2. REVIEW OF THE LITERATURE

2.1 Fetal growth - genes and the environment

Fetal growth is dependent upon various external factors. The exact influences of genes on fetal growth are still to a large extent not known. According to some studies birth weight is mainly attributable to the maternal constitution and environmental factors (8-10). The correlation coefficient for birth weight between monozygotic twins is 0.67 and for dizygotic twins 0.59 and the genetic contribution to this can therefore not be more than twice the difference of these two coefficients i.e. 0.16 (10). However, more recent studies have pointed out the importance of genes and suggest that these are responsible for a large part of the variation in birth weight (9, 11-16). Therefore, one can state that both genetic and environmental factors influence fetal growth.

The environment in which the pregnant mother lives may determine whether genetic or environmental influences play a more important role. An environmental strain may influence metabolism and lead to maternal constraint of fetal growth (17-20). When environmental strain is absent and the constrain is relaxed, other factors also take up more of the variance (18).

2.1.1 Fetal growth retardation

For the time being we lack specific measures of fetal growth retardation. There are no reliable diagnostic tests distinguishing malnourished fetuses from true prematurity and those with fetal growth retardation due to other causes. Therefore, anthropometric measures and their ratios in relation to length of gestation have been used to measure fetal growth. In the following some more commonly used anthropometric measures of fetal growth will be presented.

Birth weight is a rather crude measure of fetal growth. The measurement techniques are relatively easily standardised and birth weight is often available even in older records. In absence of other accurate measures of fetal growth it has therefore been used in many studies. Recall of birth weight and other obstetric measures has proven to be unbiased and can therefore also be used (21, 22).

Birth length is more difficult to measure adequately than birth weight, leading to more variation in this measurement (23). The measuring difficulties and the relatively small variations in birth length, indicate that birth length in itself is not a very good indicator of fetal growth. Ponderal index (birth weight/birth length3) is the close equivalent of body mass index (BMI=weight/length 2) in adults. This is a measure of thinness at birth and is thought to evaluate the ratio of soft tissue to bone mass (24).

The placenta is the gateway between the mother and the fetus. Changes in placental function i.e. in permeability will inevitably affect the nutritional supply of the fetus. Placental weight and size (area) have therefore also been used as a measure of fetal growth. In addition several other measurements and ratios have been used to measure fetal growth e.g. placental to birth weight ratio, head circumference to length at birth ratio and head circumference to birth weight ratio. None of these has however proved to be more specific and reliable than birth weight, the ponderal index at birth or placental weight.

2.1.1.1. Proportional and disproportionate growth

Impaired fetal growth can be divided into proportionate and disproportionate growth retardation (24). In proportionate growth retardation the fetus is symmetrically small i.e. the ratios between weight, head circumference and length stay the same. In disproportionate growth retardation, the ratios between head, weight and length are changed. Babies may therefore be thin (small weight compared to its length), short (short length in relation to its head circumference) or short and fat (short length in relation to its head circumference as well as being fat).

Proportionate growth retardation can be caused by severe malnutrition. Lately more attention has been paid to disproportionate growth since disproportionate growth is thought to be a better measure of nutritional supply in-utero during critical periods (25). Consistent with this, birth weight and body fat are affected more than head size and length in malnutrition (26). Disproportionate growth retardation could be a result of blood diversion and brain-sparing mechanism in-utero which may lead to thinness and shortness at birth (7, 27). Thinness, as measured by a low ponderal index at birth, is a measure of disproportionate growth which has been quite widely used.

Disproportionate growth can be induced by maternal low protein diet in animals (28). However, less data is available in humans and Zlatnik et. al. found no relationship in humans between protein intake during pregnancy and disproportionate growth (29). The significance of disproportionate growth as a measure of growth retardation with a nutritional aetiology causing later disease, has therefore been questioned (30).

2.1.2 Maternal nutrition and the fetus

Relatively little is known about the maternal influences that alter fetal and placental growth in humans (31, 32). Generally, nutrition during pregnancy is considered to be reasonably adequate in Finland today and severe malnutrition is rare. Recommendations concerning diet are commonly based on preventing disease in the mother during pregnancy, obstetric complications ( e.g. eclampsia, labour complications or gestational diabetes) or applying recommendations that have been obtained by studying diseases in non-pregnant women. The long term effects on the offspring of the diet recommended to pregnant mothers are however not well known.

Deficient nutrition leads to growth retardation. In rats, growth retardation and disproportionate growth can be induced by protein restriction. The weight of the pancreas, muscle and liver (the key organs in glucose metabolism) is hereby reduced more than the size of the brain, lung, heart, kidney or thymus (33). Fetal growth retardation could also be induced by other factors than protein restriction e.g. over-expression of the insulin like growth factor binding protein-1 protein causes lowering of birth weight and fasting hyperglycaemia (34).

In rats, dietary changes induce not only growth retardation but also persistent changes in metabolism. A maternal low protein diet has been shown to cause hypertension and impaired glucose tolerance in the offspring (28, 35-37). These alterations in glucose tolerance are permanent and the effects are transmitted through several generations (38). Intrauterine malnutrition also leads to a marked fat accumulation and aberrations in the sympathetic activity (39). Overfeeding, on the other hand, with a high fat maternal diet in pregnancy leads to alterations in cholesterol metabolism (40).

In humans, severe malnutrition is clearly harmful and reduces birth weight, placental weight, length at birth, and head circumference at birth as well as causes disease in the fetus (41, 42). The size of the fetus is affected the most from famine in the third trimester and this effect is not mediated by the shortening of gestation (43). Severe malnutrition is practically non-existent in Finland today, where an excessive intake of calories and obesity is common. The nutritional focus should be laid on the composition of the food since calorie intake is usually adequate. At the time when the children in this study cohort were born (1924-33) severe malnutrition was probably more common.

Maternal pre-pregnancy weight is a marker of the maternal nutritional status which influences pregnancy outcome. Low maternal pre-pregnancy weight is associated with lower birth weight (32). However, since obesity in the mother has been shown to increase the number of adverse events in pregnancy, not all women benefit from attempts to increase pre-pregnancy weight (44). Mothers, who themselves were of low birth weight have an increased risk of having a low birth weight baby and interventions to reduce the risks for adverse pregnancy outcomes should attempt to optimise pre-pregnant weight (45, 46).

