Volume 159 Number 12 June 15, 2004
American Journal of
EPIDEMIOLOGY
Copyright © 2004 by The Johns HopkinsBloomberg School of Public Health
Sponsored by the Society for Epidemiologic Research Published by Oxford University Press
ORIGINAL CONTRIBUTIONS
Parental Exposure to Polycyclic Aromatic Hydrocarbons and the Risk of Childhood
Brain Tumors
The SEARCH International Childhood Brain Tumor Study
S. Cordier1,*, C. Monfort1, G. Filippini2, S. Preston-Martin3, F. Lubin4, B. A. Mueller5, E. A.
Holly6, R. Peris-Bonet7, M. McCredie8, W. Choi9,†, J. Little10, and A. Arslan11
1 Institut National de la Santé et de Recherche Médicale, Unité
625, Rennes, France.
2 Neuroepidemiology Research Unit, Istituto Nazionale
Neurologico “C Besta,” Milan, Italy.
3 Department of Preventive Medicine, Norris Comprehensive
Cancer Center, Keck School of Medicine, University of Southern California, Los Angeles, CA.
4 Department of Clinical Epidemiology, Chaim Sheba Medical
Center, Sackler School of Medicine, Tel-Hashomer, Israel.
5 Public Health Sciences Division, Fred Hutchison Cancer
Research Center, Seattle, WA.
6 Department of Epidemiology and Biostatistics, University of
California, San Francisco, San Francisco, CA.
7 Unidad de Informacion y Documentacion Medicosanitaria,
Instituto Lopez Pinero, Consejo Superior de Investigaciones Cientificas, Universitat de Valencia, Valencia, Spain.
8 Department of Preventive and Social Medicine, University of
Otago, Dunedin, New Zealand.
9 Manitoba Cancer Treatment and Research Foundation,
Winnipeg, Manitoba, Canada.
10 University of Aberdeen Medical School, Aberdeen,
Scotland.
11 International Agency for Research on Cancer, Lyon,
France. Received for publication July 31, 2003; accepted for publication January 5, 2004.
Experimental evidence suggests that parental exposure to polycyclic aromatic hydrocarbons (PAH), which occurs primarily through tobacco smoke, occupational exposure, and air pollution, could increase the risk of cancer during childhood. Population-based case-control studies carried out in seven countries as part of the SEARCH Program compared data for 1,218 cases of childhood brain tumors and 2,223 controls (1976–1994). Parental occupational exposure to PAH during the 5-year period before birth was estimated with a job exposure matrix. Risk estimates were adjusted for child’s age, sex, and study center. Paternal preconceptional occupational exposure to PAH was associated with increased risks of all childhood brain tumors (odds ratio (OR) = 1.3, 95% confidence interval: 1.1, 1.6) and astroglial tumors (OR = 1.4, 95% confidence interval: 1.1, 1.7). However, there was no trend of increasing risk with predicted level of exposure. Paternal smoking alone (OR = 1.4) was also associated with the risk of astroglial tumors in comparison with nonsmoking, non-occupationally-exposed fathers. Risks for paternal occupational exposure were higher, with (OR = 1.6) or without (OR = 1.7) smoking. Maternal occupational exposure to PAH before conception or during pregnancy was rare, and this exposure was not associated with any type of childhood brain tumor. This large study supports the hypothesis that paternal preconceptional exposure to PAH increases the risk of brain tumors in humans. brain neoplasms; child; germ cells; mutation; paternal exposure; polycyclic hydrocarbons, aromatic
* Correspondence to Dr. Sylvaine Cordier, Institut National de la Santé et de la Recherche Médicale, Unité 625, Groupe d’Étude de la
Reproduction chez le Male, Université Rennes I, Campus de Beaulieu, 35042 Rennes Cedex, France (e-mail: sylvaine.cordier@rennes.inserm.fr). † Deceased.
Abbreviations: CI, confidence interval; ISCO, International Standard Classification of Occupations; ISIC, International Standard Industrial Classification of All Economic Activities; OR, odds ratio; PAH, polycyclic aromatic hydrocarbons; SEARCH,
Surveillance of Environmental Aspects Related to Cancer in Humans.
Polycyclic aromatic hydrocarbons (PAH) are a family of compounds released during incomplete combustion or pyrol-ysis of organic matter. The many sources of human exposure include tobacco smoke, air pollution, and occupational settings (1). Recent experimental evidence shows that paternal exposure to PAH before conception or maternal exposure during pregnancy can result in somatic or germinal mutations in the embryo that increase the risk of cancer during childhood (2–5). Alternatively, epigenetic modifica-tions may occur in germinal or somatic cells that may result in increased cancer risk for the conceptus (6, 7).
