Contents - Previous - Next

2. Discussion

There is good evidence that size at birth is associated with blood pressure, diabetes and coronary heart disease in adult life. These associations appear to be real and cannot be explained by confounding with socio-economic factors. These results raise many new and exciting issues. They are challenging the way in which we conceptualise the etiology of these conditions, which has usually been in terms of factors operating in adult life alone.

The fetal origins hypothesis also requires us to reorientate our thinking about fetal growth and its impairment. Perinatal epidemiology has been principally concerned with studying the immediate or relatively short-term consequences of variations in fetal growth. Reflecting the imperatives of obstetric and neonatal clinical practice, it has also tended to focus on determinants of relatively extreme perinatal outcomes that have recognisable and definable pathologies at birth or in infancy. As summarised above, however, the variation in risk of adult disease in relation to size at birth appears to be one that is across the entire range of size at birth, and is not principally a result of contrasts in risk between the pathologically small and the remainder.

2.1 Impaired fetal growth: definition and measurement

Most of the studies described above have only a very limited range of measures of perinatal outcome and pregnancy, many being restricted to birth weight alone. Relatively few of the studies of adult outcomes have data on gestational age. Where it has been possible to examine the effect of gestational age, it appears that the birth weight effects are most pronounced among those born at term. From this we can conclude that it is variation in growth rate rather than prematurity that is important.

Ponderal index and placental weight have both been looked at in a number of studies. Ponderal index appears to be more closely related to impaired glucose tolerance than birth weight. However, associations of placental weight or placental to birth weight ratio in relation to blood pressure have not found any consistent patterns. Barker and colleagues, have also explored a range of other size at birth parameters, including head and abdominal circumferences. The notion of 'head sparing' has been advanced by this group as an explanation for their finding that disproportions between head circumference and birth weight or length appear to be predictive of later outcomes. However, these findings have not been replicated by other groups.

It has been hypothesised that growth impairment in utero at different stages of gestation results in differences in size and shape at birth and in different adult pathologies. David Barker has gone as far as to speculate on the specific short and long-term consequences of "fetal undernutrition" in the first, second and third trimesters (Barker, 1995). He has suggested that undernutrition in the first trimester leads to proportionately small babies that are prone to haemorrhagic stroke, while undernutrition in the second or third trimesters leads respectively to thin babies prone to coronary heart disease and short babies at increased risk of coronary heart disease and thrombotic stroke. Although this framework provides a useful basis for discussion, direct evidence for these gestation-specific long-term effects is going to be difficult to obtain in observational studies of human populations. It is striking that even in the field of peri-natal epidemiology, the effects of malnourishment at different stages of gestation are poorly understood. It would, nevertheless, be worthwhile to conduct a systematic review of the extent to which size and shape at birth are indeed informative about the stage in gestation at which fetal undernutrition or growth impairment has occurred.

2.2 Possible explanations and mechanisms

If the associations between fetal growth impairment and later disease are indeed real, there are several explanations and mechanisms that require further elucidation. These are discussed briefly below.

2.2.1 Genetic factors There is a prima facie case to answer that common genetic factors may underlie at least part of the associations observed between size at birth and later disease. This is particularly so in the case of the associations seen with diabetes. Insulin plays a key role in fetal growth, and defects in its action are thought to be responsible for the association between size at birth and diabetes. It is thus plausible that a genetically determined defect in insulin action could lead to this association. Recent results from a study of Danish monozygotic twins, however, are not consistent with this explanation, as they showed a greater risk of non-insulin dependent diabetes in the twin who was lighter at birth (Vaag et al, 1996).

If a common genetic factor does underlie some of these associations, this requires size at birth to have an appreciable genetic component. In the perinatal literature, genetic factors are generally assumed to play a minor role in determining birth weight, it being frequently stated that less than 20% of the variance in birth weight can be accounted for by genes (Robson, 1978). Relatively little work has been done in recent years to confirm this consensus estimate based on studies conducted prior to the mid-1970s. However, one exception is the work of Magnus from Norway (Magnus, 1984 a,b) which suggests that up to 40% of the variation in birth weight could be due to genetic factors. It would be desirable for this issue to be revisited in order to try and reconcile these larger estimates from those usually quoted.

One strategy that is being used in a number of studies to address this question is to collect data on genotypes for candidate loci such as the IGF-1 and IGF-2 genes. This information, in conjunction with data on size at birth and adult phenotype/disease risk, should provide important insights.

2.2.2 Nutrition of the mother and fetus On the assumption that the associations between size at birth and later disease are a consequence, at least in part, of the in utero environment, there is a need to clarify the extent to which this can be equated with fetal or maternal nutrition. There has been a tendency in the work of the Southampton group to emphasise the importance of adequate maternal nutrition as a public health implication of this research. However, in general terms, as discussed in other papers presented at the meeting, the link between maternal nutrition and fetal growth is a complex one. Certainly, maternal diet in pregnancy appears to have only a limited impact on fetal growth, although pre-pregnancy weight is clearly important.

