Work

Phototherapy and childhood cancer: Shared risk factors

Year: 2019

Type: letter

Abstract: Dear editor, We read with interest the recent paper on phototherapy and risk of childhood solid tumors.1 While this study raises interesting questions, it is known that a number of childhood cancers have prenatal origins. Studies of newborn blood spots have shown that certain characteristic translocations of acute lymphoblastic leukemia (ALL) originate in utero during fetal hematopoiesis.2 Furthermore, the broad category of embryonal tumors (which includes retinoblastoma, Wilms tumor, neuroblastoma, rhabdomyosarcoma, hepatoblastoma and medulloblastoma) by definition arise from premature tissues of the embryo. Research has shown that the pregnancies of children who later developed cancer have a higher incidence of complications including some that may also increase the likelihood of hyperbilirubinemia and need for phototherapy. Indeed, some risk factors for severe hyperbilirubinemia are also independent risk factors for certain childhood cancers, including prematurity, low birthweight, intrauterine growth restriction, primiparity, gestational diabetes, small or large size for gestational age, and (in the case of germ cell tumors) Asian heritage.3-7 Thus, this raises the question as to whether the hyperbilirubinemia and/or phototherapy are true causes of cancer, or if observed associations are instead a result of shared risk factors. We investigated this question in a case–control study of Danish children.4 In our study, cases (diagnosed <age 20 years) were identified from the Danish Cancer Registry and controls were selected at random from the Central Population Register and matched by sex and birth year (25:1 matching rate). We broke the matching in the present analysis to have better statistical power to examine this rare exposure. We identified diagnoses of hyperbilirubinemia or neonatal jaundice or procedure codes for phototherapy occurring in the first 28 days after birth (Supporting Information Table S1) from the Danish National Patient Register or the Medical Births Registry. We did not have access to information on bilirubin concentration of children. Analyses included children born after the establishment of the National Patient Register (births 1977–2013). Our study was approved by the Danish Data Protection Agency and the University of California, Los Angeles. All newborns have elevated bilirubin concentrations compared to adults, yet only 5–15% of US newborns receive phototherapy treatment.8 Therefore we estimated the risk for cancer with having either a hyperbilirubinemia diagnosis or a procedure code for phototherapy. Because phototherapy is extremely common among children born before 34 weeks gestation, we excluded these children from analyses (n = 3,856). Analyses were conducted with unconditional logistic regression with adjustment for the matching factors, sex and birth year. To follow Auger et al.'s approach closely, we additionally adjusted for maternal age. We then stratified by birth weight and gestational age to determine whether increased risk might be explained by these factors. We did not adjust for congenital anomalies or Cesarean section as the inclusion of these factors in models may result in a collider stratification bias. We report results for cancer types with at least five exposed cases, limiting to cancers for which there were sufficient N to examine subgroups. Because embryonal tumors were rare, we combined all embryonal cancer types for analyses. Also, after Auger's approach, we restricted analyses to children whose cancer diagnosis occurred at least 2 months after the first diagnosis of hyperbilirubinemia or procedure code for phototherapy. We additionally examined whether results differed by the presence of congenital anomalies or size for gestational age (as defined previously9), but results were similar across groups. Although astrocytoma and glioma include overlapping cases, we presented both, to thoroughly investigate possible associations with central nervous system tumors. Stratification by sex or ethnicity was not possible due to small sample sizes. Results are shown in Table 1. Although the overall results among all births might suggest an elevated cancer risk with hyperbilirubinemia or phototherapy, stratified analyses showed that point estimates differed in subgroups. For most cancer types, children born preterm or low birth weight had considerably higher effect estimates for cancer than children born at term or normal birth weight, with the exception of ALL, for which estimates were higher with greater birth weight (an established risk factor for this cancer3). Although phototherapy may be underreported in our data source, in a sensitivity analysis we restricted to patients with hyperbilirubinemia who did not have a procedure code for phototherapy. Findings were not markedly different from those in the overall sample (Supporting Information Table S2). Auger and colleagues found higher point estimates for children diagnosed with cancer between ages 4–11 years. In a sensitivity analysis, we limited to that age range. Although sample sizes were limited for many cancer types, we were still not able to observe compelling increases in risk for any solid tumors among at-term and normal birthweight children (Supporting Information Table S3). If phototherapy increases risk for childhood cancer, one would expect similar point estimates across the subgroups we conducted stratified analyses for, but we did not observe this. Auger and colleagues found associations with solid tumors which they stated was particularly driven by associations with central nervous system (CNS) cancers, but we observed no compelling increase for these cancers when we restricted to term births and children with normal birth weights. Our results raise the question as to whether hyperbilirubinemia or phototherapy is truly a cause of childhood tumors, or instead a sign of adverse prenatal events predisposing to both the condition and the treatment. We cannot exclude the possibility that hyperbilirubinemia or phototherapy may have weak associations with some cancer subtypes. The two-hit hypothesis, originally suggested as an etiologic process for retinoblastoma and sometimes raised for ALL, states that a "second hit" occurring either later in the prenatal period or during postnatal life is necessary for cancer development.10 Yet, evidence of altered fetal growth observed for many childhood cancers, a surrogate for an adverse in-utero environment, supports a prenatal origin.3 The American Academy of Pediatrics provides clear guidelines for phototherapy initiation. Until there is additional and consistent evidence to suggest causation, it is vital to remember that severe hyperbilirubinemia should be treated with phototherapy. Extremely high concentrations of bilirubin may require exchange transfusion and can cause kernicterus, a rare, but preventable cause of neonatal death and mental impairment. Yours sincerely, Julia E. Heck Xiwen Huang Kara L. Calkins Yuelian Sun Jorn Olsen Beate Ritz Johnni Hansen The data that support the findings of our study are available from national Danish registries, and restrictions apply to their availability. Table S1 Diagnostic and procedure codes for neonatal jaundice and hyperbilirubinemia and phototherapy Table S2. Risk of cancer among Danish children born between 1977 and 2013 who were diagnosed with neonatal jaundice or hyperbilirubinemia but who did not receive phototherapy (n = 7,763 cases, 195,326 controls) Table S3. Risk for cancer among children born between 1977 and 2013 who were either diagnosed with neonatal jaundice or hyperbilirubinemia or who were treated with phototherapy, limiting to cancer cases diagnosed between ages 4–11 (n = 2,013 cases, 195,877 controls) Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.

Source: International Journal of Cancer

DOI: 10.1002/ijc.32701