The role of vitamins and amino acids in regulating fetal growth is not entirely clear. Low levels of vitamin B, magnesium, essential (fetal and maternal) amino acids (lysine, valine, glycine and threonine) and alpha globulin have been linked to fetal growth retardation (47-49). Although the relative influence of each nutrient has not been entirely clarified, the composition of organs and the placenta can be changed by altering the diet and its composition. In humans, a high carbohydrate intake in early pregnancy suppresses placental growth, if combined with a low dairy protein intake in late pregnancy (50). Small-for-date babies do not have any absolute deficit in the number of cells in kidney, heart or liver, it is rather the size of the cells which is affected by the growth retardation (51).

Birth weight is increased in babies born to mothers who eat fish frequently (52, 53). It has been speculated that the marine n-3 fatty acids may be the cause of the increase in birth weight, possibly by prolonging gestation (53, 54) or by increasing the ratio of thromboxanes to prostacyclins which improves placental blood flow (55, 56). However, a more recent study did not show any influence of fish oils on fetal growth (57). Birth weight can also be increased by manipulating other nutrients than seafood but maternal nutritional resources prior to pregnancy are important (58).

In order to establish the optimal diet for the fetus interventional follow-up studies are needed but results from earlier studies do however call for caution. In the 1970's in a controlled randomised trial in New York women received nutritional supplementation during pregnancy (59). A high protein supplementation increased the number of premature births and neonatal deaths. Although birth weight increased 41g in average with a controlled protein-calorie diet, this change was not significant. At one year of age no effects of the diet were seen. A high density protein supplementation may therefore not be favourable (60). Another study showed that nutrient supplementation during pregnancy leads to decreased birth weight, birth length and head circumference (61). Studies have also suggested that an elevated protein content in the diet during childhood leads to obesity and an increased risk of cardiovascular disease (62, 63). Not only protein intake during pregnancy has been implicated, a high intake of carbohydrate in late pregnancy combined with a low intake of dairy protein may also be a cause of thinness at birth (20).

Results from the Bacon Cow study suggest that only certain subgroups, which remain to be identified, would benefit from nutritional protein supplementation during pregnancy (64). Possibly, a therapeutic range exists for the optimal protein content and there may be nutritional windows where the fetus is extremely sensitive to alterations in the diet (65). Infudicious dietary restrictions during pregnancy are therefore ill-advised (49).

2.1.3 Regulation of intra-uterine growth

The growth of the fetus is a complex process which is yet not fully understood. We know that it is affected by a number of factors, among them nutrition, hormones, genes and growth factors . Correlations exist between birth weight and certain hormones, maternal glucose levels, amino acids, free fatty acids, and triglycerides but the relative influence of each factor on birth weight is still controversial (66). More is known about the risk factors for low birth weight and Table 1 shows some of the risk factors for low birth weight (67-74).

Table 1. Factors influencing fetal growth.

Parity
Maternal history of prior
low-birth weight infants
Maternal height
Maternal pre-pregnancy weight
Maternal birth weight
Gestational weight gain and caloric intake
Paternal weight and height
Retarded growth during first trimester
Maternal general morbidity
Episodic illness
Maternal stress
Racial/ethnic origin
Low social status
Cigarette smoking
Alcohol consumption

2.1.3.1 Hormones

Thyroid hormones stimulate protein synthesis and cell enlargement and are vital for fetal maturation (75). Thyroxin crosses the placenta and hence the fetus is also affected by maternal levels of the hormone. In the new-born baboon thyroid homeostasis may be reset by thyroxin in the mother's milk (76). In humans, the levels of total thyroxin correlate with the length of gestation but the level of fetal thyroxin seems to be unrelated to birth weight (77, 78). Another study showed that levels of cord thyroxin and thyroid stimulating hormone may be low in growth retarded fetuses (79). Maternal hyperthyroidism is associated with a higher risk of maternal, fetal, and neonatal complications (80).

Leptin is a recently discovered hormone which is closely related to food intake and obesity (81). It has also been implicated in the pathogenesis of Type 2 diabetes. Leptin is synthesised in utero and the concentrations in fetuses are similar to those in adults (82). Cord leptin, correlates positively with birth weight (82, 83). However, maternal leptin levels during pregnancy are poor predictors of fetal growth and therefore they may not play a role in fetal growth retardation (84).

Sex hormones are vital regulators of pregnancy and fetal growth. A lack of dihydroepiandrosterone may cause fetal growth retardation since dihydroepiandrosterone levels have been found to be lower in the smaller twin (85). Björntorp has proposed that high androgens levels may be a risk factor for Type 2 diabetes (86, 87). In support of this single testosterone injections in new-born rats lead to insulin resistance and changes in body fat distribution (88). There is also evidence that high estrogen levels during fetal life may increase the risk of breast cancer while the impact on cardiovascular disease is not known (89).

Glucocorticoids are catabolic hormones which have a negative influence on cell proliferation. The level of corticotropin releasing factor is increased in growth retarded fetuses and exposure to glucocorticoids results in an overall fetal growth retardation in both humans and animals (90-92). Glucocorticoids are therefore thought to have a role in programming adult disease. Cortisol also triggers maturation and can be used to improve survival in pre-term births. However, the long term effect of cortisol administration in humans is not known.

The insulin like growth factor-insulin axis is likely to have a substantial influence on fetal growth during the second and third trimester (79, 93). Insulin stimulates growth by several different mechanisms; increased uptake and utilization of nutrients, direct mitogenic actions and by increasing the release of other hormones and growth factors, such as insulin like growth factors and their binding proteins (93). Insulin is mainly related to fetal macrosomia (94, 95). It is the endogenous production of insulin that increases intrauterine growth rate since maternal insulin does not cross the placenta (95). Endogenous insulin is increased by maternal hyperglycaemia since glucose passes through the placenta. The insulin like growth factors interact with both insulin and growth hormone (96-98). Although growth hormone levels are high during pregnancy and growth hormone is correlated with birth weight, the greatest influence of growth hormone is thought to be on growth during adolescence and not on fetal growth (99-101).

Insulin like growth factors or somatomedins promote proliferation of a variety of cells and are essential for prenatal growth (102). Insulin like growth factor-I and insulin like growth factor binding protein-3 in the fetus are positively correlated with birth weight in humans (79, 94, 101, 103, 104). Although insulin like growth factor-II is secreted in far larger quantities by the fetus than insulin like growth factor-I, it is not involved in the control of fetal growth (101, 105, 106). However, the direct influence of high or low insulin like growth factor levels during fetal life on later disease are largely unknown.

