Early Life Events and Their Consequences for Later Disease a Life History and Evolutionary Perspective

7/28/2019 Early Life Events and Their Consequences for Later Disease a Life History and Evolutionary Perspective

1/19

Feature Article

Early Life Events and Their Consequences for LaterDisease: A Life History and Evolutionary Perspective

PETER D. GLUCKMAN,1,2* MARK A. HANSON,2 AND ALAN S. BEEDLE11Liggins Institute, University of Auckland, and National Research Centre for Growth and Development,

Private Bag 92019, Auckland, New Zealand2Centre for Developmental Origins of Health and Disease, University of Southampton, Southampton SO16 5YA,United Kingdom

ABSTRACT Biomedical science has little considered the relevance of life history theory and

evolutionary and ecological developmental biology to clinical medicine. However, the observa-

tions that early life influences can alter later disease riskthe developmental origins of health

and disease (DOHaD) paradigmhave led to a recognition that these perspectives can inform

our understanding of human biology. We propose that the DOHaD phenomenon can be consid-ered as a subset of the broader processes of developmental plasticity by which organisms adapt

to their environment during their life course. Such adaptive processes allow genotypic variation

to be preserved through transient environmental changes. Cues for plasticity operate particu-

larly during early development; they may affect a single organ or system, but generally they

induce integrated adjustments in the mature phenotype, a process underpinned by epigenetic

mechanisms and influenced by prediction of the mature environment. In mammals, an adverse

intrauterine environment results in an integrated suite of responses, suggesting the involve-

ment of a few key regulatory genes, that resets the developmental trajectory in expectation of

poor postnatal conditions. Mismatch between the anticipated and the actual mature environ-

ment exposes the organism to risk of adverse consequencesthe greater the mismatch, the

greater the risk. For humans, prediction is inaccurate for many individuals because of changes

in the postnatal environment toward energy-dense nutrition and low energy expenditure, con-tributing to the epidemic of chronic noncommunicable disease. This view of human disease from

the perspectives of life history biology and evolutionary theory offers new approaches to preven-

tion, diagnosis and intervention. Am. J. Hum. Biol. 19: 119, 2007. ' 2006 Wiley-Liss, Inc.

There has been an increased focus on thepossible role that early life events might playin the generation of later disease risk inhumans. There is a considerable literaturerelating the nature of infant feeding to laterhealth consequences (e.g., Lucas, 1991) and

also relating birth size to later risk of disease(e.g., Syddall et al., 2005). These observationsin turn arose from earlier studies linking theconditions of life in infancy to later disease riskand mortality (Barker and Osmond, 1986;Forsdahl, 1977), a link more recently con-firmed by extensive epidemiological studies(reviewed by Godfrey, 2006).

Although the epidemiological associationsare strong, a mechanistic understanding isessential in order to demonstrate that theseassociations are causal and in what way, andto identify the potential for intervention.

Mechanistically and experimentally, there isan increasing body of knowledge showing that

manipulation of the environment in the periodextending from conception to infancy can beassociated with permanent changes in physiol-ogy and/or structure. In turn, many of thesechanges are associated with permanent altera-tions in gene expression regulated by epige-

netic factors such as DNA methylation and his-tone methylation/acetylation (Gluckman andHanson, 2006a).

In this article, we place these clinical and ex-perimental observations connecting early life

Contract grant sponsor: M.A.H. is supported by theBritish Heart Foundation.

*Correspondence to: Peter D. Gluckman, Liggins Insti-tute, University of Auckland, Private Bag 92019, Auckland,New Zealand. E-mail: [email protected]

Received 1 June 2006; Revision received 17 August 2006;Accepted 29 August 2006

Published online in Wiley InterScience (www.interscience.wiley.com). DOI 10.1002/ajhb.20590

AMERICAN JOURNAL OF HUMAN BIOLOGY 19:119 (2 007)

VVC 2006 Wiley-Liss, Inc.


2/19

events with later disease consequences, gener-ally known as the developmental origins ofhealth and disease (DOHaD) phenomenon,within a wider framework encompassing life

history and evolutionary concepts. We suggestthat the DOHaD phenomenon represents themost visible manifestation of the fundamentalprocesses of developmental plasticity by whichmammals adjust their life history strategy inresponse to environmental cues during earlydevelopment.

DEVELOPMENTAL ORIGINS OF HEALTHAND DISEASE

The origin of the DOHaD concept arosefrom epidemiological observations relatingbirth size to an altered risk of cardiovasculardisease (Barker and Osmond, 1986) and type2 diabetes (Hales et al., 1991). The earliest con-cepts directly linked failure of fetal growth topermanent changes in structure and functionthat predisposed to disease. Subsequently, itwas suggested that these changes might alsohave adaptive value in a deprived (in nutri-tional terms) postnatal environment by makingthe individual permanently thrifty, e.g. bybeing small and insulin resistant. This thriftyphenotype hypothesis (Hales and Barker,

2001) was placed in contradistinction to Neelsthrifty genotype hypothesis (Neel, 1962) inwhich he had proposed that the thrifty aspectsof the contemporary individual arose from genesselected over a long time period during ourancestors huntergatherer existence. The con-sequences were however similarthe adult or-ganism would be well suited to an environmentin which nutrient supplies were limited butwould be more likely to be become unhealthy ina nutritionally rich environment.

It is now clear that small birth size is not anobligatory part of the pathway leading to in-

creased disease riskduring development,many environmental factors (cues) can inducechanges in physiology and anatomy with long-term consequences, and in humans, these maybe manifested as altered disease risk, withoutnecessarily inducing a major, or indeed any,change in birth size. This is partly evident fromthe graded relation between disease risk andbirth size that occurs across the normal birth-weight range (e.g., Barker, 1998). Moreover, itis now known that prematurity independent ofgrowth retardation is also associated with long-term metabolic consequences (Hofman et al.,

2004), and the offspring of women subjected tosevere undernutrition in early pregnancy dur-

ing the famine in the winter of 1944/1945 inthe Netherlands did not have reduced birthweight but do have an increased risk of obesity(Painter et al., 2005).

A range of epidemiological studies in manypopulations now extends the disease risk pro-file to include coronary heart disease andstroke; insulin resistance and thus type 2 dia-betes; altered endothelial function and in-creased risk of hypertension; central and pe-ripheral propensity to obesity; altered bodycomposition, including reduced bone mineraldensity which may lead to osteoporosis; andpossible effects on cognitive and emotionalfunction. There are several recent biomedicallyfocused reviews of the DOHaD concept and theobservations that support it (Gluckman andHanson, 2006a), including those published re-cently in this journal (Kuzawa and Pike, 2005),so this review will focus on the broader inter-pretation and significance of this association.

There are limitations to both the classicalthrifty genotypic and phenotypic models, al-though both have been important in the evolu-tion of understanding in the field. We havebuilt off these to consider further the potentialadaptive basis of the DOHAD phenomenon andits relationship to the processes of develop-mental plasticity. The original thrifty geno-

type hypothesis posited that the high incidenceof obesity and insulin resistance in westernpopulations was a result of ancestral selectionfor marginal nutritional environments andthat the expression of these thrifty genes re-sulted in a greater risk of obesity and insulinresistance in the presence of high energy in-take and low energy expenditure. But faminein our huntergatherer ancestors may havebeen less common than supposed (Benyshekand Watson, 2006), the search for thrifty geneshas been unproductive, and the ability to repro-duce these phenomena by manipulation of ex-

perimental animals of uniform genetic back-ground (reviewed in Gluckman et al., 2005b;McMillen and Robinson, 2005) and by short-term natural experiments such as the DutchHunger Winter (Painter et al., 2005) arguesagainst a purely genetic explanation. Nonethe-less, genetic variation will play a role in deter-mining sensitivity to environmental stimuli,and the presence of such effects as reflected inthe influence of specific polymorphisms withinthe epidemiological observations is well docu-mented (Eriksson et al., 2002; Jordan et al.,2005; Kajantie et al., 2004; Kubaszek et al.,

2004; Weedon et al., 2005). How genetic varia-tion and environmental cues may interact to

2 GLUCKMAN ET AL.

American Journal of Human Biology DOI 10.1002/ajhb


3/19

match the phenotype predictively to the envi-ronment has recently been modeled (Leimaret al., 2006).

The thrifty phenotype model developed by

Hales and Barker attempted to explain the epi-demiological relationship between birth sizeand disease risk by proposing that the fetusadapts to maternal undernutrition by growthretardation, which leaves the adult better eq-uipped to cope in a deprived rather than an en-riched postnatal environment. This adaptivemodel suggests that the organism responds toimmediate challenges by reducing growth, per-haps by mechanisms involving reduced insulinsensitivity, but must then cope postnatally withthe consequences for the rest of its life. Al-though this model can explain a relationship

between low birthweight and disease risk, ithas limitations. As noted above, it cannot easilyexplain the continuous relationship betweenbirth size and disease risk that occurs even inthe upper part of the birthweight range. Norcan it explain how early life events can inducelong-term physiological changes withoutchanges in birth size. It cannot explain develop-mental induction in systems not involvingnutritionfor example induction in the sys-tems regulating behavior, body temperature, orfluid balance for which there is good experi-mental evidence (Gate et al., 1999; Ross et al.,2005; Vickers et al., 2003). Finally, it is incon-sistent with recent clinical data suggesting thatsmall-for-gestational-age individuals are bornwith enhanced insulin sensitivity and that in-sulin resistance appears later in life (Ibanezet al., 2006; Mericq et al., 2005)these observa-tions imply that any adaptive advantage ofreduced insulin sensitivity only becomes mani-fest after the neonatal period.

