Risk of Cancer in Children of Parents Occupationally Exposed to Hydrocarbon Solvents and Engine Exhaust Fumes: A Register-Based Nested Case-Control Study from Sweden (1960-2015).

BACKGROUND: It remains unclear whether parental occupational exposure to hydrocarbon solvents (HCS) or engine exhaust fumes (EEF) is associated with higher risks of cancer in the offspring.
OBJECTIVES: Our aim was to estimate relative risks of childhood cancers associated with maternal or paternal exposure to aliphatic/alicyclic, aromatic, or chlorinated HCS or gasoline/diesel EEF.
METHODS: We conducted a case-control study in which individuals <20y old, born 1960-2014, were identified from the Swedish National Cancer Register (1960-2015) at first cancer diagnosis and matched to population controls (1 case:25 controls) on birth year and sex. Maternal and paternal occupation around the child's birth was retrieved for 9,653 cases and 172,194 controls and 12,521 cases and 274,434 controls, respectively, using information from six censuses and a nationwide register. Using the Swedish job-exposure matrix (SWEJEM), we assessed exposure to HCS and EEF (any or higher/lower). Odds ratios (ORs) and 95% confidence intervals (CIs) of 15 childhood cancer subtypes were estimated using conditional logistic regression models adjusted for several confounders.
RESULTS: Maternal exposure to aromatic HCS was associated with non-Hodgkin lymphoma (OR=1.64; 95% CI: 1.05, 2.58), aliphatic/alicyclic HCS with germ cell tumors (OR=1.52; 95% CI: 0.89, 2.59), and gasoline/diesel EEF with astrocytoma (OR=1.40; 95% CI: 1.04, 1.88), myeloid leukemia (OR=1.53; 95% CI: 0.84, 2.81), lymphomas (OR=1.60; 95% CI: 0.85, 3.02 for Hodgkin; OR=1.44; 95% CI: 0.71, 2.91 for non-Hodgkin), and epithelial tumors (OR=1.51; 95% CI: 0.93, 2.44). Paternal exposure to gasoline EEF was associated with Hodgkin lymphoma (OR=1.21; 95% CI: 1.01, 1.44) and soft tissue sarcomas (OR=1.22; 95% CI: 1.00, 1.48). No notable difference was observed between higher and lower exposure. DISCUSSION: Our findings suggest that occupational exposure to HCS or EEF, especially in the mother, may increase the risk of some childhood cancers. They add to the growing literature on adverse effects from HCS and EEF in the child, but replication of these associations in other populations is warranted. https://doi.org/10.1289/EHP11035.

Introduction

Occupational exposure to fumes from aliphatic, aromatic, or chlorinated hydrocarbon solvents (HCS) and gasoline or diesel engine exhaust fumes (EEF) has long been implicated in the etiology of adult cancers.[1-3] It is reasonable to assume that parental exposure to these carcinogens at the critical time of preconception or the perinatal and postnatal periods may increase the risk of pediatric cancer in the offspring.[4] Indeed, previous epidemiological studies reported higher risks of some childhood cancer subtypes associated with self-reported maternal and/or paternal HCS exposure. A moderately increased risk of acute lymphoid leukemia (ALL) was observed in children of mothers occupationally exposed to HCS overall[5,6] and to benzene (an aromatic HCS) specifically.[7,8] Only one study observed a potentially increased risk of ALL associated with paternal exposure to benzene.[8] These associations were not replicated in other material.[5,7-11] In a register-based assessment from Denmark, children whose mothers or fathers worked in painting or printing industries where aliphatic and aromatic HCS use is abundant were about two times more likely to be diagnosed with acute myeloid leukemia or central nervous system (CNS) tumors in comparison with controls.[12] Similarly, about 2-fold increased risks of leukemia, primarily ALL, were found in children of mothers who self-reported exposure to paints.[6,13] By contrast, paternal self-reported employment in such industries was not associated with higher risks of CNS tumors or leukemia.[6,14]

