Neuroblastoma is a neoplasm derived from embryonic neural crest cells. It is the most common tumor occurring in the first year of life and accounts for 8–10% of all childhood tumors. 1 In the United States, neuroblastoma occurs in approximately 1 of every 7,000 live births, contributing about 550 new cases each year, with an annual incidence rate of 9.2 cases per million children under 15 years of age. 1 At diagnosis, more than half of these children are under 2 years of age, and approximately 80% are under 4 years of age; rarely is neuroblastoma diagnosed in children older than 10 years of age.
The severity of neuroblastoma ranges from asymptomatic tumors that spontaneously regress to rapidly progressing tumors that respond poorly to aggressive treatment and often lead to death. 2 Clinical and biological subgroups of neuroblastoma have been characterized by the age at diagnosis, tumor stage at diagnosis, and specific genetic alterations such as MYCN oncogene amplification. 2,3 These disease subgroups are known to have prognostic significance but may also be of etiologic relevance.
Relatively little is known about etiologic factors for neuroblastoma. The young age at onset of most cases emphasizes the need to investigate exposures and events occurring before conception and during gestation in addition to genetic contributions to neuroblastoma. 4 Previous studies have reported positive associations between farm residence or parental employment in agriculture and neuroblastoma, but findings have been inconsistent. 5–8 The results from these studies, combined with reports from studies of occupational and residential pesticides and other childhood cancers and birth outcomes, 9–12 have prompted this investigation. To assess the potential effects of residential pesticides on neuroblastoma, we conducted a large case-control study of neuroblastoma and collected detailed information from parents on parental and childhood residential pesticide exposures.
Subjects and Methods
The details of this study are provided elsewhere. 13 Briefly, children diagnosed with neuroblastoma between May 1, 1992, and April 30, 1994, at one of 139 participating hospitals throughout the United States and Canada were eligible to participate if their biological mother was available for telephone interview and spoke English or Spanish. Fathers were eligible only if mothers also provided interviews. Five hundred thirty-eight case mothers (73% of eligible) and 405 case fathers (76% of eligible) provided interviews. Controls were identified through telephone random digit dialing and individually matched to cases by telephone number (area code and exchange) and on the date of birth (±6 months for cases diagnosed under 3 years of age, ±1 year for cases diagnosed over 3 years of age). Five hundred four control mothers (72% of eligible) and 304 control fathers (61% of eligible) completed interviews.
Two collaborative clinical trial groups, Children’s Cancer Group and Pediatric Oncology Group, confirmed the diagnosis of neuroblastoma and provided information on the clinical stage at diagnosis (International Neuroblastoma Staging System; INSS) and MYCN amplification status for case children. Telephone interviews with both case and control mothers and fathers collected information about family medical histories, vitamin and medication use, parental job histories, occupational exposures, residential pesticide use, and other factors.
Each parent was separately asked about pesticides used in the home and garden and by professional exterminators during the time period from 1 month before conception to the date of diagnosis. Interviews queried the type of pesticide, as well as the purpose, frequency, and timing of use relative to dates of the child’s conception and birth. Because both parents were interviewed independently, maternal and paternal reports of pesticide use or timing of use sometimes differed. Among households in which parents did not agree, it was not possible to determine which parent more accurately recalled the details about pesticide use. Thus, for each pesticide category, information from both parents was combined to create two exposure indicator variables representing the following: (1) either the mother or father reported pesticide use, but the other parent reported no use, or (2) both parents in the household reported pesticide use. Both were compared with a reference group in which both parents reported no pesticide use. Both parents reporting an exposure does not necessarily indicate that the exposure was greater but may indicate a greater probability that the exposure actually occurred. This classification method was possible only when both parents provided information.
To determine whether the timing of exposure was associated with neuroblastoma, exposures were evaluated in two time windows representing the period from 1 month before conception through the end of pregnancy and the period from birth to diagnosis. Exposures during the childhood period were evaluated only for children older than 3 months of age at diagnosis. To evaluate whether the effects of pesticide exposure differed in clinical and biological subgroups of neuroblastoma, which might define etiologically homogeneous subgroups, analyses were stratified by the child’s age, and for cases, the tumor’s stage and MYCN amplification status.