The word thrifty implies directionality in-herent in the underlying mechanisms and thisis not the case. Recent studies in infants born

after in vitro fertilization show that they growup thinner, taller, and with enhanced insulinsensitivity, demonstrating that early life eventscan adjust subsequent metabolic physiology ineither direction (Cutfield, 2005). Moreover,large infants are also at risk of other later-lifedisease consequences, such as an increasedrisk of breast cancer (Dos Santos et al., 2004)probably mediated by higher estradiol levels(Jasienska et al., 2006). We have thereforesuggested that the original observations re-lating birth size to cardiovascular and meta-bolic disease risk represent a special case of a

much broader and generalizable phenomenonthat allows the pattern of development to be

modified in either direction as a result of de-velopmental cues (Gluckman and Hanson,2004b; Gluckman et al., 2005c,d).

CLASSES OF DEVELOPMENTAL RESPONSEAND THEIR ADAPTIVE VALUE

Environmental factors acting during thephase of developmental plasticity can either actto disrupt the normal program of developmentor to modulate it (Gluckman et al., 2005d).Developmental disruption may be overtly ter-atogenic, leading to gross structural malfor-mation, or may be much more subtle. Clearly,such disruptive responses cannot be consid-ered adaptive.

The processes of phenotypic plasticity per-

mit a range of phenotypes to be expressedfrom one genotypeindeed, the concept of thereaction norm was developed to demonstrateand illustrate this potential range of pheno-types (Schlichting and Pigliucci, 1998). Thepresumption is that there is potential adapt-ive advantage for a particular phenotype in aspecific environmental situation (Eshel andMatessi, 1998), although empirical proof ofsuch enhanced fitness is a formidable chal-lenge. However, as we have detailed elsewhere(Gluckman and Hanson, 2005), it is useful todistinguish adaptive responses to environmen-tal cues that have their potential advantage inclose temporal proximity to the cue from thosewhere the cue operates in one phase but thepotential advantage is manifested in a laterphase of the life course.

Immediately adaptive responses

Traditionally, developmental plasticity haslargely been viewed in terms of the immedi-ately adaptive class of response. For example,the altered morphology of the tadpole in re-

sponse to intraspecific competition can be seenas immediately advantageous although it mayincur a cost later in life (Relyea and Hoverman,2003). Premature metamorphosis in the spa-defoot toad is another example of a responsewhere the advantage is essentially immedi-atelinked to survival if the pond is evapo-ratingbut which may have a fitness cost inreduced adult size later in life (Newman, 1992).Both these examples illustrate that the ad-vantages of an immediate alteration in devel-opmental pattern that ensures short-termsurvival must be traded off against potential

disadvantages in a later environment, andsuch adaptive arguments are generally used

3DEVELOPMENTAL CHOICE OF LIFE HISTORY STRATEGY



4/19

to explain the evolution of developmentalplasticity.

Intrauterine growth retardation and mildprematurity in the human can both be seen as

analogous immediately advantageous adapta-tions. The major nongenetic cause of intrau-terine growth impairment is a reduced nutri-tional supply across the placenta to the fetus.This may arise because of maternal or placen-tal disease or from excessive maternal con-straint. The fetus responds to reduced nutri-ent supply by reducing its anabolic demand,in part by reducing plasma concentrations ofinsulin and of plasma and tissue concentra-tions of insulin-like growth factor (IGF)-1,which are the two best-documented fetal growth-promoting hormones (Gluckman, 1995). There

is also a major redistribution of fetal blood flowto protect development of the critical organs, inparticular the fetal brain, which can lead to adegree of asymmetrical growth retardation.This reduction in fetal growth is clearly neces-sary to allow fetal survival to birth. But beingborn small has consequences. The risk of neo-natal death and morbidity is significantlyhigher (Spencer, 2003; Wilcox, 2001) and so arethe risks of subsequent cognitive impairmentand persistent growth failure (Gluckman andHarding, 1997; McCarton et al., 1996). Clearly,immediate survival has been achieved at thecost of the individual being at greater subse-quent risk.

Prematurity provides a similar example. Pre-mature birth can result from an infectiousintrauterine environment in which the fetus isat greater risk. Some prematurity also appearsto be induced by maternal behavior that mightcompromise the fetus, such as poor nutrition orsmoking, or by maternal psychosocial stress.Premature induction of labor clearly involves atradeoff between the risk of continued exposureto an abnormal intrauterine environment and

that of being born at full-term but with the in-creased risk of postnatal morbidity and mortal-ity (Pike, 2006). Preterm delivery as an imme-diately adaptive response may also benefit themothers future reproductive fitness by limitinginvestment in a particular pregnancy (Pike,2006).

Predictive adaptive responses

Conversely, there are a number of clear ex-amples where the developing organism makesphenotypic responses during development to

obtain an adaptive advantage, but where thatadaptive advantage is delayed. We have termed

these predictive adaptive responses (Gluckmanet al., 2005c).

Adult desert locusts show polyphenism, ex-isting as migratory or solitary forms in which

metabolism, wing shape, behavior, and colora-tion differ. The choice of adult phenotype isdetermined in the larval stage in response tochemical cues from the mother, but the advan-tages and appropriateness of the choice are ex-hibited only after final metamorphosis (Apple-baum and Heifetz, 1999). Clearly in this casethe constraints of developmental pathwaysrequire the choice to be made at an early stagein development, and in polyphenic species suchas the locust there are two distinct morphs,although intermediates may form. In themountain vole, the decision about the tempo of

growth and acquisition of sexual maturity ismade in relation to the season of yearanimalsborn early in the summer undergo rapidgrowth and breed early, whereas animals bornlate in the season grow slowly and overwinteras prepubertal animals of subadult size (Neguset al., 1992). In the meadow vole, coat thicknessfor life is determined before birth by maternalsignals involving melatonin, related to whetherday length is shortening or lengthening. Thethermal environment before birth and in thenest is similar, so the advantage of the choice ofcoat thickness does not appear until somemonths later when summer or winter tempera-tures are encountered (Lee and Zucker, 1988;Lee et al., 1989). In each of these examples, adevelopmental pathway is chosen in expecta-tion of a future environment and fitness is en-hanced if the predicted environment is matchedto the phenotype that has developed. It is easy toenvisage the evolutionarypathwayto the appear-ance of these predictive adaptive responses, sincethey involve some relativelypredictable feature oftheenvironment.Although these examples illus-trate a dichotomy in the mature phenotype, we

would anticipate that predictive adaptive res-ponses can also be reflected over a continuousrange of phenotypes.

Predictive adaptive responses can only beadaptive if the forecast is correct and will bemaladaptive if it is not. The retention of mecha-nisms underpinning such phenotypic memorywill be advantageous provided that the predic-tion is generally more often correct than incor-rect, although this in turn depends on the rela-tive fitness of each phenotype in the predictedand nonpredicted environments and on therate of environmental change relative to gener-

ation time (Jablonka et al., 1995; Moran, 1992;Sultan and Spencer, 2002). Modeling shows

4 GLUCKMAN ET AL.



5/19

that these processes would be particularlyvaluable in allowing an organism to survive dur-ing a transient environmental shift (Gluckmanet al., 2005c; Jablonka et al., 1995; Moran,

1992).These different types of responseimmedi-

ate or predictive, disruptive, or adaptiveto anenvironmental cue have been presented as dis-continuous, but there will be overlap. Indeed,the same environmental cue may induce pre-sumptively adaptive or disruptive responsesdepending on its magnitude. For example,maternal hyperglycemia may induce congeni-tal heart defects, a disruptive response, oralterations in fetal growth with long-term con-sequences that may be adaptive. A develop-mental response to an endocrine disruptor may

involve interactions between the hormone-liketoxin and the normal processes of developmen-tal plasticity (Heindel and Lawler, 2006), butalthough such interactions use physiologicalmechanisms they disrupt development andcannotbeconsideredadaptive.

Evaluating whether a response is adaptiveor disruptive may be difficult in a given experi-mental situation. For example, is the reductionin nephron number in sheep after maternal ex-posure to very high doses of glucocorticoids inearly gestation (Wintour et al., 2003) a process

where the steroid has disrupted the normalpattern of nephron differentiation, or is it partof some adaptive process mimicking a normalsituation where the fetus responds to mater-nal glucocorticoids crossing the placenta undersituations of maternal stress? Similarly, is thecontinuous relationship between maternalvitamin A intake and nephron number in therat (Lelievre-Pegorier et al., 1998) a dose-de-pendent disruptive effect or doesit have adaptivevalue? Equally, a fetal adaptive response mayhave both an immediate and a predictive compo-nent. The accelerated maturation of the fetus of a

malnourished mother may induce prematuredelivery, which we can consider as providingadvantage potentially benefiting both motherand fetus (Pike, 2006), but it may also be partof a predictive response allowing the organismto cope better in a later environment perceivedto be threatening.

There are now several lines of evidence thatsupport the predictive adaptive hypothesis. Inrats subjected to maternal undernutrition dur-ing pregnancy, the offspring develop a per-turbed metabolic phenotype showing obesity,hyperphagia,hyperinsulinemia, andreducedac-

tivity in open field testing, particularly if theyare fed a high-fat diet after weaning (Vickers

et al., 2000, 2003). In this case, the poor prena-tal nutrition led to prediction of a poor post-natal environment, but in reality the postnatalenvironment was nutritionally rich. Recent

data (Fig. 1) show how the prediction can be re-versed provided that the modifying cue occursduring the phase of developmental plasticity. Ifprenatally undernourished rat pups are givenleptin subcutaneously in the neonatal periodthey grow up, even on a high-fat diet, with ametabolic phenotype identical to that of ratsborn to a normally nourished mother (Vickerset al., 2005). It appears that the neonatal ratpup has been tricked into perceiving that it isfatter than it is, leptin being a hormone madeby fat. Because this neonatal stimulus occurswithin the window of developmental plasticity,

at least for metabolic physiology, it is possiblefor the prenatal prediction to be changed fromone of an adverse environment to one of anenergy-rich environment, with the consequentalterations in physiology (Gluckman et al.,2006). At a mechanistic level this may involvethe synaptogenic actions of leptin (Bouretet al., 2004) or its peripheral actions, for exam-ple on pancreatic cell apoptosis and prolifera-tion (Islam et al., 2000; Shimabukuro et al.,1998).

