twins’ appetite and eating behaviors, parental feeding practices and on aspects of the home environment in order to identify modifiable determinants of early excess weight gain. The Gemini sample was recruited in January 2008. Using birth registration data, the Office of National Statistics approached the families of all twins born in England and Wales between March and December 2007 (n = 6,754). A little over half (n = 3,435) consented to be contacted by the Health Behaviour Re- search Centre team, of whom nearly 70% (n = 2,402) completed baseline questionnaires, i.e. 36% of the original target population. Of these, 816 were opposite-sex twin pairs and 1,586 were same-sex pairs – 749 identical (MZ) and 800 nonidentical (DZ). Thirty-seven twin pairs were of unknown zygosity.

Why Twins? Studying twins provides the opportunity to investigate the relative contribution to individual differences in any given trait (such as appetite) made by genes and the environment. MZ pairs share 100% of their genes, where DZ pairs share 50% on average, but both types of twins share their environments to a very similar extent – the ‘equal environment assumption’. If MZ pairs are more similar for a given trait than DZ pairs, this suggests a heritable component to the trait in question, and the greater the difference between MZ and DZ similarity, the higher the heritability. In terms of representativeness, the Gemini cohort is broadly comparable to the general population in terms of distribution across the country, with national averages for twins in sex, zygosity, gestational age and birth weight. However, Gemini families are older, healthier (indexed by smoking rates and fruit and vegetable consumption) and more likely to be married and to be White-British than the English and Welsh general population as a whole. For a full description of the rationale, aims, method and characteristics of the participants in the Gemini study see van Jaarsveld et al. [19] .

Measures The findings discussed in this review are from data on weight and appetite collected at the first two time points: (a) at baseline when twins were on average 8.2 months old and (b) at the first follow-up when the twins were on average 15.8 months old.

158 Cooke · Llewellyn Infant Weight From birth onwards, weights were based on health professionals’ measurements recorded in each child’s Personal Health Record (the Red Book). Parents either supplied photocopied pages from this document or copied the information into the questionnaire. Additional weight measurements were requested every 3 months.

Infant Appetite Appetitive traits were measured using the Baby Eating Behavior Questionnaire (BEBQ [20]). The BEBQ was based on the Child Eating Behavior Question- naire [21] and modified to assess appetite in the first 3 months of life when infants are solely milk fed. It was completed retrospectively when twins were 8 months of age on average, and again at 15 months (in a slightly reworded form). The BEBQ comprises 17 items measuring 4 appetitive traits, the first 2 of which are ‘food approach’ behaviors [enjoyment of food (EF) and FR] and the latter 2 ‘food avoidance’ behaviors [slowness in eating (SE) and SR]: • EF (4 items), e.g. ‘My baby enjoyed feeding time’ (baseline) and ‘My child enjoys eating’ (follow-up) • FR (6 items), e.g. ‘Given the chance, my baby would always be feeding’ (baseline) and ‘Given the chance my child would eat most of the time’ (follow-up) • SE (4 items), e.g. ‘My baby fed slowly’ (baseline) and ‘My child eats slowly’ (follow-up) • SR (3 items at baseline and 5 at follow-up), e.g. ‘My baby got full up easily’ and ‘My child gets full up easily’ All items were scored on a 5-point scale as never, rarely, sometimes, often and always with mean scores for each subscale ranging from 1 to 5. Higher scores indicated higher EF and FR and lower SE and SR.

Research Questions

Studies were conducted using the Gemini cohort to investigate the following questions: • What is the direction of the association between appetite and weight gain? • Does the association remain robust when controlling for confounding factors related to the shared family environment? • Is appetite heritable in infancy? • If so, do the same genetic influences underlie appetitive traits and weight?

