Little is known about the etiology of hepatoblastoma. Because of the young age at diagnosis, several studies have looked at various birth characteristics. The purpose of our study was to investigate the incidence of hepatoblastoma in the Nordic countries and the association between selected birth characteristics and hepatoblastoma. Data from national cancer registries and birth registries in Denmark, Sweden, Norway and Finland 1985–2006 was used. Overall, 155 children with hepatoblastoma aged 0–14 years were included and individually matched to five controls drawn randomly from national population registries. The incidence rate of hepatoblastoma was 1.7 per million person-years with a predominance of boys (1.5:1). Incidence rate was highest before the age of 1 year (8.3 per million person-years). A higher risk of hepatoblastoma was found in children with birth weight <1,500 g [odds ratio (OR) = 9.5; 95% confidence interval (CI): 2.3–38.2], born preterm in week 22–32 (OR = 4.5; CI: 1.8–11.5) and Apgar scores <7 after 1 min (OR = 3.1; CI: 1.3–7.1) and 5 min (OR = 7.5; CI: 1.8–32.4). A doubling in risk was found in children who were large for gestational age (OR = 2.3; CI: 1.0–5.3). No associations were found with birth order, maternal age or maternal smoking. Our study indicates that intrauterine and/or neonatal factors are associated with increased risk of hepatoblastoma. These may include low birth weight and asphyxia leading to neonatal intensive care. Alternatively, the factors may be a consequence of hepatoblastoma developing in utero.
Hepatoblastoma is a very rare childhood cancer, with incidence rates of about 1.5 per million person-years in children aged 0–14 years in Northern Europe.1 Nevertheless, it is the most common malignant hepatic tumor in children, and usually diagnosed before the age of 4 years.2 Its etiology is largely unknown except that an increased risk of hepatoblastoma is associated with several congenital and genetic anomalies such as hemihypertrophy,3 Beckwith Wiedemann overgrowth syndrome2–5 and familial adenomatous polypsis,2, 5, 6 in addition to other rare syndromes. Previous studies have identified an association between extremely low birth weight and hepatoblastoma.7–11 Furthermore, a working group at International Agency for Research on Cancer has classified tobacco smoke (parental smoking) as a human carcinogen for hepatoblastoma in their offspring12 on the basis of previous studies.9, 13–15 No further associations are well established, as there are only few etiological studies due to the rarity of the disease. However, those studies showed positive associations between hepatoblastoma and short gestational age, eclampsia or severe preeclampsia.7 Low maternal age has also been associated with increased risk of hepatoblastoma in the offspring,13 but another study found no maternal age effect.16 It has been suggested that the association between very low birth weight and hepatoblastoma might be explained by adverse side-effects of neonatal intensive care10, 17–19 or by the increasing numbers of low birth weight babies surviving infancy.20 Some cases are diagnosed already within the first months after birth, suggesting an intrauterine origin of the cancer.8 This puts forward the hypothesis that the etiology could be related to risk factors that occur before conception, during the gestational period or soon after birth.
In our study including all cases of hepatoblastoma diagnosed in the four Nordic countries over a 22-year period, we aimed at estimating the incidence rate of hepatoblastoma in the Nordic countries as well as to investigate the possible relationship between selected birth characteristics and risk of hepatoblastoma in a register-based case–control design.
Material and Methods
The study included all children aged 0–14 years diagnosed with hepatoblastoma between January 1, 1985 and December 31, 2006, who were born and, at the time of diagnosis, were living, in Denmark, Norway, Sweden or Finland. The cases were identified in the national cancer registries and additionally in the childhood cancer registries in Sweden and Denmark. The Norwegian cases were crosschecked with the solid-tumor database of the Nordic Society of Pediatric Haematology and Oncology. Hepatoblastoma tumors were defined according to main group VIIa of the International Classification of Childhood Cancer, third edition.21
The first part of the study was a descriptive study examining the incidence rates of hepatoblastoma in the Nordic countries. The annual age-specific Nordic childhood population (with a yearly total of ∼4.4 million children) was used to estimate person-time at risk for the calculation of incidence rates.
The second part of the study was a register-based case–control study. Each case was individually matched to five randomly selected controls drawn from the background population matched by sex, age and country at the time of diagnosis for the index cases (incidence density sampling). Controls had to be alive at the time of selection and with no previous diagnosis of childhood solid tumors.
