Long-standing efforts to identify reproductive factors associated with risk of pediatric ALL and AML have not yielded clearly established etiologic factors. Associations of prior maternal fetal loss with increased risk of acute pediatric leukemias [11, 12] have not been confirmed in recent studies of pediatric [13, 14] or infant [15] leukemia. Inconsistent findings also have been reported for increased risks in firstborns [13, 14, 16, 17], young maternal age at birth ­[13–15, 18, 19], and advanced maternal age [14, 20, 21], with the latter potentially affected by changing maternal birth cohort patterns [22].

Decades of efforts to identify infectious agents causing pediatric ALL have been unsuccessful, and more recent studies of maternal medical conditions and/or treatments generally have not been replicated. It has long been postulated that a viral agent infecting a susceptible mother during pregnancy may play an etiologic role in pediatric leukemias [23]. However, reports linking in utero influenza infection [24] and Epstein–Barr virus [25] with pediatric leukemia and ALL, respectively, were not subsequently confirmed [26, 27], and specific leukemogenic infectious agents have not been identified [28–30].

Maternal hypertension has been linked with pediatric ALL [19] and AML [18], but the association has not been replicated, and no relationship has been found for preeclampsia [14] or gestational diabetes [18, 19]. Also requiring replication are findings of excess risks of pediatric common B-cell precursor leukemia in offspring of women with a previous molar pregnancy [14] and those receiving antibiotics [31] or mind-altering drugs during pregnancy [32], pediatric AML in offspring of mothers with polyhydramnios or anemia [14], and pediatric hematopoietic malignancies in offspring of mothers treated previously with infertility drugs [13, 33]. An excess of pediatric leukemia in mothers using marijuana before or during pregnancy [34] was not confirmed [35].

Diagnostic X-ray exposures from the late 1940s through the mid-1970s have been associated consistently with modest elevation in risk of pediatric leukemia, although risks have declined over time as radiation doses have decreased [36, 37]. Increased risk of pediatric leukemia was first linked with fetal exposure to abdominal or pelvic diagnostic X-ray of the pregnant mother more than 50 years ago [16]. Subsequently, additional data from the large Oxford Survey of Childhood Cancers, a medical record-based study in hospitals in the northeast United Kingdom (UK), and close to 30 other case–control studies support an overall RR of 1.4 (i.e., 40 % increased risk) for pediatric leukemia associated with diagnostic ionizing radiation exposure in utero, but the interpretation of the statistical association has been debated [37]. Data are insufficient to determine risks of pediatric leukemia in offspring of women undergoing radiotherapy for cancer, and there is no evidence linking ultrasonography during pregnancy with risk of pediatric leukemia [36].

Data evaluating leukemia risk with in utero exposure to ­environmental sources of ionizing radiation are limited. The cohort study of survivors who were in utero at the time of the atomic bombings of Hiroshima and Nagasaki “…cannot provide information on the effect of radiation on the incidence of childhood cancers” in the absence of complete information for the first 5 years of follow-up and insufficient size to assess such rare outcomes [38]. Similar limitations characterize a cohort of persons who were in utero at the time of the Chernobyl accident [39]. Existing data are not informative about pediatric leukemia risks in offspring of female airline crew, women living in proximity to nuclear testing or nuclear installations, or women exposed to residential radon, gamma, or other natural background radiation [40].

A small study linking paternal preconception occupational exposure of nuclear industry workers with excess leukemia and lymphoma risks in offspring [41] was not confirmed in a study of 39,557 children of male nuclear workers [42]. Furthermore, there was no excess of pediatric leukemia in offspring of female or male medical radiation workers in the UK [43] or the USA [44]. A meta-analysis of all studies of parental preconception occupational exposure to extremely low-frequency magnetic fields (ELF-MF) revealed a borderline significant increased risk of childhood leukemia in offspring (RR  =  1.35, 95 % CI  =  0.95–1.91) [45]. However, the studies in the meta-analysis included substantial heterogeneity, lacked detailed exposure data, and did not consider occupational chemical or other confounding exposures. In addition, the largest and most detailed epidemiologic study showed no association [45].

Risks of leukemia in offspring of parents occupationally exposed to pesticides have been extensively investigated, but few investigations have identified the specific pesticides to which the fetus has been exposed. Meta-analysis of more than 30 case–control and five cohort studies revealed similarly increased risks for both ALL and AML (RR  =  2.64, 95 % CI  =  1.00–5.00; RR  =  2.64, 95 % CI  =  1.48–4.71, respectively) with maternal occupational exposure, with higher risks in studies of farm-related exposures compared with studies of mixed or unknown place of exposure [46–49]. No consistent association was observed for childhood leukemia with paternal occupational exposure to pesticides [48].

