In the United States (US), more than 15,000 children are diagnosed with cancer annually.^1,2 Few established risk factors are identified for pediatric cancer, including specific cancer predisposition syndromes (e. g., Li-Fraumeni syndrome). Recent studies have demonstrated that birth defects are one of the strongest risk factors for pediatric cancers.^3,4,5,6,7,8 Specific chromosomal birth defects are known to increase risk of specific pediatric cancers, for example, Down syndrome-acute leukemia and chromosome 13q14 deletion syndrome-retinoblastoma.^{7,9,10,11,12,13} Additionally, there is strong evidence that non-chromosomal birth defects are associated with cancer in children.^13-17 In fact, both major non-chromosomal birth defects (e.g., non-syndromic critical congenital heart defects) and minor birth defects (e.g., rib anomalies) are associated with increased risk of cancer in children.^14,16,18 However, the etiology for most cases remains unknown.¹⁹

While most studies have shown associations of birth defects with childhood cancers, fewer have explored this association by attained age. There is some evidence that cancer risk in these children is particularly high in the very young.^7,13,20-22 For example, Dawson et al. observed 1.74 times higher risk for any cancer among children with birth defects by 4 years of age.²¹ Additionally, in a recent study using data from Oklahoma, Janitz et al. demonstrated that children <1 year-old with birth defects had the highest risk of developing cancer compared to their contemporaries without birth defects.²² Based on this evidence, our objective was to assess the risk of cancer among children with and without birth defects in Arkansas by age of attainment. We also explored associations based on birth defects by organ system and by subtypes of pediatric cancers.

Maternal demographic and perinatal characteristics of the study sample with and without birth defects are presented in Table 1. In our population, 23,341 of 629,086 (3.7%) children were diagnosed with birth defects and 1,037 (0.2%) children were diagnosed with cancer. Among children with birth defects, children with cancer were more likely to weigh >=2500 g at birth (100% vs. 79.3%). Among children without birth defects, there were significant differences found in infant sex, gestational age, maternal education and race. Without the presence of birth defects, children with cancer were more likely to be male (54.1% vs. 45.9%), born ≥ 41 weeks gestational age (9.5% vs 7.1%), have maternal education > high-school (39.6% vs. 37.7%), and have maternal race as White (82.5% vs. 76.4%).

In the present analysis of cancer risk by age of attainment among children with birth defects in Arkansas, we observed the strongest associations among the youngest children. Particularly, we observed greater risk of cancer up to five years of age among children with chromosomal birth defects. Among children with non-chromosomal birth defects, higher risk for cancer was noted primarily during the first year of life. For specific non-chromosomal birth defects and any cancer, we observed significantly higher cancer risk for infants with defects of the cardiovascular, gastrointestinal, and musculoskeletal systems. Among cancer subtypes, we observed a similar pattern of greater risk in the first year of life among children with birth defects, particularly for embryonal cancers. We also noted an increased risk for embryonal and germ cell tumors in older children (≥10 years) with non-chromosomal birth defects, although these results are limited by sample size and should be interpreted with caution.

Our study findings are consistent with other studies that have explored associations between birth defects (chromosomal and non-chromosomal) and pediatric cancers. However, except for a recent study in Oklahoma by Janitz et al., ¹⁶ majority of these studies did not report findings by attained age groups; these studies reported greater risk for cancer among children with birth defects at younger ages.^7,13,20,21 For example, Botto et al. reported the highest hazard for cancer in the first three to five years of age among children with non-chromosomal birth defects.²⁰ Agha et al. reported nearly six-times higher risk among infants with any birth defect (relative risk 5.8; 95% CI 3.7-9.1).⁷ Additionally, Dawson et al. showed doubled risk of developing cancer among children born with a birth defect, especially in the first four years of life.²¹

After exploring overall associations for both chromosomal and non-chromosomal birth defects with pediatric cancers, we focused our further analyses on non-chromosomal birth defects by organ systems and subtypes of cancers. A growing number of studies have evaluated the association between non-chromosomal birth defects and childhood cancer, though not always specific to analysis by certain age groups.^13-17 Similar to these studies, our study confirms the strong association between cardiovascular defects and neoplasms in early childhood, especially during first year of life.^28,29 After excluding children with chromosomal birth defects or leukemia, Fisher and colleagues found more than a three-fold risk of cancer in children with CHDs.¹⁵ Two other studies have shown a 28%-45% increased cancer risk among children with CHDs.^28,30 Our findings on positive associations between pediatric cancers and gastrointestinal defects are consistent with a previous study by Fisher and colleagues, in which a statistically significant association (HR 2.44) was observed between birth defects of the digestive system and pediatric cancers.¹⁵