Data on birth characteristics was derived from the population-based medical birth registries in Denmark (established 1973), Sweden (established 1973), Norway (established 1967) and Finland (established 1987), which contain mandatory continuously updated reports on all births in the respective countries. The registries were linked by the unique personal identification numbers assigned to all citizens in the Nordic countries, allowing accurate linkage. The medical birth registries include information on all births—in hospital and home births. The data from the birth registries is collected prospectively during pregnancy and at birth by the midwife. The completeness and the validity of the Nordic birth registries are considered to be excellent.22
The following birth characteristics were included in the study: birth weight, gestational age, fetal growth, Apgar score, birth order, maternal age and maternal smoking. All variables were categorized according to cut-off points determined a priori. In our study, birth weight was categorized as follows according to previous studies on birth weight and hepatoblastoma; very low birth weight (<1,500 g), low birth weight (1,500–2,499 g), normal birth weight (2,500–3,999 g) and high birth weight (≥4,000 g). Gestational age was measured as completed weeks of gestation, which in the early period was primarily determined by the date of the last menstrual period and in the later years by ultrasound in early pregnancy. This change happened gradually over time. Birth weight by gestational age was used as a proxy for fetal growth. Most growth curves are based on cross-sectional data on birth weight by gestational age based on the weight of newborns, but these curves do not necessarily reflect the normal intrauterine growth velocity, especially not in the preterm period when much of the data are from abnormal deliveries. We decided to estimate the deviations from the expected birth weight by gestational age by using fetal growth curves based on ultrasonically estimated fetal weights of infants subsequently born at term, for a combined Danish and Swedish cohort.23 Separate growth curves were used for boys and girls. The children were categorized as small for gestational age (SGA), defined as more than two standard deviations (SDs) of the mean below the normal growth curve, large for gestational age (LGA) defined as more than two SD above the normal growth curve and appropriate for gestational age (AGA) defined as the interval between SGA and LGA. We calculated the SD of the mean from our own control sample.
Both Apgar score after 1 and 5 minutes were included, because the two variables may express two different physiological conditions of primary and secondary asphyxia. The variables were converted into dichotomous variables with low Apgar score <7 and normal Apgar score ≥7.
Data on birth order was derived from number of siblings and birth year in the central population registries and was based on maternal siblings only. Data on maternal smoking was self-reported and registered by the midwife during pregnancy. In the four Nordic countries, different routines applied as to when in the pregnancy smoking habits were registered and also the categorization of dose. Thus, maternal smoking was defined as a dichotomous variable measuring if the mother had been a smoker at some time point during pregnancy.
When comparing the incidence rates between the Nordic countries, direct age-standardization was employed to adjust for the possible differences in age distributions. The World Standard Population 2000–2025 by the World Health Organization was used as a standard population.
In the second part of the study, conditional logistic regression was used to estimate odds ratios (OR) and 95% confidence intervals (CIs). Therefore, all ORs were conditioned by sex, age and country. The causal networks underlying the assumed causal relation between each of the birth characteristics and hepatoblastoma were clarified by using Directed Acyclic Graphs (DAGs).24 This was also used for identifying confounders. The adjustment for the following confounders was performed in the analyses: gestational age, maternal age, birth order and birth weight. In the footnotes of Table 2, the variables we have adjusted for in the selected analyses are presented.
AGA: appropriate for gestational age; CI: confidence interval; LGA: large for gestational age; OR: odds ratio; SD: standard deviation; SGA: small for gestational age
In the Nordic countries, 155 children were diagnosed with hepatoblastoma during the period 1985–2006. This number corresponds to an incidence rate of hepatoblastoma of 1.7 per million person-years (Table 1), which has been relatively stable in the period 1985–2006. There was a predominance of boys compared to girls with a sex ratio of 1.5:1 (Table 1). This predominance was strongest within the first 2 years of life and decreased after that age.
Table 1. Selected descriptive characteristics of hepatoblastoma in the Nordic countries, 1985–2006
During the first 5 years of life, the incidence was inversely related to the age of the child. Thus, the annual incidence rates were highest among children <1 year of age with 8.3 per million compared to 3.5 per million for 1- to 4-year-old children (Table 1). The median age of diagnosis was 19 months and 50% of the children diagnosed within the first year of life were diagnosed at 6 months of age or younger (data not shown). When looking at the Nordic countries separately, the annual age-standardized incidence rate was lowest in Finland (1.2 per million) and highest in Norway (2.0 per million) (data not shown).
In the second part of the study, we found a statistically significant association between very low birth weight <1,500 g and hepatoblastoma, whereas children with low, normal and high birth weights showed no statistical significant risk (Table 2). Age of onset of hepatoblastoma in very low birth weight babies did not differ from the age of onset in normal birth weight cases (very low birth weight: mean 30 months and median 20 months; normal birth weight: mean 26 months and median 20 months). Numbers were small for extremely low birth weight (<1,000 g), but it showed a very strong association with hepatoblastoma risk (OR = 24.7; 95% CI: 2.9–211: five cases and one control). Born preterm in week 22–32 was also associated with a higher risk of hepatoblastoma (Table 2). A statistically nonsignificantly elevated OR of 1.9 (CI: 0.6–5.6) was observed for SGA children, but based on a small number of children that were SGA. As expected, a high agreement between very low birth weight (<1,500 g) and SGA was found; of six cases with a very low birth weight, four were also SGA (data not shown).