Nature and Nurture in Early Feeding Behavior 159 The Direction of the Association between Appetite and Weight Gain As discussed, existing data on the association between appetite and weight have been largely cross-sectional and the few existing prospective studies had not specifically investigated the possibility that appetite is a consequence of weight gain rather than a cause. Using measures of appetite at 3 and 15 months and of weight at 3, 9 and 15 months, the strength of the associations in both directions were tested using path analysis [22] . Results were that all 4 appetitive traits – greater EF, speed of eating and FR, and lower SR – were prospectively related to higher weight and faster weight gain. All were relatively stable over time such that a baby with a large appetite at 3 months tended to have a large appetite at 15 months. Although weight at 9 months was also prospectively associated with appetite at 15 months, the association was far weaker, providing support for the BST – that there is a causal effect of appetite on weight.

Does the Association Remain Robust When Controlling for Potential Confounding Factors Related to the Shared Family Environment? However, even this longitudinal study could not entirely rule out the possibility that some aspect of the shared home environment of children with larger or smaller appetites might account for differences in weight gain. In order to rule out familial confounding, data from a subsample – pairs of same-sex DZ twins whose appetites were dissimilar at 3 months – were analyzed [23]. This meant that all other potentially influential variables that twins share in common (e.g. maternal food intake in pregnancy or smoking, parental weight, socioeconomic status or older siblings) were held constant and the causal effect of appetite on weight gain could be more robustly tested. Twins were deemed to be ‘discordant’ for appetite if their scores for the FR and/or SR subscales of the BEBQ at age 3 months differed by 1 standard deviation or more. This resulted in data from 228 pairs being entered into the analyses. Despite there being no significant differences in birth weight between FR- or SR-discordant pairs, by 3 months those with a greater appetite were heavier than their sibling and, by 15 months, were on average nearly 1 kg heavier, which at that age equates to approximately 10% of body weight.

The Heritability of Appetite in Infancy Using data from the baseline Gemini questionnaire when infants were on average 8 months of age, Llewellyn et al. [24] conducted an investigation to determine the extent to which the four appetitive traits measured by the BEBQ were heritable in the very earliest stages of life – the first 3 months of exclusive milk feeding.

160 Cooke · Llewellyn Table 1. Parameter estimates and 95% CIs for ACE model fitting for the BEBQ subscales

BEBQ scale Additive genetic Shared Nonshared effect (A) environment effect (C) environment effect (E)

EF 0.53 (0.43–0.63) 0.45 (0.35–0.54) 0.03 (0.02–0.04) FR 0.59 (0.52–0.65) 0.30 (0.24–0.36) 0.11 (0.10–0.13) SE 0.84 (0.79–0.86) 0.00 (0.00–0.05) 0.16 (0.14–0.17) SR 0.72 (0.65–0.80) 0.12 (0.05–0.19) 0.16 (0.14–0.17)

Intraclass correlations for MZ and DZ twins for the scale scores were calculated that showed that MZ twins were significantly more similar than DZ twins, although the size of differences varied between traits. Subsequent model-fitting analyses revealed moderate to high heritability estimates for the four traits and provided an indication of the contribution of environmental factors, both shared and nonshared by the twins (table 1 ). The genetic effect was large for SE (84%) and SR (72%) and moderate for FR (59%) and EF (53%). These results confirm the important role that genes play in appetite variation from the very beginnings of postnatal life. That the heritability estimate for feeding rate (SE) was so high is consistent with previous findings for eating speed in older children [6], and taken together with evidence of an association between feeding speed and later adiposity [10] suggests that this trait may mediate genetically determined growth rate in infancy. The 72% heritability estimate for SR is also broadly consistent with the previous estimate of 63% in 11-year-old children [3] and stresses the importance of genes in the regulation of sensitivity to hunger and fullness. The heritability of the other two appetitive traits, FR and EF, was only moderate (59 and 53%, respectively) with a greater role for environmental factors than for SE and SR. In the case of SE, there was no shared environmental effect at all, which is also consistent with the findings of Carnell and Wardle [3] in 11-year-old children.