Findings from epidemiological studies of pediatric leukemia in offspring of parents exposed to heavy metals (including arsenic), polychlorinated biphenyls, dioxin, indices for outdoor air pollution, traffic density, drinking water disinfection by-products, and specified and unspecified solvents are inadequate to estimate risks [50] using definitions proposed by the US National Academy of Sciences [51]. Limited evidence has linked pediatric leukemia in offspring with paternal preconception exposure to motor vehicle emissions, driving or inhaled particulate hydrocarbons [52] or with maternal exposure to unspecified solvents [50].

The role of maternal smoking in the development of pediatric leukemia has been evaluated for several decades, with a few studies reporting an increased risk. However, a meta-analysis [53] and subsequent studies concluded that maternal smoking during pregnancy is not a major risk factor for childhood acute leukemia [54]. Paternal preconception smoking has been investigated less extensively, but a recent meta-analysis of nine studies showed a modest association with childhood ALL (RR  =  1.17, 95 % CI  =  1.04–1.30) [55].

A comprehensive assessment of pediatric leukemia risk and parental alcohol consumption in 33 case–control studies did not strongly support an association [56]. Paternal preconception consumption of alcohol also has not been linked with elevated risk [57]. However, a meta-analysis of 21 case–­control studies concluded that maternal consumption of alcohol during pregnancy was associated with increased risk of pediatric AML (RR  =  1.56, 95 % CI  =  1.13–2.15), but not ALL [58]. Reasons why in utero exposure to alcohol increases AML but not ALL risk are unknown.

The possible role of maternal diet or vitamin supplements in risk of acute leukemia has only begun to be investigated systematically, and to date only a few studies have reported results. Three investigations found reduced risks of pediatric leukemia in offspring of mothers who consumed higher levels of fruits and vegetables during pregnancy [4, 59, 60]. Spector et al. found that maternal consumption of foods that inhibit DNA topoisomerase II was linked with increased risk of infant AML [4]. A recent meta-analysis [61] showed weak evidence of a reduced risk of developing childhood ALL associated with maternal folate use before pregnancy, and other studies also found reduced risk with maternal use of iron supplements [32, 62].

High birth weight has been linked consistently with increased risk of ALL, and recent studies suggest a similar relationship for AML. A meta-analysis of 31 studies demonstrated significantly increased risks of ALL (RR  =  1.23, 95 % CI  =  1.15–1.32) associated with birth weight of ≥4 kg, and seven of these studies showed a similar relationship for AML (RR  =  1.40, 95 % CI  =  1.11–1.76) [63]. It has been suggested that accelerated growth rather than high birth weight per se is etiologically linked with ALL [64], but further epidemiologic investigations are needed. Low birth weight (<1.5 kg) has been associated with increased risk of AML (RR  =  1.49, 95 % CI  =  1.03–2.15) but not ALL [63].

Large case–control studies examining the association of breast-feeding and risk of childhood acute leukemias have reported 9–20 % decreased risk for ALL, but a less consistent picture for AML, ranging from no reduction to 23 % reduction in risk [65, 66]. A meta-analysis found that ever having been breast-fed was associated with a 9 % reduced risk of ALL based on 17 studies and no reduction in risk of AML based on nine studies [67]. Although few studies have examined risks for combined breast-feeding and milk supplementation, one reported a moderately increased risk of childhood leukemia among those infants who received milk supplementation with breast-feeding more than 50 % of the time [68].

Clusters (a group of cases representing an excess intensity within the population-at-risk which is unlikely to be due to chance) of pediatric leukemia have been described for decades, but no infectious or other causal agent(s) have been linked with these clusters [69, 70]. A comprehensive, broad-based epidemiologic and laboratory investigation of the largest pediatric acute leukemia cluster to date, diagnosed in residents of Churchill County, Nevada, was unable to identify a causal agent [71].

Two mechanisms have been suggested by which infectious agents might be associated with the age peak from 2 to 5 years in the common form of pediatric ALL. Greaves postulated a two-hit model of leukemogenesis, with the first event or mutation occurring in rapidly dividing immature B cells in utero and the second event arising in early childhood as a result of delayed exposure to infectious agents [72]. Epidemiologic assessment of surrogate measures related to the timing of exposure to infectious agents, such as daycare attendance, birth order and interval between births, and the timing, frequency, and severity of infections in the first 2 years of life generally supports a reduced risk of ALL associated with early and substantial daycare attendance [73] and social activity in the first year of life [74], but findings for other measures are not as consistent [75]. However, findings of significantly more clinically diagnosed infectious episodes during infancy, particularly in the neonatal period, and ­earlier age of onset in children who subsequently developed ALL compared with controls suggest that a dysregulated immune response to infection increases risk of developing pediatric ALL [76]. Kinlen proposed that childhood ALL occurs as a rare response to a specific, albeit currently unknown viral agent(s), particularly when there is mixing of rural (or other low-density) populations with urban (or other high-density) populations [77]. Summarizing multiple extreme examples of population mixing in Britain, Kinlen observed significant short-term excesses of pediatric leukemia [78], although others reviewing the population mixing studies have disagreed with the conclusions drawn by Kinlen [79].

