Paternal Occupational Exposure to Pesticides or ... - Semantic Scholar

Archives of Environmental & Occupational Health, Vol. 61, No. 3, 2006 Copyright © 2007 Heldref Publications

Paternal Occupational Exposure to Pesticides or Herbicides as Risk Factors for Cancer in Children and Young Adults: A Case-Control Study From the North of England Mark S. Pearce, PhD; Donna M. Hammal, MSc; M. Tevfik Dorak, PhD; Richard J.Q. McNally, PhD; Louise Parker, PhD

ABSTRACT. Researchers in numerous studies have suggested that preconception paternal occupational exposures to various substances, including pesticides and herbicides, may be involved in the etiology of childhood cancers. Using data from the Northern Region Young Persons’ Malignant Disease Registry, the authors investigated whether paternal occupations likely to involve such exposures, as recorded at the time of a child’s birth, were associated with children’s cancer risk. The authors matched cases (n = 4,723), on sex and year of birth, to controls from 2 independent sources: (1) all other patients from the registry with a different cancer and (2) 100 cancer-free individuals per case from the Cumbrian Births Database. An inverse association existed, particularly in males, between lymphoid leukemia and paternal occupations with likely exposures to pesticides and/or herbicides. However, this was not significant after stratifying by residential status (urban versus rural). Results do not support a role for preconception paternal occupational exposures to pesticides or herbicides in the etiology of childhood cancer. KEYWORDS: agriculture, cancer, epidemiology, herbicides, paternal exposures, pesticides

R

esults of numerous studies have suggested that preconception paternal occupational exposures may have a role in the etiology of childhood cancers.1 In particular, occupations with exposures to pesticides and herbicides have been associated with a number of specific childhood cancers; however, results have been inconsistent, with both positive and negative associations reported.2–27 The childhood cancers reported to be associated with paternal exposure to pesticides or herbicides include acute leukemia,2–9 renal tumors,9,11,12 brain tumors,13–16 central nervous system tumors,9,14,17–19 and Ewing’s sarcoma.20,21 However, other investigators22–25 have failed to find any such

associations and have pointed to methodological difficulties in defining exposure windows.22 Researchers26,27 in 2 recent studies investigated the potential link between paternal pesticide exposures and risk of germ cell tumors in offspring and found no support for such a relationship overall but reported significant negative associations in boys only. We tested the hypothesis that an association exists between paternal occupations likely to result in exposures to pesticides or herbicides at the time of birth, classified using a job-exposure matrix, and risk of childhood cancer in children and young adults (aged 0–24 years at diagnosis) in the

Mark S. Pearce, Donna M. Hammal, M. Tevfik Dorak, and Richard J.Q. McNally are with the School of Clinical Medical Sciences (Child Health), University of Newcastle upon Tyne, England. Richard J.Q. McNally is also with the Institute of Health and Society, University of Newcastle upon Tyne, England. Louise Parker is with the Department of Paediatrics, Dalhousie University, Halifax, Canada. 138

Archives of Environmental & Occupational Health

North of England, using data from the Northern Region Young Persons’ Malignant Disease Registry (NRYPMDR)28,29 and the Cumbrian Births Database.30 We tested this hypothesis in males and females separately. METHODS The NRYPMDR is a population-based registry that has recorded information on young people (aged younger than 25 years) considered residents in the North of England who have been diagnosed with malignancies and benign central nervous system tumors since 1968.28,29 The registry is located within the Newcastle Hospitals Trust, which is the regional specialist center for cancer in children and adolescents. It currently holds information on more than 6,000 cases of malignant disease. Registration with the registry is not mandatory, but cases are identified from multiple sources; consultants throughout the region notify the registry of any malignancies in children and young adults, and death certificates and hospital admissions are regularly scrutinized. Data are periodically cross-checked with regional and national cancer registries to ensure that the information is as accurate and complete as possible. Overall ascertainment of cancers is believed to be more than 95% complete.28 The study region represents a mixture of several heavily populated urban areas and widespread rural communities. It has a population of 3.1 million that is predominantly white (ethnic minorities account for less than 2%). Study Subjects We undertook case-control analyses for cases of primary cancers. Cases were patients diagnosed with cancer between 1968 and 2000. We coded tumors for morphology and site using the International Classification of Diseases for Oncology (ICDO-2)31 and classified them using categories from the International Classification of Childhood Cancer (ICCC).32 We selected 2 independent sets of controls. First, for each particular disease group, we selected as controls all other patients from the NRYPMDR with different cancer types, with the same sex and year of birth as each case; this allowed for repeated controls. Second, we randomly selected 100 controls per case from live births recorded on the Cumbrian Births Database (CBD) who were not diagnosed with cancer, again matched on sex and year of birth. The CBD, described in detail elsewhere,30 holds birth registration details, from birth certificates, on all children born to mothers resident in the county of Cumbria, in the North of England, between January 1, 1950, and December 31, 1993. The county of Cumbria contributes to the NRYPMDR and includes approximately 16% of the population of the north of England. In common with the rest of the study region, Cumbria is a mixture of both urban and rural areas but is among the most sparsely populated counties in England, with industriMay/June 2006, Vol. 61, No. 3