2.2 Childhood growth

2.2.1 Factors influencing childhood growth

A model differentiating between three discrete, but related periods of postnatal growth: fetal/infant, childhood and pubertal phases, has been put forward by Karlberg (107). During these periods, growth is regulated by different hormonal control systems.

The infant period is commonly seen as a continuation of fetal life and the insulin like growth factor system is proposed to be the most important factor regulating growth. The exact timing when the growth hormone system takes precedence over the insulin like growth factor system in regulating growth is not known. Growth hormone is most important in regulating growth during adolescence but it can be modified by insulin like growth factor levels and nutritional availability (107, 108). Sex steroids dominate the pubertal growth period. During the points of intersection between these phases the child could be more sensitive to adverse influences (109).

Growth in childhood is affected by both environmental and genetic influences (110-113). Although childhood growth usually follows the growth trajectories, there are considerable inter-individual variations which cannot be attributed to any specific disease. The ability to predict weight at one year based upon birth weight, is therefore limited (114). Later in childhood the degree of obesity is more predictive since obese children tend to be obese later in life (115, 116) This tendency is stronger in men than in women (115, 117). If the parents of obese children also are obese, this more than doubles the risk of obesity of the child later in life. Obese children with non-obese parents are however at low risk for later obesity (116). On the other hand being short or underweight at the age of 7 predisposes to shortness and thinness in adult life (118).

An unfavourable intra-uterine environment is related to obesity in adolescence. Famine in utero during the two first trimesters predisposed to obesity among 300 000 men at the age of 19 years. However, famine in the third trimester was not related to subsequent obesity (119). High birth weight in those not exposed to famine has been proposed to be associated with obesity in adolescence but the predictive power is poor (120, 121).

The maximum attainable height is likely to be determined by individual genes but this height has not yet been reached since average height has increased for generations (122). The increase in height is therefore mainly due to an improvement in environmental factors (26). Evidence exists that short stature in adult life which is associated with an increased risk of CHD could be determined in utero (123-126). Stature is also influenced by nutritional intake in childhood. In support of this, the Boyd Orr's study of 3,762 pre-war children in Britain shows an association between family food expenditure, housing conditions and height in childhood (127). The adult height is also influenced by other factors such as social class, parity, smoking during pregnancy and mother's height (128).

Obesity in childhood is a predictor of an increased risk of mortality later in life (129). Paradoxically, both short and tall children have increased mortality (130). Many cardiovascular risk factors can be identified already in children, like obesity and severe insulin resistance (131, 132). These cardiovascular risk factors tend to persist from childhood into adult life and they often cluster in the same individuals (133). Not only the risk of cardiovascular disease is influenced by diet in childhood, childhood energy intake has also been linked to an increased risk of cancer (134).

2.2.2 Catch-up growth

A fetus experiencing growth retardation during pregnancy leading to a smaller size at birth, may achieve a similar body size compared with other children of the same age later in life. This is achieved through accelerated growth i.e. catch-up growth. In a Swedish study, 87% of the children born small for gestational age fully caught up with their peers within 2 years of life (135). Catch-up growth can be seen in twins where the co-twin leaner at birth, will catch up in weight already at the age of one year (136). However, not all children catch up and the risk of short stature is increased in growth retarded fetuses (137-139). Infants born small-for-date have a 7-fold increased risk of short stature in adolescence (139-141).

Despite the normalisation of childhood height and weight in growth retarded fetuses, there is evidence that catch-up during the nursing period is detrimental later in life (33, 65). In rats, the combination of prenatal undernutrition with retarded fetal growth, and good postnatal nutrition with accelerated growth, leads to striking reductions in life span (142). The cause of this is not known but may be due to the fact that fetal growth retardation leads to reduced cell numbers, which are subsequently overgrown due to the limited cell mass (143). Another possibility is that catch-up growth alters body composition in later life. Babies who are thin at birth have a smaller muscle mass (7). It is possible that if they develop a high BMI in childhood they have a disproportionately high fat mass. Persisting alterations in hormone levels ( e.g. insulin, glucocorticoids, insulin like growth factors and growth hormone) could also underlie the adverse effects of catch-up growth.

2.3 The fetal growth hypothesis

There are considerable variations in CHD incidence world-wide which cannot be explained by traditional risk factors (7). In Norway, Forsdahl observed a relationship between infant mortality and CHD, leading to the hypothesis that poor living conditions are a risk factor for CHD (144). Later studies have confirmed the relationship between infant mortality and CHD (145-147).

Forsdal's hypothesis triggered a number of studies trying to link socio-economic conditions to cardiovascular disease and adult health. Both positive (148-154) and non-conclusive (155-157) findings have been published. Supporting the hypothesis, is the finding that death rates among London civil servants were higher in those who are shorter in stature, which could be due to inferior early environment (158). Also, stroke and CHD correlate more strongly with neonatal mortality than postneonatal mortality, a finding which would indicate that intra-uterine factors are responsible for the increased risk of CHD (146, 159).

Traditionally the genetic component of diseases has been addressed with the help of twin studies. A greater concordance of diseases in monozygotic twins than in dizygotic twins has been taken as evidence of a genetic component of the disease. Since the concept of programming evolved, the validity of concordance in classical twin studies as a method for establishing the genetic component of disease has been questioned (160). Monozygotic twins are however less similar in birth weight than dizygotic and therefore the concordance of disease in monozygotic twins is not due to concordance in birth weight which questions the previous argument (161). The growth regulation is however more complicated in twins and might therefore differ from the ones acting in singleton pregnancies. There is evidence that the fetal origins hypothesis does not apply for the fetal growth retardation experienced by twins as the laws of Mendelian inheritance are more clearly followed when maternal constraint is not active (18, 162).

Barker and colleagues have proposed that impaired fetal growth, expressed as low birth weight and a low ponderal index at birth, is a risk factor for cardiovascular disease and the metabolic syndrome (159). According to the original hypothesis, the growth retardation is due to deficient nutrition. To date, the theory encompasses also other stimuli than malnutrition as the cause of growth retardation e.g. changes in the hormonal setting. The mechanism by which growth retardation or other stimuli can cause later disease has been called programming but the mechanisms for this are largely unknown (163).