Do the Same Genetic Influences Underlie Appetitive Traits and Weight? The high heritability of appetitive traits in infancy raises the question of whether the genes responsible are also implicated in weight gain. The first common variant to be implicated in adiposity in adults and children was the FTO gene, with adults carrying the high-risk version being on average ∼ 3 kg heavier than those with the low-risk version. A large study of children also found that carrying the high-risk allele was associated with lower satiety sensitivity and that this mediated the relationship between FTO and weight [25] . Many more obesity- related genes have been identified such that it is possible to calculate an

Nature and Nurture in Early Feeding Behavior 161 individual’s risk score. A score comprising 28 known obesity-related variants was calculated for the same sample of children; as genetic risk of obesity increased, SR decreased, and SR mediated some of the association between the genetic risk score and weight [25–27]. These studies support the idea that ‘obesity genes’ influence weight via their effects on appetite. However, they included only a very small number of known variants. Twin studies allow researchers to explore the extent to which all the genes that influence appetite are broadly the same as those that influence weight using inferential modeling. This hypothesis was tested in the Gemini cohort, with the finding that approximately one third of the genes that influence certain appetitive traits (SE and SR) also influence weight [28] . Associations between weight and the other two traits measured (FR and EF) were too small to permit the modeling of common pathways. This may be because very young infants have so little scope to influence their intake and it may be that at later ages the magnitude of the observed association would be larger. Again, the results support the BST suggesting that in the plentiful food environment of the 21st century, those with a genetic predisposition to be less sensitive to their internal signals of satiety may be more likely to overeat and ultimately to gain excess weight.

Environmental Influences on Early Eating Behavior

A large body of research had examined the impact of environmental factors on the eating behavior of children of 2 years of age upwards, but little is currently known about their impact on appetite in infancy. The influential factors for infants are likely to include the intrauterine environment in terms of under- or overnourishment, impacting on gestation and birth weight, but parental feeding practices – what, when and how children are fed – will also play a critical role in the development of eating behaviors – both positively and negatively. For example, the symbiotic feeding relationship afforded by breastfeeding may help to develop infants’ self-regulatory abilities, where the common practice of encouraging a formula-fed infant to finish the bottle may override these. Feeding to soothe rather than only when an infant is hungry may promote excess intake in those predisposed to FR [29] . Once solid foods are introduced, rapid exposure to a wide variety of foods is associated with less rejection of novel food and may be especially beneficial for food-avoidant children [30]. Parental control over the quantity and type of food eaten has been shown to influence the eating behaviors of older children, and may also be important in infancy. In the Gemini cohort, mothers of infants with a lower birth weight and/or smaller appetite were more likely to pressure their

162 Cooke · Llewellyn child to eat more and less likely to restrict their intake than mothers of infants with a larger appetite. In another study, restriction was driven by parental perception of their child’s risk of overweight and their belief that infants cannot recognize their own feelings of hunger and satiety [31] . Pressure to eat was associated with concern about their infant’s low appetite and risk of underweight. Given that parents typically underestimate their children’s weight, there is a danger that normal weight or even overweight children may be pressurized into overeating resulting in unhealthy weight gain and diminishing self-regulation. Given that the first years of life are increasingly seen as a critical period for the development of lifelong eating habits, a greater understanding of the environmental influences on infant appetite is an important aim for future research.

Conclusions and Implications

It is clear that like weight, appetitive traits are highly heritable. These traits are ob- servable from the earliest stages of life and predict weight gain prospectively. The finding that common genes affect weight partly via appetite explains why some individuals are more susceptible to environmental triggers to overconsume. There is a tendency to assume that if behavioral traits are highly heritable, they are not easily amenable to change. Despite the strong influence of genetic factors on appetite, environmental interventions might reduce the expression of these traits in infants at the high-risk extremes. For example, advising mothers to offer smaller, more frequent feeds might be helpful for infants who become full very quickly and using a slower-flowing teat could reduce eating speed in infants who appear less able to regulate their intake in response to satiety cues. Toddlers who have trouble recognizing when they are full would benefit from smaller portions and a higher proportion of low-energy foods in their diet. For those who are particularly food responsive, keeping unhealthy foods out of the home environment would pay div- idends. In later childhood, it might be possible to intervene to teach at-risk children better self-regulation by training them to recognize and act upon their internal feelings of hunger and fullness. This is a relatively unexplored area at present, but is a crucial topic for future research with important implications for obesity prevention.

