EARLY NUTRITION AND LATER DIABETES RISK Mikael Knip and Hans K. Åkerblo m H ospital f or Children and Adolescents, Uni v ersit y o f Helsinki, P.O. Box 281, FI-00029 HUCH , Helsinki , Finland Abs tra c t: Early feeding may modify the risk of both type 1 (T1D) and type 2 diabetes ( T2D ) later in life. The info r mation generated so far is, however, c ontroversial. When evaluating studie s on the impact of early feeding on r isk of later diabetes , the data hav e to be assessed critically and possible c onfounding factors have to be considered. The study design may induce biases and there are considerable differences in early feeding practices a c r oss v ari ous cou ntri es an dcu lt u r es. Accordin g l y it ma y not be possible t o g eneralise observations based on o ne population. Lon g breastfeedin g , exclusive breastfeedin g in particula r , and su pp lementation with vitamin D in infancy have been reported to conf er partial protection against beta-cell f f autoimmunit y and T1D. In contra s t, earl y exposure to cow’s milk proteins and cereals and heav y wei g ht i n infanc y have been implicated as r isk factors for T1D. Long breastfeeding has also been observed to protect against T2D in aboriginal populations. Poor fetal nutrition r esulting in low birth weight has been identified as a factor contributing to l at e r in su lin r es i s tan ce an d T2D . R ece nt d ata in d i c at e that cu rr e nt o verweight and obesity are stronger d e te rminant so f in su lin r es i s tan ce t han birth weight among preschool child r en. High-nutrient diet and rapid g rowth in earl y infanc y have been reported to adversel y pro g ramme the principal components of the metabolic s y ndrome includin g insulin res i s tan ce an d T2D . It i s an i m portant scientific and public-health m m o bjective to define protective and predisposing effects of early nutrition o n the development of diabetes, since early feeding can potentially be mod ifi ed t o minimi se th e ri s k o f lat e r c hr o ni cd i se a ses. Key words : Beta-cell autoimmunity, breastfeed i ng, cereals, cow’s milk, growth, i nsulin resistance, type 1 diabetes, type 2 diabetes 14 2 E ARLY NU TRITI ON A N D LATER DIABETE S RI S K 14 3 1. INTRODUCTION Diabetes is characterised by increased blood glucose c oncentrations. These are regulated by a loop comprising two c omponents, the insulin-secreting b e t a ce ll s an d th e in su lin- se n s iti ve t issues, i.e. liver, muscle and adipose tissue, which respond to insulin. Loss of blood g lucose control is due t o beta-cell failure, resultin g in i nsulin deficiency, insulin resistance of the target tissues or a c ombination of both. Type 1 diabetes (T1D) is perceived as a chronic i mmune-mediated disease with a subclinical prodrome characterised by s elective loss of insulin-producin g beta c ells in the pancreatic islets in genetically susceptible persons (Knip 2002). The preclinical disease phase is asymptomatic and may last from a few months up to more than 10 years. Acute clinical onset, ketoacidosis and dependency of exo g enous insulin are characteristic features, and the y result fro m absolute insulin deficienc y . T y pe 2 diabetes (T2D) is t y picall y a m etabolic disorder of obese middle-aged or old people with slow c linical onset and non-insulin de p endence ( Wilkin 2001 ) . The p atients s uffer from relative insulin deficienc y because of im p aired insulin action c ombined with the inabilit y of the beta cells to compensate for the r educed insulin sensitivity. T2D i s a component of the metabolic s yndrome characterised by obe s i ty, insulin resistance, lipid abnormalities, hypertension and cardiovascular disease. Both T1D and T2D are caused b y the interaction of a series of g enetic and environmental factors. A lthou g h data available indicate that t here is a stronger genetic predisposition to T2D than to T1D, more is k nown about the genetic background of the latter with a major part of the disease susceptibility conferred by HLA genes on the short arm of c hromosome 6. Life st y le character i stics, such as obesit y and ph y sical i nactivity, are well established exogenous risk factors for T2D, while the environmental determinants