Another consideration recently recognized isthe importance of the interaction between theeffect of the prenatal cues on the phenotypeand its consequent effect on the response to thepostnatal environment. For example, in ratsthe prenatal environment determines both pe-ripheral and central sensitivity to high-energynutritional intake imposed after weaning(Vickers et al., 2000). The effects of the inducedphenotype can be exacerbated (Hales andOzanne, 2003; Vickers et al., 2000) or amelio-rated (Jimenez-Chillaron et al., 2006; Stofferset al., 2003; Vickers et al., 2005; Wyrwoll et al.,2006) by postnatal interventions, indicating a

window of physiological plasticity that extendsbeyond the intrauterine period. In this discus-sion, we have simplistically described only onedevelopmental environment and only one ma-ture environment, but it would be more realis-tic to envisage a cascade whereby a series of ge-notype-environment interactions during onestage in development determines the settingfor interactions with the environment at thenext stage, and so on.

There is increasing evidence that predictiveadaptive responses are underpinned by epige-netic changes. These may be manifest as al-

tered gene expression and consequent alteredregulatory control or, given the importance of




6/19

epigenetic mechanisms to developmental biol-ogy, as organizational change and altered tis-sue development. There is a growing body of lit-

erature relating prenatal manipulation by hor-monal or nutritional stimuli to methylation ofDNA andchanges in chromatin structure affect-ing the expression of genes underlying thealtered phenotype. For example, maternalbehavior in rats can be manipulated to leave theoffspring as adults with altered stress responsesand behavior. This is associated with alterationsin the expression of the glucocorticoid receptorin the hippocampus, which in turn are causedby alterations in DNA methylation in a pro-moter sequence for the gene. The altered pheno-type can be reversed by chemical manipulation

of histone acetylation, one of the consequencesof DNA methylation (Weaver et al., 2004). In

another experimental model, the offspring offemale rats exposed to a low-protein diet showchanges in expression of several genes, includ-

ing that for the glucocorticoid receptor, in theliver, heart, muscle, and fat (Bertram et al.,2001); such changes are associated with alteredDNA methylation and can be reversed by folicacid supplementation during the period ofmaternal malnutrition (Lillycrop et al., 2005).We have preliminary evidence that the reversalof the effects of prenatal undernutrition by post-natal leptin treatment (Vickers et al., 2005) isaccompanied by coordinated changes in expres-sion and promoter methylation of key genes.

There is increasing interest in epigenetictransmission across generations (Chong and

Whitelaw, 2004; Drake and Walker, 2004),although the mechanisms and their relevance

Fig. 1. Reversal of developmental induction by neonatal leptin treatment. Rat dams were fed ad libitumthroughout pregnancy (AD) or undernourished throughout pregnancy (UN). Pups from UN mothers were cross-fos-tered to AD mothers after birth to standardize postnatal feeding, and all offspring were fed a high-fat diet afterweaning. The neonates were treated with either saline (S) or recombinant rat leptin (L) on postnatal days 313.Total body fat and metabolic profile as reflected by fasting plasma leptin, insulin and C-peptide (a marker of pan-creatic insulin secretion) concentrations were measured in adult females (postnatal day 170). Untreated prenatallyundernourished animals (UN S) were obese, hyperinsulinemic and hyperleptinemic as adults compared with con-trol animals (AD S); development of this metabolic phenotype was prevented by neonatal leptin treatment (UN L).

Values are means6SEMfor eight animals pergroup(redrawnfrom data in Vickers et al., 2005).

6 GLUCKMAN ET AL.



7/19

to reported transgenerational inheritance inhumans (Kaati et al., 2002; Pembrey et al.,2006) are poorly understood. There is experi-mental evidence that developmental induction

can be transmitted from the F0 generation tothe F1 generation and beyond through the maleand/or female lines after hormonal (Drakeet al., 2005) or nutritional (Benyshek et al.,2006; Zambrano et al., 2005) cues. Alterna-tively, there are indirect means by which envi-ronmental influences can be reflected acrosstwo or more generationsfor example, theuterine size of a female may be influenced byher experience as a fetus (Ibanez et al., 2003)and this in turn may influence the size of herown progeny.

THE DOHaD PARADIGMAN ADAPTIVEPERSPECTIVE

We have used the concept of predictive ad-aptive responses to explain the observationsthat led to the development of the DOHaD par-adigm (Gluckman et al., 2005b). We have sug-gested that the developing organism senses itsmetabolic milieu and adjusts its physiologicalhomeostatic setpoints according to the environ-ment it forecasts will exist after birth. If thesignals in the developmental phase suggest

limited nutrient availability, then the organismwill adjust its developmental trajectory suchthat the mature individual has a metabolic ho-meostasis better adapted for survival in asparse environment. Any association betweenthe outcome and birth size is thus an epipheno-menon of the relationship between nutrientavailability and the induction of predictive ad-aptive responses. Such a model can explain theobserved continuous associations between birthsize and disease risk, how disease risk may beindependent of changes in birth size, and howdevelopmental induction can occur in other

physiological systems.An important feature of such a model is thatthe alterations in regulatory systems inducedby the developmental cues may be subtle andnot observable under basal conditionstheymay affect the sensitivity of a physiologicalresponse underpinning a specific trait in theadult and thus make the organism more or lesssusceptible to disease in an extreme postnatalenvironment. Lifestyle disease in the humanthen occurs when the individual lives in anenvironment beyond their forecasted andinduced physiological homeostatic rangewe

term this the match-mismatch model. Wherethe organisms physiology is matched by devel-

opmental processes to the environment inwhich it lives in later life, its risk of disease islower; when it is mismatched, its risk is higher(Fig. 2). Clearly, both immediate and predictive

components may play a role in the final pheno-type and there will be a degree of interactionbetween the two, particularly if the environ-mental cues are more extreme. For example,we have elsewhere suggested that predictiveresponses play a role in determining the role ofprenatal nutrition in the timing of menarche,although this effect is modified by nutritionalstatus in the peripubertal period (Gluckmanand Hanson, 2006b).

Implicit in this discussion is the view thatthe DOHaD phenomenon is underpinned byphysiological mechanisms that have evolved in

mammals to provide adaptive advantage. Thereare several reasons, based on a large body of ex-perimental evidence reviewed in detail else-where (Gluckman et al., 2005b), why we believethis to be so. First, it is very easy to demon-strate this phenomenon in a wide variety ofmammalian species with diverse life historystrategies, there being data from experimentalmanipulation in mice, rats, sheep, pigs, guineapigs, and non-human primates and from epide-miological studies in humans. Second, relation-ships similar to the human epidemiologicalfindings can be seen in animals even in the ab-sence of maternal manipulation, suggesting aphenomenon that represents physiological plas-ticity rather than pathology. Third, a relativelysimilar phenotype can be induced by a varietyof nutritional or hormonal manipulations act-ing either early or late in developmentfor ex-ample, a trend to obesity, insulin resistance,and endothelial dysfunction can be inducedby high-fat diets, low-protein diets, maternalglobal undernutrition, and maternal glucocorti-coid exposure at differing times in gestation insheep, rats, mice, and guinea pigs. That these

different cues converge on a common outcomereflecting an integrated response of the fetus toa perceived threatening postnatal environmentsuggests that the induced phenotype has valuefor the fitness of the offspring. Finally, our dem-onstration that the integrated phenotypeinduced by prenatal nutritional insult can becompletely reversed by a simple postnatal endo-crine manipulation (Fig. 1; Vickers et al., 2005)argues for a single mechanism underpinning abroad range of developmental effects.

Stearns and Ebert (2001) have usefully sum-marized criteria for distinguishing adaptive

from nonadaptive processes. One such criteriondefines an adaptation as a change in a phe-




8/19

notype that occurs in response to a specificenvironmental signal . . . resulting in improvedgrowth, survival, or reproduction. We havelinked prenatal nutritional insult to a definedchange in phenotype; later in this review, weexplain why we believe that this phenotypeimproves survival in a sparse postnatal envi-ronment.

To summarize, our model invokes establishedconcepts from comparative biology to extend

the thrifty hypotheses by proposing that thedeveloping organism senses its environmentthroughout development and adjusts its physi-ology to be appropriate for the environment inwhich it predicts that it will live until andthrough reproductive age. These mechanismsconstitute predictive adaptive responses thatare influenced by the normal range of environ-ments experienced by the organisms mother.The fetus may also make some immediately ad-aptive responses if they assist in surviving untilbirth, and these responses may be reflected inbirth size. In this way, developmental plasticity

tunes physiological homeostatic settings in waysthat are much faster and potentially more sen-

sitive to changing conditions than those achievedby classical genetic mechanisms alone. For anypostnatal environment there will be a set ofphysiological settings associated with optimalfitness for that environment, and in the case ofhumans a risk of disease will exist when welive beyond the range of the expected environ-ment.

THE FETAL ENVIRONMENT AND MATERNAL

CONSTRAINT

The consequences of alterations in the de-velopmental environment will depend on howthe fetus perceives that environment and theaccuracy of the forecasts it makes. The mam-malian embryo and fetus has the particularchallenge of interpreting the maternalpla-cental transduction of environmental informa-tion, a process that may not operate with com-plete fidelity.

