Association of Parental, Child, and Environmental Factors with the Occurrence of Childhood Leukemia in Upper Egypt

Background: Leukemia is the most common malignant tumor in children, responsible for nearly one-third of all childhood cancers. However, the exact risk factors of childhood leukemia are


INTRODUCTION
lobally, childhood cancer is considered the ninth leading cause of childhood disease burden. (1) Leukemia is the most common type of childhood cancer, representing nearly one-third of all pediatric cancers, and acute lymphocytic leukemia (ALL) is the most common subtype of leukemia. (2,3) In Egypt, childhood leukemia represents 35.6% of all incidents of childhood cancer. (4) ALL is the most common subtype. Most of the affected children are four years or younger. (5) Early diagnosis of cancer is a crucial priority in health care. Many pediatric leukemia symptoms are similar to symptoms caused by non-leukemia mild diseases. This lack of specific symptoms complicates the diagnostic challenge faced by front-line clinicians, highlighting the need for recognizing the possible signs and symptoms of this disease. (6,7) Childhood leukemia's etiology remains largely unclear. Many epidemiological studies associate certain genetic, childbirth, parental, and environmental factors with leukemia risk. Leukemia is assumed to develop in response to multi-factorial interactions between genetic and environmental factors. (8) A positive family history of cancer may predispose children to the development of leukemia. (9) A meta-analysis and review reported subtypespecific leukemia associations with both low and high birth weights. (10) Maternal history of fetal loss and birth by cesarean section were also associated with an elevated risk of childhood leukemia. (11,12) The oldest or youngest G Original Article extremes of parental age reportedly played a role in the epidemiology of specific leukemia subtypes. (13) Regarding environmental factors, American and Canadian case-control studies associated exposure to postnatal diagnostic radiation with an increased childhood leukemia risk. (14,15) Parental smoking or passive smoke exposure of mother or child also posed a potential risk for child leukemia. (16,17) Growing evidence revealed positive associations between childhood leukemia and childhood exposure to household or agriculture pesticides. (18)(19)(20)(21) Published Egyptian studies on risk factors of childhood leukemia are limited, (16) and no previous study explored these risk factors in Upper Egyptian children. This study aimed to identify potential child, parental, and environmental factors that contribute to leukemia risk among Egyptian children in Assiut Governorate.

Study design and setting and sampling
We conducted a case-control study in Assiut Governorate in Upper Egypt. The capital city is Assiut, about 375 km south of Cairo. In 2018, the total population of the Assiut Governorate was approximately 4,527,565.
The sample size was calculated for the case-control design using the Open Epi program, version 3.01, updated in 2013, based on an odds ratio of 2.01 (22) , a confidence interval of 95%, a power of 80%, and control:case ratio of 1:1. The calculated sample size was 156 cases and 156 controls. The sample size was rounded to 170 cases and 170 matched controls.
Inclusion criteria for the cases were leukemic children aged ≤18 years old and recently diagnosed (within one year) at the South Egypt Cancer Institute (SECI) at Assiut University. The leukemia diagnosis was based on a hematological and histopathological examination conducted at SECI. The control group included children matched with cases by age (within a 6-month range) and gender. The control subjects were children diagnosed with mild illness and free from hematological problems (based on complete blood count examination). We recruited the controls from children attending the general outpatient clinic of the Children's University Hospital at Assiut University.

Data collection
Data collection was carried out from February 2017 to March 2019. We collected data for both cases and controls using a semi-structured interview questionnaire completed by the children's mothers. The type of leukemia was obtained from patient sheets in the pediatric unit in SECI. The following data about both controls and cases were collected from the questionnaires: o Personal and birth characteristics and medical history of children, including age, gender, residence, birth order, weight at birth, breastfeeding, daycare attendance, history of hospital admission before the current illness, history of prior tumor occurrence, previous exposure to diagnostic radiation, and mode of the child delivery. (16,17,(23)(24)(25)(26)(27)(28) We assessed exposure to diagnostic radiation among cases before illness and among controls up to the time of the questionnaire. o Parental and family factors, including father's and mother's ages at the time of childbirth, parental consanguinity, parent educational level/occupation, family income in Egyptian pounds (LE), and positive family history of cancer and/or leukemia. (16,23,27,29) o Maternal histories of abortion and radiation exposure before this pregnancy. (