al areas, including the Sellafield nuclear installation, concentrated in the southwest portion of the county. Paternal Occupation We coded paternal occupation, as routinely recorded on the birth certificate, which was available for births on the CBD and is obtained by the registry whenever possible, according to the 1990 Standard Occupational Classification (SOC)33 and hence by the paternal social class (on the basis of occupation) at the time of the child’s birth. This classification assigns numeral designations of I through V (I representing the professional, and presumably the most advantaged, occupational social class; V representing the unskilled, and presumably the least advantaged, occupational social class; II through IV as intermediate occupational social class designations, with III divided into subclasses on the basis of manual or nonmanual occupation). We excluded those for whom paternal occupation was unavailable from birth certificates, or for which there was no suitable occupational code (including “unemployed” and “student”). We identified births with a recorded paternal occupation likely to include exposure to pesticides or herbicides (SOC codes 160 and 169 [farm owners and managers, forestry managers, and horticulturists], 594 [gardeners, groundskeepers], 595 [horticultural trades], 900–902 [farm workers], 904 [forestry workers]). Two of the authors (LP and MSP) made this classification independently of the data on the basis of how likely it was considered that occupations within each of the occupational codes in the 1990 classification included occupational exposures to pesticides or herbicides. We obtained urban or rural status using census data for each mother’s residence, as recorded on the birth certificate. Statistical Analysis We investigated whether the risk of cancer during childhood or young adulthood was increased when paternal employment recorded at birth was in an occupation likely to result in pesticide or herbicide exposure. In performing the analyses, we considered all cases, then males and females separately. We also assessed risk of childhood leukemia in the 0–6 years age group, the range of ages that represents the local childhood peak for childhood leukemia. Using multivariable conditional logistic regression, we estimated odds ratios (ORs) and corresponding 95% confidence intervals (95% CIs), adjusting for potential confounding factors. We used the statistical software package Stata, version 8.0 (Stata Corp, College Station, TX), for all statistical analyses. RESULTS Paternal occupation at birth, derived from birth certificates, was available for 4,727 individuals at diagnosis. We excluded the remaining 413 (9%) cases with primary diagnoses in the study period because of missing or invalid 139

Table 1.—Distribution of Cases and Controls by Paternal Occupational Social Class and Urban Versus Rural Residential Status Variable Social class I II III nonmanual III manual IV V Other† Residence Urban Rural

Cases

Registry control*

CBD control*

211 620 431 2273 668 393 115

9686 27918 20383 104373 30311 17794 6015

23481 73427 38229 184003 75713 38304 46341

3709 323

168518 14565

323976 104866

*

Includes repeated controls. Includes armed forces, students, and unemployed.

†

paternal occupational details. We likewise excluded 7,006 (2%) of the 289,602 live births recorded in the CBD who did not develop cancer. Table 1 shows the details of the number of cases and controls by paternal occupational social class and residential status (urban or rural). Table 2 shows the details of the number of cancer patients and controls, by diagnostic group, with known paternal occupations for all diagnostic groups with at least 5 cases over the study period. We did not include diagnostic groups with fewer than 5 cases (but do contribute to the totals). Of cases, 90 were born after December 31, 1993, the final date for which births were recorded on the CBD, and hence have only registry controls. Age 0–24 years Table 3 presents the ORs and corresponding 95% CIs for all ICCC diagnostic groups and subcategories with more than 5 cases, and at least 2 exposed cases, over the study period. Male offspring of fathers likely to have been occupationally exposed to pesticides or herbicides had a significantly lower risk of lymphoid leukemia than either registry or CBD controls. Negative associations, although nonsignificant, also appeared when we restricted the analysis to both urban (OR 0.38; 95% CI 0.12–1.21 using registry controls; OR 0.51; 95% CI 0.17–1.48 using CBD controls) and rural (OR 0.39; 95% CI 0.12–1.26 using registry controls; OR 0.40; 95% CI 0.13–1.26 using CBD controls). Using registry controls, we observed increased risks with likely paternal exposure to pesticides or herbicides for the “neuroblastoma and ganglioblastoma” and “fibrosarcoma, neurosarcoma, and other fibromatous neoplasms” categories, but again these were restricted to males and were not confirmed when using CBD controls. We saw a number of significantly reduced risks when using CBD controls but not when using registry controls (see Table 3). We observed similar results when adjusting for paternal occupational social class. None of these results remained statistically significant (results not presented) 140