The international case-control study coordinated by the International Agency for Research on Cancer as part of the SEARCH (Surveillance of Environmental Aspects Related to Cancer in Humans) Program offered us the opportunity to test the hypothesis that parental occupational exposure to PAH affects children’s risk of brain tumors.
MATERIALS AND METHODS
The detailed design of the SEARCH study at each center has been described in previous publications on the pooled data(8–11) and on data from individual centers (12–16). Briefly, concurrent population-based case-control studies were conducted in nine study centers (Sydney, Australia; Israel; Paris, France; Winnipeg, Canada; Milan, Italy; Valencia, Spain; Los Angeles, California; San Francisco, California; Seattle, Washington) in seven countries. The eligible cases comprised all children in those regions first diagnosed during the study period with a primary malignant tumor of the brain (International Classification of Diseases, Ninth Revision, code 191) or cranial nerves (International Classification of Diseases, Ninth Revision, code 192.0). The upper age limit at case diagnosis was 14 years in Europe and Australia and 19 years in Israel and the United States. The cases were diagnosed between 1976 and 1994; 65 percent of them were diagnosed between 1985 and 1989. Physicians and families were contacted by mail, and in-person inter-views with the mother usually took place at the patient’s home. Interviews were completed for 1,218 of the 1,627 eligible cases (75 percent). Histologic confirmation of the diagnosis was obtained for 90 percent of participating cases; the other cases were assigned to a morphologic group based on the radiologic diagnosis.
Randomly selected controls from the general population were either pair-matched (six centers) or frequency-matched (three centers) to cases by sex and birth year (or by age at one center). The different study centers used several different sampling frames to contact and recruit control participants; these included census data, telephone directories, and random digit dialing methods. Eligible families were those known to have at least one eligible child according to the protocol applied in each center. Of the 2,950 eligible control families contacted, 2,223 (75 percent) were interviewed.
Mothers, and fathers whenever possible, were interviewed in their homes (or by telephone for some fathers) according to a structured questionnaire. Questionnaires used were similar at all centers. A number of suspected risk factors for childhood brain tumors were investigated, including expo-sure to tobacco smoke and parental occupational history. Specifically, all jobs held for at least 1 month during the 5-year period before the child’s birth were recorded, and parents were asked to describe the tasks performed at each job site and to report the type of industry, the number of hours worked weekly, and the products handled. Overall, 48 percent of the fathers who were employed during the 5-year period participated in the occupational interviews (55 percent among cases, 45 percent among controls). This percentage was much higher at the US centers (81 percent), where fathers were usually interviewed by telephone after the interview with the mother. When fathers did not partici-pate, answers about paternal occupational history provided by the mothers were used in the analysis.
Occupations were coded according to the International Standard Industrial Classification of All Economic Activities (ISIC) (17) and the International Standard Classification of Occupations (ISCO) (18). Exposure to PAH was estimated with a job exposure matrix first developed for an interna-tional collaborative study of laryngeal cancer in Southern Europe (19). Because the original job exposure matrix did not contain all of the ISIC-ISCO combinations found in our study, the group of experts who created it prepared an exten-sion so that every industry-occupation combination in our study had a PAH exposure level. The calendar period consid-ered was 1974 onward. Exposure categories were defined as follows—level 1: job-related exposure is not higher than that for the general population; level 2: the job may entail a cumulative exposure higher than that for the general popula-tion; level 3: the job may entail exposure to levels definitely higher than those for the general population, but an impre-cise job description does not permit discrimination between the exposed and those not exposed. This category was further subdivided according to the a priori probability of exposure (level 3a: probability less than one third; level 3b: probability between one third and two thirds; level 3c: prob-ability greater than two thirds). There were two additional categories—level 4: the job entails exposure to the specific agent at levels clearly higher than those of the general popu-lation; and level 5: the job entails exposure to the specific agent, and exposure is known to be particularly high. Expo-sure levels were thus defined by both the probability of expo-sure and the intensity of expoexpo-sure. That is, as the category level increases, the likelihood of misclassification decreases and the intensity of exposure increases. If a causal associa-tion exists, risk should increase for each category. To obtain sufficient numbers in each category, we combined the orig-inal classifications into three groups: no exposure (level 1), medium exposure (levels 2–3b), and high exposure (levels
≥3c). Persons employed in several occupations during the 5-year period were classified according to their highest level of exposure.
Statistical analysis
Because different matching strategies were used at the different study centers, odds ratios were estimated by uncon-ditional logistic regression, which allowed more efficient use of all of the controls in the histologic subgroup analysis. Post hoc matched strata were created for centers that used frequency matching by generation of a reference date for controls, defined as the date on which the control child attained the age of the corresponding case at diagnosis. A comparison of results from matched conditional analyses and unmatched analyses for the same subset of study partic-ipants allowed us to evaluate the influence of matching vari-ables that could not be taken into account in the unmatched analysis, such as area of residence. Since risk estimates were similar, we decided to report only the results from unmatched analyses.