Disclosure Statement

The authors declare that no financial or other conflict of interest exists in relation to the contents of the chapter.

Nature and Nurture in Early Feeding Behavior 163 References

1 Elks CE, Den HM, Zhao JH, et al: Variability 13 Li R, Fein SB, Grummer-Strawn LM: Asso- in the heritability of body mass index: a sys- ciation of breastfeeding intensity and bottle- tematic review and meta-regression. Front emptying behaviors at early infancy with in-

Endocrinol (Lausanne) 2012; 3: 29. fants’ risk for excess weight at late infancy.

2 Speliotes EK, Willer CJ, Berndt SI, et al: As- Pediatrics 2008; 122(suppl 2):577–584. sociation analyses of 249,796 individuals re- 14 Silventoinen K, Rokholm B, Kaprio J, Sø- veal 18 new loci associated with body mass rensen TI: The genetic and environmental

index. Nat Genet 2010; 42: 937–948. influences on childhood obesity: a systematic 3 Carnell S, Wardle J: Appetite and adiposity in review of twin and adoption studies. Int J

children: evidence for a behavioral suscepti- Obes (Lond) 2010; 34: 29–40.

bility theory of obesity. Am J Clin Nutr 2008; 15 Johnson L, Llewellyn CH, van Jaarsveld

88: 22–29. CHM, et al: Genetic and environmental in- 4 French SA, Epstein LH, Jeffery RW, et al: Eat- fluences on infant growth: prospective analy- ing behavior dimensions: associations with sis of the Gemini twin birth cohort. PLoS

energy intake and body weight: a review. Ap- One 2011; 6:e19918.

petite 2012; 59: 541–549. 16 Neale B, Mazzeo SE, Bulik CM: A twin study 5 Barkeling B, Ekman S, Rossner S: Eating be- of dietary restraint, disinhibition and hunger: haviour in obese and normal weight 11-year- an examination of the Eating Inventory old children. Int J Obes Relat Metab Disord (Three Factor Eating Questionnaire). Twin

1992; 16: 355–360. Res 2003; 6: 471–178. 6 Llewellyn CH, van Jaarsveld CHM, Boniface 17 Carnell S, Haworth CMA, Plomin R, Wardle D, et al: Eating rate is a heritable phenotype J: Genetic influence on appetite in children.

related to weight in children. Am J Clin Nutr Int J Obes (Lond) 2008; 32: 1468–1473.

2008; 88: 1560–1566. 18 Haworth CM, Carnell S, Meaburn EL, et al: 7 Temple JL, Legierski CM, Giacomelli AM, et Increasing heritability of BMI and stronger al: Overweight children find food more rein- associations with the FTO gene over child-

forcing and consume more energy than do hood. Obesity (Silver Spring) 2008; 16: 2663– nonoverweight children. Am J Clin Nutr 2668.

2008; 87: 1121–1127. 19 van Jaarsveld CHM, Johnson L, Llewellyn 8 Fisher JO, Birch LL: Eating in the absence of CH, Wardle J: Gemini: a UK twin cohort hunger and overweight in girls from 5 to with a focus on early childhood weight trajec-

7 years of age. Am J Clin Nutr 2002; 76: 226– tories, appetite and the family environment.