Information on risks of childhood cancer from parental EEF exposure is scarce. A recent Danish register–based study indicated that maternal occupational exposure to diesel EEFs led to higher risks of astrocytoma but not leukemia in the child.[15] However, previous studies based on self-reports of occupation or EEF exposure found no elevated risk for CNS tumors[16] but did find increased risks of ALL from parental,[17] paternal,[11,18] or maternal exposure.[11,19] Children whose fathers were exposed to gasoline EEF also appeared to be at risk of non-Hodgkin lymphoma in one study.[11]

It is challenging to draw conclusions from previous investigations because most findings were based on a few exposed parents who commonly self-reported exposure to HCS and/or EEF after their child’s diagnosis. In addition, maternal exposure, some HCS and EEF or their components, and other cancer types were inadequately examined. To overcome some of these limitations, we conducted a large case–control study using register-based data and a Swedish job-exposure matrix (SWEJEM). We aimed to estimate relative risks of childhood cancer by subtype associated with maternal or paternal occupational exposure to HCS (aliphatic/alicyclic, aromatic, or chlorinated) and gasoline and diesel EEF around the offspring’s birth.

Methods

Childhood Cancer Cases and Noncancer Controls

In this nested population-based case–control study, we identified incident cases of childhood cancer from the Swedish National Cancer Register, which includes diagnoses established by means of clinical or laboratory investigations on a nationwide scale since 1958.[20] We included children, ages 0–19 y, with a first cancer diagnosis in the register between 1960 and 2015. All identified childhood cancer cases were born in Sweden between 1960 and 2014. We excluded those with benign non-CNS tumors using two algorithms primarily based on the National Board of Health and Welfare’s (Socialstyrelsen) classification of benign tumors according to the calendar year of diagnosis. For those diagnosed in 1992 or earlier, tumors were considered benign if the third digit of the histopathology code (PAD/C24.1) was not a “6,” except for tumors with a combination of International Classification of Diseases (ICD), seventh revision (ICD-7) or ICD-O/2 and/or PAD/C24.1 (pathology) codes listed in Table S1. For those diagnosed in 1993 or later, tumors were considered benign if the fifth digit in the morphology code was not a “3.” The second rule did not apply to all CNS tumors, intracranial and intraspinal germ cell tumors, or leukemias. In addition, Hodgkin and other lymphomas were not excluded if the morphology code indicated suspicion of malignancy.

Following the International Classification of Childhood Cancer, third edition (ICCC-3), main and extended classification,[21] we categorized cancers into lymphoid leukemia, myeloid leukemia (including myeloproliferative and myelodysplastic syndromes), Hodgkin and non-Hodgkin lymphoma, ependymoma, astrocytoma and other gliomas (including the groups IIIb and IIId from the ICCC-3[21] and morphology code 9425), CNS embryonal tumors (including medulloblastoma), neuroblastoma, retinoblastoma, renal tumors, hepatic tumors, bone tumors, soft-tissue sarcomas, germ cell tumors, and epithelial tumors. Childhood cancers in the National Cancer Register were coded using four ICD versions, and codes were grouped using the best information available at the time of diagnosis aiming to mimic the groups in ICCC-3.[21-23]

Up to 25 controls born in Sweden without a cancer diagnosis in the National Cancer Register at the time of the corresponding case’s diagnosis were sampled from the Total Population Register[24] and individually matched to cases on birth year and sex using incidence density sampling. We excluded cases and controls with a diagnosis of Down or a neurocutaneous syndrome (e.g., neurofibromatosis or tuberous sclerosis) in the National Medical Birth Register (data available since 1973), National Patient Register (1964 and onward),[25] or National Cause of Death Register (1952 and onward)[26] to focus on offspring without a cancer-predisposing syndrome. All remaining children were linked to their biological mothers and/or fathers using the Multi-Generation Register, which has excellent coverage for individuals born since 1932 and living in Sweden anytime since 1961.[27] Children born outside Sweden were excluded to ensure unbiased ascertainment of parents and their occupational exposures. The unique personal identification number[28] assigned by authorities at birth was used to link records across registers.