when we stratified the analysis by urban versus rural status, ensuring that rural cases were matched to rural controls and urban cases matched to urban controls. Age 0–6 years In all, 435 cases of lymphoid leukemia were diagnosed in children aged younger than 7 years, of which 16 were offspring of fathers likely to have been occupationally exposed to pesticides or herbicides. A significantly reduced risk of lymphoid leukemia with likely paternal exposure to pesticides or herbicides remained when we restricted the analysis using CBD controls to cases aged younger than 7 years at diagnosis (OR 0.37; 95% CI 0.24–0.57). We observed similar results when we excluded from the analysis cases and controls from the village of Seascale, near the Sellafield nuclear installation (OR 0.38; 95% CI 0.25–0.57) and after adjusting for paternal occupational social class (OR 0.46; 95% CI 0.27–0.78). However, this relationship was not significant when stratified by urban versus rural status (OR 0.55; 95% CI 0.26–1.16 for urban status; OR 1.15; 95% CI 0.61–2.17 for rural status). We also saw a nonsignificant reduced risk when using registry controls (OR 0.76; 95% CI 0.50–1.15). We observed no other significant associations between other diagnostic categories and likely paternal exposure to pesticides or herbicides for this age group. In the sex-specific analyses, significantly reduced risk of lymphoid leukemia in this age group appeared with likely paternal exposure to pesticides or herbicides for boys using both sets of controls (OR 0.44; 95% CI 0.22–0.89 using registry controls; OR 0.23; 95% CI 0.11–0.46 using CBD controls). We also saw a reduced risk, of borderline statistical significance, for girls when we used registry controls (OR 0.58; 95% CI 0.34–1.00), whereas we observed a nonsignificant increased risk when using CBD controls (OR 1.22; 95% CI 0.72–2.07). Age 7–24 years We saw significant inverse associations between likely paternal exposure to pesticides or herbicides and risk of Hodgkin’s disease (OR 0.39; 95% CI 0.25–0.61), NonHodgkin’s lymphoma (OR 0.20; 95% CI 0.08–0.55), gonadal germ cell tumors (OR 0.25, 95% CI 0.11–0.56), malignant melanoma (OR 0.38; 95% CI 0.17–0.85) and carcinomas categorized as “other and unspecified carcinomas” (OR 0.30; 95% CI 0.15–0.60). In all cases, these results are similar to those seen when considering the entire age range of the registry (0–24 years) (see Table 3). COMMENT Researchers2–21 previously linked paternal occupational exposure to pesticides and/or herbicides with a number of childhood cancers. We examined the role of likely prebirth paternal occupational exposure to pesticides or herbicides Archives of Environmental & Occupational Health

Table 2.—Distribution of Cancer Cases and Controls by Disease Type (ICCC groupings), Case–Control Group, and Likely Paternal Occupational Exposure to Pesticides or Herbicides in 4,723 Children in the North of England Cases Diagnostic group Leukemia Lymphoid Acute nonlymphocytic Chronic myeloid Other specified Unspecified Lymphomas and reticuloendothelial neoplasm Hodgkin’s disease Non–Hodgkin’s lymphoma Burkitt’s lymphoma Unspecified lymphoma CNS tumor Ependymoma Astrocytoma Primitive neuroectodermal tumor Other gliomas Other specified intracranial and intraspinal neoplasms Unspecified CNS tumors Sympathetic nervous system tumors Neuroblastoma and ganglioneuroblastoma Other sympathetic nervous system tumors Retinoblastoma Renal tumors Wilms’s tumor, rhabdoid, and clear cell sarcoma Renal carcinoma Hepatic tumors Hepatoblastoma Hepatic carcinoma Malignant bone tumors Osteosarcoma Chondrosarcoma Ewing’s sarcoma Other specified malignant bone tumors Soft tissue sarcomas Rhabdomyosarcoma and embryonal sarcoma Fibrosarcoma, neurofibrosarcoma, and other fibromatous neoplasms Other specified soft tissue sarcomas Unspecified soft tissue sarcomas Germ cell, trophoblastic, and other gonadal neoplasms Intracranial and intraspinal germ cell tumors Other and unspecified nongonadal germ cell tumors Gonadal germ cell tumors Gonadal carcinomas Other and unspecified malignant gonadal tumors Carcinomas and other malignant neoplasms Adrenocortical carcinoma Thyroid carcinoma Nasopharyngeal carcinoma Malignant melanoma Skin carcinoma Other and unspecified carcinomas All cancers