Analysis was restricted to parents who reported having had at least one occupation outside the home during the 5-year period before the birth. Adjustment for risk estimates was made for the matching variables: center, sex, and age at diagnosis of the child (0–4, 5–9, 10–14, or ≥15 years). Other potential confounders such as parental age or number of years of schooling did not modify risk estimates by more than a few percentage points and were not retained in the adjustment.
Tobacco smoking was introduced into the analysis either as a confounder or as an additional source of exposure to PAH for estimation of the joint effects of tobacco smoke and occupational exposure. Both paternal and maternal smoking were defined as dichotomous variables for each period considered because of the absence of quantitative informa-tion for the father. The period of interest for paternal expo-sure to PAH (from occupation or from smoking) was the period before conception, whereas maternal exposure was defined as exposure before conception or during pregnancy. Analyses were performed for all brain tumors and for the histologic subgroups for which we had adequate numbers of cases: astroglial tumors (International Classification of
Diseases for Oncology codes 9380–9382, 9384, 9400–9421,
9424–9429, 9431–9440, and 9442); primitive neuroecto-dermal tumors (codes 9470–9473 and 9501); and other glial tumors, including ependymoma (codes 9391 and 9392), oligodendroglioma (code 9450), and ganglioglioma (code 9505). The subgroup analyses used all control participants.
RESULTS
More than half of these childhood brain tumors were clas-sified as astroglial (52 percent); 21 percent were clasclas-sified as primitive neuroectodermal, and 12 percent were classified as other glial tumors (table 1). The remaining group included various types of other intracranial tumors. Nearly half of the cases came from the Seattle, San Francisco, and Los Angeles centers on the West Coast of the United States. Maternal age at the child’s birth and parents’ educational level were very
similar for cases and controls. Similarly, the average number of occupations reported during the 5-year period before birth did not differ significantly between cases (1.52 (standard error, 0.03) for fathers; 1.30 (standard error, 0.03) for mothers) and controls (1.46 (standard error, 0.02) for fathers; 1.25 (standard error, 0.02) for mothers).
Occupations considered by the job exposure matrix to entail high exposure to PAH in our population were (by decreasing numbers) motor-vehicle or aircraft engine mechanics, toolmakers and machine-tool operators, some farmers and agricultural workers, welders, some construc-tion workers (roofers, cement finishers, etc.), cooks, equip-ment operators, plumbers and pipe fitters, and sheet-metal workers. Occupations considered to involve exposure at a low level were motor-vehicle drivers, warehouse porters, machinery fitters, bricklayers or paviours, some categories of agricultural workers, printers, and janitors.
Maternal occupational exposure to PAH during pregnancy was rare and was not associated with a significant excess of any type of childhood brain tumor (table 2). Odds ratios were greater than 1 for every tumor group among mothers who smoked and were occupationally exposed in comparison with nonsmoking, non-occupationally-exposed mothers, but these increases disappeared after adjustment for paternal preconceptional exposure to PAH (from smoking or from the occupation). When maternal occupational exposure before conception was considered, no increase in risk was observed for any tumor group.
Paternal occupational exposure to PAH before the child’s conception was associated with increased risks of all child-hood brain tumors (odds ratio (OR) = 1.3, 95 percent confi-dence interval (CI): 1.1, 1.6) and astroglial tumors (OR = 1.4, 95 percent CI: 1.1, 1.7) (table 3). However, there was no evidence of an increase in risk with increasing exposure level either in two categories (medium and high) or in the original five levels of exposure (all tumors—level 1 (referent): OR = 1; level 2: OR = 1.4 (95 percent CI: 1.1, 1.8); level 3a: OR = 1.2 (95 percent CI: 0.8, 1.7); level 3b: OR = 1.2 (95 percent CI: 0.9, 1.7); level 3c: OR = 1.5 (95 percent CI: 1.1, 2.2); level 4: OR = 1.3 (95 percent CI: 1.1, 1.7); level 5: OR = 1.2 (95 percent CI: 0.6, 2.4)). The risk associated with occupational exposure in the astroglial tumor group increased when fathers who smoked were excluded (OR = 1.7, 95 percent CI: 1.3, 2.3). The magnitude of the effect of smoking alone (OR = 1.4) was smaller than the risk associated with occupational exposure with (OR = 1.6) or without (OR = 1.7) smoking, but there was no evidence that risk increased among fathers who smoked and were occupationally exposed. Center-specific risk estimates for paternal occupational exposure to PAH ranged from 1.1 (Los Angeles) to 2.3 (Valencia). There was no statistically significant heterogeneity between centers (χ2 = 11.04, p = 0.20). When risks were estimated separately for the four groups of age at diagnosis (0–4, 5–9, 10–14, and ≥15 years), there was no evidence of interaction between age and paternal exposure to PAH.