The relation between residential pesticide exposures and neuroblastoma was evaluated by estimating the odds ratios (ORs) for both parents reporting use (ORB) and either parent reporting use (ORE) and 95% confidence intervals (CIs) using unconditional logistic regression. The matching factor, child’s age, was assessed as a continuous variable and as a group of indicator variables categorizing age into 6-month strata. Because the continuous form of age adequately accounted for the effects of age and met the linearity requirements of logistic models, it was included in all models. The potential confounding effects of race, household income, maternal age, and education were also evaluated. Of these, only household income was retained in the final analyses, based on the change in the OR with its inclusion and on its weak association with pesticide use and neuroblastoma in these data.
To assess potential bias due to missing paternal information, analyses were conducted that used only maternal reports of pesticide use and stratified by whether the father completed an interview.
Demographic characteristics were similarly distributed between cases and controls in the total study (Table 1). Twenty-five per cent of cases and 39% of controls were excluded because the father was not interviewed. Mothers of these children had slightly lower household income and education, were younger, and were less likely to be white compared with those included. Thirty cases and 18 controls were also excluded because of missing information about household income, leaving 390 cases and 296 controls available for the residential pesticide analyses.
In the total study population, pesticide use was common from the month before conception to the reference date; 65% of parents reported using pesticides in or around the home, 47% reported using lawn and garden pesticides, and 29% reported professional extermination. The proportion of mothers and fathers reporting use of these pesticides in their individual interviews was similar, but the level of agreement when both parents participated ranged from weak [home pesticides Kappa (K) = 0.2] to moderate (extermination K = 0.6, garden pesticides K = 0.4). Parental agreement was similar for cases as compared with controls. In general, subjects whose parents agreed in their reports of pesticide use were demographically similar to those whose parents disagreed, although parents in disagreement about extermination were of slightly lower income.
When both parents reported use of pesticides, the use of professional extermination was associated with neuroblastoma [ORB = 1.4 (95% CI = 0.9–2.1)], as were household pesticides [ORB = 1.6 (95% CI = 1.0–2.3)] and garden pesticides [ORB = 1.7 (95% CI = 0.9–2.1)]. These associations were not evident when only one parent reported use (Table 2). Effect estimates associated with each type of pesticide were slightly stronger for use during the childhood period than during pregnancy; however, pesticide use tended to be correlated across the two time periods. In most households, when at least one parent reported pesticide use during pregnancy, pesticide use was also reported during childhood (63% for home pesticides and 62% for garden pesticides), limiting the ability of this study to evaluate further the independent effects of pesticide exposure during specific time periods.
Although few parents reported the brand or chemical name of the pesticides used, many reported the purpose of pesticide use. The most commonly reported purpose for using pesticides in the home was to control for ants or roaches (41%), which was associated with neuroblastoma [ORB = 1.8 (95% CI = 1.0–3.1)]. Less frequently reported reasons for using home pesticides included treatment for flies (10%), fleas (11%), and termites (<1%). Among the specific types of pesticides used in the garden, herbicides were more strongly associated with neuroblastoma [ORB = 1.9 (95% CI = 1.1–3.2)] than were insecticides [ORB = 1.3 (95% CI = 0.7–2.3)]. Few parents reported the reason for professional extermination, which precluded meaningful analysis.
The effects of pesticides were generally stronger among children diagnosed at 1 year of age or older than among those diagnosed under 1 year of age (Table 3). Home pesticides were not notably associated with neuroblastoma in children under 1 year of age [ORB = 1.2 (95% CI = 0.7–2.2)]; however, among children 1 year of age and older the effect was nearly doubled [ORB = 1.9 (95% CI = 1.1–3.2)]. This pattern was found for specific uses of pesticides, such as the treatment of ants, roaches, and flies in the home. Garden pesticides were also more strongly associated with neuroblastoma among the older children [ORB = 2.2 (95% CI = 1.3–3.6)] than among children under 1 year of age [ORB = 1.0 (95% CI = 0.6–2.0)], including garden pesticides specified as herbicides or insecticides. This age-specific pattern was not evident in relation to professional extermination.