2.4 CHD

2.4.1 Fetal growth and CHD

The first studies reporting an association between birth weight and CHD came from Hertfordshire and Sheffield, United Kingdom (164). In the cohort studies from Hertfordshire and Sheffield, CHD mortality among 13249 men decreased progressively with increasing birth weight (165, 166). The results from 5585 women in Hertfordshire were similar, although the relationships are not as strong as in men (167). Table 2 shows some of the studies reporting on the association between fetal growth and CHD.

The Caerphilly study including 1258 men recruited between 1979 and 1983 and followed up prospectively for 10 years, showed that birth weight was related to both fatal and non-fatal CHD. In addition fibrinogen showed an inverse relationship to birth weight. Social factors in childhood did not change these relationships and neither did socio-economic status (SES) or conventional risk factors in adulthood (168, 169).

In the Nurses' health study, a cohort of 70 297 women were followed from 1976 to 1992. Birth weights were assessed retrospectively through a questionnaire (170). The main outcome measures were non-fatal cardiovascular disease, coronary revascularisation, and stroke. The study showed an inverse relationship between birth weight and both CHD and stroke.

The Uppsala study of 14611 men had a 97% follow up over the entire life course of the cohort born 1915-29. Among these men, a 1000 g increase in birth weight led to a proportional reduction in the rate of CHD of 0.77 (0.67-0.90). No confounding effects of socio-economic factors were observed (171).

Some evidence against an association between birth weight and CHD has been found in a study from the Swedish twin registry (172). In this study twins of lower birth weight did not have increased mortality compared to singleton births of higher birth weight. The shorter twin did however have increased mortality compared to the taller one. The regulation of the growth in twins is more complicated than singletons, however, and this makes the results difficult to interpret.

The increased risk of CHD resulting from low birth weight could be mediated by classical CHD risk factors. Supporting this, birth weight was associated with the components of the metabolic syndrome in the San Antonio Heart Study of 564 young adults (173). Low birth weight is also associated with a higher waist to hip ratio (174).

The effects of an adverse intra-uterine milieu span longer than just one generation i.e. death rates of the current generations depend on the circumstances in early life of previous generations. The infant mortality in the 1920's and 1930's correlates with the CHD mortality of these birth cohorts later in life (144, 145). Also, if the infant's birth weight is small in relation to its mother's birth weight, this leads to an increased risk of mortality in the infant (175). Low birth weight is associated with increased risk of cardiovascular mortality later in life not only for the infant but also for its parents. This increased risk of mortality from cardiovascular disease and all causes in the parents cannot be explained by social, environmental, behavioural, and physiological risk factors (176).

The causes of the intergenerational effects are still not known. A recent study shows that mothers who themselves were born small-for-date have an increased risk of giving birth to an infant that is also small-for-date (177). Studies imply that early environmental programming of the hypothalamo-pituitary-adrenal axis in the mother during her own fetal life could affect hormone levels during pregnancy and thereby influence the next generation (25, 178). The relative importance of the genetic versus the environmental influences on the intergenerational effects remain to be determined.

2.4.2 Childhood growth and CHD

The negative effects of obesity on health and cardiovascular disease are well documented (179-181). Many CHD risk factors can be identified in children; obesity, insulin resistance and hyperinsulinaemia and they often cluster in the same individuals (131, 132). Obese children have increased levels of CHD risk factors and obesity is associated with increased mortality later in life (129, 182). Prevention of CHD by reducing obesity in young children is therefore of importance (183, 184).

Not only obesity affects the risk of CHD, also stature which is dependent on childhood growth, is related to overall CHD mortality (130). However, stature in childhood is related to socio-economic conditions as well as the well established CHD risk factors, e.g. cholesterol, blood pressure and dietary fat intake (185, 186). Therefore, the magnitude of the role of the socio-economic factors in childhood on CHD risk is uncertain (187).

Table 2. Studies investigating the relationship between fetal growth and CHD


Authors

n

Sex Country Age or year of birth Dependant variable Independent variable Direction of association Adjustment
Barker
(164)

5654

M UK 1911-30 birth weight birth weight fatal CHD fatal CHD ns inverse SMR weight at 1 year, breast feeding
Barker
(166)

1586

M UK 1907-24 birth weight ponderal fatal CVD fatal CVD inverse inverse SMR SMR

Frankel
(168)

1258

M UK 45-59 birth weight non-fatal or fatal CHD inverse age, adult BMI
Leon
(171)

14611

M Sweden 1915-29 birth weight ponderal fatal CHD fatal CHD inverse inverse - -
Martyn
(165)

13249

M UK 1907-30 birth weight fatal CHD inverse SMR
Osmond
(167)

15727

M+F UK 1911-30 birth weight birth weight birth weight birth weight fatal CHD fatal CVD fatal CHD fatal CVD ns inverse inverse inverse women, SMR women, SMR men, SMR men, SMR
Rich-Edwards
(170)

70297

F US 30-55 birth weight birth weight birth weight non-fatal MI non-fatal CHD CABG or PTCAa) ns inverse inverse multivariateb) multivariate b) multivariate b)
Stein
(421)

517

M+F India 38-60 birth weight non-fatal or fatal CHD or CABG inverse age, sex  

Ponderal= ponderal index, b)multivariate=adjusted for BMI, smoking, hypertension, cholesterol, family history of MI under 60, menopausal state, use of postmenopausal hormones, SMR= relationships expressed as standardised mortality ratios, UK = United Kingdom, US= United States, a)PTCA= percutaneous transluminal coronary angioplasty, CABG= coronary artery bypass graft, CVD= cardiovascular disease, ns= not significant, M= male, F= female.

2.5 Type 2 diabetes

Contrary to the decrease in CHD incidence in most western countries, the global incidence of diabetes is increasing rapidly. The estimated number of Type 2 diabetes subjects in Europe in 1994 was 16 million, a figure which is expected to rise to 24.4 million by the year of 2010 resulting in a 50% increase of Type 2 diabetes cases in 16 years (188).

Type 2 diabetes is caused by both environmental and genetic factors (189). The environmental effects on the incidence of diabetes have been estimated to be largely between 60% and 90% (190). Of these, obesity and a sedentary life style are the most important (191-195). However, twin and family studies have found proof of a strong genetic component with a high concordance rate of Type 2 diabetes among monozygotic twins (196-199).

Both insulin resistance as well as deficient insulin secretion are early metabolic disturbances in Type 2 diabetes and the severity of these increases in the course of the disease (192, 193, 200-202). There is still controversy whether insulin resistance or insulin secretion is the primary metabolic defect (203).

