This study shows that children who experience severe acute malnutrition and develop kwashiorkor, marasmic-kwashiorkor or marasmus have pre-existing developmental differences at birth, as reflected by substantial differences in birth weight. This observation was consistent over the 30 years in the population under study.
This finding provides a developmental explanation for the disparate syndromes of severe malnutrition. It implies that underlying developmental attributes play a role in determining which syndrome is expressed when children are exposed to severe undernutrition. We propose that the developmental environment induces plastic changes affecting the development of metabolic control that can manifest as distinctly different phenotypes under extreme conditions. In the rodent, very different metabolic phenotypes with associated epigenetic changes can be induced by different prenatal nutritional conditions and these offspring respond very differently to postnatal manipulations such as leptin administration 
or exposure to high fat diets 
. Persistent epigenetic changes into adulthood have been reported in people exposed to famine in utero
. The epigenetic state of children in a Western population at birth can be related to mothers' nutritional intake during pregnancy and in turn to later patterns of body compositional development 
It has been suggested that developmental plasticity is maintained in mammals to allow the fetus and infant to adjust their development to anticipated future nutritional and other environments. Two classes of developmentally plastic response have been described 
. The first is immediately required to promote survival of the disadvantaged fetus; growth retardation is such a response. The second class is delayed or predicted responses where the developmental trajectory is altered for predicted advantage to promote survival until puberty and the opportunity to reproduce. The latter class, referred to as a predictive adaptive response, may become disadvantageous later in life if the prediction, based on inadequate transplacental nutrition, is for a low nutritional environment and the offspring faces nutritional excess. However, when the disadvantage presented by such an environment occurs after the reproductive phase, it is likely to be essentially invisible in evolutionary terms. Both classes of response can be induced in the same individual, giving rise to the often reported association between low birthweight and later disease risk. There is considerable experimental support for such a model but its applicability to humans has not been previously directly demonstrated.
The differences in metabolic and clinical attributes between marasmus and kwashiorkor provide a clear demonstration of the pre-pubertal fitness advantage of a predictive adaptive response induced by poor transplacental nutrition, as indicated by lower birthweight. The children born with lower birthweight had a metabolic response to severe undernutrition that aided survival, resulting in the more benign syndrome of marasmus. This finding provides the first direct evidence in humans in support of the fitness-enhancing effects in childhood of anticipatory responses in utero.
Nineteen million children worldwide every year become severely wasted 
. Some of these present with marasmus. If those who develop this syndrome of wasting without oedema possess similar metabolic characteristics as observed in our studies, then we propose that they represent a group with a higher risk of developing obesity and co-morbidities in later childhood, adolescence and adulthood upon exposure to a liberal dietary energy intake 
. Across the globe such exposure is almost ubiquitous, and exists even in poor countries because of the impacts that globalization and economic transformation have on making food energy more available at an affordable price to almost all their population 
. On the other hand, individuals with the metabolic architecture seen in kwashiorkor patients are at risk of developing the oedematous syndrome when subjected to severe undernutrition in childhood. Thus, in famine situations, greater attention may need to be paid to children of higher birthweight as they may be at greater risk. We further propose that survivors of kwashiorkor will have a differential and lesser risk of obesity and its cardio-metabolic co-morbid pathologies in adult life.
As this was a retrospective cohort there is a danger of misclassification. However, we sought to reduce this bias by reviewing all admission clinical findings and derived independent diagnoses of kwashiorkor, marasmus and marasmic-kwashiorkor using the Wellcome Classification. Birthweight recall by mothers of the subjects when subjects had been admitted as children might have been inaccurate. However, the validity of such recall has been established in our setting (13). Errors of recall are likely to occur in a way that does not depend on diagnosis, and so are likely to blunt rather than generate birth weight differences between diagnoses. Because the syndromes use weight at admission in their definition, the observed association between syndrome and birth weight may result from a mixture of two effects – an in utero developmental effect and the tracking of weight as children age. However, using the presence of oedema alone as the basis of analysis gives a similar conclusion suggesting that the finding reported here is not due to any confounding effect of birthweight on later weight or height gain.