Estimated associations between neuroblastoma and professional extermination or garden pesticides were slightly stronger among case children with amplified MYCN status than those with normal MYCN status; however, the confidence intervals in the two groups overlapped substantially (Table 4). This was also the case for comparison of pesticide effects in early-vs late- stage neuroblastoma compared with controls. MYCN amplification and late-stage disease were correlated with the age at diagnosis (1 year and older).
We evaluated the independent effects of home pesticides, garden pesticides, and professional extermination by adjusting each for the others. Adjustment for multiple pesticide use had no effect on the overall results (data not shown). Home pesticide use was often reported in addition to garden pesticides (77%) or professional exterminators (74%), but there was no evidence of interaction among these three categories of residential pesticides.
Parents reporting garden pesticide use were asked who applied the chemical. Many mothers and fathers reported that the father applied the pesticides (57% and 78%, respectively). When the mother reported that she applied the pesticides, the neuroblastoma effect estimate was higher [OR = 2.2 (95% CI = 1.3–3.8)] than when she reported the father applying pesticides [OR = 1.5 (95% CI = 1.1–2.1)] or when the father reported that he applied the pesticide [OR = 1.1 (95% CI = 0.8–1.5)]. To evaluate the effect of missing paternal pesticide data, we used only the pesticide information reported by the mother, stratified by the presence of paternal information. The association between home pesticides and neuroblastoma was similar when paternal information was [OR = 1.3 (95% CI = 0.9–1.7)] and was not [OR = 1.3 (95% CI = 0.8–2.1)] available. Results for extermination were also similar. The association with garden pesticides, however, was slightly higher when the father’s information was available [OR = 1.8 (95% CI = 1.3–2.5)] than when it was not [OR = 1.2 (95% CI = 0.7–2.2)], after adjustment for maternal age, education, and income.
When pesticides are used, children have great potential for exposure, owing to the amount of time they spend on the floor, in the yard, and with pets, which are all associated with higher levels of pesticides. 14–16 Residential pesticide use during pregnancy and childhood was common in this study. Pesticides used in the home by parents and professional exterminators were essentially insecticides. Garden pesticides included both herbicides and insecticides.
Pesticides used in both the home and garden were moderately associated with increased neuroblastoma. The strongest associations were for garden pesticides in children diagnosed after 1 year of age. The stronger effect in older children may reflect either a longer period of pesticide exposure and time for the action of pesticides to contribute to neuroblastoma or different effects of pesticides in an etiologic pathway marked by an older age at diagnosis. Mobile children can access more floor and yard areas where pesticides have been applied, compared with infants, who are typically restricted to a protected area and may have lower opportunity for exposure. Alternately, the clinical heterogeneity of neuroblastoma may reflect etiologic differences whereby pesticides may act through a biological pathway more common to the type of neuroblastoma in older children. Age, tumor stage, and MYCN amplification status are commonly used to distinguish clinical subgroups of neuroblastoma. 3 Because the effects of pesticides were not greatly modified by MYCN amplification status, these data do not support the potential for pesticides to act through a pathway directly involving MYCN amplification. Further evaluation of the effects of pesticides in biological subgroups of disease was limited by sparse data representing combinations of age, stage at diagnosis, and MYCN amplification status and lack of data on other biological characteristics.
Although pesticides could be involved in the development of neuroblastoma by initiating either germ-line or somatic cell mutations, few are known mutagens. 17–19 Nonmutagenic theories include the possibility that pesticides are involved in carcinogenesis through tumor promotion. Some pesticides (for example, chlordane, lindane, and chlorpyrifos) are known to affect the immune system , potentially contributing to neuroblastoma development by decreasing regulation of cell proliferation or surveillance for dysfunctional or undifferentiated neural crest cells. 20–23 Other pesticides demonstrate estrogen-mimicking properties or otherwise disrupt endogenous hormonal activity. These pesticides could alter the proliferation or differentiation of neural cells that continue to be regulated by hormones into the neonatal period. 19,24–31
Although the mechanisms described are plausible means by which pesticides could contribute to the development of neuroblastoma, further interpretation of the positive yet imprecise associations reported here requires consideration of measurement and design issues. In this study, exposure measurement was based on parents’ self-report of residential pesticide use. This measurement assumes that parents can remember and report their prior use of pesticides and that the quality of recall and reporting does not differ by case status. Most parents in this study were asked to recall pesticide use from 1 to 5 years earlier. While this interval is shorter than for most cancer studies, it was still a long time to remember the details about the pesticides used, especially details concerning the timing of exposure relative to pregnancy and the child’s birth.