Statistical analysis
We analyzed the data using SPSS version 25. Chi-square tests (ꓫ 2 ) compared the distribution of frequencies among different groups . Fisher's exact test determined expected counts <5. Based on testing data normality for quantitative variables, Mann-Whitney U test was applied. Multivariate logistic regression analysis identified childhood leukemia predictors. The presence of leukemia (yes/no) was considered the dependent variable. The significant variables resulting from the bivariate analysis were entered as explanatory variables in the multivariate regression model . Odds ratios were calculated to measure the association between the different risk factors and childhood leukemia at 95% confidence intervals. A pvalue less than 0.05 was considered statistically significant.

Ethical considerations
The Assiut Medical Ethics Committee approved the study. Researchers complied with the International Guidelines for Research Ethics. The mothers of the recruited children provided written informed consent. Illiterate mothers provided informed consent in the presence of a witness. The study aim was explained individually to each mother before filling out the questionnaire. We preserved the anonymity and confidentiality of all obtained information. The authors have no conflict of interest to declare.

RESULTS
Among leukemic children (n = 170) included in the study, the most common leukemia subtype was ALL (81.8%), followed by acute myeloid leukemia (AML) (17%); chronic myeloid leukemia represented 1.2% of the studied cases, as illustrated in Figure 1. The first reported complaints of leukemic illness were recurrent fevers (74.1%), joint pain (28.2%), hematological problems (24%), and abdominal enlargement/pain (12.9%), as shown in Figure 2. Table 1 lists the personal, birth and medical histories of both cases and controls included in the study. Among the cases, nearly 63% of childhood leukemia cases were males, and more than one-third of cases (37.6%) were 5 years or younger. Childhood leukemia was significantly associated with rural residence, abnormal birth weight, and previous exposure of the child to diagnostic radiation (p-value ≤ 0.001). However, leukemia was not significantly associated with birth order, delivery mode, breastfeeding, daycare attendance, or childhood history of other cancers. More than one response was allowed. Hematological problems were as follows: pallor, epistaxis, ecchymosis, and petechia.   Table 2 presents parental criteria, family history of malignancy, and monthly household income among studied children. There were no significant differences between cases and controls regarding the ages of the mother and father at the time of childbirth, the father's educational level, parental consanguinity, the mother's work, and previous fetal loss before the index child. However, childhood leukemia was significantly associated with maternal education; 58.8% of case-patient mothers were illiterate than 41.2% of control patient mothers (pvalue = 0.004). Childhood leukemia was significantly associated with the mother's preconception exposure to radiation, the father working as a farmer, a family history of leukemia, and a family history of other types of malignancies (pvalue < 0.05). The median monthly household income among leukemic cases (800 LE) was significantly lower than that among controls (1200 LE) (p-value < 0.001). Table 3 shows the relationship between childhood leukemia and exposure to environmental factors. A significantly higher proportion of leukemic cases were exposed to paternal smoking and other home smokers than exposure among controls (p-value < 0.001). Moreover, a significantly higher proportion of leukemic cases were exposed to household and/or agricultural pesticides than exposure in controls (pvalue < 0.05). Similarly, residence near power line/substations (≤200 M) was significantly associated with childhood leukemia (p-value = 0.002). On the other hand, there was no significant association between living near benzene stations, main roads, or factories and childhood leukemia (p-value > 0.05).   Table 4 shows the multivariate logistic regression model for childhood leukemia risk factors. Predictors of childhood leukemia exposure to agricultural pesticides (OR = 5.369), household pesticides (OR = 2.009), or previously exposed to diagnostic radiation (OR = 3.951). Moreover, the significant predictors of