231. Twin Res Hum Genet 2010; 13: 72–78. 9 Webber L, Hill C, Saxton J, et al: Eating be- 20 Llewellyn CH, van Jaarsveld CHM, Johnson haviour and weight in children. Int J Obes L, et al: Development and factor structure of

(Lond) 2009; 33: 21–28. the Baby Eating Behaviour Questionnaire in

10 Agras WS, Kraemer HC, Berkowitz RL, the Gemini birth cohort. Appetite 2011; 57: Hammer LD: Influence of early feeding style 388–396.

on adiposity at 6 years of age. J Pediatr 1990; 21 Wardle J, Guthrie CA, Sanderson S, Rapoport

116: 805–809. L: Development of the Child Eating Behav- 11 Agras WS, Kraemer HC, Berkowitz RL, et al: iour Questionnaire. J Child Psychol Psychia-

Does a vigorous feeding style influence early try 2001; 42: 963–970.

development of adiposity. J Pediatr 1987; 110: 22 van Jaarsveld CHM, Llewellyn CH, Johnson 799–804. L, Wardle J: Prospective associations between 12 Wright CM, Parkinson KN, Drewett RF: appetitive traits and weight gain in infancy.

How does maternal and child feeding behav- Am J Clin Nutr 2011; 94: 1562–1567. ior relate to weight gain and failure to thrive? 23 van Jaarsveld CHM, Boniface D, Llewellyn Data from a prospective birth cohort. Pediat- CH, Wardle J: Appetite and growth: a longi-

rics 2006; 117: 1262–1269. tudinal sibling analysis. JAMA Pediatr 2014;

168: 345–350.

164 Cooke · Llewellyn 24 Llewellyn CH, van Jaarsveld CHM, Johnson 28 Llewellyn CH, van Jaarsveld CHM, Plomin R, L, et al: Nature and nurture in infant appetite: et al: Inherited behavioral susceptibility to analysis of the Gemini twin birth cohort. Am adiposity in infancy: a multivariate genetic

J Clin Nutr 2010; 91: 1172–1179. analysis of appetite and weight in the Gemini

25 Wardle J, Carnell S, Haworth CM, et al: Obe- birth cohort. Am J Clin Nutr 2012; 95: 633– sity associated genetic variation in FTO is 639. associated with diminished satiety. J Clin En- 29 Stifter CA, Anzman-Frasca S, Birch LL: Par-

docrinol Metab 2008; 93: 3640–3643. ent use of food to soothe infant/toddler dis- 26 Llewellyn CH, Trzaskowski M, van Jaarsveld tress and child weight status: an exploratory

CHM, et al: Satiety mechanisms in genetic study. Appetite 2011; 57: 693–699.

risk of obesity. JAMA Pediatr 2014; 168:338– 30 Moding KJ, Birch LL, Stifter CA: Infant tem- 344. perament and feeding history predict infants’

27 Wardle J, Llewellyn CH, Sanderson S, Plomin responses to novel foods. Appetite 2014; 83: R: The FTO gene and measured food intake 218–225.

in children. Int J Obes (Lond) 2009; 33: 42–45. 31 Gross RS, Mendelsohn AL, Fierman AH, Messito MJ: Maternal controlling feeding

styles during infancy. Clin Pediatr 2011; 50: 1125–1133.

Nature and Nurture in Early Feeding Behavior 165

Complementary Feeding: Taste, Eating Behavior and Later Health

Fewtrell MS, Haschke F, Prescott SL (eds): Preventive Aspects of Early Nutrition. Nestlé Nutr Inst Workshop Ser, vol 85, pp 167–168, (DOI: 10.1159/000441156) Nestec Ltd., Vevey/S. Karger AG., Basel, © 2016