Validation of paternal exposure to PAH as derived from maternal reports about the father’s occupational history was possible in a subset of interviews from the US centers. For 103 families (90 control families and 13 case families),
information on paternal occupational history was obtained from both mothers and fathers. Among controls, 12 fathers were misclassified by the mother’s report: six unexposed fathers were classified as exposed to PAH, and six exposed fathers were classified as unexposed. This provides an esti-mate of the amount of misclassification introduced by using mothers’ reports. Among cases’ families, numbers were too small to infer the amount of misclassification. Assuming nondifferential error, we could estimate the impact of this misclassification by computing a corrected odds ratio for all tumors combined (OR = 1.4) as a Mantel-Haenszel combina-tion of stratum-specific estimates (OR = 1.3 (from answers given by the fathers) and OR = 1.5 (the corrected odds ratio derived from answers given by the mothers)).
We also restricted the analysis to US centers, from which 44 percent of the cases originated and which used similar control selection and interview procedures. The risk esti-mates obtained (OR = 1.3, 95 percent CI: 1.0, 1.7) were very close to those observed in the rest of the study population (OR = 1.4, 95 percent CI: 1.1, 1.7) for paternal exposure to PAH.
DISCUSSION
Our results provide evidence that paternal preconceptional exposure to PAH from either tobacco smoking or occupa-tional exposure could increase the risk of brain tumors, espe-cially astroglial tumors, in children. No similar effect was observed for maternal smoking or maternal occupational exposure to PAH before conception or during pregnancy.
These results from a large representative sample corrob-orate our previous findings from European centers (20). They accredit, for a narrower class of chemicals (the PAHs), the hypothesis proposed by Fabia and Thuy (21) in 1974 and confirmed (22–27) and refuted (28, 29) by subsequent studies. Since then, experimental findings have suggested mechanisms for these associations in humans.
Much of this evidence is based on biologic markers of PAH binding to DNA. Such markers were not available in our epidemiologic study. The absence of direct measure-ments of PAH exposure from all possible sources precludes too strong an interpretation of our findings. Of the three prin-cipal sources of PAH exposure in industrialized countries— TABLE 1. Sociodemographic and medical characteristics of cases and controls, SEARCH* international
childhood brain tumor study, 1976–1994
Cases (n = 1,218) Controls (n = 2,223) Odds
ratio 95% CI* No. % No. % Morphologic subgroup Astroglial 623 51.5 Primitive neuroectodermal 259 21.4 Other glial 148 12.2 Other intracranial 179 14.8 Unknown 9 Study center Sydney, Australia 82 6.7 164 7.4 Israel 300 24.6 574 25.8 Paris, France 75 6.2 113 5.1 Winnipeg, Canada 45 3.7 83 3.7 Milan, Italy 90 7.4 318 14.3 Valencia, Spain 86 7.1 170 7.7
Los Angeles, California 304 25.0 315 14.2
San Francisco, California 102 8.4 205 9.2
Seattle, Washington 134 11.0 281 12.6
Sex
Male 651 53.5 1,216 54.7 1
Female 567 46.6 1,007 45.3 1.0 0.9, 1.2
Age (years) at diagnosis†
0–1 159 13.1 247 11.1 1 2–4 291 23.9 500 22.6 1.0 0.7, 1.2 5–9 368 30.2 697 31.5 0.9 0.7, 1.2 10–14 267 21.9 530 23.9 0.9 0.7, 1.1 15–19 133 10.9 242 10.9 0.9 0.6, 1.2 ≥20 0 7 Table continues
tobacco smoke, occupational exposure, and air pollution— we could estimate exposure for only the first two.