Ethical permission for this study was obtained from the Regional Ethics Review Board in Stockholm (original 2011/634-31/4, amendments 2014/417-32 and 2016/27-32), which waived the need to request informed participant consent. All analyses were conducted in line with the Declaration of Helsinki.

Parental Occupational Exposure to Hydrocarbon Solvents and Engine Exhaust Fumes

We assessed parental occupational exposure to HCS and EEF around the child’s birth. The employment history of parents was retrieved from six censuses[29] conducted by Statistics Sweden in 1960, 1970, 1975, 1980, 1985, and 1990 for children born 1960–1996 and the Longitudinal Integrated Database for Health Insurance and Labor Market Studies (LISA)[30] for children born 1997–2014. Although LISA data were available on an annual basis after 2001, parental occupation during the early study period was obtained from the census or LISA record nearest to a child’s birth as detailed in Table S2. Occupations in censuses and LISA were recorded using three-digit codes based on the Nordic Classification of Occupations (NYK) and the 1996 Swedish Standard for Classification of Occupations (SSYK 96), respectively.[29,30] These were adapted from the 1958 and 1988 editions of the International Standard Classification of Occupations (ISCO-58/88-COM), respectively.[31] We excluded cases and controls whose mother or father’s employment could not be identified because they were unemployed or because their employment was not registered or could not be classified using the coding system (Figure 1).

Figure 1.

Diagram shows the derivation of the analytical sample used in analyses of maternal (left panel) and paternal exposures (right panel).

Parental occupational exposure to HCS and EEF was evaluated using a job-exposure matrix, SWEJEM. SWEJEM originated from the widely used Finnish matrix (FINJEM),[32] which was adapted to the Swedish occupational context and covers occupations coded in both census and LISA data.[33] SWEJEM include exposure probability and exposure level during 12 calendar periods from 1945 to 2018.[33] Exposures of interest were aliphatic/alicyclic, aromatic, and chlorinated HCS, as well as gasoline and diesel EEF. Gasoline, benzene, polycyclic aromatic hydrocarbons (PAH), and carbon monoxide were assessed separately to determine agent-specific effects and allow for comparisons with previous studies. Exposure to gasoline and benzene was ascertained in parents of children born 1960–1996, when their use was more common.[34] The data used to define the exposure variables used in this study originated from measurements in occupational settings. SWEJEM was evaluated in several studies assessing exposure in different Swedish populations.[35,36]

In the main analyses, we classified children of parents employed in an occupation with nonzero probability (P) or level (L) of exposure in SWEJEM as exposed. In another analysis, we examined occupations with higher and lower exposure separately. For each occupational exposure and occupation title, information on prevalence of exposures within an occupation and level of exposure in SWEJEM could vary by calendar period. For census data, information from SWEJEM representing the periods 1960–1974, 1975–1984, 1985–1994, and 1995–1996 was used. For LISA data, we used SWEJEM information on prevalence and level for the periods 1998–2000, 2001–2003, 2004–2006, 2007–2009, 2010–2012, and 2013–2015. For each occupational exposure, we used the median of the product of P and L as a cutoff value to classify occupation titles into those with higher and lower exposure. The median was estimated separately among mothers and fathers of exposed controls for the whole study period (Table S3). Occupational titles with calendar period specific above or equal to the median were considered of higher exposure, whereas those with lower than the median were regarded of lower exposure to the pollutant of interest (Table S4).

Other Variables

We collected data on potential confounders from nationwide registers. From the Total Population Register, we received information on a child’s birth year, sex (male or female), region of residence at diagnosis (grouped into six health care regions: Stockholm including Gotland, Uppsala-Örebro, West, South, Southeast, or North), and birth order (first, subsequent birth, or missing). We obtained maternal and paternal birth year (to calculate age at childbirth) and country of birth (Sweden or other) from the same register.