Registry control* n Exposed

n

CBD control* Exposed

n

Exposed

1007 744 203 38 10 12 806 526 243 22 11 843 85 290 112 186 143 27 183 164 19 71 143 130 13 22 13 9 245 140 8 79 16 320 117 87

34 23 7 3 1 0 32 23 8 0 1 33 2 12 6 7 6 0 10 10 0 3 4 4 0 1 0 1 7 3 0 0 0 11 4 6

42118 30947 8821 1609 362 379 36538 23417 11539 872 529 36632 3692 12638 4649 8195 6203 1255 8264 7246 1018 3603 8264 6126 742 1056 586 470 13191 7403 337 4469 828 16329 6208 4640

1506 1129 303 52 12 10 1223 809 362 30 8 1262 110 450 144 286 227 45 278 247 31 110 256 230 26 43 26 17 442 252 8 151 27 560 204 155

95099 70800 18999 3500 900 900 77099 49900 23599 2200 1100 78598 7600 27199 10600 17599 13100 2500 16200 14400 1800 6700 12800 11600 1200 2100 1200 900 24099 13600 800 7899 1600 30500 11200 8600

7878 5574 1811 327 64 102 7293 4929 2096 154 23 6714 639 2362 852 1440 1184 237 1232 1093 139 468 973 861 112 183 105 78 2244 1310 79 671 168 2769 988 819

101 15 437 25 59 284 58 11 632 6 71 19 183 63 290 4727

0 1 14 1 1 8 3 1 21 0 2 0 8 2 9 171

4501 980 22039 1200 2499 15466 2442 432 29117 221 3389 1021 9262 3030 12194 216480

168 33 801 41 102 513 124 21 1119 10 131 31 372 112 463 7628

9300 1400 39799 2500 4600 26499 5100 1100 56500 400 6500 1900 17100 5400 25200 440794

853 109 3637 216 387 2417 511 106 5311 31 576 173 1535 453 2543 38833

Note. ICCC International Classification of Childhood Cancer; CBD Cumbrian Births Database; CNS central nervous system. * Includes repeated controls.

on the risk of cancer in children and young adults, and explored the findings for girls and boys separately. A negative association between risk of lymphoid leukemia and paternal occupations likely to have involved exposure to pesticides or herbicides, however, was no longer significant after stratification by urban versus rural status. Likewise, we observed a May/June 2006, Vol. 61, No. 3

number of significantly reduced risks when using CBD controls but not in the stratified analysis. The major strengths of this study are its size and population base. We included more than 4,500 cases of cancer in children and young adults from the NRYPMDR, which is believed to be more than 95% complete. We used 2 large, 141

142


0.61–1.22 0.50–1.15 0.45–2.09 0.72–8.15 0.84–1.73 0.87–2.03 0.51–2.11 0.79–1.61 0.19–3.33 0.66–2.13 0.70–3.78 0.50–2.30 0.49–2.58 0.85–3.11 0.94–3.48 0.51–5.33 0.26–1.92 0.29–2.15 0.40–1.80 0.19–1.94 0.52–1.74 0.38–2.80 0.85–4.58 0.50–1.46 0.41–1.68 0.35–3.74 0.55–1.34 0.18–3.05 0.53–2.22 0.20–3.48 0.42–1.59 0.86–1.17

0.86 0.76 0.97 2.45 1.21 1.33 1.03 1.13 0.81 1.18 1.63 1.07 1.13 1.62 1.81 1.65 0.70 0.79 0.84 0.61 0.95 1.03 1.97 0.85 0.83 1.14 0.86 0.74 1.08 0.84 0.82 1.00