To improve the validity of the exposure classification, we considered whether only one parent or both parents reported exposure. Both parents’ agreement that a pesticide was used would not necessarily indicate greater exposure than reports by only one parent, but confirmation from both parents increased our confidence that the reported pesticide was really used. Thirty-five per cent of the couples in our data disagreed about whether pesticides were used, underscoring the problem of error in recall or reporting. Yet, because concordance in pesticide reporting between mothers and fathers did not differ by case status, it does not appear that case parents conferred with each other or otherwise put greater effort toward recalling pesticide use than control parents. Thus, it seems unlikely that differential recall based on motivation of case parents might have resulted in an overestimate of the exposure effect in this study. 32 Nevertheless, we do not know whether the category that requires both parents to report is truly more or less sensitive to recall bias than exposures reported by only one parent.
Although requiring both parents to contribute to the exposure classification may have reduced bias in the effect estimates, it also reduced the study size by excluding children with information from only one parent. To assess potential selection bias owing to missing paternal information, we conducted analyses using only the mothers’ reports of pesticides, stratified by whether or not the father provided an interview. Although the association between mother’s report of garden pesticide use and neuroblastoma was higher when the father was interviewed than when he was not, effect estimates associated with home pesticide use and professional extermination were similar between the two groups. It remains unclear why maternal reports of garden pesticide use would differ by the joint distribution of case status and paternal participation. It is possible that because fathers were more likely to apply garden pesticides in these data, if the father was absent from the home, either garden pesticides were not used or the mother was less likely to remember their use.
Even when parents accurately report the use of pesticides and the timing of use with respect to pregnancy or childhood, use does not necessarily confer exposure to the child. The chemical properties of the pesticides, method of application, use of protective equipment, activity of the child, and consequently the exposure pathways (dermal, ingestion, and inhalation) may be important determinants of the type and degree of pesticide exposure. Furthermore, pesticides are not a homogenous group of chemicals. Grouping home pesticides by the method of application (sprays, solids, and traps) did not reveal substantial differences in associations with neuroblastoma; effect estimates for each application method followed patterns similar to those of general home pesticide use, but estimates were much less stable. Although some parents reported the brand name or the purpose for using pesticides, few reported the chemical. Because multiple chemicals may be found in products of the same brand, there was not enough information to evaluate the effects of specific chemicals in these data. There was also no way to account for pesticides used in other settings such as daycare or for pesticide exposure from environmental sources such as ground or surface water or dusts from regional spraying. Exposure assessment methods that account for details surrounding exposure specific to the home and individual are necessary to advance our understanding of the effects of pesticides. 33
Few other studies have evaluated the possible relation between pesticides and neuroblastoma in any detail. In a prospective study of farming families in Norway, Kristensen et al5 inferred pesticide exposure through farm residence and pesticide purchase records. This study reported elevated rates of neuroblastoma among children whose families resided on farms around the time of birth [OR = 2.5 (95% CI = 1.0–6.1)]. Among the few studies of occupational exposures, only one collected specific information about pesticide exposure, reporting increased risk of neuroblastoma when either the mother or father used pesticides on the job. 34 The other occupational studies, which produced equivocal results, based pesticide exposure on the parent’s job title rather than specific pesticide exposure information. 6–8
This study was large and appeared to accurately represent children with neuroblastoma. Case children were referred through Children’s Cancer Group and Pediatric Oncology Group, which are estimated to see nearly 95% of all children diagnosed with neuroblastoma in the United States. 35 Furthermore, the age distribution of children in this study was similar to that of the children with neuroblastoma identified through the Surveillance, Epidemiology, and End Results Program, a population-based registry system representing 10% of the U.S. population. 36 Nevertheless, it is possible that the use of random digit dialing to select controls may have resulted in similar distributions of pesticide exposure among cases and controls, based on pesticide use patterns by geographic region. If this occurred in our study, the effect of pesticides estimated may be underestimated.