2.5.1 Fetal growth and Type 2 diabetes

The association between birth weight and Type 2 diabetes was first shown in the Hertfordshire study where birth weight was collected from original birth records. In oral glucose tolerance tests of 370 men aged 64 years, the 2 hour glucose value was inversely associated with birth weight (204). There was however no relationship between birth weight and fasting glucose. Contrary to the men, in 297 women from Hertfordshire there was a relationship between birth weight and both fasting glucose, 2 hour glucose and 32-33 split proinsulin (205). Birth weight was also related to fasting insulin concentrations in women. Table 3 summarises results from studies reporting on the relationship between fetal growth and glucose metabolism.

Also in the Preston study of 140 men and 126 women, there was an association between birth weight and 2 hour glucose values. The prevalence of the metabolic syndrome, closely associated with Type 2 diabetes, among 64-year-old men with birth weights < 2.95 kg was 22% and fell progressively with increasing birth weight. Among men with a birth weight > 4.31 kg, the prevalence was 6% (206). Fasting plasma pro-insulin concentrations fell with increasing birth weight but fasting plasma insulin was not related to birth weight. Subjects with the metabolic syndrome also tended to have a low ponderal index at birth (207).

In the Uppsala study 1333 men had an oral glucose tolerance test at the age of 50 years and were followed up for Type 2 diabetes at the age of 60 years. A strong inverse linear trend between ponderal index at birth and prevalence of diabetes at the age of 60, was found. Ponderal index correlated only weakly with insulin concentrations (208).

The relationships between fetal growth and later disease could be confounded by external influences such as selective loss to follow-up. At the age of 70 years, when the men in the Uppsala were re-examined, the associations between fetal growth and diabetes had changed. At this age only 709 men attended the examination. The monotonic inverse relation seen in the earlier follow-up had been replaced by an inverse U-shaped relation between ponderal index at birth and glucose intolerance. The relation between ponderal index and insulin sensitivity was U-shaped. The relation was also obscured by gestation as the relationships were opposing in pre-terms compared with full-term births (209).

The Health Professionals Follow-up Study of 22846 men assessed both birth weight and diabetic status of the individuals through a questionnaire. The reported birth weight measurements were validated by asking both the study persons and their mothers. A relatively good correlation was obtained on comparing the results of the birth weight answers between these groups, r= 0.71 (p<0.001). Results showed that the lifetime cumulative incidence of diabetes decreased from 7.6% in the birth weight group < 2495 g to 4.2 % in the group > 4536 g (210).

Vestbo studied 620 subjects of which 303 were offspring of diabetic patients (211). The results showed a significant inverse relationship between birth weight and fasting glucose (p=0.008) but the authors concluded that birth weight had only a minor influence on the different risk factors. However, no adjustments for length of gestation were made.

Table 3. Studies reporting on the relationship between fetal growth and glucose metabolism.

Table 3

The relationship between size at birth and diabetes has not been linear in all studies. In the Pima-Indians, where Type 2 diabetes prevalence is extremely high, a U-shaped relationship was found (4). The age adjusted prevalence for birth weights < 2500 g, 2500-4499 g, and ³ 4500 g were 30%, 17%, and 32%, respectively. It was suggested by the authors that selective survival of low birth weight infants, which are genetically predisposed to insulin resistance and diabetes, provides an explanation for the observed relation between low birth weight and diabetes and the high prevalence of diabetes. The high incidence of Type 2 diabetes in high birth weight children was likely to have been caused by a high incidence of gestational diabetes in the mothers.

An association between fetal growth and glucose tolerance is detectable in children according to some studies (212), but not all (213). A low ponderal index at birth is associated with higher 30 minute plasma glucose concentrations in 7-year old children. The insulin levels again were lower when adjusting for glucose levels (212). Another study did not find any relationship between low birth weight and glucose intolerance at the age of 10-11 years (213). Possibly, childhood obesity could be a stronger determinant of insulin level and insulin resistance than size at birth (213). Another explanation is that the results of this study could have been affected by the onset of puberty in the subjects.

2.5.1.1 The effects of transitory famine

Severe long-standing malnutrition causes a variety of disease and shortens life expectancy. Temporary malnutrition during pregnancy due to unfavourable conditions like war or other similar circumstances, mimic the effects of an intervention where the pregnant women and infants are exposed to a low calorie diet. Results from two such historical opportunities to study the actual effects of famine on later disease have been published. In the Second World War, during the Leningrad siege and the Dutch Hunger winter, caloric intake was restricted for several months and the pregnant women and the infants were exposed to famine.

The Leningrad study compares three groups of individuals, one exposed to famine in utero, a second group exposed to famine in infancy and a third group who were born in an area outside the city served as controls (214). The timing of their date of birth in relation to the siege of Leningrad, determined whether the subjects were exposed in-utero or in infancy. No data exist on birth measurements or the weight of the mothers. The results show that there was no difference between the subjects exposed to starvation in utero and those starved during infant life. A possible source of bias is that data on the exposure of the individuals is lacking. In addition, both the in-utero group and the infancy-group were exposed to malnutrition in infancy. The Leningrad study did however find an association between adult stature and raised concentrations of glucose and insulin 2 hours after a glucose load-independently of siege exposure. Since birth weight is a predictor of stature, this would indirectly imply that fetal growth as well as childhood growth may be related to glucose tolerance (139-141).

The Dutch Hunger Winter study examined 702 individuals for whom birth records were available and who had been exposed to malnutrition at some stage during pregnancy. Since the famine was clearly delineated in time, it was possible to compare the effects of famine during different trimesters of gestation. Malnutrition during the first two trimesters in pregnancy predisposed to obesity (119). Both 2 hour glucose values and insulin during an oral glucose tolerance test were elevated in people exposed to famine in gestation (215). This finding supports the association between undernutrition during pregnancy and CHD since 2 hour glucose values are an indicator of increased cardiovascular mortality (216).

2.5.1.2 High birth weight and maternal risk of diabetes

Although high birth weight seems favourable for the new-born, the same is not automatically true for the mother. Studies have shown, that excessive birth weight is associated with subsequent increased risk of Type 2 diabetes in the mother (217-219). There is a possibility that this could be influenced by e.g. gestational diabetes. A common complication of gestational diabetes is macrosomia and the abnormal glucose tolerance is also associated with other perinatal complications (220). There are however no long term follow-up studies of high birth weight individuals born to mothers with gestational diabetes which could clarify the prognosis of these individuals. Therefore more studies are needed before the prognosis of these babies can be established.