DISCUSSION
Childhood leukemia is the most common cancer among children. This case-control study's objective was to identify potential associations between child, family, and environmental factors and the risk of leukemia among Egyptian children ≤18 years. In the current study, the most common leukemia subtype was ALL, which represented about 82% of leukemia, and the most affected age group was children 5 years or younger. This is consistent with American, Brazilian, Libyan, and Canadian reports showing that ALL formed about three-fourths or more of leukemia cases in children and teens, and childhood leukemia was more common in children less than four years of age. (23,(32)(33)(34) Greaves explained the disease occurrence at an early age by in utero exposure to possible maternal and perinatal risk factors. (35) The earliest presentations of childhood leukemia are nonspecific symptoms. These nonspecific symptoms may mislead general practitioners. The first complaints that led to seeking health care in this study were fever (74%), joint/ bone pain (28%), hematological problems such as pallor and petechia (24%), and abdominal enlargement (12.9%). The common symptoms and signs were similar to other studies but with different orders of frequency. A metaanalysis study showed that the five predominant signs and symptoms in more than 50% of children were hepatomegaly (64%), splenomegaly (61%), pallor (54%), fever (53%), and bruising (52%) (6) . A retrospective Iraqi study reported the initial presenting features of childhood leukemia were fever (81.8%), pallor (67.3%), hepatosplenomegaly (56.4%), lymphadenopathy (49.1%), bone pain (40%), anorexia (36.4%), and mucocutaneous bleeding (27.3%). (36) Reports vary on the relationship of residence with childhood leukemia. Our study showed that residence was not a risk factor for childhood leukemia. Consistent with our report, there was no significant impact of residence (urban/rural) on the incidence of childhood leukemia in another Egyptian case-control study carried out in 2016. (16) In contrast, there is a theory that children living in rural areas tend to have higher leukemia rates. This might be explained by the immunologically active rural environment. (30) In the present study, children with an abnormal birth weight (less than 2500g and more than 4000g) had a significantly higher risk for developing childhood leukemia. Similarly, other studies reported that high birth weights or both high and low extremes of birth weights (U-shaped) elevate the risk of childhood leukemia. (9,10) High birth weights might be associated with a higher rate of cell proliferation and, subsequently, an increase in the precursor cells at risk for malignant changes. (9) Children with previous exposure to diagnostic radiation had a significantly higher risk of developing childhood leukemia. Consistent with this result, American and Canadian case-control studies associated exposure to postnatal diagnostic radiation with an increased risk of childhood leukemia. (14,15) In the current study, protective childhood factors, such as breastfeeding and daycare attendance, were not significantly associated with childhood leukemia. These results disagreed with an American case-control metaanalysis and review studies. These studies showed that breastfeeding and daycare attendance were associated with reduced childhood leukemia risk. (26,27,37) The protective role of daycare attendance could be explained by Greaves' delayed infection hypothesis. (35) Daycare attendance increases the occurrence of common infections. Common infections early in life may play a protective role against acute leukemia. In the current study, the protective effects of breastfeeding and daycare attendance may be concealed; a vast majority of studied mothers of both cases and controls breastfed their children and did not send their children to daycare centers.
The mode of delivery was not significantly associated with childhood leukemia. This finding disagreed with an Egyptian case-control study that reported the association between pediatric leukemia and cesarean section (16) . Also, an American case-control study reported significantly increased odds of ALL for cesarean section, while there was no significant effect of cesarean section on AML risk. (28) Likewise, there was no significant association between parental age (father and mother age) at the time of delivery and childhood leukemia. This finding disagreed with several studies showing that advanced parental age is associated with childhood risk of ALL. (13,16,38) Maternal age was associated with the risk of childhood AML in a Ushaped manner at both oldest and youngest extremes. (13) Maternal educational level was not a predictor of pediatric leukemia. Variations in the association of childhood leukemia with maternal education were reported. Childhood leukemia was significantly associated with low maternal education in India. (31) In contrast, an Egyptian study reported that children of mothers with secondary or higher educational levels had significantly higher odds for developing acute leukemia (16) . This study explained the association of maternal education level with childhood leukemia as a reflection of an association with western lifestyle. (16) Other studies showed a link between childhood leukemia and higher socioeconomic status. (42)(43) Father