Children with a low 5 min Apgar score, especially children with a score of 0–5, had a higher overall risk of childhood cancer that was diagnosed before 6 months of age. We also observed higher risks for several main childhood cancers like Wilms’ tumour. The associations were independent of country, child sex, child birth characteristics (birth weight, birth order and gestational age) and maternal factors (age, education and smoking during pregnancy). However, we found no association between low Apgar score and risk of leukaemia and other lymphatic/haemapoietic malignancies.
The theory of ‘developmental-origins of health and disease’ proposes the importance of in utero environment for long-term human health.22
We observed that children with a low Apgar score between 0 and 5 had a higher overall cancer risk than those with an optimal score (9 or 10). The mechanism underlying this observation is, however, unclear. A low Apgar score is a marker of a suboptimal fetal environment20
or other factors that prevent the child from achieving a high score. From a programming perspective, it is interesting to observe that the effect of a low Apgar score on overall cancer risk was the strongest for cancers diagnosed before 6 months of age. Tumours from brain/nervous system, endocrinal glands, kidney and leukaemia/lymphomas were among the most frequent diagnosed cancers during this period. This observation is in line with suggestions from previous studies that in utero exposures to insulin-like growth factors,8
may play a role for the relationships between other birth outcomes and many childhood cancers, or childhood cancer risk in general. A low Apgar score probably shares aetiology with cancers initiated in fetal life, and different biological pathways may operate for the association between Apgar score and childhood cancers. For example, neonatal treatments related to low Apgar scores may increase the risk of some childhood cancers.36
The best evidence for fetal origins of childhood cancer has been available for leukaemia.7
Our findings suggest that those observations may operate through the mechanisms that do not affect the Apgar score, and similar interpretations apply to other lymphatic/haemapoietic neoplasms. The associations between a low Apgar score and several other specific childhood cancers in our study are noteworthy. For example, the highest risk of a low Apgar score was obtained for Wilms’ tumour, which is in line with observations in two register-based studies (restricted to only girls in one of the studies)39
but not in another case–control study.41
Hypoxia, as indicated by a low score, may result in cell damage that subsequently leads to Wilms’ tumour.42
Alternatively, neonatal treatments provided to neonates with a low Apgar score may also increase the risk of Wilms’ tumour.36
Hepatoblastoma is reported to be associated with factors like low birth weight,44
smoking during pregnancy or young maternal age.45
A recent study showed a reverse association between birth order and retinoblastoma.9
However, the observed elevated risks of both hepatoblastoma and retinoblastoma after adjustment might indicate an independent role of a low Apgar score for these two childhood cancers.
The observed associations between low Apgar scores and childhood cancer risk were not explained by other adverse birth outcomes, which have been widely used as the proxy indicators of fetal environment to explain fetal origins for a number of adult diseases.22
As expected, a low Apgar score was more common among children with adverse birth outcomes, which often correlate with childhood cancers.7–9
However, the elevated risks related to a low score were observed in almost all subgroups of baseline characteristics, including but not restricted to pregnancies with adverse birth outcomes. Furthermore, the associations were consistent according to country and maternal factors under investigation.
The most important strengths of our study include the prospective longitudinal design, large sample size and detailed data on other covariates. The registry system in the Nordic countries provides both a complete case ascertainment and accurate linkage with other data, which allow complete follow-up with least impact of misclassification error.27
The rarity of childhood cancer makes population-based epidemiological studies very difficult. Much of the heterogeneity of previous results might be due to the small sample sizes and lack of control for factors related to the child or the mother. Our data enabled us to do a more in-depth investigation by examining risks in subgroups. The cohort design based on prospectively collected high-quality data minimised the impact of information or recall bias.
One limitation of our study is that we lack information on risk factors after birth. However, factors associated with a low Apgar score, such as related neonatal treatments, may lie in the pathways between exposure and outcome, and should not necessarily be controlled for in the analyses.46
A second limitation is that we cannot rule out the confounding of factors like environmental exposures after birth. Third, the case numbers for several childhood cancers are small, although the total population included over 5 million children.
To conclude, our findings support the developmental-origins hypothesis of childhood cancer. An association between a low Apgar score and childhood cancer does not prove a causal role of the components that make up the Apgar score but strengthens the relevance of viewing the prenatal time period as a causal time window of interest. A low Apgar score may reflect a pathological pregnancy that could share causes with childhood cancers, or childhood cancers may have a clinical onset that starts during fetal life. In the first situation, a low Apgar score may also be associated with cancer risk in adulthood. In addition to being a widely accepted assessment tool in neonatal care, the Apgar score may indicate programming effects of fetal environment on further health, suggesting that its role in clinical practice and public health may reach beyond its current use.