Table 2. Birth characteristics and risk of hepatoblastoma in the Nordic countries, 1985–2006
Being LGA was statistically significantly associated with an increased risk of hepatoblastoma (Table 2). Sensitivity analysis of fetal growth by birth weight was performed (data not shown), showing that the LGA children with a birth weight ≥4,000 g could have a higher risk of hepatoblastoma (OR = 2.2; CI: 1.0–4.6) compared to AGA children with a birth weight below 4,000 g. Furthermore, the AGA children who weighed ≥4,000 g had a reduced risk, which could explain, why no significant association was found in the analysis of birth weight alone (OR = 0.9; CI: 0.6–1.5). It was also found that LGA children with hepatoblastoma are more likely to be diagnosed within the first year of life compared to AGA and SGA children (data not shown).
The analyses of Apgar score (Table 2) showed that there was a higher risk of hepatoblastoma for children with low Apgar score (<7) at 1 and 5 min after birth compared to children with a normal Apgar score (≥7). Children with a low 1 min Apgar score had a 3 times higher risk of hepatoblastoma compared to children with a normal Apgar score and children with a low 5 min Apgar score had a nearly 8 times higher risk of hepatoblastoma compared to children with a normal Apgar score. After additional adjustment for birth weight and gestational age, the risk of hepatoblastoma remained 3 times higher with a low 1 min Apgar score and the risk of hepatoblastoma was 7.5 times higher with a low 5 min Apgar score compared to children with a normal Apgar score. However, it should be noted that the numbers are very small.
Neither birth order nor maternal age or maternal smoking showed evidence of statistically significant associations with hepatoblastoma (Table 2). However, it should be noted that maternal age above 40 years were associated with a 2 times higher risk of hepatoblastoma, but this association was not statistically significant, which may be due to small numbers in this age group.
The results of our study show an incidence rate of hepatoblastoma that was similar to the incidence rates of previous studies.1, 25 Also the distinct patterns by sex and age found in the previous studies are confirmed by our study. Furthermore, we found that the Nordic incidence rate was relatively stable during the study period in contrast to other studies who found an increasing incidence rate from 1970s to 1990s.1, 26
Previous studies have found an association between very low birth weight and hepatoblastoma, which was confirmed by our study.7, 8, 10, 27 The biological mechanisms for the observed associations between low birth weight, preterm birth, low Apgar score and higher risk of hepatoblastoma are unclear. However, there are various plausible hypotheses. First, adverse side effects of factors in the neonatal treatment such as oxygen supplementation,28–30 ionizing radiation and phthalates in the medical devices18 could explain the association. Furthermore, an interaction between low birth weight and neonatal care treatment could cause the adverse side effects of neonatal care treatment to operate more powerfully in preterm children with very low birth weight.7
Second, the observed association might reflect reverse causality, indicating that these children are born preterm, with low birth weight and poor health resulting in a low Apgar score because of in utero initiated carcinogenesis. The tumor could start developing in utero, because of developmental disturbances occurring during organogenesis, which permits a deregulated continuation of proliferation resulting in a mass of immature tissue recognized as a hepatoblastoma tumor.3 Supportive of this hypothesis is, first, the fact that the incidence rate of hepatoblastoma is inversely correlated to age with the highest incidence rate within the first year of life; it is even possible that the incidence peak is already in the prenatal period. Second, there are several reports concerning hepatoblastoma tumors detected in stillborn fetuses31, 32 and within the first days after birth in live born children,32–35 which also implies that some hepatoblastoma tumors develop in utero.
This is the first study that has observed associations between accelerated growth and hepatoblastoma. Other studies have mainly focused on very low birth weight in preterm children instead of high birth weight or growth as a combined measure. However, there are indications of an increased risk of hepatoblastoma in children with high birth weight compared to children with normal birth weight in two previous studies.8, 13 But the results were not statistically significant in regard to high birth weight, which could be explained by very few cases in the high birth weight category.
High birth weight has been associated with increased risk of other childhood cancers,36–38 and two different hypotheses have been suggested as plausible explanations of the associations between accelerated growth, high birth weight and childhood cancer. First, high birth weight could be seen as an indication of greater number of cells, resulting in more cell divisions which could increase the vulnerability to carcinogens.39 Second, the insulin-like growth factor 1 might play a role, because the hormone is associated with increased growth and, furthermore, it has the potential to stimulate the proliferation of cancer cells in utero.36, 40 Since high birth weight per se was not associated with hepatoblastoma, it is expected that the association between LGA children and hepatoblastoma may only partly be explained by the mass of the tumor.