The amount of information on residential pesticides in this study has allowed a more detailed evaluation of the relation between pesticides and neuroblastoma than previous studies. The differing effects of pesticide in age subgroups of neuroblastoma suggest that pesticides may be involved in etiologic pathways specific to the types of neuroblastoma in older children.
1. Brodeur GM, Castleberry RP. Neuroblastoma. In: Pizzo PA., Poplack DG, eds. Principles and Practice of Pediatric Oncology. Philadelphia: Lippincott-Raven, 1997.
2. Brodeur GM. Neuroblastoma and other peripheral neuroectodermal tumors. In: Fernbach DJ, Vietti TJ, eds. Clinical Pediatric Oncology, 4th ed. St. Louis: Mosby Year Book, 1991.
3. Brodeur GM, Maris JM, Yamashiro DJ, Hogarty MD, White PS. Biology and genetics of human neuroblastoma. J Pediatr Hematol Oncol 1997; 19: 93–101.
4. Robison LL, Daigle A. Control selection using random digit dialing for cases of childhood cancer. Am J Epidemiol 1984; 120: 154–156.
5. Kristensen P, Andersen A, Irgens LM, Bye AS, Sundhem L. Cancer in offspring of parents engaged in agricultural activities in Norway: incidence and risk factors in the farm environment. Int J Cancer 1996; 65: 39–50.
6. Bunin GR, Ward E, Kramer S, Rhee CA, Meadows AT. Neuroblastoma and parental occupation. Am J Epidemiol 1990; 131: 776–780.
7. Spitz MR, Johnson CC. Neuroblastoma and paternal occupation: a case-control analysis. Am J Epidemiol 1985; 21: 924–929.
8. Wilkins JR, Hundley VD. Paternal occupational exposure to electromagnetic fields and neuroblastoma in offspring. Am J Epidemiol 1990; 131: 995–1107.
9. Daniels JL, Olshan AF, Savitz DA. Pesticides and childhood cancer. Environ Health Perspect 1997; 105: 1068–1077.
10. Zahm SH, Ward MH. Pesticides and childhood cancer. Environ Health Perspect 1998; 106: 893–908.
11. Kristensen P, Irgens LM, Andersen A, Bye AS, Sundheim L. Birth defects among offspring of Norwegian farmers. Epidemiology 1997; 8: 537–544.
12. Seever LE, Arbuckle TE, Sweeney A. Reproductive and developmental effects of occupational pesticide exposure: the epidemiologic evidence. Occup Med 1997; 12: 305–325.
13. Olshan AF, DeRoos AJ, Teschke K, Neglia JP, Stram DO, Pollock BH, Castleberry RP. Neuroblastoma and parental occupation. Cancer Causes Control 1999; 10: 539–549.
14. Lewis RG, Fortmann RC, Camann DE. Evaluation of methods for monitoring the potential exposure of small children to pesticides in the residential environment. Arch Environ Contam Toxicol 1994; 26: 1–10.
15. Davis JR, Brownson RC, Garcia R. Family pesticide use in the home, garden, orchard and yard. Arch Environ Contam Toxicol 1992; 22: 260–266.
16. Fenske RA, Black KG, Elkner KP, Lee C, Methner M, Soto R. Potential exposure and health risks of infants following indoor residential pesticide applications. Am J Pubic Health 1990; 80: 689–693.
17. Kale PG, Petty BT, Walker S, Ford JB, Dehkordi N, Tarasia S, Tasie BO, Kale R, Sohni YR. Mutagenicity testing of nine herbicides and pesticides currently used in agriculture. Environ Mol Mutagen 1995; 25: 148–153.