2.5.2 Childhood growth and Type 2 diabetes

Epidemiological studies have well documented the importance of preventing obesity in the attempt to reduce the incidence of Type 2 diabetes. Obesity in childhood is associated with abnormal glucose tolerance and often tracks into adult life leading to an increased risk of Type 2 diabetes (221-223). Other risk factors, i.e. elevated blood pressure, cholesterol and impaired glucose tolerance are also identifiable in children with an increased risk of the metabolic syndrome and they also track into adult life (224-227). There is evidence that a low birth weight in combination with obesity in childhood is more detrimental than just sheer obesity in a child of normal birth weight (228).

2.6 Blood pressure

Hypertension is one of the most common non-communicable diseases in Western societies. Although the prevalence depend on the criteria which have been tightened in recent years, there are indications that the prevalence of hypertension is decreasing in many western countries (229-233).

Many risk factors for hypertension are same as for CHD and Type 2 diabetes e.g. obesity, diet and lack of exercise. Similar preventive measures are therefore required for these diseases which often cluster in the same individual as the metabolic syndrome. Since hypertension is an important risk factor for CHD and stroke, some of the CHD risk associated with fetal growth retardation may be mediated by hypertension.

Hypertension is likely to be initiated early in life by some event or stimuli since early high blood pressure (BP) values are predictive of later hypertension and there is evidence for blood pressure tracking (234, 235). The blood flow is pulsatile and arterial pressure depends not only on lumen size but also on arterial stiffness (236). Fetal growth influences arterial stiffness and has consequently been proposed as a possible initiating event leading to hypertension (237).2.6.1 Fetal growth and blood pressure

The association between birth weight and BP was originally first shown in 5362 36-year old British men and women (238). The relationship between birth weight and BP is initially positive in new-borns (239-241) but later a change to an inverse relationship occurs as has e.g. been shown in the (242-249), Japan (250), Jamaica (222), Sweden (251, 252), Israel (253), the US (173, 210, 254), the Netherlands (255), Croatia (256), Italy (257), New Zealand (258). and France (259). On average, the change in systolic BP per 1 kg change in birth weight is approximately 2-3 mmHg (Table 4) (260).

The Nurses' Health Study II examined 71 100 women between 30 and 55 years of age and birth weight was found to be inversely related to BP (254). Also the Health Professionals Follow-up Study of 22 846 men showed an inverse relationship between birth weight and adult hypertension (210). Birth weight in these studies was however assessed by recall and not collected from actual birth records.

In a Swedish study of 149 378 conscripts Nilsson et al. showed that birth weight is inversely related to systolic BP. Systolic BP fell by 0.8 mmHg per every one kg increase in birth weight. Rapid growth in childhood combined with a low birth weight strengthened this association (261).

Levine et.al. found that twins of lower birth weight showed a more rapid rate of rise in blood pressure during infancy. At one year, the increase in blood pressure exceeded that in body weight. Despite greater differences in birth weights between monozygotic twins, there were smaller differences in systolic blood pressure at 1 year. This suggests that intrauterine environmental factors related to birth weight are important in determining blood pressure in infancy and that growth in infancy cannot abolish this association (262). On the other hand, a prospective randomised study of 758 infants weighing under 1850 g at birth, nutritional intake and growth performance in infants was not linked to blood pressure at 7.5-8 years of age (263)

Table 4. Hazard ratios for CHD according to size at birth and height at age 7 years.

Table 4

In some studies, the detected associations between birth weight and blood pressure are more clear in adults i.e. the regression coefficients are larger than in children. As was first postulated by Folkow, this could be due to amplification of hypertension throughout life after the initiating event (264). Of the studies showing an association between birth weight and BP some found that this association becomes stronger in adulthood (242, 257, 265, 266) but others did not find any evidence for amplification (246, 255). The rise of BP throughout life is, however, consistent with the theory of amplification.

If birth weight is associated with blood pressure, then it should also be related to the hemodynamics characteristics of the blood pressure system. In one study where BP was the highest in people with the lowest birth weight, birth weight was also related to arterial compliance measured by pulse wave velocity (267). However, in one study, left ventricular mass, which is increased by hypertension, was not related to birth weight (259).

Despite the fact that most studies have found evidence for a relationship between birth weight and BP, it has been proposed that this effect is negligible (246). Birth weight accounts for less than 5 % of the risk of hypertension compared with 12% from adult BMI which therefore is more important (246). The relationship has also been found to be relatively weak in other studies (268) but the relationship between birth weight and BP is strengthened if adjustment is made for later BMI (250, 255, 268). Possibly, the effects of growth retardation and later obesity are additive i.e. they are independent risk factors for hypertension (269). Differing levels of obesity and other cardiovascular risk factors may explain the differing strength of association between fetal growth and BP.

Inconsistencies in the relationship between fetal growth and BP exist and remain unexplained. In neonates, the relation between birth weight and BP is positive (239, 240, 270, 271). Other studies show a positive or negative relationship depending on the method of adjustment (253, 272-274). Launer found a U-shaped relationship between birth weight and BP (275). Some studies failed to show any relationship between birth weight and BP(244, 276-278). However, in a review of studies comprising of more than 66 000 subjects birth weight was inversely associated with BP in most of the studies (260).

2.6.2 Childhood growth and blood pressure

The link between obesity and hypertension is well recognised. Childhood obesity often precedes adult obesity and therefore increases the risk of hypertension. High-risk individuals can be identified early in childhood and the development of obesity in theses children predict higher levels of cardiovascular risk factors including elevated blood pressure (224, 225, 279).

The consensus is, that prevention of cardiovascular diseases should begin early i.e. in childhood but some controversies still remain about the efficacy and timing of preventive efforts (183). A twin study of 56 monozygotic and 29 same-sex dizygotic twin pairs showed that the genetic influence on risk factors was moderate and that the potential to modify risk profiles during the transition from childhood to adolescence was substantial (280).

In a study of 333 British children aged 9 to 11 years, birth weight showed a significant, graded, positive association with flow-mediated dilation (281). This indicates that low birth weight is associated with impaired endothelial function in childhood, an early key event in atherogenesis.