of T1D have remained more speculative ( Åkerblom et al. 2002 ) . Fetal malnutrition has been im p licated as a risk factor for T2D ( Hales and Barker 1992 ) . Recent observations indicate, h owever, that current overwei g ht is a stron g er determinant of insulin r esistance than birth weight among young children (Wilkin et al. 2002). Short breastfeeding, early exposure to cow’s milk proteins and c ereals, failing supplementation with vitamin D and heavy weight in i nfanc y have been implicated as risk factors of T1D related to earl y feeding (Virtanen and Knip 2003). Short exclusive breastfeeding, high- 144 M ikael Knip and Hans K. Å kerblom nutrient diet and early growth acceler a t ion in infancy have been reported t o confer increased risk for T2D (Pettitt et al. 1997, Singhal and Lucas 2004 ) . These risk factors will be discussed below . 2. LIMITATIONS OF STUDIES ON EARLY FEEDING - LATER DISEASE RELATIONS The role of early feeding in the development of later chronic diseases has remained controversial in many respects despite an extensive research activity in this field. These controversies may at least partl y reflect differences in stud y desi g n. Man y studies have been r etrospective and such studies mi g ht be associated with substantial recall biases. Intervention studies which potentially provide the strongest evidence for a possible relationship between early feeding practices and s ubsequent health are so far very f e w in number in this area, partly due t o the lon g follow-up time needed when assessin g the consequences of i nfant feeding on later health . There are also considerable differences in early feeding practices be t wee n v ari ous cu lt u r es an dcou ntri e s. The frequency and duration of breastfeedin g varies conspicuousl y , and similarl y the use of h y drol y sed formulas appears to be ver y frequent in some countries, while it is extremely rare in other countries. About 15% of the infants in Colorado, U SA, are ex p osed to cereals as their fi r st foreign proteins, whereas this is exceptional e.g. in Finland. Accordingly observations made in one c ountr y are not necessaril yg eneralisabl e to other g eo g raphical re g ions. Some of the controversies may be explained by differences in exposure and outcome measures. A minority of the studies dealing with breastfeeding have differentiat e dbe t wee n e x c l us i ve an d t o tal breastfeedin g , althou g h the biolo g ical effects of breastfeedin g must be r elated to whether the infant is g iven breast milk onl y or supplemental food in addition to breast milk. The end of exclusive breastfeeding leads i n most cases to exposure to cow’s milk, and accordingly it may be d iffi cu lt t ode fin ew h e th e r a lat e r a dver se h e alth e ff ec t i s r e lat ed t os h o rt exclusive breastfeedin g or alternativel y to earl y exposure to cow’s milk. H ence it would be important to defin e the dietary exposure in detail both qualitatively and quantitatively when assessing later health consequences o f infant feeding . E ARLY NU TRITI ON A N D LATER DIABETE S RI S K 14 5 3. EARLY FEEDING AND RISK OF TYPE 1 DIABETES 3.1 Breastfeeding and exposure to cow’s milk A series of retrospective studies have implied that breastfeeding m ay protect from T1D, but there are also reports suggesting no protective effect and even some indicating a predisposing effect of breastfeeding (see Virtanen and Knip 2003). Gerstein conducted already i n 1994 a meta-anal y sis of all acceptable retrospective studies available by that time to assess whether early feeding practices had any influence o n the risk of T1D. That meta-anal y sis showed that breastfeedin g for a s horter period than 3 months was associat e d with a 1.4-fold risk for T1D , whereas exposure to cow’s milk before the a g e of 4 months conferred a 1.6-fold disease risk. Prospectiv es t ud i es o n th e a ssoc iati o n be t wee n early feeding and the appearance of beta-cell autoimmunity, i.e. diabetes- associated autoantibodies, have also provided partly conflicting results. Three p ros p ective studies from the U n i ted States, Germany and Australia did not observe an y effect of breastfeedin g /earl y exposure to cow’s milk o n the appearance