A limited number of studies investigating an infectious etiology for pediatric acute leukemia have reported a protective effect from vaccination in infancy or early childhood, with risks varying according to the type of vaccine and age at immunization [80]. A meta-analysis of epidemiologic studies examining atopy and risk of childhood leukemia revealed a 31 % significantly reduced risk of ALL based on six studies and no significant association for AML based on two studies [81]. Inverse associations were seen for ALL with asthma, eczema, and hay fever, but the authors caution about ­overinterpretation of the results in light of the limited number of studies, substantial heterogeneity, and potential misclassification.

In general, studies of low-dose postnatal diagnostic radiographic exposures (estimated organ doses range from 0.01 to 6 mGy) have not observed increased risks of pediatric leukemia; the results were based on questionnaire responses and estimated doses were not evaluated [82]. To date, there have been no reports assessing pediatric cancer risks in children who have undergone one or more higher-dose diagnostic imaging procedures, such as computed tomographic (CT) examinations (estimated organ doses from CT scans of the chest, abdomen, brain, spine, and face range from 10 to 80 mGy) or repeated radiologic examinations during the neonatal period, although studies of the former are underway.

Earlier cohort studies of infants or children irradiated for benign conditions were generally too small to estimate risk of pediatric leukemia accurately [83], but radiotherapy during childhood for tinea capitis was associated with a subsequent moderately increased risk of leukemia mortality (RR mortality  =  2.3, estimated bone marrow dose  =  30 mGy) [84]. Two of three earlier small studies found a dose–response relationship between estimated radiation dose to the bone marrow from radiotherapy for treatment of pediatric cancer and subsequent risk of secondary leukemia. A larger more recent investigation did not confirm this relationship; however, excess secondary leukemia was evident—likely due to chemotherapy treatment [85], a strong and well-established risk factor for acute leukemia. More information on the relationship between chemotherapeutic agents and acute leukemia risk is provided in the section of this chapter on adult leukemia.

The follow-up investigation of Japanese atomic bomb survivors, the most important epidemiological study of radiation-exposed populations for quantitative risk assessment, demonstrated that survivors who were less than 10 years old at exposure had moderate-to-high relative risks for both ALL and AML [86]. Relative and absolute risks were higher among children than adults, and these risks peaked within 10 years. The absolute risk for incidence was estimated to decrease by about 5 % for each year’s increase in age, and risk for ALL among females was about 40 % of that among males.

Results from other studies of postnatal environmental ionizing radiation and risk of pediatric acute leukemias are less clear. Leukemia and lymphoma risks were found to be increased in persons under age 25 among populations living near nuclear fuel reprocessing or weapons production plants in the UK (specifically Sellafield and Dounreay), but not among populations residing close to nuclear plants generating electricity; however, environmental radiation levels measured in proximity to these facilities were considered too low to ascribe to the radiation exposures from these plants [87]. Residential exposure to radon was linked with elevated risk of pediatric ALL in an ecological study but was not associated with an increased risk in studies in the USA [88], Germany [89], or the UK [90]. Although there was no overall association of residential radon with pediatric AML, a borderline increase was observed in children ages 2 or older [91].

Initial reports of two- to threefold excess risks of pediatric leukemia in children residing in homes with high levels of 60-Hz magnetic field exposures from residentially proximate power lines were assessed further in large studies with more extensive and direct measurements. Results from pooled analyses of these high-quality studies revealed that pediatric leukemia risks were not increased among children residing in homes with power frequency magnetic field exposures under 0.4 μT (which included more than 99 % of residences internationally), but were twofold increased among the <1 % of children living in homes with exposures ≥0.4 μT [92]. Reasons for the increased risk are unclear, particularly since experimental studies have not linked power frequency magnetic fields with carcinogenesis [93].

Residential use of insecticides demonstrated a dose–response relationship with total childhood leukemia [94] and ALL [95], but exposures to insecticides in agricultural settings were not associated with risk, nor was use of ­herbicides or fungicides in any setting [50]. Limited evidence supports a relationship of pediatric leukemia with childhood residential exposure to nearby high traffic density, car repair garages or gasoline stations [50], and with frequency of household use of petroleum products, solvents, and paint [5, 50, 96, 97]. No clear association was observed with pediatric leukemia risk and exposure to drinking water disinfection by-products, drinking water nitrate, residence near hazardous waste disposal sites, or exposure to environmental tobacco smoke [50].

Only a few small studies have assessed risk of pediatric leukemia with dietary components, and replication of the results is needed. Findings were mixed for consumption of cured meats [98, 99], and reduced risks were associated with intake of bean curd and vegetables in a study in Taiwan [99] and with consumption of foods containing vitamin C and/or potassium in a study in California, particularly during the earliest years of life [98]. Concern about a report linking intramuscular vitamin K with risk of pediatric leukemia [100] was dispelled following a pooled analysis showing no association [101].