Lurbe and colleagues performed ambulatory BP measurements of 332 children aged 6-16 years who were not growth retarded. Birth weight was inversely related to daytime systolic BP and systolic BP load when controlled for sex, current height, ponderal index at birth, and age. The differences between the birth weight groups in BP became more clear with increasing age which supports the concept of amplification (282).

2.7 Programming

2.7.1 Evidence of programming

Programming is the process where a stimulus or insult at a critical period of development induces long lasting significant changes in cells, as proposed by Lucas (163). These changes can only be induced during critical periods in the development and if the stimulus is applied before or after the critical period, no permanent changes will occur. The existence of critical periods in the development are supported by human studies (26).

Evidence for programming comes from animal experiments where hormonal programming is readily demonstrated. In the 1960's Barraclough showed that female rats can be made sterile and their sexual behaviour changed with testosterone injections on the fifth day postpartum (283). The function of the pituitary gland and the ovaries is normal but secretion of gonadotropin from the hypothalamic region has been disturbed. If the injection is given on the 20th day postpartum, no permanent effects on the rat will be seen. There is convincing evidence from many studies that metabolism can be permanently changed by early stimuli, not only by androgens, but also by other stimuli during critical periods (40, 284-288).

In rats, nutritional programming can be induced by a low-protein diet which causes irreversible changes in metabolism. Feeding rats a low protein diet during pregnancy results in disproportionate intrauterine growth and hypertension in the young rats (28), increased glucocorticoid-inducible glutamine synthetase activity (289) and lower activities of the glutathione synthetic enzyme gamma-glutamylcysteine synthetase in liver and lung (290). Also elevated plasma renin activity can be seen in rats exposed to a low protein diet during pregnancy (291). These changes can not be reversed by changing the diet later in life.

Another example of programming is the determination of sweat glands in humans. The number of sweat glands is determined during the first three years of life. All humans have roughly the same number of sweat glands when they are born but the environment in which they live during the first years will determine the number available later in life. The number of sweat glands cannot be changed later in life (7).

Hormone levels are affected irreversibly in growth retarded human fetuses and this may be linked to later disease. Umbilical corticotropin-releasing hormone levels have been found to be elevated in growth retarded human fetuses (91). Also adrenocorticotropic hormone and dihydroepiandrosterone levels are higher in growth retarded fetuses (91). Dihydroepiandrosterone is an indicator of adrenarche which may be a risk factor for Type 2 diabetes (86, 87). Recent findings support the concept of programming of androgen abnormalities by fetal growth (85).

2.7.2 Mechanisms

The Barker hypothesis originally stated that adult disease is programmed by deficient nutrition during pregnancy leading to fetal growth retardation but nowadays it also covers other mechanisms. Deficient nutrition could have a direct effect on the organs in the body or affect the endocrine system. This may affect organ function or reset the hormonal levels which may thereby lead to disease later in life.

2.7.2.1 CHD

The effects of classical risk factors such as dyslipidemia, hypertension and diabetes on the incidence of CHD are well recognised and their relationship to CHD is indisputable. Classical risk factors, however, do not account for all of the incidence of CHD and the fetal growth hypothesis could therefore explain some of this variation. Fetal growth retardation may increase the risk of CHD by increased incidence of dyslipidemia, hypertension, and Type 2 diabetes in the individuals.

High serum cholesterol levels are associated with an increased risk of CHD. The increased risk of CHD associated with fetal growth retardation could also be mediated by dyslipidemia caused by a defect in liver function. In support of this, higher levels of cholesterol, triglycerides, fibrinogen, factor VII and Apo A-1 have been found in children with low birth weight (211, 292-294). Liver size and cell number are decreased in rat offspring whose mothers have been fed a low protein diet during pregnancy (33). Changes in liver enzymes activities associated with glucose metabolism have been detected in offspring exposed to maternal low protein diet during pregnancy. Hepatic glucokinase activity was reduced by about 50% and phosphoenolpyruvate carboxykinase was increased by 100% in the offspring despite an adequate protein diet during nursing (295, 296). These findings are remarkable since neither glucokinase nor phosphoenolpyruvate carboxykinase are expressed until after birth (142).

A change in insulin like growth factor-1 levels during pregnancy has been implicated as a cause of fetal growth retardation. A possible resetting of insulin like growth factor-I levels is supported by the finding that insulin like growth factor-I levels are reduced in patients with arteriosclerosis (297). However, a later study found no evidence of reduced insulin like growth factor-I levels in patients with the metabolic syndrome (298).

2.7.2.2 Type 2 diabetes

Both a defect insulin secretion and insulin resistance are needed for the development of manifest Type 2 diabetes. If fetal growth retardation is the cause of Type 2 diabetes, it should also be associated with the defects in muscle metabolism, such as insulin resistance, or deficient beta cell function which predispose to Type 2 diabetes. Another major defect found in Type 2 diabetes is an increased hepatic glucose output (299, 300).

Animal experiments clearly show the adverse effects of a low protein diet during pregnancy. A low protein diet during pregnancy causes reduced beta cell proliferation and islet vascularisation, reduced islet size leading to deficient insulin secretion and glucose intolerance,(36, 37, 301-309). Overnutrition later in life adds to the effects of undernutrition in pregnancy and have been shown to cause higher plasma insulin levels in male rats (310).

Changes in hormone levels may be the underlying cause of the programming of Type 2 diabetes. The use of glucocorticoids cause insulin resistance and therefore changes in glucocorticoid metabolism are one possible candidate (311). Glucose metabolism in the fetus may also be affected by the same mechanisms which have been implied to cause hypertension i.e. exposure to glucocorticoids during pregnancy. In support of this, prenatal glucocorticoid exposure has been shown to cause hyperglycaemia in the rat offspring (312). This hyperglycaemia was caused by administration of carbenoxolone which is an inhibitor of 11beta-hydroxysteroid dehydrogenase. The changes in cortisol levels may be caused by increased activity of the hypothalamo-pituitary-adrenal axis (313).

The studies trying to link birth weight with insulin resistance and insulin secretion in humans have not been as conclusive as the animal experiments. The nutritional deficiency during pregnancy could damage the beta-cells in the pancreas and, in support of this, small-for-date fetuses have been shown to have decreased beta-cell mass and vascularisation (314). Yet in humans, some studies find that birth weight is associated with insulin secretion, while others do not (314-317). Another study showed that a low ponderal index at birth is associated with insulin resistance later in life (318). More recent studies do however indicate that insulin resistance may have a role in programming disease since it has been shown that babies born small-for-date are more insulin resistant later in life (319, 320). There was however, no relationship between fetal growth and glycogen synthase which has been implicated in the pathogenesis of insulin resistance and Type 2 diabetes (321).