Fetal nutrition is not the same as maternaldietary intake. The provision of fetal nutrients

depends on the maternal nutritional environ-ment, her energy expenditure, her behavior

Fig. 2. The matchmismatch paradigm of metabolic disease. The developing organism senses maternally trans-mitted environmental cues, such as undernutrition, during prenatal and early postnatal life. Developmental plas-ticity in response to these cues modifies the default trajectory defined by the inherited fetal genome and epigenomeaccording to whether the environment is perceived as adequate (dark background) or deprived (light background),resulting in adjustment of metabolic setpoints. If the eventual mature environment, whether adequate or deprived,matches the prediction then the risk of metabolic disease in later life is low. If there is a mismatch between the pre-dicted and actual mature environments, particularly if the mature environment is richer than anticipated, thenthe risk of metabolic disease is enhanced.

8 GLUCKMAN ET AL.



9/19

and health, her metabolic and cardiovascularfunction, placental health and function, andfetal endocrine status. A further feature ofmammalian development is that fetal growth

is constrained by a number of factors. Theterm maternal constraint was introduced byOunsted (1965) to describe nongenetic mater-nally determined environmental and physio-logical influences that limit fetal growth. Theconcept derived from the classical work of Wal-ton and Hammond (1938) which showed thatthe size of the fetus was significantly affectedby the size of the mother, as opposed to the fetalgenotype. Later work has confirmed this to bethe case in several monotocous mammalianspecies. Indeed, in human pregnancies origi-nating from egg donation and surrogate moth-

erhood, birth size is more tightly correlatedwith recipient stature than with donor stature(Brooks et al., 1995).

The mechanisms underlying maternal con-straint are poorly understood. In some way, ma-ternal size determines nutrient availability tothe fetus. Whether the constraint mechanismlies in the uterine vasculature or in placentalstructure and function is not yet clear, but ingeneral shorter mothers who have smaller pel-vises have smaller babies. An alternative expla-nation might lie in parentoffspring conflict, aconcept originated by Trivers (1974) and ex-tended by Haig to include the observations thatthe genes for IGF-2 and the IGF-2 clearancereceptor are imprinted in the mouseIGF-2being expressed in the fetus from the paternalallele and the receptor being expressed fromthe maternal allele in the placenta (Haig andGraham, 1991). The conflict argument suggeststhat the fetus drives fetal growth under pater-nal influences, while the mother attempts tolimit fetal growth by clearing IGF-2 from the fe-tal circulation via the placenta. However, inhumans the IGF-2 receptor is not imprinted,

and in species that have a long gestation with along fetal as opposed to embryonic phase IGF-1and insulin are the primary determinants offetal growth in late gestation and thus of birthsize (Gluckman and Pinal, 2003). However, thereare some limited data suggesting that the pla-centa might also clear IGF-1 (Gluckman andPinal, 2003). Whatever the mechanism, non-genetic factors including maternal constrainthave been estimated to contribute over 50% ofthe variance in birth size in humans (Morton,1955; Ounsted and Ounsted, 1973).

Maternal constraint is of particular impor-

tance in human evolution. Maternal heightand pelvic proportions are tightly correlated.

As a result of the adoption of an upright pos-ture, pelvic proportions were altered. Further,the large brain of our species meant that wehad to adopt a secondarily altricial life course

to ensure the head could exit the pelvic canal.Even so, whereas the chimpanzee head can re-adily exit the pelvis, the human head can only

just exit the maternal pelvic canal and onlythen by some complex rotations (Trevathan,1988). Thus, maternal constraint has been thekey to limit fetal growth and allow our speciesto reproduce successfully.

The concept of maternal constraint has beenextended to include other nongenomic and in-trinsic physiological mechanisms that impacton birth size and developmental trajectory, al-though whether all rely on the same underly-

ing mechanisms is not clear (Gluckman andHanson, 2004a). Such effects are seen in primi-parous women, who generally have smallerbabies, in women of young maternal age, andin multiple pregnancies. There is no a priorireason why the mechanisms by which thesedifferent situations constrain fetal growth needbe identical. Indeed, the reduced growth oftwins is obviously a reflection of limits on thenutrient supply that can cross from mother tofetus. The constraint on fetal size in the firstpregnancy is likely to be due to poorer perfu-sion of the uterine bed but can also be inter-preted in life history terms as a mechanism bywhich a mother limits her investment in herfirst offspring, which generally would havehappened at a young age, in expectation of sub-sequent pregnancies. Recent studies indeedshow that the effect of parity on maternal con-straint is an important influence on relation-ships between birth size and later outcomes:truncal obesity is greater in first-born children(Stettler et al., 2000) and the inverse relation-ship between blood pressure and birth size isstronger in first-borns (Hardy et al., 2004).

The reduced size of infants born to adolescentmothers appears to be due to continued diver-sion of nutrients to support maternal develop-ment (Wallace et al., 2004) and again can beargued in life history terms as a strategy to en-hance maternal survival, both by protecting hernutritional status and because her pelvis doesnot reach maximal dimensions until some yearsafter menarche. As women now delay their firstpregnancy, at least in western societies, we areseeing a greater number of pregnancies at ad-vanced maternal agethese babies are also ofreduced birthweight (Gilbert et al., 1999).

Parenthetically, variation in birthweightmight also be affected by subtle changes in




10/19

gestational length. There is experimental, andincreasing clinical, evidence that mothers un-dernourished or stressed early in pregnancymay give birth after shorter gestational

lengths (reviewed in Gluckman et al., 2005a).This is an alternative strategy that the fetusmay use to respond to a threatening situation,and in experimental studies it is associatedwith accelerated activation of the hypothala-micpituitaryadrenal (HPA) axis (Bloomfieldet al., 2003).

The net effect of maternal constraint is gen-erally to limit nutrient availability to the fetus.Thus one does not anticipate a linear relation-ship between birth size and maternal nutri-tional status. As fetal conditions improve, fetalgrowth increases asymptotically (at least in terms

of lean body mass). There are exceptionsforexample, the fetal overgrowth syndromes suchas the BeckwithWeidemann syndrome whereboth alleles of IGF-2 are expressed by the fetusinstead of one being silenced by parental im-printing. In infants of diabetic mothers, the fe-tus is heavier due to increased fat mass second-ary to the adipogenic effects of increased fetalinsulin that is released in response to the in-creased transplacental glucose transfer (Schwartzand Teramo, 2000). There is a much smallereffect on lean body mass. Such infants arelarge, but have an abnormal body compositionwith increased adiposity and suffer pathologi-cal consequences. Historically, their poor sur-vival might have excluded them from the nor-mal pattern of human development. Thus, withthe exception of gestational diabetes, fetalgrowth in virtually all human pregnancies islimited by maternal constraint, even whenbirth weight is within the normal range. Thislimits the environmental range that the fetuscan detect and therefore predict. This in turnhas the effect of limiting the range of nutri-tional environments to which the physiology

of the organism can respond in postnatal life(Gluckman and Hanson, 2004b; Gluckmanand Hanson, 2004c).

Analyses of fitness in wild populations(Charmantier and Garant, 2005) confirm thatenvironmental factors play a greater role inphenotypic variation in populations under poorconditions than under good conditions. Such anobservation is compatible with the relation-ships we have described. The asymptotic rela-tionship seen under conditions when the fetusis exposed to a good environment will meanthat trait variation in such fetuses is primarily

a reflection of genetic variation. Conversely,when the fetus is making predictions based on

a poor environment, the individual lies on asteeper part of the predicted/actual relation-ship curve (Fig. 3) and one would anticipategreater variation in the trait as an adult. The

variation is enhanced further because of thevarious ways in which maternal nutritional in-formation may be misinterpreted by the fetus,for example after maternal or placental dis-ease, thus giving rise to a greater risk of inac-curacy in the prediction.

DEVELOPMENTAL MISMATCH ANDPATTERNS OF DISEASE

In a population living in a stable environ-ment, evolution will have matched physiologyto that environment. But the developmentalenvironment may vary and thus shift the posi-tioning of the range of environments an indi-vidual can live in as an adult (discussed inDeWitt and Scheiner, 2004). One way in whichthis can occur is through predictive adaptation.If the prediction is correct then there will be agood match between the phenotype adoptedand the environment in which the organismwill later live, and resulting fitness will behigh. If the prediction is poor, there will be amismatch between the environment experi-

enced and the phenotype induced. We havesuggested that concepts of developmental mis-match can provide a useful perspective on theDOHaD phenomenon (Gluckman et al., 2005a).

Impaired fetal growth is an indirect reflectionof an impaired fetal environment arising fromexcessive maternal constraint, the presence ofa severe maternal environment, or maternal orplacental disease. The fetus sees all these sit-uations as reflecting impaired nutrition, pre-dicts a deprived postnatal environment and ad-

justs its metabolic phenotype accordingly.But most of these situations lead to faulty

predictions, because the postnatal environ-ment is adequate. Thus, the individual willhave some mismatch between its maturephysiology and the environment it inhabits.The greater the mismatch, the greater is therisk of adult disease.

Such a paradigm can explain why popula-tions undergoing rapid nutritional transitionsuch as in India and China are witnessing amassive increase in the prevalence of metabolicand cardiovascular disease. Although the intra-uterine environment can only change slowlyacross generations (Gluckman and Hanson,

2005), the postnatal environment can changevery rapidly. Thus, a stunted mother will se-

10 GLUCKMAN ET AL.



11/19

verely constrain the growth of her fetus andinduce a metabolic phenotype appropriate for a

nutritionally deprived environment, but thepostnatal environment will not match thosesettings. Indeed, Figure 3 would suggest thatdevelopment in a severely constrained settingreduces the upper limit of the range of post-natal environments that can be tolerated with-out risk of disease. Such a paradigm explainsthe increase in lifestyle disease that occurswhen a population migrates rapidly from theless developed to the developed world, for ex-ample from Ethiopia to Israel (Cohen et al.,1988). One implication of such a model is that,even if there is optimal fetal growth, maternal

constraint always limits the postnatal environ-ment that the fetus can predict and prepare for.