of beta-cell autoimmunity, while a Finnish birth c ohort stud y reported that short exclusive breastfeedin g and earl y exposure to cow’s milk predisposed young children to progressive beta- c ell autoimmunit y in the form of positivit y for all four autoantibod y s pecificities analysed (see Virtanen and Knip 2003). A pilot intervention study conducted mainly in Finland eliminating exposure to cow’s milk proteins over the first 6-8 months of life i ndicated that the early dietary intervention resulted in a reduction of 40- 6 0% in the cumulative frequency of all diabetes-associated autoantibodies exce p t GAD antibodies ( see Kni p 2003 ) . Based on the experience from the pilot stud y a randomised, controlled and double- blinded trial proper was initiated in 2002 to answer the question whethe r i t is possible to reduce the frequency of diabetes-associated autoantibodies and/or clinical T1D by the age of 6 years and the c umulative incidence of T1D by the age of 10 years by weaning to a h i g hl y h y drol y sed formula over the first 6-8 months of life. This Trial to R educe IDDM in the Genetically at Risk (TRIGR) will recruit 2032 i nfants with at least one affected family member and HLA -co nf e rr ed s usceptibility to T1D by the end of April 2006, and accordingly the final 14 6 M ikael Knip and Hans K. Å kerblom answer will be available at the earliest in year 2012 illustrating the need for long follow-up periods in prospective and intervention studies exploring the association between early feeding and later health. 3 .2 Early exposure to cereals Two recent prospective studies have indicated that early exposure to c ereals ma y increase the risk of seroconversion to positivit y for diabetes- associated autoantibodies (Norris et al. 2003, Zie g ler et al. 2003). The A merican report suggested that both early (before the age of 4 months) and late exposure (at the age of 7 months or later) to cereals were a ssoc iat ed w ith an in c r e a sed ri s k of beta-cell autoimmunity, while the f G erman stud y implied that an increased risk was related to exposure to c ereals before the age of 3 months. In addition the American survey i ndicated that both gluten-containing and non-gluten-containing cereals co nf e rr ed an in c r e a sed ri s k f o r be ta- c ell autoimmunity. Neither of the s tudies reported an y data on the amount of cereals the infants were exposed to at various a g es. Earl y exposure to cereals are a g ainst the i nfant n u triti o n r eco mm e n d ati o n s in all develo p ed countries and occurs o nly rarely. Accordingly one may ask whether early exposure to cereals i s a proxy of other baby care practices predisposing to T1D. 3.3 Vitamin D supplementation The daily requirement of vitamin D can not be met in infancy without oral supplementation, particularl y not in the winter time with decreased daylight. A European r etrospective multicentre survey i ndicated that vitamin D supplementation during infancy was associated with a reduced risk of T1D ( EURODIAB 1999 ) . A Finnish birth cohort s tud y showed that vitamin D supplementation in the infant period was r elated to a decreased risk of T1D , whereas suspicion of rickets b y the age of 2 years increased the risk of later diabetes (Hyppönen et al. 2001). T hese observations suggest that vitamin D supplementation in infancy c onfers partial protection against T1D . 3.4. Growth in infancy Increased wei g ht g ain in infanc y has been associated with an e nhan ced ri s k o f T1D in c a se - co nt r ol studies (see Virtanen and Knip E ARLY NUTRITION AND LATER DIABETES RISK 147 2003). In a Finnish surve y current wei g ht in the hi g hest quartile at the a g e of 3 and 6 months resulted in a 1.5-fold risk of T1D later in c hildhood (Hyppönen et al. 1999). Whether accelerated linear growth in i nfancy contributes to an increased ri s k o f T1D ha s r e main ed co ntr ove r s ial . 4. EARLY FEEDING AND RISK OF TYPE 2 DIABETES 4.1 Breastfeeding Pettitt et al. reported in 1997 that exclusive breastfeeding for the first 2 months of life was associated with a significantly reduced rate of T2D among Pima Indians. A more recent study among Native C anadians indicated that breastfeedin g for more than 12 months reduced t he risk of subsequent T2D before the a g e of 18 y ears to about one fifth o f the average risk (Young et al. 2002). It remains open, whether the protective effect of breastfeeding on the development of T2D is also o perative among Caucasians, since so far no such observations have been reported. 