It was found that LGA children with hepatoblastoma are more likely to be diagnosed within the first year of life compared to AGA and SGA children. This could imply that the mechanisms behind development of hepatoblastoma in LGA children are happening at accelerated pace, compared to hepatoblastoma in AGA and SGA children. Additionally, it could indicate that the tumors in LGA children develop in utero. However, the younger age at diagnosis in LGA children might also reflect screening, as some LGA children could have features of Beckwith Wiedemann syndrome, which is known to be associated with accelerated growth.41 Children with Beckwith Wiedemann syndrome are at higher risk of neoplasms including hepatoblastoma4 and, therefore, they are followed more carefully for these outcomes.
In contrast to previous studies,9, 13–15 no association was seen between maternal smoking and hepatoblastoma in our study. A working group at the International Agency for Research on Cancer recently classified the link between parental smoking and hepatoblastoma in their offspring to be well documented with sufficient evidence (Class 1 carcinogen) within their Monograph program on the evaluation of carcinogenic risks to humans.12 We performed a meta-analysis of all five studies (including the present one) and the pooled risk estimate was 1.7 (CI: 1.1–2.7) showing a significant association (using Rothman's EPISHEET based on an algorithm by Fleiss42) (Fig. 1). However, there was substantial heterogeneity (p for homogeneity was 0.02). Our prevalence of smoking mothers was somewhere in between the other studies, i.e., 19%, with other studies ranging from 12 to 44%. There was also no systematic pattern when comparing studies assessing smoking data by interview compared to extracting the information from birth registries or hospital records.
The smoking status of the mother in our study was registered in the medical birth registries before the child was diagnosed, which minimizes the risk of recall bias compared to previous studies in which data on exposure was obtained retrospectively. A risk of desirability bias should also be considered when interpreting the results. However, at the same time, there is a chance that some of the women were registered as smokers, but then actually smoked very little during pregnancy, since several studies have indicated that a measurable proportion of women stop smoking when pregnant. Although we had some missing data on smoking status, it is unlikely that this caused a substantial dilution of the association. Data on paternal smoking were not available, but it would have been interesting to include this parameter as a previous study has found a stronger association with hepatoblastoma when both parents smoke.15
The most important strength of our study is the long study period in the four countries combined, resulting in one of the largest studies for hepatoblastoma until today. Furthermore, childhood tumor registration in the Nordic countries provides a unique tool for addressing these questions, as the Nordic cancer registries have a long lasting history of high-quality registration; they are mandatory and regulated by law; and they are validated or supplemented with other independent data sources, such as pathology registries, hospital discharge registries, death certificate registries and independent childhood cancer registries, which ensure a virtually 100% coverage.
Our study has a high degree of validity, because it is population and register based, and thereby has a reduced risk of selection and recall bias. The risk of recall bias is furthermore reduced, because data on exposure was registered prospectively and was therefore independent of disease status.
However, there are some concerns to consider, when interpreting the results. Risk estimation could have been influenced by confounders we had no information on such as malformations or overgrowth syndromes. The sample size remains small, due to the rarity of the tumor, thus, CIs of the risk estimates are broad. Furthermore, misclassification of diagnosis is possible. Several children were diagnosed with hepatoblastoma after the age of 5 years. Hepatoblastoma is extremely rare after the age of 5 years, which could indicate that these children may have had, e.g., hepatocellular carcinoma.
In conclusion, this Nordic population-based study provides some evidence that hepatoblastoma might have a prenatal origin and adds to the increasing scientific evidence of an association between very low birth weight and hepatoblastoma. Furthermore, our study has also contributed new knowledge of an association between accelerated growth and the risk of hepatoblastoma, which has not been found in previous studies. Our study did not confirm previously reported associations between maternal smoking and risk of hepatoblastoma. The potential links with neonatal exposure require further studies, as their confirmation will increase our understanding of the etiology of these childhood tumors. Preferably, future studies should include medical records to explore prenatal and neonatal supportive care. Furthermore, genome-wide association or gene-targeted studies may clarify whether children with hepatoblastoma have adverse drug metabolism and if so, adjustment of future neonatal treatments could reduce the risk of this early childhood cancer.
We thank Mr. Aslak Harbo Poulsen and Ms. Pernille Frederiksen (Institute of Cancer Epidemiology, Copenhagen) for their IT support. Kjeld Schmiegelow holds the Danish Childhood Cancer Foundation Professorship in Paediatrics and Paediatric Oncology.