Summary on Complementary Feeding: Taste, Eating Behavior and Later Health

The aim of this section of the workshop was to address whether complementary feeding (CF) practices in a holistic sense may have the potential to improve later health and developmental outcomes, and prevent adverse outcomes such as obesity and other noncommunicable diseases. The first presentation considered the evidence that ‘nutritional’ aspects of CF – mainly timing and content – influence later health outcomes and concluded that, although data are limited in quantity and quality, delaying the introduction of CF until 4 months may protect against later obesity and possibly allergy. The difficulties in designing and conducting research studies investigating effects of CF were highlighted, including maternal willingness to be randomized to use practices which were not originally planned and attrition with longer- term follow-up. A number of common themes were identified throughout the workshop. Firstly, the need for a holistic approach to CF was emphasized given the complex interplay between nutrition, feeding behavior and psychological factors; a ‘one- size-fits-all’ approach to CF is not feasible or sensible given variation between infants, their environments and cultural factors. Secondly, it was highlighted that our traditional weaning practices (such as feeding to soothe, frequent large meals and forced eating) and our taste and flavor preferences, all evolved to be suitable in conditions where food is scarce; they are not appropriate for modern obesogenic environments and this may result in later problems such as distinguishing hunger from other distress cues, ignoring satiety signals, disliking co- erced foods and preferring unsuitable sweet, energy-dense foods. Despite these issues, and a clear genetic component to aspects of taste, flavor preferences and appetite, parents and caregivers can both buffer and amplify environmental influences, and a number of research studies were presented that demonstrate how interventions can successfully modify outcomes. For example, flavor preferences can be modified by exposures during pregnancy and early infancy from breast milk but also infant formulas, and these preferences can persist. Parenting behavior can also favorably influence infant feeding practices and infant growth. Thus, in a randomized controlled trial designed to promote responsive parenting, infants of mothers who received advice on sleep routines, soothing, recognizing hunger and fullness cues, and delaying solids beyond 4 months had slower weight gain in the first 12 months. Twin studies suggest that some aspects of infant appetite are highly heritable, including eating speed and satiety, whilst appetite is probably causally related to weight gain. This raises the intriguing possibility that it may be possible to identify individuals who are at risk of overeating in an obesogenic environment and intervene to prevent adverse outcomes. Mary S. Fewtrell

168 Fewtrell Subject Index

ACHOIS trial 65 parental preferences 25 Allergy Behavioral susceptibility theory breast milk and prevention 9, 11, 31, (BST) 156 32 Breastfeeding complementary feeding timing allergy prevention in infants 9, 11, 31, effects 117, 118 32 early development 2 early flavor experiences in food early dietary factors 21–23 acceptance 139, 140 early origins 5, 6 metabolome studies 90, 91 epidemiology 21, 29, 35, 36 maternal diet impact 8 CD, see Celiac disease natural resolution of food allergy 36 Celiac disease (CD), complementary obesity association 6, 7 feeding timing effects 118 overabundance evolutionary CF, see Complementary feeding adaptation 4, 5 Childhood Obesity Project probiotics and prevention 9, 50, 51 (CHOP) 82–84, 86, 87, 91, 94, 97, 98 sensitization route 30 CHOP, see Childhood Obesity Project tolerance induction, see Oral tolerance Cognition, complementary feeding tolerance window in infant feeding for studies of outcomes 119, 150, 151 allergy reduction 10–12 Complementary feeding (CF) type 2 response 2, 3 Complementary Feeding Utility weaning food impact 25 Index 152 ALSPAC study, see Complementary definition 114 feeding early flavor experiences in food Amino acid metabolism, see Infant acceptance 141, 142 formula later health effect studies Appetite, see also Obesity ALSPAC study appetitive model of obesity 154 cognition 150, 151 genetic basis 157 data-driven dietary patterns 148–150 Baby food infant feeding guidelines allergy prevention 23 scores 147 market 22, 23 subjects 146, 147