We retrieved information on parental education (, 10–12, of completed education, or missing) from censuses (data available 1960, 1970, 1975, 1980, 1985, and 1990) or LISA (). For children born before 1990, when educational level was sporadically collected in censuses, parental education was imputed from the census closest to birth or LISA. We also collected a calendar period-specific indicator of parental occupational social class, which was further grouped into blue-collar worker, lower-level white-collar worker, upper-level white-collar worker, or unclassified/unknown socioeconomic status using the codes specified in Table S5. Parental history of cancer was defined as any cancer diagnosis in the National Cancer Register before their child’s birth. Maternal self-reported smoking during the first trimester of pregnancy (yes, no, or missing) was obtained from the National Medical Birth Register, with data available from 1982 onward.

Statistical Analysis

Maternal and paternal occupational exposures were analyzed separately. We estimated tetrachoric correlations to examine the co-occurrence of exposures in cases and controls. We used conditional logistic regression models to estimate odds ratios (ORs) and corresponding 95% confidence intervals (CIs) of childhood cancer, any and by type, associated with each of the parental occupational exposures. We used the binary definition of exposure (exposed vs. unexposed) in primary analyses and compared higher and lower exposure to no exposure in a secondary analysis. Models were inherently adjusted for the matching variables (child’s age at diagnosis and sex) and further controlled for child’s birth order and region of residence at diagnosis and the parental variables (maternal or paternal, depending on the analysis) of country of birth and age, education, and cancer history at child’s birth. Estimates from models adjusted only for the matching variables were nearly identical to the fully adjusted models; hence, only the latter are reported. Missing values for birth order, parental education, and maternal smoking were marked as a separate category and included in all models. Because values were missing for a small proportion of subjects (), we did not impute values to avoid unnecessary computational complexity.

Sensitivity Analyses

We performed several sensitivity analyses to examine whether our findings were susceptible to biases, primarily exposure misclassification. We repeated the main analyses considering only children born around a census year or in 2000 and later to identify parental occupation more accurately. In another analysis, we included only children whose parents’ employment could be assessed before or at birth. Moreover, in an additional analysis we defined parents as exposed if they worked in occupations with exposure probability in SWEJEM (vs. in the main analysis). To reduce residual confounding by socioeconomic status, we reran our models considering only children of parents classified as blue-collar workers. We also further adjusted for maternal first trimester smoking to better account for differences in lifestyle factors among parents of cases and controls. The latter analysis was restricted to children born from 1982 onward. Furthermore, to check whether associations changed over time or by source of employment information, we stratified our analyses into two groups comparing children born 1960–1996 (census) and those born 1997–2014 (LISA). Last, we conducted a sensitivity analysis including parents who were excluded from the main analysis because their employment information could not be assessed, considering them as unexposed.

Data management and analyses were performed using SAS software (version 9.4; SAS Institute Inc.), and R (version 4.2.0; R Foundation for Statistical Computing).

Results

We compared maternal occupational exposures between 9,653 childhood cancer cases and 172,194 controls and paternal exposures between 12,521 cases and 274,434 controls (mean age at diagnosis 9 y; female 46% in both analyses). Sociodemographic characteristics of children and their parents were similar between cases and controls in both maternal and paternal analyses (Table 1). In comparison with unexposed parents, those exposed to any of the HCS were commonly employed in blue-collar jobs, whereas differences in socioeconomic status among parents exposed and unexposed to EEF were less pronounced (Table S6).

Table 1

Demographic and other characteristics of childhood cancer cases, their matched controls, and parents.