4.2 Fetal and infant growth Accordin g to the fetal ori g ins h y pothesis intrauterine malnutrition predisposes the individual to various components of the metabolic s yndrome including cardiovascular disease, hypertension, obesity, dyslipidaemia and T2D (Barker 2002). A series of retrospective studies h a ve s h ow n that th e rat eo f th e above mentioned components of the m etabolic s y ndrome are inversel y related to birth wei g ht. Over the last few years evide n c e has accumulated indicating that early infant weight gain and current weight are more strongly associated with insulin resistance in young childr e n than birth weight. A survey in 5- y ear-old British children indicated that current wei g ht was the s trongest determinant of insulin resistance both in boys and girls, and n either birth weight nor postnatal weight gain were independent predictors of insulin resistance (Wilkin et al. 2002). 14 8 M ikael Knip and Hans K. Å kerblom The g rowth acceleration h y pothesis su gg ests that earl y infant g rowth, wei g ht g ain in particular, is th eco mm o n de n o minat o r o f th e m etabolic syndrome (Singhal and Lucas 2004). According to this h ypothesis the association previously observed between birth weight and t he components of the metabolic syndrome would be the consequence of t he phenomenon that infants born small for g estational a g e experience t he fastest weight gain in early infancy. Singhal et al. reported in 2003 t hat early feeding with a nutrient-enriched formula in preterm infants r esulted in higher circulating conce n t rations of 32-33 s p lit p roinsulin, a m arker of insulin resistance , in adolescence compared to feedin g with banked breast milk or a standard formula. The 32-33 split proinsulin c oncentrations in the p eri p heral ci rcu lati o n o f th e a do l esce nt swe r e r elated to the weight change over the first 2 weeks of life with the h ighest levels in those with the fastest weight gain. Another analysis of t h es am eco h o rt s h owed that th e r ew a s an in ve r se co rr e lati o n be t wee n flow-mediated endothelium-dependent d i lation in the brachial artery and e arly postnatal weight gain suggesting that high weight gain over the first 2 weeks of life p redis p oses to im p aired vascular endothelial function i n adolescence (Sin g hal et al. 2004). Insulin resistance has been shown r ecentl y to be associated with endothelial d y sfunction (Hsueh et al. 2004 ). 5. CONCLUSIONS AND FUTURE DIRECTIONS There is definitely a need to expand our present knowledge of the c onsequences of early feeding on later health and disease. Accumulated e vidence su gg ests that earl y exposure to complex dietar y proteins ma y be a risk factor for T1D. One can s peculate that early weight gain in i nfancy may induce beta-cell stress making these cells more susceptible t o cellular damage and thereby increasing the risk of T1D. Regula r v itamin D supplementation in infanc y ma y not onl y protect the child fr o m ri c k e t sbu t al so fr o m lat e r T1D . Preliminary data indicate that exclusive breastfeeding for at least 2 m onths and prolonged overall breastfeeding protects against T2D, at l east among aboriginal populations. High-nutrient diet and accelerated g rowth, wei g ht g ain in particular, in earl y infanc y ma y be a risk facto r for the development of the metabolic syndrome and T2D. E ARLY NU TRITI ON A N D LATER DIABETE S RI S K 149 These observations argue in favour of strong promotion of e xclusive breastfeeding over the first few months of life in practically all i nfants. It has been repeatedl y shown that wei g ht g ain is slower on breastfeeding than on formula feeding, and already that consequence m ay reduce the subsequent risk of both T1D and T2D. The adverse e ffects of early weight gain and the fact that the energy content of formulas is usuall y hi g her than that of breast milk raises