Thus, the issue is the relative change (mismatch)from developmental to adult environment, not

whether the fetal environment is deprived oradequate, and the model is therefore relevantto the current obesity epidemic in the devel-oped world.

PRENATAL AND POSTNATAL EFFECTS

Most of the discussion above has focused onthe nutritional and energetic environment.This is inevitable, given that in life historyterms energetics are a very important compo-nent of the environment and nutritional manip-ulation has been the major experimental sys-

tem studied. However, the same model canapply to other dimensions of the environment,

Fig. 3. Consequences of a mature environment outside the predicted range. The zone of maximal fitness result-ing from the different developmental trajectories determined by environmental cues in early life is shaded. Note

the nonlinearity of the relationship, arising from maternal constraint acting to limit the size of the fetus at deliveryand setting the upper limit of the postnatal environment which the species can tolerate without the risk of meta-bolic disease even when maternal constraint is minimal (horizontal heavy broken line). Fetuses A and B are shownat the upper limits of their range of fitness. Fetus A has set its developmental trajectory in expectation of a rela-tively deprived environment. If the mature environment is richer than predicted, the risk of metabolic disease isincreased. In contrast, fetus B has developed in expectation of an adequate environment, the extent of unpredictedexcess is smaller, and a higher level of postnatal nutrition is needed for the risk of metabolic disease to beincreased. As increasing affluence raises the richness of the mature environment and because of the limiting effectof maternal constraint, a greater proportion of even optimally nourished fetuses in utero are mismatched to theirpostnatal environment (modified from Gluckman and Hanson, 2005).




12/19

in particular stress which in comparative termsmay reflect both intraspecific competition andpredator risk. There is ample data suggestingthat glucocorticoid exposure in utero as a result

of elevated maternal glucocorticoid levels caninduce long-term changes in the glucocorticoidaxis (ORegan et al., 2001) that may have short-term adaptive valuefor example, by tuningstress responses during the population cycle ofthe snowshoe hare (Boonstra et al., 1998).There is also increasing evidence that nutri-tional and hormonal cues intersect. In experi-mental animals, the phenotypes arising frommaternal glucocorticoid exposure and from ma-ternal nutritional manipulation are very simi-lar (Bertram and Hanson, 2001). The enzyme11b-hydroxysteroid dehydrogenase type 2 in

the placenta normally inactivates maternalglucocorticoids to limit fetal exposure. Under-nutrition reduces 11b-hydroxysteroid dehy-drogenase type 2 activity in the rat (Bertramet al., 2001) and low birthweight is associatedwith reduced enzymatic activity in the humanplacenta (reviewed in Seckl and Meaney, 2004),suggesting a possible common mechanism. Theglucocorticoid receptor in the brain and liverappears to be a target of epigenetic regulationby maternal factors (Lillycrop et al., 2005;Weaver et al., 2004). The same paradigm canalso explain the induction of long-term changein other physiological systems, such as thoseregulating temperature (Diamond, 1991), thirst(El-Haddad et al., 2004) and reproduction(Cooper et al., 1996; Gluckman and Hanson,2006b).

The initial cues may act at differing points inthe window of plasticity. For each organ, therewill be a time in development after which plas-ticity is clearly no longer possible. Equally,some cues may act very early in developmentthere is, for example, experimental evidencethat changing the maternal environment in the

pre-implantation period can have long-termeffects on metabolic and cardiovascular func-tion (Kwong et al., 2000). Similarly, in experi-mental in vitro fertilization, changing the cul-ture conditions can induce long-term develop-mental changes (Walker et al., 1996; Younget al., 2001). Some may be disruptive in naturebut others may be a reflection of evolved adapt-ive processes. In the agouti mouse, changingnutrient balance in the periconceptional periodalters for life the methylation status of the mu-tant gene underlying the agouti phenotype(Cooney et al., 2002; Wolff et al., 1998), again

suggesting that the periconceptional period is astage in which epigenetic change can be in-

duced. The limited human data suggests thatnutritional status at conception also has im-portant influences on both gestational length(Susser and Stein, 1994) and birth size (Neggers

and Goldenberg, 2003; Villar and Rivera, 1998).Conversely, there is both experimental (re-

viewed above) and clinical evidence to suggestthat the postnatal period is also one in whichlonger-term effects can be induced or the effectsof earlier cues can be modulated. In humans,there is much data suggesting that the natureof infant feeding has long-term metabolic, car-diovascular and cognitive effects (Lucas et al.,1998; Singhal et al., 2003; Singhal et al., 2004;Stettler et al., 2005). Parenthetically, the termprogramming was first adopted to describesuch linkages and the same term has been used

extensively in discussions of the fetal originsparadigm. We have avoided using it, as we andothers see conceptual limitations in its use. Itimplies a hardwired process, whereas theeffects observed are clearly much more flexible.In general, bottle-feeding appears to have rela-tively deleterious effects. Such studies have ledsome to conclude that enhanced infant growthitself induces long-term consequences (Singhaland Lucas, 2004). This may indeed be so, but itis difficult to separate out this effect from theearlier and inevitable constraint in utero whichthen sets the scene for such accelerated infantgrowth to occur. Other studies have suggestedthat, rather that rapid infant growth being themost significant factor, it is prolonged impairedinfant growth followed by rapid growth inchildhood (Bhargava et al., 2004; Erikssonet al., 2003). Although these different conclu-sions have led to some considerable debate as tothe importance of different patterns of growth,one consistent theme emergesthat constrainedfetal or infant growth followed by nutrition-enhanced infant or later childhood growth hasmetabolic consequences. Both patterns are

compatible with the predictive model and thedeleterious effect of nutritional mismatch sug-gested earlier. The issue of whether these twophases of perinatal development are independ-ent or interdependent is more complex and re-quires further research.

DOHaD, EVOLUTION, AND LIFE HISTORYSTRATEGY

There has been much recent progress in thefield of ecological developmental biology as wellas a greater integration of developmental biol-

ogy into the evolutionary synthesis (Gilbert,2001; West-Eberhard, 2003). Much of this

12 GLUCKMAN ET AL.



13/19

paper has suggested that the phenomena re-flected in the experimental and clinical obser-vations associated with the DOHaD paradigmare generalizable and related to basic processes

of development.We would suggest that the processes of devel-

opmental plasticity, in particular those havinglonger-term potential adaptive significance,have been central to the survival of mamma-lian species through variable environments.They would have been particularly importantin a generalist species such as the human. De-velopmental plasticity acts to increase thepotential for phenotypic variation, allowing ashift in the positioning of the individuals phys-iology within the reaction norm to optimize itsphenotype for a particular environment and

increase its fitness (survival and reproductivesuccess). In this way, a greater proportion ofindividuals will survive a transient environ-mental shift and greater genotypic variationcan be maintained (Gluckman and Hanson,2005). The concept that an environmental stim-ulus can shift the organism into a differentdevelopmental pathway (in the case of theDOHaD paradigm, a pathway anticipating asparse or threatening postnatal environment)was classically visualized by Waddington (1942,1957) as canalization towards alternativeend-states, with constraint on variation of aparticular end-state pictured as the steepnessof the local gradients of the epigenetic land-scape. More recent usage of the term canaliza-tion, also termed auto-regulation by Schmal-hausen (1949), has focused on the impliedaspect of constraint of phenotypic variationagainst the effects of genetic or environmen-tal perturbations (Debat and David, 2001).

Although developmental plasticity and envi-ronmental canalization act in opposite direc-tions on phenotypic variation, they both act topreserve genetic variationthe former by pro-

moting variation of the phenotype in a chang-ing environment, the latter by constrainingvariation of an evolved optimal phenotype in astable environment. The classic experiments ofBelyaev and his coworkers, who demonstratedthat phenotypic variation could be exposedwhen an apparently homogeneous populationof foxes was selected for one behavioral charac-teristic, tameness, are particularly relevantto our model of integrated responses in thatsuch selection appeared to act on componentsof the HPA axis but resulted in the appearanceof variation in a wide range of morphological,

physiological, and behavioral traits (Trut,1999).

Human development does not involve thegeneration of the distinct morphs describedabove for the desert locust, but it can nonethe-less be envisaged in an analogous way. The

human embryo, fetus and neonate senses itsfuture nutritional status via signals from itsmother and makes an integrated response tothose signals. This response is manifest as arange of phenotypes later in life, with conse-quences dependent on the energy density of theenvironment in which the individual lives asan adult.

We propose that such plasticity is a result ofwell-integrated responses regulated by a lim-ited number of key genes whose expression isaffected by the developmental cue and whichthen regulate subsequent patterns of gene

expression and tissue organization. Whethersuch regulation is positive (conceptually, acti-vation of plasticity) or negative (conceptually,deactivation of canalization) remains to bedetermined. The existence of genes for plastic-ity is a subject of debate (DeWitt and Scheiner,2004), whereas candidate canalization genesacting at the whole-organism level have beenidentified, for example HSP90 in Drosophila(Queitsch et al., 2002). However they are re-gulated, these integrated responses determinethe subsequent phenotype and life course.