1.42 1.47 2.00 0.70 0.98

0.75 — 1.16

0.64

3.89

1.33 0.75 1.19 0.93

3.27 1.34 1.52

2.15 2.38

1.34 0.83 0.87 1.16 0.79 1.63 0.58 0.68

0.65 0.44 1.33 2.34 1.16

0.18–11.2 0.45–4.79 0.46–8.78 0.17–2.89 0.79–1.22

0.33–1.70 — 0.57–2.39

0.28–1.44

1.48–10.20

0.58–3.05 0.18–3.09 0.55–2.55 0.23–3.85

0.76–14.1 0.41–4.35 0.47–4.95

0.92–5.02 1.01–5.57

0.76–2.35 0.30–2.25 0.51–1.48 0.28–4.87 0.29–2.14 0.58–4.58 0.14–2.38 0.17–2.79

0.39–1.09 0.22–0.89 0.54–3.30 0.54–10.2 0.72–1.87

Registry control Male OR CI

0.50 0.94 — 0.85 1.02

1.18 1.15 0.74

1.15

0.56

0.26 0.45 0.70 1.14

0.81 0.29 0.32

1.17 1.32

1.31 1.38 1.46 — 1.59 1.62 1.61 1.69

1.15 1.22 0.58 2.62 1.27

OR

CI

0.07–3.65 0.38–2.31 — 0.40–1.83 0.82–1.27

0.28–4.95 0.35–3.76 0.42–1.30

0.56–2.36

0.08–4.14

0.04–1.89 0.06–3.28 0.26–1.90 0.27–4.76

0.11–6.03 0.04–2.09 0.04–2.32

0.42–3.26 0.47–3.69

0.68–2.50 0.50–3.84 0.91–2.34 — 0.77–3.31 0.38–6.95 0.64–4.04 0.60–4.76

0.72–1.84 0.72–2.07 0.14–2.39 0.31–22.20 0.74–2.19

Female

Note. CBD Cumbrian Births Database; CNS central nervous system. Results listed in bold are significant at p .05.

Leukemia Lymphoid Acute nonlymphocytic Chronic myeloid Lymphoma and reticuloendothelial neoplasm Hodgkin’s disease Non–Hodgkin’s lymphoma CNS tumor Ependymoma Astrocytoma Primitive neuroectodermal tumor Other glioma Other specified intracranial and intraspinal neoplasma Sympathetic nervous system tumors Neuroblastoma and ganglioneuroblastoma Retinoblastoma Renal tumor Wilms’s tumor, rhabdoid, and clear cell sarcoma Malignant bone tumor Osteosarcoma Soft–tissue sarcoma Rhabdomyosarcoma and embryonal sarcoma Fibrosarcoma, neurofibrosarcoma and other fibromatous neoplasm Germ cell, trophoblastic and other gonadal neoplasm Gonadal germ cell tumor Gonadal carcinoma Carcinomas and other malignant neoplasms Thyroid carcinoma Malignant melanoma Skin carcinoma Other and unspecified carcinomas All cancers

Diagnostic group

All cases OR CI

0.32 0.43 0.42 0.29 0.38

0.27 0.56 0.33

0.33

0.71

0.29 0.21 0.37 0.38

0.62 0.39 0.45

0.71 0.81

0.38 0.27 0.44 0.29 0.48 0.56 0.46 0.40

0.38 0.37 0.36 0.59 0.35

0.08–1.33 0.20–0.92 0.10–1.73 0.14–0.59 0.33–0.45

0.13–0.57 0.17–1.81 0.21–0.53

0.19–0.58

0.31–1.64

0.14–0.62 0.07–0.66 0.20–0.68 0.14–1.04

0.20–1.99 0.14–1.06 0.16–1.21

0.36–1.40 0.41–1.60

0.24–0.59 0.12–0.60 0.31–0.63 0.07–1.20 0.27–0.87 0.23–1.39 0.22–0.99 0.16–0.98

0.27–0.55 0.24–0.57 0.17–0.77 0.14–2.46 0.24–0.51

All cases OR CI

0.56 0.57 0.57 0.28 0.35

0.27 — 0.57

0.24

1.27

0.42 0.25 0.44 0.34

1.02 0.63 0.70

0.72 0.83

0.29 0.25 0.29 0.40 0.29 0.51 0.25 0.15

0.29 0.23 0.47 0.42 0.29

0.07–4.30 0.18–1.83 0.18–1.83 0.07–1.15 0.28–0.43

0.12–0.62 — 0.18–1.83

0.11–0.55

0.50–3.25

0.18–0.95 0.06–1.04 0.21–0.95 0.08–1.41

0.24–4.33 0.20–2.01 0.22–2.24

0.29–1.78 0.34–2.07

0.16–0.55 0.09–0.68 0.16–0.52 0.10–1.67 0.11–0.78 0.16–1.64 0.06–1.03 0.02–1.05

0.17–0.49 0.11–0.46 0.19–1.16 0.06–3.14 0.17–0.48

CBD control Male OR CI

0.23 0.37 — 0.29 0.43

0.25 0.57 0.28

0.50

0.22

0.10 0.16 0.29 0.43

0.35 0.18 0.21

0.71 0.79

0.53 0.30 0.64 — 0.73 0.67 0.69 0.70

0.52 0.58 0.23 0.97 0.46

OR

Table 3.—Unadjusted Odds Ratios (ORs) and 95% Confidence Intervals (CIs) for Each International Classification of Childhood Cancer (ICCC) Diagnostic Group (Ages 0–24), by Sex