169 composition effects Flavor perception macronutrients 119, 120 early flavor experiences in food micronutrients 120 acceptance salt 120 amniotic fluid and breast milk 139, feeding method effects 120, 121 140 mechanisms 115, 116 complementary feeding 141, 142 methodology 115 infant formula 140, 141 timing of introduction impact flavor-flavor learning 131 allergy 117, 118 neurocircuitry 136 celiac disease 118 olfaction 136, 137 cognition 119 prenatal development of taste and obesity 116, 117 odor sensitivity 136–138 recommendations 114, 115 taste sensitivity and acceptance biology 138, 139 Developmental origins of health and Food acceptance, see Flavor perception disease (DOHaD) 72 Food allergy, see Allergy DOHaD, see Developmental origins of Formula, see Infant formula health and disease GDM, see Gestational diabetes mellitus EAT study 32 Gemini study Eating behavior, see Feeding practices environmental influences on early Epicutaneous immunotherapy (EPIT), eating behavior 162, 163 tolerance induction in food allergy 49, infant weight and appetite 159 50 overview 157, 158 Epigenetics research questions childhood obesity 76–78 appetite and weight gain early development in relation to association 160 childhood adiposity and body appetite heritability 160, 161 composition 72, 73 family environment effects 160 metabolic programming genetics of appetitive traits and biomarkers 77 weight 161, 162 overview 76 overview 159 EPIT, see Epicutaneous immunotherapy twin cohort 158 EPOCH study 85 Gestational diabetes mellitus (GDM) interventions 64, 65 Feeding practices obesity association 60, 61 environmental influences on early eating behavior 162, 163 Healthy Moms study 63, 64 learning from parents 125–127, Hygiene hypothesis 7 132 obesity risks IGF-1, see Insulin-like growth factor-1 feeding to soothe 128, 129 Infant formula flavor-flavor learning 131 amino acid metabolism in infants learned likes and dislikes 129–131 aromatic amino acids 95, 96 portion size 128, 129 branched-chain amino acids 92–95 social influence to facilitate childhood obesity significance 97 tasting 131, 132 glutamine 96, 97

170 Subject Index childhood obesity studies 86 research questions early flavor experiences in food appetite and weight gain acceptance 140, 141 association 160 hypoallergenic formula in food allergy appetite heritability 160, 161 prevention 50, 51 family environment effects 160 plasma amino acid response 90, 91 genetics of appetitive traits and protein intake studies of obesity weight 161, 162 clinical trials 103–107 overview 159 growth rates with high protein 102 twin cohort 158 low-protein formula rationale 107 gut bacteria 21 needs versus actual intakes 102, infant protein intake studies, see Infant 103 formula overview 92 maternal obesity Inflammation, overabundance diabetes, see Gestational diabetes evolutionary adaptation 4, 5 mellitus Insulin, childhood obesity epidemiology 60, 61 prediction 84–86, 90 gestational weight gain 61–64 Insulin-like growth factor-1 (IGF-1), mechanisms of offspring childhood obesity role 82–86, 90 effects 75–78 prepregnancy obesity 66 Leptin overabundance evolutionary allergy levels 7 adaptation 4, 5 childhood obesity prediction 85 Olfaction, see Flavor perception LIMIT study 63, 74 Oral tolerance development 30, 31 Metabolic programming, see Epigenetics induction in food allergy baked milk and egg diet 37, 38 Obesity epicutaneous immunotherapy 49, 50 allergy association 6, 7 oral immunotherapy amino acid metabolism and childhood clinical trials 39–44 obesity risks 97 dosing schedule 38, 39 appetite genetic basis 157 immunoglobulin E monoclonal appetitive model 154 antibody combination biomarkers for childhood obesity therapy 45–47 prediction 82–87 multiple foods 45 complementary feeding timing overview 38 effects 116, 117 response 39 early dietary factors 20 safety 45 eating behavior, see Feeding practices sublingual immunotherapy 45, 48, epidemiology 19, 20 49 epigenetics of childhood mechanisms 36 obesity 76–78 Gemini study Physical activity, childhood body environmental influences on early composition effects 74, 75 eating behavior 162, 163 Probiotics, food allergy prevention 9, 50, infant weight and appetite 159 51 overview 157, 158 Protein intake, see Infant formula

Subject Index 171 SLIT, see Sublingual immunotherapy Vitamin D, gestational status and Sublingual immunotherapy (SLIT), offspring obesity 73, 74 tolerance induction in food allergy 45, 48, 49 Weaning, see also Complementary feeding TASR38, flavor perception role 139 allergy impact 25 Taste, see Flavor perception parent focus group results 23–25 Tolerance, see Oral tolerance patterns 24

172 Subject Index