Life-history theory is based on considerationof the allocation of resources to somatic effort(growth and survival) and reproductive effort,with the assumption that selection operates tomaximize reproductive fitness by determiningthe optimal allocation between somatic andreproductive investment. A corollary is that in-vestment in post-reproductive longevity doesnot affect reproductive fitness and is not fa-vored by selection. Our model suggests that ifthe fetus predicts an uncertain future, whichmay be reflected in being born smaller or pre-maturely, it will invest less in somatic effort,

possibly by lower investment in maintenanceand repair (Kirkwood, 2002), and hence thoseborn smaller tend to die earlier (Kajantie et al.,2005). Associated with this there may be lessinvestment in systems in which there is someredundancy, for example fewer nephrons (Hin-chliffe et al., 1992; Langley-Evans et al., 2003)and fewer neurons in some regions of the brain(Mallard et al., 2000; Tolsa et al., 2004)obser-vations confirmed both experimentally and clin-ically. Because the fetus predicts a shorter life itwill invest less in musculoskeletal growth, pre-disposing to sarcopenia and osteopenia, both

features of the human born smaller (Sayer andCooper, 2005). There is advantage when living




14/19

in a poor nutritional environment to layingdown fat when possible and to having periph-eral insulin resistance, both features well dem-onstrated in humans. In experimental animals,

this potential advantage can be seen in hyper-phagia (Vickers et al., 2000), altered appetiteregulation (Plagemann et al., 2000) and apreference for a high-fat diet (Bellinger et al.,2004). Given the relationship between thenutritional and HPA axes, it would not besurprising that the HPA axis is altered undersuch conditions, as is reported in both humans(Phillips et al., 1998) and animals (Seckl andMeaney, 2004), and this change may also con-tribute to the obese phenotype.

Resource allocation to reproductive effort, re-flected in age of maturity and fecundity, will

also be affected by developmental cues. Inhumans, good nutrition in childhood leads toearlier puberty (Karlberg, 2002). Conversely,low birthweight leads to accelerated sexualmaturation if postnatal energetics allow (Cooperet al., 1996; Gluckman and Hanson, 2006b;Ibanez et al., 2000; Karlberg, 2002), althoughfecundity is lower in both females (Ibanezet al., 2003) and males (Cicognani et al.,2002; Francois et al., 1997) of low birth-weight; both these observations are in accord-ance with predictions of life history theory(Stearns and Koella, 1986). Lower testoster-one levels in males may also represent a trade-off between reproductive and immune func-tions (Muehlenbein and Bribiescas, 2005). Fur-thermore, early maturation may itself involvea trade-off between offspring quantity andquality in the subsequent generation (Coalland Chisholm, 2003; Wallace et al., 2004), per-petuating a cycle of low birthweight.

Two extreme examples are illustrated inFigure 4a fetus exposed to an optimal nutri-tional environment in utero and a fetus ex-posed to a less than optimal environment. Al-

though the definitive life courses will inevitablybe modified by subsequent environmental expo-sure, consideration of the developmental compo-nent alone in the context of the immediatelyadaptive and predictive responses made in thealternative environments allows us to under-stand the biology of those born smaller.

While we have focused on the extremes tomake this point, we propose that there is a con-tinuous range of human metabolic morphsrepresenting a suite of integrated responses tothe environmental cues received in utero or bythe neonate which establish the setpoints of

the metabolic and related systems. We suggestthat these different developmental trajectories

are set by altered expression, presumably regu-lated by epigenetic processes, of key regulatorygenes. Although the specific nature of theseprocesses remains to be determined, compo-

nents of the HPA axis seem to be key candi-dates. Depending on the postnatal environment,such responses may manifest in an altered riskof disease.

Such an integrated suite of somatic and re-productive responses to early cues forecastinga sparse or threatening postnatal environmentwill have evolved to promote individual sur-vival to reproductive age and avoid extinctionof the lineage, and at a population level willpreserve genetic variation. But in an enrichedpostnatal environment this mismatched devel-opmental trajectory results in metabolic dis-

easemechanisms to ameliorate such diseasewould not have evolved since the adverse con-sequences occur predominantly in the post-reproductive period. In this way the predictionsof the match-mismatch model are not dissimi-lar from those of antagonistic pleiotropy (Wil-liams, 1957), which model suggests that allelesconferring fitness advantage early in life willbe selected for even if they are deleterious laterin life. The fundamental difference between thetwo models is that an adaptive mechanisminvolving phenotypic plasticity will allow re-versibility at the individual (Vickers et al.,2005) level should environmental conditionsimprove (or be perceived to have improved).

It has been argued that although induction ofthriftiness in the offspring has adaptiveadvantage, such advantage is predominantly forthe mother, reducing the demand of the currentpregnancy and lactation cycle and enhancingher future reproductive fitness (Wells, 2003).There are several reasons why we believe this tobe a flawed argument. First, predictive res-ponses cause an integrated change in the off-springs phenotype and subsequent life history

across several dimensions rather than a reduc-tion in birth size alone, an observation not read-ily explained by the maternal fitness hypothesis.Second, such reduction in birth size is not anobligatory partof the suite of alterations inducedby an adverse prenatal environment. Third, weagain point to data showing that one of the hall-marks of thriftinessinsulin resistancedoesnot appear until well after birth in small-for-gestational-age individuals (Ibanez et al., 2006;Mericq et al., 2005), implying little immediateadvantage for the mother.

Humans evolved in environments very dif-

ferent from those in which we now live. Analy-sis of paleoclimatic data (Potts, 1998) offers

14 GLUCKMAN ET AL.



15/19

support for the view that hominin evolutionoccurred in response to change over long (sev-eral generations) rather than short (seasonalor single-generation) time frames. In such asituation, forecasting (Bateson et al., 2004)of the environment would have been valuable,and the accuracy of the cue reaching the fetusmay have been enhanced by phenotypic iner-tia integrating the signal across multiple gen-erations to reduce the effects of short-term fluc-tuations (Kuzawa, 2005). It is probable that

enriched environments were experienced onlyrarely and that selection favored the operationof maternal constraint and the continued oper-ation of ancestral mechanisms of adaptive de-velopmental plasticity because they matchedthe individual to the postnatal environmentmost of the time. Together with the relativelyshort life expectancy of Paleolithic and Neo-lithic humans (Austad, 1994), this meant thatthese processes were usually adaptive. How-ever, in the enriched and rapidly changingnutritional environments of post-industrial so-ciety the risk of postnatal mismatch between

the developmentally induced phenotype andthe actual conditions experienced has increased

dramatically. Lifestyle diseases reflect that welive in a manner and in an environment beyondthe range we evolved to live within, a rangethat remains limited by maternal constraint.What clinicians observe as the DOHaD phe-nomenon is actually part of a fundamental pro-cess by which the organism adjusts its lifecourse strategy in response to environmentalinfluences acting early in development.

CONCLUSIONS

Attempts to explain the original observationsof a relationship between birth size and laterrisk of heart disease and of the long-term im-pact of various infant feeding regimens havemoved from consideration of specific situationsto a much broader emphasis on how life historyand evolutionary biology intersect with studiesof human disease. Major research questions re-main to be resolved: the genotypic determinantsof environmental sensitivity during develop-ment, the identification of the key regulatorygenes, the fundamental epigenetic processes

involved in phenotypic plasticity, the basis ofthe adjustment and integration of the various

Fig. 4. Integrated life history responses to developmental cues. Two extreme examples of developmental trajec-tories are shown, one in response to a perceived adequate nutritional environment and the other in response to aperceived deprived environment.




16/19

components of the life course strategy, and thecomplexities surrounding the role of the infantperiod in modifying earlier phenotypic choices.It is clear that life history biology offers much

to inform our understanding of human chronicnoncommunicable disease. It shifts the diseaseparadigm from one of external causation to oneof the match or mismatch between the evolutio-narily and developmentally defined constitu-tion of the organism and the environment inwhich it finds itself. This bridges the gapbetween the traditional medical and anthropo-logical models. The developmental componentoffers approaches to prevention and the expand-ing knowledge of epigenetics offers approachesto intervention.

At a pragmatic level, this perspective means

that different strategies may be more appropri-ate for disease reduction in different popula-tions (Gluckman et al., 2005b). Thus, in devel-oped countries the focus must remain on post-natal interventions to limit energy intake andincrease energy expenditure. In populations indeveloping countries a greater focus on the wellbeing of the mother may be important to allowsubsequent generations to maintain metabolichealth during a socioeconomic transition to anenvironment of higher energy density. Impor-tant approaches in this setting include delayingfirst pregnancy until the mother is fully grown,say at least 45 yr after menarche, ensuringadequate nutritional status at conception, andfocusing on maternal health and nutrition dur-ing pregnancy.

ACKNOWLEDGMENTS

We thank Patrick Bateson, Hamish Spencer,David Raubenheimer, and Chris Kuzawa forthoughtful discussions.

LITERATURE CITED

Applebaum SW, Heifetz Y. 1999. Density-dependent physio-logical phase in insects. Annu Rev Entomol 44:317341.

Austad SN. 1994. Menopause: An evolutionary perspective.Exp Gerontol 29:255263.

Barker DJ, Osmond C. 1986. Infant mortality, childhoodnutrition, and ischaemic heart disease in England andWales. Lancet 1:10771081.

Barker DJP. 1998. In utero programming of chronic disease.Clin Sci 95:115128.

Bateson P, Barker D, Clutton-Brock T, Deb D, DUdine B,Foley RA, Gluckman P, Godfrey K, Kirkwood T, MirazonLahr M, McNamara J, Metcalfe NB, Monaghan P,Spencer HG, Sultan SE. 2004. Developmental plasticityand human health. Nature 430:419421.

Bellinger L, Lilley C, Langley-Evans SC. 2004. Prenatal ex-posure to a maternal low-protein diet programmes a pref-

erence for high-fat foods in the young adult rat. Br J Nutr92:513520.

Benyshek DC, Johnston CS, Martin JF. 2006. Glucose me-tabolism is altered in the adequately-nourished grand-offspring (F3 generation) of rats malnourished duringgestation and perinatal life. Diabetologia 49:11171119.

Benyshek DC, Watson JT. 2006. Exploring the thrifty geno-types food-shortage assumptions: A cross-cultural com-parison of ethnographic accounts of food security amongforaging and agricultural societies. Am J Phys Anthropol131:120126.