CI

0.03–1.66 0.13–1.00 — 0.13–0.67 0.34–0.54

0.03–1.80 0.18–1.84 0.15–0.50

0.23–1.06

0.03–1.61

0.01–0.74 0.02–1.14 0.11–0.79 0.10–1.77

0.05–2.58 0.03–1.33 0.03–1.55

0.26–1.95 0.28–2.18

0.28–1.01 0.07–1.24 0.40–1.02 — 0.36–1.51 0.16–2.80 0.28–1.72 0.25–1.95

0.33–0.84 0.34–1.00 0.06–0.93 0.12–7.59 0.26–0.82

Female

independent control groups, both with routinely collected data, eliminating any potential reporting bias. Further, the method of exposure classification was identical for both the registry and CBD groups, ensuring that any misclassification of occupations would be nondifferential. Findings differed between the control groups; in particular, we identified a number of significant negative associations using CBD controls but not registry controls. Results were similar after adjusting for socioeconomic status. None of these associations remained significant in the analyses stratified by urban versus rural status. The county of Cumbria, where all births recorded in the CBD occurred, is primarily a rural county with a higher than average risk for occupational exposures to pesticides or herbicides because of the nature of work there, as opposed to the more urban areas in the east of the study region. It is probable that this difference in the control groups led to the significantly reduced risks before stratification seen when using CBD controls. Draper et al34 identified excess risk of childhood leukemia in isolated and rural areas relative to urban areas. In particular, Kinlen35 observed significant excess risk of leukemia in rural communities with high levels of population mixing, which suggests an infective basis for the disease. Pearce et al36 used the same data as we did to identify an increased risk of childhood leukemia among the offspring of men considered likely to have high levels of occupational contact with other individuals. In the current study, we also found an increased risk of childhood leukemia in relation to paternal employment in occupations with likely exposures to electromagnetic fields,37 although this relationship was only detected in males. This study also has some limitations, many of which apply to most of the studies in this field. Because ascertainment of likely prebirth or preconceptional paternal exposures depended on paternal occupational codes from birth certificates rather than personal interviews, it was necessary to exclude at the outset all individuals (both cases and controls) with missing or invalid paternal occupation codes. This is most likely to include individuals whose fathers were either unemployed at the time of birth or were not recorded on the birth certificate. (Most likely, this represents births in the lowest socioeconomic group.) It is not possible to quantify whether these exclusions may have introduced bias into the study. We used job titles as a surrogate for likely exposure, resulting in a lack of data on levels, duration, or timing of exposure, specific pesticide exposures, and possible changes in exposure types and levels over the long study period. Paternal employment at the time of birth may not accurately reflect actual prebirth or preconceptional exposures, particularly in situations where the father may have changed jobs or where the person named on the birth certificate may not be the child’s biological father. It is also possible that individuals may be misclassified for paternal occupational exposure because of ambiguity regarding common job titles and varying levels and likelihood of exposure, which may account for our findings (although this is less likely with primarily agricultural occupations than for many May/June 2006, Vol. 61, No. 3

other areas of employment). We also used a large number of statistical tests, which may have resulted in an inflated type I error rate and raised the possibility of spurious results. However, the small numbers in some of the cancer groupings considered, particularly in the stratified analysis, could have led to a lack of statistical power for some diagnostic groupings. We also were unable to examine other factors, such as maternal occupational exposures and environmental or lifestyle factors in childhood or in early adult life. Our findings, consistent with those of previous research,23–25 including a large study from the UK,24 failed to support the hypothesis that paternal exposure to pesticides or herbicides is associated with an increased risk of childhood cancer. We found a reduced risk of lymphoid leukemia among males, regardless of which control group was used. Researchers in 3 previous studies26,27,38 have noted inverse associations between paternal pesticide exposure and risk of germ cell tumors in male offspring. We confirmed this, but it was apparent only when using CBD controls and not after stratifying for urban versus rural status. We also observed significant inverse associations for Hodgkin’s disease, NonHodgkin’s lymphoma, malignant melanoma, and carcinomas categorized as “other and unspecified.” In all cases, we saw these associations only when using CBD controls, a group more likely to live in rural areas. Although it is likely that the lack of initial matching on urban versus rural status may have accounted for these significant inverse associations, paternal pesticide exposure previously has been linked with a sex ratio significantly different from that normally seen at birth39 and reduced semen quality and quantity.40,41 The only significant positive association was for neuroblastoma and ganglioblastoma. This was limited to males, was seen only when using registry controls, and was based on a small number (6) of exposed male cases. However, it is consistent with a previously reported increased risk of neuroblastoma in relation to paternal occupational exposure to insecticides.42 In conclusion, we could find no support for the hypothesis of a relationship between preconception or prebirth paternal occupational exposure to pesticides or herbicides and risk of childhood cancer. In light of the fact that the rural nature of many such exposures is associated with rural settings, urban versus rural status should be accounted for in analyses of this potential association wherever possible. ********** The North of England Children’s Cancer Research Fund and Children with Leukemia supported this study. The Newcastle Hospitals NHS Trust funds the Northern Region Young Person’s Malignant Disease Registry. The authors thank Richard Hardy for computing support, Claire Hamilton for assistance with accessing data from the Registry, and Katharine Kirton for assistance with the Cumbrian Births Database. Requests for reprints should be sent to Dr Mark S. Pearce, Paediatric and Lifecourse Epidemiology Research Group, School of Clinical Medical Sciences (Child Health), University of Newcastle upon Tyne, Sir James Spence Institute, Royal Victoria Infirmary, Newcastle upon Tyne, NE1 4LP. E-mail: [email protected]