The nutritional milieu of the fetus is of course ultimately dependent upon the mother. However, it is also dependent upon placental function which plays an important role not just in being a conduit between the mother and fetus, but also has major effects upon the mother's physiology through its endocrine capacities (Haig, 1993). As illustrated in the down regulation of growth found in the offspring of mothers with severe diabetes in pregnancy (Aerts et al, 1996), the relationship between fetal growth and nutritional substrate is not a simple one of "the more the better". Additionally, studies of the Pima Indians suggest that exposure to abnormally high levels of glucose in utero actually predispose to diabetes in the offspring through non-genetic mechanisms (Pettitt et al, 1986).

2.3 'Programming': permanent effects on structure and function

The current term used to describe the notion of in utero environmental determinants of later disease is 'programming'. Lucas (1991) is one of the few people who have attempted to define this concept with any rigour. He defines 'programming' in the context of the fetal origins hypothesis as two things:

(i) induction, deletion, or impaired development of a permanent somatic structure as a result of a stimulus or insult operating at a critical period;

(ii) physiological 'setting' by cm early stimulus or insult at a 'sensitive' period, resulting in long-term consequences for function.

Earlier work concerned with the long-term consequences of gestational diabetes on the health of the offspring foreshadowed many of those ideas. It is particularly interesting to compare the contemporary definitions of programming with Freinkel's concept of fuel-mediated teratogenesis (Freinkel and Metzger, 1979; Freinkel, 1980) that was defined as:

"...permanent changes in habitue (that is, anthropometric modifications) or in endocrine or neuroendocrine metabolism by abnormal fuel presentations during the period of intrauterine development of the terminally differentiated cells that determine these functions" (Freinkel, 1980, p 1032)

Both definitions incorporate the idea of permanent changes in structure or function of an organism consequent upon specific environmental conditions in utero at particular stages of gestation.

One obvious example of this type of programming would be if fetal growth impairment resulted in suboptimal numbers of cells or structures that in adult life might be expressed as increased susceptibility to disease. Deficits in numbers of pancreatic beta-cells or in numbers of nephrons in the kidney at birth could be seen to predispose to diabetes (Aerts et al, 1996) and hypertension (Mackenzie and Brenner, 1995) respectively, as it is believed that the total complement of these structures is fixed in utero.

Animal models in which manipulation of maternal nutrition or physiology has resulted in permanent changes in the physiological characteristics of the offspring have been developed. These include rat models showing an association between level of protein in the maternal diet and blood pressure (Langley and Jackson, 1994) and pancreatic function (Snoeck et al, 1990) in the offspring and the induction of abnormalities in insulin action and responsiveness in the offspring of mothers made diabetic in pregnancy - an effect that appears to last over several generations (Aerts and Van Assche, 1979).

2.4 Implications and research priorities

At the present time the fetal origins hypothesis should be regarded mainly as an area for basic big-medical research. It is difficult at this stage to draw strong public health conclusions with the current limited state of knowledge.

What is relatively clear, however, is that any disease susceptibilities induce in utero are likely to be modulated by adult circumstances and characteristics, in particular obesity. Studies have already found evidence that the associations of size at birth with blood pressure (Leon et al, 1996), non-insulin dependent diabetes (Lithell et al, 1996) and coronary heart disease (Gunnell et al, in press) are strongest among those who are relatively obese. If this is correct, as seems very plausible, drawing rigid distinctions between fetal and later life effects may prove to be as sterile and unproductive as the pursuit of a rigid nature/nurture dichotomy. The interesting public health questions are likely to revolve around how early and later life factors interact.

From a global perspective, one of the really pressing issues raised by the fetal origins hypothesis concerns the long-term consequences of fetal growth in many areas of the world where mothers are living at sub-optimal levels of nutrition. This issue is made all the more important if there is indeed an interaction with adult life style factors. It may be that nutritional impairment in utero due to poor nutritional plane of the mother, coupled with the development of energy dense and westernised diet and life style in adult life could lead to particularly adverse health trends. This combination characterises the life-trajectories of many people in developing and industrialising countries today. It is thus a priority to undertake research on the fetal origins hypothesis in populations that are undergoing such health transitions.

Finally, if we are at the stage of clarifying the basic scientific issues raised by the fetal origins hypothesis, we need to be clear about the priorities for research. In addition to the already existing work on animal models in which the effects on the offspring of manipulation of maternal diet and physiology are studied, other models should be developed to look at the determinants of cell number and function and the modulation of this by environmental determinants in utero. With respect to epidemiological work in humans, it is now important to conduct more sophisticated studies that allow dissection of the separate influences of well known determinants of size at birth (eg parity, maternal age, maternal constraint and maternal adiposity) on later life characteristics of the offspring such as blood pressure.

Contents - Previous - Next