Our estimate of occupational exposure to PAH was indi-rect. Exposure to PAH in the workplace typically occurs via products or combustion fumes originating from incomplete combustion, pyrolysis, or pyrosynthesis of organic matter. The list of occupations entailing PAH exposure higher than that in the general population was identified some time ago, but individual assessment of exposure requires detailed workplace descriptions or ad hoc measurements. Self-reports could identify circumstances of exposure to oil or coal products or combustion products, but they are limited by the lack of relative or objective benchmarks against which to judge one’s working conditions (30); this results in very variable sensitivities. An additional problem in our data is that only 48 percent of working fathers provided answers about their own occupational exposures. This explains why our assessment was preferably based on the evaluation provided by a job exposure matrix, assuming that answers from the mothers would be of better quality with regard to the father’s occupational history than when mothers were
reporting specific exposures present in the father’s work-place. Use of job exposure matrices introduces a high level of misclassification, since the heterogeneity within job and industry and individual exposure determinants are not accounted for, but this misclassification would be expected to be similar among cases and controls. We used different levels of exposure corresponding to increasing levels of both probability of exposure in a given job and intensity of expo-sure. However, there was no suggestion of increasing risk with increasing exposure levels. The absence of a dose-response pattern may have several alternative explanations: exposure misclassification, no association between PAH exposure and cancer risk, or a nonlinear shape in the relation, as discussed below.
One may also question the validity of applying a job expo-sure matrix originally developed in Southern Europe to other parts of the world such as Australia, Israel, or the US West Coast. However, these countries have similar levels of indus-trial development and similar occupational processes, and all circumstances of occupational exposure to PAH have been TABLE 1. Continued
* SEARCH, Surveillance of Environmental Aspects Related to Cancer in Humans; CI, confidence interval. † Age at diagnosis for cases; age at reference date for controls.
Cases (n = 1,218) Controls (n = 2,223) Odds
ratio 95% CI
No. % No. %
Maternal age (years) at child’s birth
<20 89 7.3 125 5.6 1
20–34 1,026 84.3 1,900 85.6 0.8 0.6, 1.1
≥35 102 8.4 194 8.7 0.8 0.6, 1.2
Unknown 1 4
Maternal education (years)
<8 173 14.2 289 13.0 1
8–10 252 20.7 493 22.2 0.9 0.7, 1.1
>10 792 65.1 1,439 64.8 0.8 0.6, 1.0
Unknown 1 2
Paternal age (years) at child’s birth
<20 26 2.2 33 1.5 1
20–34 925 78.4 1,675 77.1 0.8 0.5, 1.3
≥35 229 19.4 464 21.4 0.7 0.4, 1.2
Unknown 38 51
Paternal education (years)
<8 158 13.4 271 12.5 1
8–10 259 21.9 510 23.5 0.9 0.7, 1.1
>10 763 64.7 1,391 64.0 0.9 0.7, 1.1
Unknown 38 51
Employment during the 5-year period Father No 120 9.9 172 7.7 1 Yes 1,098 90.1 2,051 92.3 0.9 0.7, 1.2 Mother No 281 23.1 514 23.1 1 Yes 937 76.9 1,709 76.9 1.0 0.9, 1.2
considered by experts when updating the job exposure matrix.
Using a validation subsample, we could verify that the use of maternal reports for paternal occupational history had only a minimal impact on the overall risk estimate.
Occupational exposure to PAH was frequent: overall, 40 percent of the control fathers were classified as probably exposed and 19 percent as more highly exposed; these percentages are similar to those estimated with this job expo-sure matrix in previous applications (20, 31). Expoexpo-sure to PAH occurred mainly among blue-collar workers, and the higher proportion of case fathers exposed might have been related to a lower educational level among cases than among controls. Nonetheless, adjustment for parental education did not modify the estimates.
The link observed between the risk of childhood brain tumors (mainly astroglial) and paternal exposure to PAH, either occupational or from smoking, together with the absence of an association with maternal smoking before or during pregnancy (11), tends to suggest that paternal precon-ceptional exposure rather than maternal passive exposure to tobacco smoke is the risk factor involved. It also corrobo-rates the observation by Ji et al. (32) that the risk of child-hood cancer, including brain tumors, increases with duration of paternal smoking in the absence of maternal smoking. This was also confirmed in a review of papers on parental smoking and childhood cancers conducted by Boffetta et al. (33). Postnatal and prenatal exposures from parental
smoking or occupations are very correlated. Previous reports from situations where different periods of exposure could be studied did not show an impact of postnatal exposures (32); a role of these exposures in our study could not be assessed but cannot be excluded.