********** 143

References 1. Little J. Epidemiology of Childhood Cancer. IARC Scientific Publications, No. 149. Lyon: International Agency for Research on Cancer; 1999. 2. Laval G, Tuyns AJ. Environmental factors in childhood leukaemia. Br J Ind Med. 1988;45:843–844. 3. Buckley JD, Robison LL, Swotinsky R, et al. Occupational exposures of parents of children with acute nonlymphocytic leukemia: a report from the Children’s Cancer Study Group. Cancer Res. 1989;49:4030–4037. 4. Buckley JD, Buckley CM, Ruccione K, et al. Epidemiological characteristics of childhood acute lymphocyctic leukemia: analysis by immunophenotype. Leukemia. 1994;8:856–864. 5. Magnani C, Pastore G, Luzzato L, Terracini B. Parental occupation and other environmental factors in the etiology of leukemias and nonHodgkin’s lymphoma in childhood: a case-control study. Tumori. 1990;76:413–419. 6. Infant-Rivard C, Sinnett D. Preconceptional paternal exposure to pesticides and increased risk of childhood leukaemia. Lancet. 2001;354:1819. 7. Meinert R, Schuz J, Kaletsch U, Kaatsch P, Michaelis J. Leukemia and non-Hodgkin’s lymphoma in childhood and exposure to pesticides: results of a register-based case-control study in Germany. Am J Epidemiol. 2000;151:639–646. 8. Heacock H, Hertzmann C, Demers PA, et al. Childhood cancer in the offspring of male sawmill workers occupationally exposed to chlorophenate fungicides. Environ Health Perspect. 2000;108:499–503. 9. Kristensen P, Andersen A, Irgens LM, Bye AS, Sundheim 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. 10. Flower KB, Hoppin JA, Lynch CF, et al. Cancer risk and parental pesticide application in children of agricultural health study participants. Environ Health Perspect. 2004;112:631–635. 11. Fear NT, Roman E, Reeves G, Pannett B. Childhood cancer and paternal employment in agriculture: the role of pesticides. Br J Cancer. 1998;77:825–829. 12. Sharpe CR, Franco EL, de Camargo B, et al. Parental exposures to pesticides and risk of Wilms’ tumor in Brazil. Am J Epidemiol. 1995;141:210–217. 13. Cordier S, Lefeuvre B, Filippini G, et al. Parental occupation, occupational exposures to solvents and polycyclic aromatic hydrocarbons and risk of childhood brain tumours (Italy, France, Spain). Cancer Causes Control. 1997;8;688–697. 14. Wilkins JR, Koutras RA. Paternal occupation and brain cancer in offspring: a mortality-based case-control study. Am J Ind Med. 1988;14:299–318. 15. Cordier S, Mandereau L, Preston-Martin S, et al. Parental occupations and childhood brain tumors: results of an international case-control study. Cancer Causes Control. 2001:12:865–874. 16. van Wijngaarden E, Stewart PA, Olshan AF, et al. Parental occupational exposure to pesticides and childhood brain cancer. Am J Epidemiol. 2003;157:989–997. 17. Wilkins JR, Sinks T. Parental occupation and intracranial neoplasms of childhood: results of a case-control study. Am J Epidemiol. 1990;132:275–292. 18. Kuijten RR, Bunin GR, Nass CC, Meadows AT. Parental occupation and childhood astrocytoma: results of a case-control study. Cancer Res. 1992;52:782–786. 19. Feychting M, Plato N, Nise G, Ahlbom A. Paternal occupational exposures and childhood cancer. Environ Health Perspect. 2001;109:193–196. 20. Holly EA, Aston DA, Ahn DK, Kristiansen JJ. Ewing’s bone sarcoma, paternal occupational exposure and other factors. Am J Epidemiol. 1992;135:122–129. 21. Valery PC, McWhirter W, Sleigh A, et al. Farm exposures, parental occupation, and risk of Ewing’s sarcoma in Australia: a national case-control study. Cancer Causes Control. 2002;13:263–270.