2.7.2.3 Hypertension

Defects in placental function may lead to fetal growth retardation and raised blood pressure in the offspring. The placenta acts as a barrier against substances as well as a supply of nutrients between the mother and the fetus. Therefore, a defective placental function will influence the growth of the fetus. Recent twin studies show that the relationship between birth weight and blood was not abolished in analysis of monozygotic twins (322, 323). This suggests that fetal growth retardation is caused by mechanisms acting in the fetoplacental unit and that maternal nutrition may not be as important. Possible hormones include growth hormone, insulin, insulin-like growth factor I, glucocorticoids, catecholamines and angiotensin II (324).

Dexamethasone infusion during pregnancy causes hypertension in rat offspring (325). The same phenomenon can be demonstrated in sheep where undernutrition at 22-29 days of pregnancy did induce hypertension but there was no effect of undernutrition at 59-66 days of gestation on the prevalence of hypertension (326). This has led to the hypothesis that glucocorticoid exposure in pregnancy induces hypertension later in life (327).

Placental 11 beta-hydroxysteroid dehydrogenase normally protects the fetus from the effects of endogenous maternal glucocorticoids which cause fetal growth retardation (90, 325). Synthetic glucocorticoids, such as dexamethasone, are not substrates of 11-beta-hydroxysteroid dehydrogenase and therefore the passage of these substances through the placenta is not impaired (325). Down-regulation of 11-beta-hydroxysteroid dehydrogenase by a low protein diet lowers the 11-beta-hydroxysteroid dehydrogenase activity and thereby increases the exposure of the fetus to endogenous maternal glucocorticoids and causes growth retardation as well as hypertension in the offspring (289, 328). The down-regulation in the activity of 11-beta-hydroxysteroid dehydrogenase may be caused even by a mild protein restriction, whilst activities of other glucocorticoid-insensitive control enzymes are unchanged (289). Inhibiting 11-beta-hydroxysteroid dehydrogenase in adrenalectomized pregnant rats had no effect on birth weight or blood pressure supporting the hypothesis that glucocorticoids secreted from the adrenal are responsible for the raised blood pressure (329). Hypersensitivity to glucocorticoids is primarily thought to be associated with undernutrition in mid to late gestation (291). In humans, 11-beta-hydroxysteroid dehydrogenase activity is also related to birth weight (330).

Insufficient nutrition during pregnancy may cause direct changes in organ structure and size (267). Low maternal haemoglobin during pregnancy is linked to BP (331) as well as birth weight (332). According to the Brenner hypothesis (also called the nephron number hypothesis), a decreased number of nephrons leads to hypertension (333-336). Studies in rodents show that maternal protein intake during gestation is directly related to the relative weight and the number of nephrons formed (35, 335, 337-340). Dexamethasone infusion during pregnancy increases blood pressure and impairs renal development (341). Also consistent with the Brenner hypothesis, the total number of nephrons is reduced in IUGR infants (342). However, a human study of Type 2 diabetes patients, showed no association between low birth weight and low kidney weight or low birth weight and few and/or small glomeruli (343).

Arterial stiffness is an important component in hypertension because of the pulsatile nature of blood flow and it is determined by intra-uterine factors and genetic factors (237, 344-346). Arterial compliance predicts CHD mortality and morbidity (347, 348). Peripheral resistance is increased by arterial stiffness which is determined, not only by systolic and diastolic pressure, but also sodium intake, serum high-density lipoprotein cholesterol, daily energy expenditure in physical activity, and serum insulin levels. (349). It has been proposed that low birth weight impairs the synthesis of elastin in the walls of the aorta and large arteries. This deficiency can lead to permanent changes in the mechanical properties of these vessels and lead to hypertension (237). A study of 337 men and women showed that impairment of fetal growth is associated with decreased compliance in the conduit arteries of the trunk and legs and raised blood pressure in adult life(267). Low birth weight fetuses show loss of end diastolic blood flow velocity (350). The aortic vessel wall diameters are also smaller in children born small for their gestational age (p < 0.01) (241).

Alterations in the hypothalamo-pituitary-adrenal axis have also been proposed as the cause of growth retardation and hypertension. Lack of dietary protein blunts spontaneous pulsatile growth hormone release, attenuates growth hormone responsiveness to growth hormone releasing factor 1 challenge and reduces pituitary growth hormone content and size. Suppression of growth hormone release is mediated at least in part by increased insulin like growth factor secretion (351). Rats exposed to a maternal low-protein diet also exhibited increased glycerol-3 phosphate dehydrogenase activity in the hypothalamus, whereas their pyruvate kinase activity was not changed. These effects could be reversed by metyrapone (35).

2.7.2.4 The impact of genes

There is a familial aggregation of CHD, Type 2 diabetes and hypertension and environmental risk factors have a clear role in predisposing to these diseases. The role of genetic factors is not fully known. So far, the genetic studies have been unrewarding, although there are indications of genetic effects predisposing to Type 2 diabetes and dyslipidemia (352-354).

There is also a familial aggregation of infants born small for their gestational age (355, 356). Recent findings show the importance of genetic factors for fetal growth as paternal factors have an influence on the risk of being born SGA (356). Possible candidates are angiotensin II receptor AT1 or insulin like growth factor-I gene defects which have both been associated with fetal growth retardation (102, 357, 358). Mutations in the angiotensin converting enzyme-gene have also been shown to modify the effects of fetal growth retardation on glucose metabolism (359).

Despite the familial aggregation of being born small-for-date and cardiovascular disease, no major genetic link between these have been found. Candidate gene studies have been unrewarding but mutations in the glucokinase gene, which is a key regulator of glucose metabolism in the pancreas, have been shown to result in reduced birth weight (360, 361). Twin studies show that the association between low birth weight and Type 2 diabetes in twins is at least partly independent of genotype. The association may be due to intrauterine malnutrition although it cannot be excluded that the association between low birth weight and IGT could be due to a coincidence with a certain genotype causing both low birth weight and IGT in some subjects (362). More prospective studies are needed to study the impact of genes on fetal growth and adult disease.