Several reports based on PAH adducts provide indirect support for the hypothesis of a role of paternal preconcep-tional exposure. A recent paper shows a correlation between molecular markers of PAH exposure and sperm damage among men with occupational exposure (5). Finette et al. (34) examined T lymphocytes in the umbilical cord blood of newborns and reported an increase in “illegitimate” genomic mutations in the hypoxanthine-guanine phosphoribosyl-transferase reporter gene; they attributed it to maternal expo-sure to passive cigarette smoke, but it may have had paternal causes. Benzo[a]pyrene diol epoxide-DNA adducts have been detected in preimplantation embryos from couples in which both parents smoked but also in cases where only the father smoked (2). In addition, animal evidence shows an increased incidence of brain tumors in the progeny of male rats exposed to chemicals such as ethylnitrosourea before mating (35). Paternal but not maternal germ-line mutations were recently observed among laboratory mice exposed in situ to air pollution around steel mills (which release large amounts of PAH) (4). These correlations suggest that muta-tional events occurring during spermatogenesis may be the cause of the increased tumor incidence in the offspring. However, none of the studies in the published literature TABLE 2. Odds ratios for childhood brain tumors according to maternal occupational exposure to polycyclic aromatic hydrocarbons during pregnancy and joint exposure from occupation and active smoking, SEARCH* international childhood brain tumor study, 1976–1994
* SEARCH, Surveillance of Environmental Aspects Related to Cancer in Humans; OR, odds ratio; CI, confidence interval. † Adjusted for study center, child’s age, sex, and (in the upper portion of the table) maternal smoking during pregnancy. ‡ Reference category.
Exposure to polycyclic aromatic hydrocarbons
Tumor type
All tumors Astroglial
(n = 475) Primitive neuroectodermal (n = 208) Other glial (n = 119) No. of cases (n = 937) No. of controls (n = 1,709) OR*,† 95% CI* No. of cases OR† 95% CI No. of cases OR† 95% CI No. of cases OR† 95% CI Occupational exposure during
pregnancy Not exposed‡ 882 1,611 1 444 1 202 1 110 1 Exposed 46 90 1.0 0.7, 1.4 26 1.1 0.7, 1.8 5 0.5 0.2, 1.2 7 1.2 0.5, 2.6 Medium 27 54 1.0 0.6, 1.7 12 0.9 0.5, 1.7 3 0.5 0.1, 1.5 7 2.0 0.9, 4.5 High 19 36 1.0 0.5, 1.7 14 1.4 0.8, 2.7 2 0.5 0.1, 2.0 0 Unknown 9 8 5 1 2
Joint exposure from smoking and occupation
No exposure from either‡ 729 1,304 1 362 1 170 1 88 1
Exposure from smoking but not
occupation 152 305 0.9 0.7, 1.2 82 1.0 0.8, 1.3 31 0.8 0.5, 1.2 22 1.1 0.7, 1.7 Exposure from occupation but
not smoking 33 68 0.9 0.6, 1.4 19 1.1 0.6, 1.9 2 0.2 0.1, 0.9 5 1.1 0.4, 2.8 Exposure from both smoking
and occupation 13 22 1.2 0.6, 2.5 7 1.3 0.5, 3.0 3 1.2 0.4, 4.2 2 1.5 0.3, 6.6
explored the likelihood of epigenetic tumor causation after PAH exposure; hence, this possibility cannot be ruled out (6, 7).
In our study, the risk of brain tumors in children was greater among fathers who were occupationally exposed than among fathers who smoked, in comparison with fathers who had neither exposure. However, this risk did not increase when fathers were exposed both occupationally and through smoking or when the level of estimated occupational exposure increased. Ji et al. (32), in studying paternal smoking, noted that patterns of increase in risks of childhood cancer were less consistent with the number of cigarettes smoked per day (analogous to intensity) than with the dura-tion of smoking. These observadura-tions may well be related to the saturation in DNA adduct formation that has been demonstrated at high levels of PAH exposure (36).
In summary, the findings of our large study support the hypothesis that paternal preconceptional exposure to PAH may increase the risk of brain tumors in humans. Moreover, several reports linking PAH exposure and germ-cell damage provide indirect support for this association. It would be difficult to obtain large-scale confirmation that this mecha-nism operates in a portion of childhood brain tumors—by measuring PAH-DNA adducts, for example. Nonetheless, attempts at this sort of biologic confirmation must be incor-porated into future studies.
ACKNOWLEDGMENTS
This research was supported by the SEARCH Program of the International Agency for Research on Cancer and by grants to individual collaborators, as follows—Paris, France: the French National League against Cancer; United States: grant CA47082 from the National Institutes of Health (identification of cases in Los Angeles, California, was supported by grant CA17054 and contract NO1-CN-25403 from the National Institutes of Health and subcontract 050(C-G)-8709 from the California Department of Health Services); Israel: grant R01-CA-51117-03 from the US National Cancer Institute; Milan, Italy: the “Oncology” Applied Project of the National Research Council and the Italian Association for Cancer Research; Valencia, Spain: the Spanish Fund for Health Investigation and the Health and Consumers’ Council of Valencia; Sydney, Australia: the New South Wales Cancer Council.
The authors thank P. Pisani, L. Troschel, and I. Belletti, who kindly provided updated information for the job expo-sure matrix, and J. A. Cahn for skillful assistance in revising the manuscript.