144

22. Pearce MS, Parker L. Paternal employment in agriculture and childhood kidney cancer. Pediatr Hematol Oncol. 2000;17:223–230. 23. Mutanen P, Hemminki K. Childhood cancer and parental occupation in the Swedish family-cancer database. J Occup Environ Med. 2001;43:952–958. 24. McKinney PA, Fear NT, Stockton D. Parental occupation at periconception: findings from the United Kingdom Childhood Cancer Study. Occup Environ Med. 2003;60:901–909. 25. Rodvall Y, Dich J, Wiklund K. Cancer risk in offspring of male pesticide applicators in agriculture in Sweden. Occup Environ Med. 2003;60:798–801. 26. Chen Z, Stewart PA, Davies S, et al. Parental occupational exposure to pesticides and childhood germ-cell tumors. Am J Epidemiol. 2005; 162:858–867. 27. Chen Z, Robison L, Giller R, et al. Environmental exposure to residential pesticides, chemicals, dusts, fumes, and metals, and risk of childhood germ cell tumors. Int J Hyg Environ Health. 2006;209: 31–40 28. Cotterill SJ, Parker L, Malcolm AJ, et al. Incidence and Survival for cancer in children and young adults in the Northern Region Young Person’s Malignant Disease Registry. Br J Cancer. 2000;83: 397–403. 29. Pearce MS, Parker L, Windebank KP, et al. Cancer in adolescents and young adults aged 15–24 years: a report from the North of England Young Person’s Malignant Disease Registry, UK. Pediatr Blood Cancer. 2005;45:687–693. 30. Parker L, Smith J, Dickinson H, et al. The creation of a database of children of workers at a nuclear facility: an exercise in record linkage. Appl Occup Environ Hyg. 1997;12:40–46. 31. Percy C, van Holten V, Muir CE. International Classification of Diseases for Oncology. Second Edition. Geneva: World Health Organization; 1990. 32. Kramárová E, Stiller CA, Ferlay J, et al. International Classification of Childhood Cancer (IARC Technical Report). Lyon, France: International Agency for Research on Cancer; 1996. 33. Office of Population Census and Surveys. Standard Occupational Classification Volumes 1–3. London, England: Her Majesty’s Stationery Office; 1990. 34. Draper G J, Vincent T J, O’Connor C M, Stiller C A. Socio-economic factors and variation in incidence rates between county districts. In: Draper G, ed. The Geographical Epidemiology of Childhood Leukemia and Non-Hodgkin’s Lymphomas in Great Britain, 1966– 1983. London, England: HMSO; 1991:37–46. 35. Kinlen LJ. Epidemiological evidence for an infective basis in childhood leukaemia. Br J Cancer. 1995;71:1–5. 36. Pearce MS, Cotterill SJ, Parker L. Fathers’ occupational contacts and risk of childhood leukemia and non-Hodgkin lymphoma. Epidemiology. 2004;15:352–356. 37. Pearce MS, Hammal DM, Dorak MT, McNally RJQ, Parker L. Paternal occupational exposure to radiation and electro-magnetic fields as risk factors for cancer in children and young adults. A casecontrol study from the North of England. Pediatr Blood Cancer. In press. 38. Møller H. Work in agriculture, childhood residences, nitrate exposure, and testicular cancer risk: a case-control study in Denmark. Cancer Epidemiol Biomarkers Prev. 1997;6:141–144. 39. Mocarelli P, Gerthoux PM, Ferrari E, et al. Paternal concentrations of dioxin and sex ratio of offspring. Lancet. 2000;355:1858–1863. 40. Swan SH. Semen quality in fertile US men in relation to geographical area and pesticide exposure. Int J Anrol. 2006;29:62–68. 41. Narmol-Maneiro L, Fernandez-D’Pool J, Sanchez BJ, et al. Seminal profile in workers exposed to cholinesterase inhibitor insecticides. Invest Clin. 2003;44:105–117. 42. Kerr MA, Nasca PC, Mundt KA, et al. Parental occupational exposure and risk of neuroblastoma: a case-control study (United States). Cancer Causes Control. 2000;11:635–643.