18. Hayes J, Laws E. Handbook of Pesticide Toxicology. San Diego: Academic Press, 1991.
19. Velazquez A, Xamena N, Creus A, Marcos R. Mutagenic evaluation of the organophosphorus insecticides methyl parathion and triazophos in Drosophila melanogaster
. J Toxicol Environ Health 1990; 31: 313–325.
20. Vianio H, Husgafvel-Pursianinen K. Mechanisms of carcinogenesis and molecular epidemiology. Br J Hosp Med 1996; 56: 162–170.
21. Repetto R, Baliga SS. Pesticides and the immune system: the public health risks. Executive summary. Cent Eur J Public Health 1996; 4: 263–265.
22. Thomas PT, Busse WW, Kerkvliet NI, Luster MI, Munson AE, Murray M, Roberts D, Robinson M, Silkworth J, Sjoblad R, Smialowicz R. Immunologic effects of pesticides. In: Baker SR, Wilkinson CF, eds. The Effects of Pesticides on Human Health: Advances in Modern Environmental Toxicology. Princeton: Princeton Scientific Publishing, 1990.
23. Brooks BO, Sullivan JB. Immunotoxicology. In: Sullivan JB, Krieger GR, eds. Hazardous Materials Toxicology: Clinical Principles of Environmental Health. Baltimore: Williams and Wilkins, 1992.
24. Crisp TM, Clegg ED, Cooper RL, Wood WP, Anderson DG, Baetcke KP, Hoffmann JL, Morrow MS, Rodier DJ, Schaeffer JE, Touart LW, Zeeman MG, Patel YM. Environmental endocrine disruption: an effects assessment and analysis. Environ Health Perspect 1998; 106 (suppl 1): 11–56.
25. Kelce WR, Stone CR, Laws SC, Gray LE, Kemppainen JA, Wilson EM. Persistent DDT metabolite p,p`-DDE is a potent androgen receptor antagonist. Nature 1995; 375: 581–585.
26. Chapin RE, Stevens JT, Hughes CL, Kelce WR, Hess RA, Datson GP. Endocrine symposium overview: modulation of reproduction. Fundam Appl Toxicol 1996; 29: 1–17.
27. Colborn T, vom Saal FS, Soto AM. Developmental effects of endocrine-disrupting chemicals in wildlife and humans. Environ Health Perspect 1993; 101: 378–384.
28. Colborn T. The wildlife/human connection: modernizing risk decisions. Environ Health Perspect 1994; 102: 55–59.
29. Faivrebauman A, Rosenbau E, Pujmirat J, Grousell D, Tixiervidal A. Differentiation of fetal mouse hypothalamic cells in serum free medium. Dev Neurosci 1981; 4: 118–129.
30. Rasmussen JE, Torres-Aleman I, MacLusky NJ, Naftolin F, Robbins RJ. The effect of estradiol on the growth pattern of estrogen receptor-positive hypothalamic cell lines. Endocrinology 1990; 126: 235–240.
31. Hirschfeld S, Helman L. Diverse roles of insulin-like growth factors in pediatric solid tumors. In Vivo 1994; 8: 81–90.
32. Werler MM, Pober BR, Nelson K, Homes LB. Reporting accuracy among mothers of malformed and nonmalformed infants. Am J Epidemiol 1989; 129: 415–421.
33. Olshan AF, Daniels JL. Invited commentary: pesticides and childhood cancer. Am J Epidemiol 2000; 151: 647–649.
34. Michaelis J, Haaf HG, Zollner J, Kaatsch P, Krummenauer F, Berthold F. Case control study of neuroblastoma in West-Germany after the Chernobyl accident. Klin Padiatr 1996; 208: 172–178.
35. Ross JA. Severson RK. Pollock BH. Robison LL. Childhood cancer in the United States: a geographical analysis of cases from the Pediatric Cooperative Clinical Trials groups. Cancer 1996; 77 (1): 201–207.
36. Gurney JG, Davis S, Severson RK, Fang JY, Ross JA, Robison LL. Trends in cancer incidence among children in the U.S. Cancer 1996; 78: 532–541.
Participating Principal Investigators: Children’s Cancer Group