Bertram C, Trowern AR, Copin N, Jackson AA, WhorwoodCB. 2001. The maternal diet during pregnancy programsaltered expression of the glucocorticoid receptor and type2 11b-hydroxysteroid dehydrogenase: Potential molecu-lar mechanisms underlying the programming of hyper-tension in utero. Endocrinology 142:28412853.

Bertram CE, Hanson MA. 2001. Animal models and pro-gramming of the metabolic syndrome. Br Med Bull 60:103121.

Bhargava SK, Sachdev HS, Fall CHD, Osmond C, Lakshmy R,Barker DJP, Biswas SK, Ramji S, Prabhakaran D,

Reddy KS. 2004. Relation of serial changes in childhoodbody-mass index to impaired glucose tolerance in youngadulthood. N Engl J Med 350:865875.

Bloomfield FH, Oliver MH, Giannoulias D, Gluckman PD,Harding JE, Challis JRG. 2003. Brief undernutrition inlate-gestation sheep programmes the hypothalamic-pitu-itary adrenal axis in adult offspring. Endocrinology 144:29332940.

Boonstra R, Hik D, Singleton GR, Tinnikov A. 1998. Theimpact of predator-induced stress on the snowshoe harecycle. Ecol Monogr 68:371394.

Bouret SG, Draper SJ, Simerly RB. 2004. Trophic action ofleptin on hypothalamic neurons that regulate feeding.Science 304:108110.

Brooks AA, Johnson MR, Steer PJ, Pawson ME, Abdalla HI.1995. Birth weight: Nature or nuture? Early Hum Dev 42:

2935.Charmantier A, Garant D. 2005. Environmental quality andevolutionary potential: Lessons from wild populations.Proc R Soc B 272:14151425.

Chong S, Whitelaw E. 2004. Epigenetic germline inheri-tance. Curr Opin Genet Dev 14:692696.

Cicognani A, Alessandroni R, Pasini A, Pirazzoli P, Cassio A,Barbieri E, Cacciari E. 2002. Low birth weight for gesta-tional age and subsequent male gonadal function. J Pediatr141:376379.

Coall DA, Chisholm JS. 2003. Evolutionary perspectives onpregnancy: Maternal age at menarche and infant birthweight. Soc Sci Med 57:17711781.

Cohen MP, Stern E, Rusecki MJ, Zeidler A. 1988. Highprevalence of diabetes in young adult Ethiopian immi-grants to Israel. Diabetes 37:824828.

Cooney CA, Dave AA, Wolff GL. 2002. Maternal methylsupplements in mice affect epigenetic variation and DNAmethylation of offspring. J Nutr 132:2393S2400S.

Cooper C, Kuh D, Egger P, Wadsworth M, Barker D. 1996.Childhood growth and age at menarche. Br J ObstetGynaecol 103:814817.

Cutfield W. 2005. IVF children are taller with increasedIGF-I, IGF-II and IGFBP-3 levels suggesting altered geneticimprinting. In Proceedings of the 3rd International Con-gress on Developmental Origins of Health and Disease,Toronto, Canada, November 1619, 2005.

Debat V, David P. 2001. Mapping phenotypes: Canalization,plasticity and developmental stability. Trends Ecol Evol16:555561.

DeWitt TJ, Scheiner SM. 2004. Phenotypic plasticity: Func-tional and Conceptual Approaches. Oxford: Oxford Uni-versity Press.

Diamond J. 1991. Pearl Harbor and the emperors physiolo-gists. Nat Hist 100:27.

16 GLUCKMAN ET AL.



17/19

Dos Santos SI, de Stavola BL, Hardy RJ, Kuh DJ,McCormack VA, Wadsworth ME. 2004. Is the associationof birth weight with premenopausal breast cancer riskmediated through childhood growth? Br J Cancer 91:519524.

Drake AJ, Walker BR. 2004. The intergenerational effectsof fetal programming: Non-genomic mechanisms for theinheritance of low birth weight and cardiovascular risk.J Endocrinol 180:116.

Drake AJ, Walker BR, Seckl JR. 2005. Intergenerationalconsequences of fetal programming by in utero exposureto glucocorticoids in rats. Am J Physiol 288:R34R38.

El-Haddad MA, Desai M, Gayle D, Ross MG. 2004. In uterodevelopment of fetal thirst and appetite: Potential forprogramming. J Soc Gynecol Investig 1:123130.

Eriksson JG, Forsen T, Tuomilehto J, Osmond C, Barker DJ.Early adiposity rebound in childhood and risk of Type 2 di-abetes in adult life. Diabetologia 2003; 46:190194.

Eriksson JG, Lindi V, Uusitupa M, Forsen TJ, Laakso M,Osmond C, Barker DJ. 2002. The effects of thePro12Ala polymorphism of the peroxisome proliferator-

activated receptor-g2 gene on insulin sensitivity andinsulin metabolism interact with size at birth. Diabetes51:23212324.

Eshel I, Matessi C. 1998. Canalization, genetic assimilationand preadaptation: A quantitative genetic model. Genet-ics 149:21192133.

Forsdahl A. 1977. Are poor living conditions in childhoodand adolescence an important risk factor for arterioscle-rotic heart disease? Br J Prev Soc Med 31:9195.

Francois I, de Zegher F, Spiessens C, DHooghe TM, Van-derschueren D. 1997. Low birth weight and subsequentmale subfertility. Pediatr Res 42:899901.

Gate JJ, Clarke L, Lomax MA, Symonds ME. 1999. Chroniccold exposure has no effect on brown adipose tissue innewborn lambs born to well-fed ewes. Reprod Fertil Dev11:415418.

Gilbert SF. 2001. Ecological developmental biology: Devel-opmental biology meets the real world. Dev Biol 233:112.

Gilbert WM, Nesbitt TS, Danielsen B. 1999. Childbearingbeyond age 40: Pregnancy outcome in 24,032 cases. ObstetGynecol 93:914.

Gluckman PD. 1995. The endocrine regulation of fetalgrowth in late gestation: The role of insulin-like growthfactors. J Clin Endocrinol Metab 80:10471050.

Gluckman PD, Beedle AS, Hanson MA, Vickers MH. Leptinreversal of the metabolic phenotypeevidence for the roleof developmental plasticity in the development of the met-abolic syndrome. Horm Res (in press).

Gluckman PD, Cutfield W, Hofman P, Hanson MA. 2005a.The fetal, neonatal and infant environmentsThe long-term consequences for disease risk. Early Hum Dev 81:.

5159.Gluckman PD, Hanson MA. 2004a. Maternal constraint offetal growth and its consequences. Semin Fetal NeonatalMed 9:419425.

Gluckman PD, Hanson MA. 2004b. The developmental ori-gins of the metabolic syndrome. Trends Endocrinol Metab15:183187.

Gluckman PD, Hanson MA. 2004c. Living with the past:Evolution, development, and patterns of disease. Science305:17331736.

Gluckman PD, Hanson MA. 2005. The Fetal Matrix: Evolu-tion, Development, and Disease. Cambridge: CambridgeUniversity Press.

Gluckman PD, Hanson MA, editors. 2006a. DevelopmentalOrigins of Health and Disease. Cambridge: CambridgeUniversity Press.

Gluckman PD, Hanson MA. 2006b. Evolution, developmentand timing of puberty. Trends Endocrinol Metab 17:712.

Gluckman PD, Hanson MA, Morton SMB, Pinal CS. 2005b.Life-long echoesA critical analysis of the developmentalorigins of adult disease model. Biol Neonate 87:127139.

Gluckman PD, Hanson MA, Spencer HG. 2005c. Predictiveadaptive responses and human evolution. Trends Ecol

Evol 20:527533.Gluckman PD, Hanson MA, Spencer HG, Bateson P. 2005d.

Environmental influences during development and theirlater consequences for health and disease: Implicationsfor the interpretation of empirical studies. Proc R Soc B272:671677.

Gluckman PD, Harding JE. 1997. Fetal growth retarda-tionUnderlying endocrine mechanisms and postnatalconsequences. Acta Paediatr Scand 86:6972.

Gluckman PD, Pinal CS. 2003. Regulation of fetal growthby the somatotrophic axis. J Nutr 133:1741S1746S.

Godfrey K. 2006. The developmental origins hypothesis:Epidemiology. In: Gluckman PD, Hanson MA, editors. De-velopmental Origins of Health and Disease. Cambridge:Cambridge University Press. pp 632.

Haig D, Graham C. 1991. Genomic imprinting and the

strange case of the insulin-like growth factor II receptor.Cell 64:10451046.

Hales CN, Barker DJ. 2001. The thrifty phenotype hypoth-esis. Br Med Bull 60:520.

Hales CN, Barker DJ, Clark PM, Cox LJ, Fall C, Osmond C,Winter PD. 1991. Fetal and infant growth and impairedglucose tolerance at age 64. BMJ 303:10191022.

Hales CN, Ozanne SE. 2003. The dangerous road of catch-up growth. J Physiol 547:510.

Hardy R, Wadsworth MEJ, Langenberg C, Kuh D. 2004.Birthweight, childhood growth, and blood pressure at 43years in a British cohort. Int J Epidemiol 33:121129.

Heindel JJ, Lawler C. 2006. Role of exposure to environmentalchemicals in developmental origins of health and disease.In: Gluckman PD, Hanson MA, editors. DevelopmentalOrigins of Health and Disease. Cambridge: Cambridge

University Press. pp 8297.Hinchliffe SA, Lynch MRJ, Sargent PH, Howard CV, Van-Velzen D. 1992. The effect of intrauterine growth retar-dation on the development of renal nephrons. Br J ObstetGynaecol 99:296301.

Hofman PL, Regan F, Jackson WE, Jefferies C, Knight DB,Robinson EM, Cutfield WS. 2004. Premature birth andlater insulin resistance. N Engl J Med 351:21792186.