REFERENCES
1. International Programme on Chemical Safety. Selected non-heterocyclic polycyclic aromatic hydrocarbons. (Environmen-tal health criteria no. 202). Geneva, Switzerland: World Health
TABLE 3. Odds ratios for childhood brain tumors according to paternal preconceptional occupational exposure to polycyclic aromatic hydrocarbons and joint exposure from smoking and occupation, SEARCH* international childhood brain tumor study, 1976– 1994
* SEARCH, Surveillance of Environmental Aspects Related to Cancer in Humans; OR, odds ratio; CI, confidence interval. † Adjusted for study center, child’s age, sex, and (in the upper portion of the table) paternal smoking.
‡ Reference category. Exposure to polycyclic aromatic
hydrocarbons
Tumor type
All tumors Astroglial
(n = 565) Primitive neuroectodermal (n = 233) Other glial (n = 134) No. of cases (n = 1,098) No. of controls (n = 2,051) OR*,† 95% CI* No. of cases OR† 95% CI No. of cases OR† 95% CI No. of cases OR† 95% CI Occupational exposure Not exposed‡ 560 1,174 1 281 1 130 1 67 1 Exposed 484 784 1.3 1.1, 1.6 254 1.4 1.1, 17 93 1.1 0.8, 1.5 62 1.4 1.0, 2.0 Medium 253 420 1.3 1.1, 1.6 126 1.3 1.0, 1.6 54 1.2 0.9, 1.7 34 1.5 0.9, 2.3 High 231 364 1.4 1.1, 1.7 128 1.5 1.2, 1.9 39 1.0 0.7, 1.5 28 1.3 0.8, 2.1 Unknown 54 93 30 10 5
Joint exposure from smoking and occupation
No exposure from either‡ 285 584 1 126 1 76 1 34 1
Exposure from smoking but not
occupation 273 590 1.1 0.9, 1.3 155 1.4 1.1, 1.9 53 0.8 0.6, 1.2 32 1.1 0.6, 1.7 Exposure from occupation but
not smoking 211 317 1.4 1.1, 1.8 112 1.7 1.3, 2.3 39 1.0 0.6, 1.5 27 1.4 0.8, 2.5 Exposure from both smoking
and occupation 272 464 1.4 1.1, 1.7 142 1.6 1.2, 2.1 54 1.0 0.7, 1.5 34 1.3 0.8, 2.1
Organization, 1998.
2. Zenzes MT, Puy LA, Bielicki R, et al. Detection of benzo[a]pyrene diol epoxide-DNA adducts in embryos from smoking couples: evidence from transmission by spermatozoa. Mol Hum Reprod 1999;5:125–31.
3. Perera F, Hemminki K, Jedrychowski W, et al. In utero DNA damage from environmental pollution is associated with somatic gene mutation in newborns. Cancer Epidemiol Biomarkers Prev 2002;11:1134–7.
4. Somers CM, Yauk CL, White PA, et al. Air pollution induces heritable DNA mutations. Proc Natl Acad Sci U S A 2002;99: 15904–7.
5. Gaspari L, Chang S-S, Santella RM, et al. Polycyclic aromatic hydrocarbon-DNA adducts in human sperm as a marker of DNA damage and infertility. Mutat Res 2003;535:155–60. 6. Walker BE, Kurth LA. Multi-generational carcinogenesis from
diethylstilbestrol investigated by blastocyst transfers in mice. Int J Cancer 1995;61:249–52.
7. Gaudet F, Hodgson JG, Eden A, et al. Induction of tumors in mice by genomic hypomethylation. Science 2003;300:489–92. 8. Preston-Martin S, Pogoda JM, Mueller BA, et al. Results of an international case-control study of childhood brain tumors: the role of prenatal vitamin supplementation. Environ Health Per-spect 1998;106:887–92.
9. McCredie M, Little J, Cotton S, et al. SEARCH international case-control study of childhood brain tumours: role of index pregnancy and birth, and mother’s reproductive history. Paediatr Perinat Epidemiol 1999;13:325–41.
10. Cordier S, Mandereau L, Preston-Martin S, et al. Parental occu-pations and brain tumors: results of an international case-control study. Cancer Causes Control 2001;12:865–74. 11. Filippini G, Maisonneuve P, McCredie M, et al. Relation of
childhood brain tumors to exposure of parents and children to tobacco smoke: the SEARCH international case-control study. Int J Cancer 2002;100:206–13.
12. Filippini G, Farinotti M, Lovicu G, et al. Mothers’ active and passive smoking during pregnancy and risk of brain tumours in children. Int J Cancer 1994;57:769–74.
