(View plain language abstract)
There is increasing awareness that childhood circumstances, particularly diet, influence the disease risk in adulthood. Direct evidence from long-term follow-up studies with detailed dietary data in childhood is sparse and proxy measures, such as final adult height, have been used instead. Taller individuals generally have an increased risk of developing cancer and this association may be specific to the leg length component of stature. Peak growth in leg length is pre-pubertal while peak growth in trunk is post-pubertal and so stronger associations with leg length have led to speculation that pre-pubertal diet may be most important in determining adult cancer risk. Stature in childhood and adulthood is dependent on genetic and environmental factors. However, childhood stature may better reflect pre-pubertal growth influencing exposures than adult height, which reflects a combination of childhood growth and also age and duration of pubertal maturation.
We have examined associations between childhood stature (measured at ages 2-14) and adult cancer risk in a large cohort (N=4,999) followed up for 59 years. In addition to leg and trunk length we have explored associations with foot length (pre-pubertal peak growth) and shoulder breadth (post-pubertal peak growth). Data are also available on childhood circumstances, including detailed dietary data, adult IGF levels, and genetic data.
Associations between childhood stature and adult cancer were generally weak although consistent with those reported elsewhere. Specifically, taller childhood stature was associated with a slightly increased, but non-significant risk of cancer, particularly breast (trunk length OR 1.26; CI 1.00, 1.60) and prostate (foot length OR 1.22; CI 0.86, 1.75), although there was no evidence that cancer risk was particularly strongly associated with any specific component of stature.
Further analyses were carried out to explore potential mechanisms underlying these associations. All components of childhood stature were positively associated with measures of childhood circumstances (breast-feeding, birth order, household income, household food expenditure, social class, crowding, number of children in the household, and household diet) and associations were stronger for components with pre-pubertal (foot and leg) rather than post-pubertal (shoulder and trunk) peak growth. This provides support for the use of stature as a proxy for childhood circumstances, particularly those occurring pre-pubertally. It has been hypothesised that dietary influences on the adult insulin-like growth factor (IGF) system may underlie stature-cancer associations. However there was only weak evidence of an association of increased IGFBP-2 levels in adults whose feet were large as children. Analysis with genetic variants gave no evidence for a strong effect of genotype on childhood stature. Variants in IGFBP-3 were associated with IGF-II, but not IGF-I levels.
Although cancer risk is increased in larger children, it is important to note that an opposite association of decreasing cardiovascular disease (CVD) with increasing stature has also been reported and we have replicated this negative association in these data. The conclusion from these analyses is therefore not that we should intervene to reduce childhood growth but that anthropometric measurements are useful tools in the exploration of childhood circumstances, particularly diet, and in identifying critical periods of growth in terms of later disease risk.
Taller adults are more likely to develop cancer than their smaller contemporaries. Moreover it is generally adults with longer legs rather than longer trunks who have this pattern of disease risk. Being tall or having long legs does not affect the likelihood of developing cancer per se. Rather height and leg length, which are determined by genetics and also by childhood circumstances such as diet, are used as proxy measures for diet when direct data are not available. Children grow throughout childhood and adolescence and children who are well-nourished are likely to be taller. Different parts of the body grow more at different times during adolescence; specifically legs and feet grow more before puberty while trunk and shoulders grow more after puberty. It has therefore been hypothesised that stronger leg length-cancer associations indicate that early (pre-pubertal) diet is most important in determining future disease risk. One disadvantage of using adult stature to represent childhood diet is that it is also affected by the timing, duration and intensity of puberty.
We therefore explored associations between adult cancer risk and childhood height measured at ages 2-14 in a large cohort (N=4,999) followed up for 59 years. We also examined associations with childhood leg and foot length (most growth before puberty) and shoulder breadth and trunk length (most growth after puberty) to explore periods of growth that particularly affect cancer risk.
Adults who were larger as children (as indexed by all measurements) were slightly more likely to develop cancer, especially breast and prostate, than those who were smaller. However, the associations were weak, although still consistent with those reported elsewhere. There was no evidence that any particular component of height was more important in determining cancer risk.
We also explored associations between childhood circumstances (breast-feeding, birth order, household income, household food expenditure, social class, crowding, number of children in the household, and household diet) and stature. All childhood factors were positively associated with all measures of stature although associations were stronger for leg and foot length than for trunk length and shoulder breadth, which suggests that leg and foot length are the best proxy measures of (pre-pubertal) childhood diet and other circumstances.
It has been suggested that childhood diet affects levels of insulin-like growth factor (IGF), which in turn affects both childhood growth and future disease risk. We looked at associations between childhood stature and adult IGF levels in a sub-sample of the cohort but found little evidence to support any strong relationship. There was no strong evidence for association between genetic variants and childhood stature, despite some evidence that genetic variants have an effect on some IGF levels.
Taller adults, particularly those with longer legs, have also been found to have a lower risk of cardiovascular disease (CVD) and we have replicated this finding. Childhood stature is therefore of interest in terms of exploring childhood circumstances such as diet and identifying critical periods of growth for future disease risk; it is not a risk factor that we should intervene to reduce.