Early fetal growth and risk factors for cardiovascular diseaseBMJ 2014; 348 doi: http://dx.doi.org/10.1136/bmj.g175 (Published 23 January 2014) Cite this as: BMJ 2014;348:g175
- Catherine E M Aiken, academic clinical fellow,
- Gordon C S Smith, professor
- 1Department of Obstetrics and Gynaecology, University of Cambridge, Box 223, Rosie Hospital and NIHR Cambridge Comprehensive Biomedical Research Centre, Cambridge CB2 0SW, UK
- Correspondence to: G C S Smith
Some readers’ principal response to the paper by Jaddoe and colleagues (doi:10.1136/bmj.g14) may be bemusement.1 Why would anyone link growth of the fetus in the first trimester to risk factors for the cardiovascular killers of middle age? However, the rationale for the study is based around two large bodies of work that have concluded that fetal growth restriction is associated with an increased risk of cardiovascular—and many other—diseases in later life,2 and also that fetal growth is profoundly influenced by conditions in the first trimester of pregnancy.3 Jaddoe and colleagues report new associations between apparently poor growth in the first trimester and a range of cardiovascular risk factors measured in school age children. Are the associations likely to be real? What mechanisms might be operating if they are? Lastly, what should we be doing about it?
Many aspects of this carefully conducted prospective cohort study support the validity of the conclusions. However, as the authors acknowledge, the possibility remains that some of their significant associations may have arisen by chance. False positive findings, or type1 statistical errors, are always a risk in studies with many different outcomes and a large number of statistical tests. The many potential classifications of both outcomes and exposures are a further complication. Furthermore, the new study is one of many arising from the same cohort,4 and we know that repeated hypothesis testing increases the risk of type 1 errors still further. Although assembling such a feast of data but then denying oneself any more than a single morsel of analysis would be perverse, the conclusions of this study should be treated as hypothesis generating. The pre-existing body of work, however, suggests that Jaddoe and colleagues’ findings are not data driven and will be reproducible.
Mechanisms that may explain the link between fetal growth and later cardiovascular risk factors have been explored extensively over the past 30 years, in both human cohorts and animal models.5 The idea that subtle influences on physiological systems occur early in development and are later magnified by the effects of growth and ageing to produce pathological phenotypes is well established. Studies have identified putative mechanisms—specifically, altered DNA methylation, mitochondrial DNA instability, and increased exposure to oxidative stress.6 7 Such mechanisms are likely to explain the trans-generational effects of certain environmental challenges in human populations.8 However, despite the plausibility and attractiveness of the developmental programming hypothesis in interpreting the results of Jaddoe and colleagues’ study, consideration must be given to other explanations for a smaller than expected fetus in early pregnancy. A fetus may measure small for dates if ovulation occurred later than usual in the menstrual cycle. Hence, a smaller than expected fetus could be a marker for reproductive disorders in the mother, such as polycystic ovary syndrome, which predisposes women to prolonged menstrual cycles. Given that polycystic ovary syndrome is associated with the metabolic syndrome, and is also likely to have an important genetic element,9 a small fetus or baby may be a marker of maternal genetic characteristics that might be inherited by the baby and lead to the associations described in the paper. Although Jaddoe and colleagues corrected estimates of fetal size for cycle length, the subtle effects of dating discrepancies are a methodological problem with many such studies. Consistent with the idea that reproductive dysfunction and cardiovascular disease could have a common genetic predisposition, we know that the birth weight of the infant is also inversely associated with the risk of cardiovascular disease in the mother and in the mother’s parents.10 11 An extensive body of work now exists that supports the view that delivery of a small baby is a marker of maternal cardiovascular dysfunction.12 Disentangling the effects on the fetus of maternal environmental stresses from the effects of maternal genetic and epigenetic predisposition to disease will be a major challenge for future studies.
For doctors, the pertinent question is whether these early effects can be modified. Can we identify interventions that might improve the early life environment and promote “normal” growth trajectories? On the basis of the current analysis, these interventions might be needed in the first trimester of pregnancy, during embryogenesis. Given the potential for interventions to cause serious harm at this stage of pregnancy, compelling evidence of safety will be needed before their evaluation. Hence, the appropriate response in the short term is that we need a deeper understanding of the strength, nature, and mechanisms of the reported associations before rushing to intervene. Studies such as those of Jaddoe and colleagues and previous work suggest that researchers need to recruit cohorts of women in the very earliest stages of pregnancy or, ideally, before conception to ensure that the initiating events are captured.1 3 For future analyses of the determinants of the health of our children and the adults that they become, a key message may lie in the words written by Oscar Hammerstein II and immortalised by Julie Andrews: “Let’s start at the very beginning, a very good place to start.”
Cite this as: BMJ 2014;348:g175
Competing interests: We have read and understood the BMJ Group policy on declaration of interests and declare the following interests: none.
Provenance and peer review: Commissioned; not externally peer reviewed.