the issue w h e th e r th e r e i s a n eed t o r educe th e e nergy content of infant formulas in t he develo p ed countries. This is a q uestion that needs further studies, preferably in the form of intervention trials implying that we have to wait f o r th ede finit e an swe r f o r so m e tim e. REFERENCES Åkerblom HK, Vaarala O, H y öt y H, Ilonen J, Kni p M. Environmental factors in the etiolo gy of t y pe 1 diabetes. Am J Med Genet 2002;115:18-29. Barker DJ. Fetal programming of coronary he art d i se a se. Tr e n ds En doc rin o l M e ta b 2002;13:364-368.Hales C, Barker DJP. Typ e 2 (non-insulin dependent) diabetes m ellitus: the thrifty genotype hypothesis. Diabetologia 1992;35: 595-601. Gerstein HM. Cow’s milk exposure and Type 1 diabetes mellitus ; a critical overview of the clinical literature. Diabetes Care 1994 ; 17:13-19. Hales C, Barker DJP. Type 2 (non-insuli n dependent) diabetes mellitus: the thrifty g enot y pe h y pothesis. Diabetolo g ia 1992;35: 595-601. Hsueh WA, L y on CJ, Quinones MJ. Insulin re s istance and the endothelium. Am J Med 2004;11 7: 109 - 11 7 . H y ppönen E, Kenward MG, Virtanen SM, Piitulainen A, Virta-Autio P, Knip M, Åkerblom HK , the Childhood Diabetes in Finland Stud y Group. Infant feedin g , earl y wei g ht g ain and risk of t y pe 1 diabetes. Diabetes Care 1999 ; 22:1961-1965. Hyppönen E, Läärä E, Reunanen A, Järvelin MR , Virtanen SM. Intake of vitamin D and r isk of type 1 diabetes: a birth cohort study. Lancet 2001;358:1500-1503. Knip M. Natural course of preclinical typ e 1 diabetes. Horm Res 2002; 57, Suppl. 1: 6- 11. Knip M. Cow’s milk and the new trials for prevention of type 1 diabetes. J Endocrinol Invest 2003;26 : 26 5- 26 7. Norris JM, Barri g a K, Klin g ensmith G, Hoffman M , Eisenbarth GS , Erlich HA , Rewers M: Timin g of initial cereal exposure in infanc y and risk of islet autoimmunit y . JAMA 2003 ; 290:1713-1720. Pettitt DJ , Forman MR , Hanson RL , Know l er WC, Bennett PH. Breastfeeding and i ncidence of non-insulin-dependent diabetes mellitus in Pima Indians. Lancet 1997 ; 350:166-168 . Singhal A, Cole TJ, Fewtrell N, Deanfield J, Lu c as A. Is slower early growth beneficial for long-term cardiovascular health? Circulation 2004;109:1108-1113. 1 50 M ikael Knip and Hans K. Å kerblom Singhal A, Fewtrell M, Cole TJ, Lucas A. Low nutrient intake and early growth for late r i nsulin resistance in adolescents born preterm. Lancet 2003;361:1089-1097. Singhal A, Lucas A. Early origins of ca r diovascular disease: is there a unifying h ypothesis? Lancet 2004;363:1642-1645. The EURODIAB Substudy 2 Study Group. Vitamin D supplement in early childhood and r isk for T y pe I (insulin-dependent) diabetes mellitus. Diabetolo g ia 1999;42:51-54. W ilkin TJ. The accelerator h y pothesis: wei g ht g ain as the missin g link between T y pe I and T y pe II diabetes. Diabetolo g ia 2001;44: 914-922. W ilkin TJ, Metcalf, BS, Murphy NJ, Kirkby J, Jeffery AN, Voss LD. The relative c ontributions of birth weight, weight chang e , and current weight to insulin resistance i n contemporary 5-year olds: The EarlyBird Study. Diabetes 2002;51:3468-3472. V irtanen SM, Knip M. Nutritional risk predictors of β - cell autoimmunity and type 1 diabetes at a young age. Am J Clin Nutr 2003;78:1053-1067. Y oung TK, Martens PJ, Taback SP, Sellers EA C , Dean HJ, Cheang M, Flett B. Type 2 diabetes mellitus in children: prenatal and earl y infanc y risk factors amon g Native C anadians. Arch Pediat r Adolesc Med 2002 ; 156:651-655. Z ie g ler A-G, Schmid S, Huber D, Hummel M, B onifacio E: Earl y infant feedin g and ris k o f developin g T y pe 1 diabetes-asso c iated autoantibodies. JAMA 2003 ; 290:1721- 1728 . . EARLY NUTRITION AND LATER DIABETES RISK Mikael Knip and Hans K. Åkerblo m H ospital f or Children and Adolescents, Uni v ersit y o f Helsinki,. Virtanen and Knip E ARLY NUTRITION AND LATER DIABETES RISK 147 2003). In a Finnish surve y current wei g ht in the hi g hest quartile at the a g e of 3 and 6 months resulted in a 1.5-fold risk. qualitatively and quantitatively when assessing later health consequences o f infant feeding . E ARLY NU TRITI ON A N D LATER DIABETE S RI S K 14 5 3. EARLY FEEDING AND RISK OF TYPE 1 DIABETES 3.1