We also observed increased risk for sarcomas from ages 5-9 (HR 3.9; 95% CI 1.4-11.0). This is consistent with a previous study in Canada that found higher incidence rates for soft tissue sarcoma (1.2 vs. 0.6/100,000 person-years).⁷ A recent study of pooled data from four US states (TX, MI, NC, and AR) showed associations between spina bifida and different forms of sarcomas (non-RMS soft tissue sarcoma).¹⁷ Furthermore, our findings on non-chromosomal birth defects and CNS tumors were consistent with previous literature; a population-based study of 5.2 million children in Norway and Sweden found high standardized incidence ratios (SIR) for brain/nervous system cancers in children with birth defects (SIR 2.5-Norway; 1.3-Sweden).³¹ However, we highlight the highest risk during 1 to 4 years-old (HR 10.3, 95% CI 5.4-19.5).³¹

To our knowledge, only the study by Janitz et al. included age-of-attainment analyses of non-chromosomal birth defects, as ours.¹⁶ Of note, our results for non-chromosomal birth defects and pediatric cancer across age groups are consistent with this study. For example, similar to our results, cancer risk among children with birth defects was highest before age six years. Also consistent with our findings concerning the associations of non-chromosomal birth defects grouped by organ system and any cancer, they observed greatly increased cancer risk in the first year of life for infants with cardiovascular (HR 22.5; 95% CI 11.0-46.1), gastrointestinal (HR 21.1; 95% CI 7.0-63.2) or genitourinary defects (HR 15.6; 95% CI 5.0-48.6). In their analysis of cancer subtypes among infants with any non-chromosomal defect, they reported increased risk only for all cancers combined and CNS tumors (HR 25.0; 95% CI 4.8-130.0), whereas we also observed increased risk for hematologic malignancies, embryonal tumors, and germ cell tumors. These differences may be due to limited number of cancer cases in both studies or by differences in the underlying population.

The mechanisms for the association between birth defects and risk of pediatric cancers are largely unknown. One hypothesis by Narod et al. suggested that somatic alterations during earlier phases of embryogenesis can affect multiple developing tissues and thus increase the risk of both birth defects and cancer in the affected child.^5,22 This could be particularly important for malignancies that occur early in life. That said, other mechanisms including novel cancer predisposition syndromes or unidentified environmental risk factors that lead to both phenotypes. Finally, from a developmental perspective, birth defects and cancer may share common etiological pathways in some children, and this may explain the stronger association in early childhood.³²

Our study has several strengths. First, we utilized valid and reliable population based registry-linked data.³³ Registry data can reduce the biases associated with clinical or interview-based studies,^14,34 and provide the most accurate assessment of cancer occurrence in the general population.³³ Hence we consider this as a robust data source for our study. Another strength of our study is the use of age-stratified modelling to explore age-specific birth defect-cancer associations.

Our study must also be considered in light of certain limitations. Although we were able to evaluate cancer and birth defects in a large population-based sample, our co-occurring case counts were small, which limited the assessment of less common phenotypes, as well as relatively modest associations as seen in older age groups. A second limitation is possible loss to follow up for children with birth defects who relocated to another state after birth. However, a previous study has shown that the migration of these children is non-differential in nature based on presence or absence of birth defect.³⁵

Another potential issue to consider is that children diagnosed with cancer earlier in life (e.g., <2 years of age) may have received greater diagnostic scrutiny and could have received a birth defect diagnosis as a consequence of the cancer diagnosis. We limited the potential for this by focusing our assessment on major birth defects that are commonly ascertained regardless of cancer diagnosis. Additionally, other assessments have evaluated the impact of this potential bias by excluding children diagnosed with cancer at early ages, and in sensitivity analyses, these exclusions did not impact the observed birth defect-cancer associations.¹⁷ Conversely, it has been hypothesized that children with birth defects are also likely to get an earlier diagnosis of cancer due to more robust disease surveillance. In another recent paper, authors investigated stage at cancer diagnosis by birth defects status and did not observe earlier diagnoses of cancer among children with birth defects.³⁶ Based on these findings, we propose that differential diagnostic scrutiny or disease surveillance does not substantially impact our findings.

In conclusion, this study adds to the growing body of evidence that children with non-syndromic birth defects have a higher risk of cancer compared to children without birth defects. Additionally, we have demonstrated that these associations are much stronger for young children with birth defects. While the mechanisms underlying these associations are not clear and likely to be different according to a specific birth defect-cancer pattern, it is hoped that our study can inform future cancer surveillance strategies among younger children living with birth defects.