Ibanez L, Ferrer A, Marcos MV, Hierro FR, de Zegher F.2000. Early puberty: Rapid progression and reduced finalheight in girls with low birth weight. Pediatrics 106:7274.

Ibanez L, Ong K, Dunger DB, de Zegher F. 2006. Early de-velopment of adiposity and insulin resistance followingcatch-up weight gain in small-for-gestational-age chil-dren. J Clin Endocrinol Metab 91:21532158.

Ibanez L, Potau N, Enriquez G, Marcos MV, DeZegher F.2003. Hypergonadotrophinaemia with reduced uterineand ovarian size in women born small-for-gestational-age. Hum Reprod 18:15651569.

Islam MS, Sjoholm A , Emilsson V. 2000. Fetal pancreaticislets express functional leptin receptors and leptin stim-ulates proliferation of fetal islet cells. Int J Obes RelatMetab Disord 24:12461253.

Jablonka E, Oborny B, Molnar I, Kisdi E, Hofbauer J,Czaran T. 1995. The adaptive advantage of phenotypicmemory in changing environments. Philos Trans R SocLond B 350:133141.

Jasienska G, Ziomkiewicz A, Lipson SF, Thune I, Ellison PT.2006. High ponderal index at birth predicts high estradiollevels in adult women. Am J Hum Biol 18:133140.

Jimenez-Chillaron JC, Hernandez-Valencia M, Lightner A,

Faucette RR, Reamer C, Przybyla R, Ruest S, Barry K,Otis JP, Patti ME. 2006. Reductions in caloric intake and




18/19

early postnatal growth prevent glucose intolerance andobesity associated with low birthweight. Diabetologia 49:19741984.

Jordan KM, Syddall H, Dennison EM, Cooper C, Arden NK.2005. Birthweight, vitamin D receptor gene polymorphism,

and risk of lumbar spine osteoarthritis. J Rheumatol 32:678683.

Kaati G, Bygren LO, Edvinsson S. 2002. Cardiovascular anddiabetes mortality determined by nutrition during parentsand grandparents slow growth period. Eur J Hum Genet10:682688.

Kajantie E, Rautanen A, Kere J, Andersson S, Yliharsila H,Osmond C, Barker DJP, Forsen T, Eriksson J. 2004. Theeffects of the ACE gene insertion/deletion polymorphismon glucose tolerance and insulin secretion in elderly peo-ple are modified by birth weight. J Clin Endocrinol Metab89:57385741.

Kajantie E, Osmond C, Barker DJP, Forsen T, PhillipsDIW, Eriksson JG. 2005. Size at birth as a predictor ofmortality in adulthood: A follow-up of 350 000 person-years. Int J Epidemiol 34:655663.

Karlberg J. 2002. Secular trends in pubertal development.Horm Res 57 (Suppl 2):1930.

Kirkwood TBL. 2002. Evolution of ageing. Mech AgeingDev 123:737745.

Kubaszek A, Markkanen A, Eriksson JG, Forsen T,Osmond C, Barker DJP, Laakso M. 2004. The associationof the K121Q polymorphism of the plasma cell glycopro-tein-1 gene with type 2 diabetes and hypertension de-pends on size at birth. J Clin Endocrinol Metab 89:20442047.

Kuzawa CW. 2005. Fetal origins of developmental plastic-ity: Are fetal cues reliable predictors of future nutritionalenvironments? Am J Hum Biol 17:521.

Kuzawa CW, Pike IL, editors. 2005. Fetal origins of devel-opmental plasticity (Wiley-Liss Plenary Symposium). AmJ Hum Biol 17:1118.

Kwong WY, Wild AE, Roberts P, Willis AC, Fleming TP.2000. Maternal undernutrition during the preimplanta-tion period of rat development causes blastocyst abnor-malities and programming of postnatal hypertension.Development 127:41954202.

Langley-Evans SC, Langley-Evans AJ, Marchand MC.2003. Nutritional programming of blood pressure and re-nal morphology. Arch Physiol Biochem 111:816.

Lee TM, Spears N, Tuthill CR, Zucker I. 1989. Maternalmelatonin treatment influences rates of neonatal devel-opment of meadow vole pups. Biol Reprod 40:495502.

Lee TM, Zucker I. 1988. Vole infant development is influ-enced perinatally by maternal photoperiodic history. AmJ Physiol 255:R831R838.

Leimar O, Hammerstein P, van Dooren TJM. 2006. A newperspective on developmental plasticity and the princi-

ples of adaptive morph determination. Am Nat 167:367376.Lelievre-Pegorier M, Vilar J, Ferrier ML, Moreau E,

Freund N, Gilbert T, Merlet-Benichou C. 1998. Mild vita-min A deficiency leads to inborn nephron deficit in therat. Kidney Int 54:14551462.

Lillycrop KA, Phillips ES, Jackson AA, Hanson MA, BurdgeGC. 2005. Dietary protein restriction of pregnant ratsinduces and folic acid supplementation prevents epige-netic modification of hepatic gene expression in the off-spring. J Nutr 135:13821386.

Lucas A. 1991. Programming by early nutrition in man.Ciba Found Symp 156:3850.

Lucas A, Morley R, Cole TJ. 1998. Randomised trial of earlydiet in preterm babies and later intelligence quotient.BMJ 317:14811487.

Mallard C, Loeliger M, Copolov D, Rees S. 2000. Reducednumber of neurons in the hippocampus and the cerebel-

lum in the postnatal guinea-pig following intrauterinegrowth-restriction. Neuroscience 100:327333.

McCarton CM, Wallace IF, Divon M, Vaughan HG Jr. 1996.Cognitive and neurologic development of the premature,small for gestational age infant through age 6: Compari-

son by birth weight and gestational age. Pediatrics 98:11671178.

McMillen IC, Robinson JS. 2005. Developmental origins ofthe metabolic syndrome: Prediction, plasticity, and pro-gramming. Physiol Rev 85:571633.

Mericq V, Ong KK, Bazaes RA, Pena V, Avila A, Salazar T,Soto N, Iniguez G, Dunger DB. 2005. Longitudinalchanges in insulin sensitivity and secretion from birth toage three years in small- and appropriate-for-gesta-tional-age children. Diabetologia 48:26092614.

Moran NA. 1992. The evolutionary maintenance of alterna-tive phenotypes. Am Nat 139:971989.

Morton NE. 1955. The inheritance of human birth weight.Ann Hum Genet 20:125 134.

Muehlenbein MP, Bribiescas RG. 2005. Testosterone-medi-ated immune functions and male life histories. Am J Hum

Biol 17:527558.Neel JV. 1962. Diabetes mellitus: A thrifty genotype rendered

detrimental by progress? Am J Hum Genet 14:353362.Neggers Y, Goldenberg RL. 2003. Some thoughts on body

mass index, micronutrient intakes, and pregnancy out-come. J Nutr 133:1737S1740S.

Negus NC, Berger PJ, Pinter AJ. 1992. Phenotypic plastic-ity of the montane vole (Microtus montanus) in unpre-dictable environments. Can J Zool 70:21212124.

Newman RA. 1992. Adaptive plasticity in amphibian meta-morphosis. BioScience 42:671678.

ORegan D, Welberg LL, Holmes MC, Seckl JR. 2001. Glu-cocorticoid programming of pituitary-adrenal function:Mechanisms and physiological consequences. Semin Neo-natol 6:319329.

Ounsted M. 1965. Maternal constraint of foetal growth in

man. Dev Med Child Neurol 7:479491.Ounsted M, Ounsted C. 1973. On Fetal Growth Rate: ItsVariations and Their Consequences. London: HeinemannMedical Books.

Painter RC, Roseboom TJ, Bleker OP. 2005. Prenatal expo-sure to the Dutch famine and disease in later life: Anoverview. Reprod Toxicol 20:345352.

Pembrey ME, Bygren LO, Kaati G, Edvinsson S, North-stone K, Sjostrom M, Golding J, the ALSPAC study team.2006. Sex-specific, male-line transgenerational responsesin humans. Eur J Hum Genet 14:159166.

Phillips DI, Barker DJ, Fall CH, Seckl JR, Whorwood CB,Wood PJ, Walker BR. 1998. Elevated plasma cortisol con-centrations: A link between low birth weight and the in-sulin resistance syndrome? J Clin Endocrinol Metab 83:757760.

Pike IL. 2006. Maternal stress and fetal responses: Evolu-tionary perspectives on preterm delivery. Am J Hum Biol17:5565.

Plagemann A, Waas T, Harder T, Rittel F, Ziska T, Rohde W.2000. Hypothalamic neuropeptide Y levels in weaning off-spring of low-protein malnourished mother rats. Neuro-peptides 34:16.

Potts R. Environmental hypotheses of hominin evolution.1998. Yearbk Phys Anthropol 41:93136.

Queitsch C, Sangster TA, Lindquist S. 2002. Hsp90 as a ca-pacitor of phenotypic variation. Nature 417:618624.

Relyea R, Hoverman JT. 2003. The impact of larval preda-tors and competitors on the morphology and fitness of ju-venile treefrogs. Oecologia 134:596604.

Ross M, Desai M, Guerra C, Wang S. 2005. Prenatal pro-gramming of hypernatremia and hypertension in neonatal

lambs. Am J Physiol Regul Integr Comp Physiol 288:R97R103.

18 GLUCKMAN ET AL.



19/19

Sayer A, Cooper C. 2005. Fetal programming of body com-position and musculoskeletal development. Early HumDev 81:735744.

Schlichting CD, Pigliucci M. 1998. Phenotypic Evolution: AReaction Norm Perspective. Sunderland, MA: Sinauer

Associates.Schmalhausen IL. 1949. Factors of Evolution. New York:

Documents

Early Life Events and Their Consequences for Later Disease a Life History and Evolutionary Perspective