occupation, especially agricultural work, had a significant impact on the risk of childhood leukemia. Children of farmer fathers had a significantly higher odds ratio for having childhood leukemia. Mexican and Iranian case-control studies reported similar results. (39,40) Previous abortion before the index child was not significantly associated with childhood leukemia. Along that same line, there were no significant links between fetal loss and childhood leukemia in Indian and American casecontrol studies. (31,41) Contrary results were reported in a Greek case-control study conducted on 1099 ALL cases and 131 AML cases. Statistically significant exposure and disease subtype-specific associations of previous miscarriage(s) were exclusively associated with AML, and stillbirths were associated with ALL in the Greek study. (11) A history of fetal loss may reflect genetic predisposition, abnormal intrauterine environment, or the effects of a common environmental exposure. (41) During the preconception period, mothers' exposure to diagnostic radiation was not a risk factor for childhood leukemia. An American case-control study reported similar results; exposure to x-rays during the maternal preconception was not a risk factor for AML. (14) Family monthly income less than 2000 LE was a risk factor for developing childhood leukemia in the present study. Similarly, other studies reported a relationship between family poverty and the occurrence of childhood leukemia. A Malaysian case-control study found that family income ≥2000 RM resulted in a significantly lower risk for childhood leukemia. (29) An Iranian case-control study reported a negative association between acute childhood leukemia and family income. (42) A Norwegian cohort study showed that family poverty during the first two years of life was a risk factor for developing lymphoid leukemia before the age of 15 years (odds ratio: 1.72, 95% CI: 1.11-2.64). (43) There is a growing evidence for the role of potential genetic factors in the occurrence of childhood leukemia (8) . In this study, a family history of leukemia and/or other types of malignancies resulted in a significantly higher odds ratio for childhood leukemia. These results were consistent with Greek, Iranian, Malaysian, and Canadian case-control studies reporting a familial contribution to childhood leukemia. (9,29,40,44) The relation between parental smoking and pediatric leukemia remains unclear. In the current study, there were significant differences between cases and controls. The cases had significantly higher proportions of paternal smoking and/or other smokers' presence at home (p-values 0.047 and <0.001, respectively). However, exposure to smoking was not a risk factor for childhood leukemia with adjusted regression. A Chinese case-control study reported contradictory results; in this report, childhood exposure to smoking significantly increased the risk of developing acute pediatric leukemia. (17) Also, father smoking was a risk factor for acute leukemia in Malaysian children. (29) In this study, the proportion of children exposed to household and/or agriculture pesticides was significantly higher in the leukemia cases than that of the control group. These results are consistent with other studies that linked exposure to household pesticides with an elevated risk of childhood leukemia. (17,20,45) Consistent with these studies, a Costa Rican case-control study found an association between elevated childhood leukemia incidence and maternal insecticide use in the home and pesticide spraying on nearby farms before and after the child's birth. (46) An Italian case-control study showed an increase in leukemia risk among children residing close to arable crops. This study emphasized the need for further investigations into the role of passive exposure to herbicides and pyrethroids in disease occurrence in the majority of cases. (18) The risk of pediatric leukemia upon exposure to certain types of pesticides, either in utero or after birth, could be explained by the probability of pesticide inhibition topoisomerase II and oxidative stress production, which lead to single-and double-strand DNA breaks and, consequently, chromosomal aberrations in hematopoietic stem and progenitor cells. However, this evidence supports the triggering of infant leukemia, but not childhood leukemia, which involves further postnatal events for overt disease. (47) Regarding residence of the children close to power lines, residence within 200 m of power lines was not a significant predictor of childhood leukemia. This result disagreed with results from Malaysian and British casecontrol studies in which children living within 200 m of power lines/substations had an increased risk of childhood leukemia. (29,48) CONCLUSION AND RECOMMENDATIONS In the current study, environmental, genetic, and social factors, including exposure to pesticides and diagnostic radiation, low family income, positive family history of cancer or leukemia, and the father's occupation as a farmer play a role in childhood leukemia. Parents should be educated about the safe use and storage of pesticides, alarming signs for childhood leukemia, and the avoidance of unnecessary diagnostic radiation exposure in children.

Limitation of the study
The study results are vulnerable to recall bias, as the case-control study design was applied.