13. Cordier S, Iglesias MJ, Le Goaster C, et al. Incidence and risk factors for childhood brain tumors in the Ile de France. Int J Cancer 1994;59:776–82.
14. Peris-Bonet R, Abad Garcia F, Melchor Alos I, et al. Childhood cancer incidence registration in the province of Valencia, Spain 1983–1990. J Epidemiol Biostat 1996;1:107–13.
15. McCredie M, Maisonneuve P, Boyle P. Antenatal risk factors for malignant brain tumours in New South Wales children. Int J Cancer 1994;56:6–10.
16. McKean-Cowdin R, Preston-Martin S, Pogoda JM, et al. Paren-tal occupation and childhood brain tumors: astroglial and prim-itive neuroectodermal tumors. J Occup Med 1998;40:332–40. 17. United Nations. International Standard Industrial Classification
of All Economic Activities. New York, NY: United Nations, 1971.
18. International Labour Organization. International Standard Classification of Occupations. Geneva, Switzerland: Interna-tional Labour Organization, 1968.
19. Ferrario F, Continenza D, Pisani P, et al. Description of a job exposure matrix for sixteen agents which are or may be related to respiratory cancer. In: Hogstedt C, Reuterwall C, eds.
Progress in occupational epidemiology. Amsterdam, the Neth-erlands: Elsevier Science Publishers BV, 1988:379–82. 20. Cordier S, Lefeuvre B, Filippini G, et al. Parental occupation,
occupational exposure to solvents and polycyclic aromatic hydrocarbons and risk of childhood brain tumors (Italy, France, Spain). Cancer Causes Control 1997;8:688–97.
21. Fabia J, Thuy TD. Occupation of father at time of birth of chil-dren dying of malignant diseases. Br J Prev Soc Med 1974;28: 98–100.
22. Peters JM, Preston-Martin S, Yu MC. Brain tumors in children and occupational exposure of parents. Science 1981;213:235– 7.
23. Hemminki K, Saloniemi I, Salonen T, et al. Childhood cancer and parental occupation in Finland. J Epidemiol Community Health 1981;35:11–15.
24. Johnson CC, Annegers JF, Frankowski RF, et al. Childhood nervous system tumors—an evaluation of the association of paternal occupational exposure to hydrocarbons. Am J Epide-miol 1987;126:605–13.
25. Wilkins JR 3rd, Sinks T. Parental occupation and intracranial neoplasms of childhood: results of a case-control interview study. Am J Epidemiol 1990;132:275–92.
26. Olsen JH, De Nully Brown P, Schulgen G, et al. Parental employment at time of conception and risk of cancer in off-spring. Eur J Cancer 1991;27:958–65.
27. Feingold L, Savitz DA, John EM. Use of a job exposure matrix to evaluate parental occupation and childhood cancer. Cancer Causes Control 1992;3:161–9.
28. Hakulinen T, Salonen T, Teppo L. Cancer in the offspring of fathers in hydrocarbon-related occupations. Br J Prev Soc Med 1976;30:138–40.
29. Zack M, Cannon S, Loyd D, et al. Cancer in children of parents exposed to hydrocarbon-related industries and occupations. Am J Epidemiol 1980;111:329–36.
30. Teschke K, Olshan AF, Daniels JL, et al. Occupational expo-sure assessment in case-control studies: opportunities for improvement. Occup Environ Med 2002;59:575–94.
31. Stücker I, Bouyer J, Mandereau L, et al. Retrospective evalua-tion of the exposure to polycyclic aromatic hydrocarbons: com-parative assessments with a job exposure matrix and by experts in industrial hygiene. Int J Epidemiol 1993;22(suppl 2):S106– 12.
32. Ji B-T, Shu X-O, Linet MS, et al. Paternal cigarette smoking and the risk of childhood cancer among offspring of non-smok-ing mothers. J Natl Cancer Inst 1997;89:238–44.
33. Boffetta P, Tredaniel J, Greco A. Risk of childhood cancer and adult lung cancer after childhood exposure to passive smoke: a meta-analysis. Environ Health Perspect 2000;108:73–82. 34. Finette BA, O’Neill JP, Vacek PM, et al. Gene mutations with
characteristic deletions in cord blood T lymphocytes associated with passive maternal exposure to tobacco smoke. Nat Med 1998;4:1144–51.
35. Tomatis L, Cabral JR, Likhachev AJ, et al. Increased cancer incidence in the progeny of male rats exposed to ethylni-trosourea before mating. Int J Cancer 1981;28:475–8.
36. Van Schooten FJ, Godschalk RW, Breedijk A, et al. 32
P-postla-belling of aromatic DNA adducts in white blood cells and alve-olar macrophages of smokers: saturation at high exposures. Mutat Res 1997;378:65–75.