	Maternal analysis		Paternal analysis
	Cases(n=9,653)	Controls(n=172,194)	Cases(n=12,521)	Controls(n=274,434)
Child’s age at cancer diagnosis [y (mean SD)]	8.7 ± 6.3	8.5 ± 6.3	9.1 ± 6.3	9.2 ± 6.3
Child’s sex [n (%)]
Female	4,461 (46.2)	79,476 (46.2)	5,774 (46.1)	126,541 (46.1)
Male	5,192 (53.8)	92,718 (53.8)	6,747 (53.9)	147,893 (53.9)
Child’s region of residence at diagnosis [n (%)]
Stockholm-Gotland	2,075 (21.5)	35,973 (20.9)	2,469 (19.7)	52,942 (19.3)
Uppsala-Örebro	1,910 (19.8)	37,139 (21.6)	2,596 (20.7)	61,017 (22.2)
West	1,853 (19.2)	32,899 (19.1)	2,446 (19.5)	52,421 (19.1)
South	1,710 (17.7)	29,595 (17.2)	2,207 (17.6)	47,417 (17.3)
Southeast	1,146 (11.9)	19,272 (11.2)	1,486 (11.9)	31,686 (11.5)
North	959 (9.9)	17,316 (10.1)	1,317 (10.5)	28,951 (10.5)
Child’s birth year [median (IQR)]	1988 (1977, 1997)	1990 (1982, 1998)	1984 (1972, 1994)	1983 (1972, 1993)
Child’s birth order [n (%)]
First-born	4,515 (46.8)	76,895 (44.7)	5,299 (42.3)	112,962 (41.2)
Subsequent birth	5,138 (53.2)	95,299 (55.3)	7,211 (57.6)	161,441 (58.8)
Missing	—	—	11 (0.1)	31 (0.0)
Parental age at child’s birth [y (mean SD)]	28.8 ± 5.2	29.0 ± 5.1	31.4 ± 6.1	31.3 ± 6.1
Parental country of birth [n (%)]
Sweden	8,738 (90.5)	156,456 (90.9)	11,343 (90.6)	247,145 (90.1)
Other	915 (9.5)	15,738 (9.1)	1,178 (9.4)	27,289 (9.9)
Parental years of education [n (%)]
≤9	2,367 (24.5)	38,139 (22.1)	4,112 (32.8)	92,937 (33.9)
10–12	4,660 (48.3)	84,899 (49.3)	5,692 (45.5)	124,563 (45.4)
≥13	2,594 (26.9)	48,459 (28.1)	2,600 (20.8)	54,530 (19.9)
Missing	32 (0.3)	697 (0.4)	117 (0.9)	2,404 (0.9)
Parental socioeconomic class [n (%)]
Blue-collar worker	3,717 (38.5)	69,875 (40.6)	6,734 (53.8)	146,594 (53.4)
Lower-level white-collar worker	3,698 (38.3)	61,168 (35.5)	2,980 (23.8)	66,031 (24.1)
Upper-level white-collar worker	995 (10.3)	18,432 (10.7)	1,631 (13.0)	34,390 (12.5)
Unclassified/unknown	1,243 (12.9)	22,719 (13.2)	1,176 (9.4)	27,419 (10.0)
Parental history of cancer at child’s birth [n (%)]	235 (2.4)	4,114 (2.4)	66 (0.5)	837 (0.3)
Maternal first trimester smokinga [n (%)]	(n=6,456)	(n=129,286)	(n=6,859)	(n=148,587)
Yes	1,043 (16.2)	21,583 (16.7)	1,128 (16.4)	25,377 (17.1)
No	4,833 (74.9)	96,821 (74.9)	5,085 (74.1)	109,865 (73.9)
Missing	580 (9.0)	10,882 (8.4)	646 (9.4)	13,345 (9.0)

Note: —, no data; IQR, interquartile range; SD, standard deviation.

Evaluated for children born 1982 or later when maternal smoking data was available in the National Medical Birth Register.

A higher proportion of fathers than mothers was exposed to HCS or EEF (Tables 2 and 3). Exposure to gasoline and diesel EEF was very likely to co-occur (correlation coefficients ; Table S7). The number of exposed cases and controls by parent’s occupation and the median probability and level by exposure are listed in Tables S8 and S9, respectively.

Table 2

Odds ratios of childhood cancer subtypes associated with maternal occupational exposure to hydrocarbon solvents or engine exhaust fumes around the time of the child’s birth.

Cancer	Maternal occupational exposure	Cases (n)	Controls (n)	Exposed cases [n (%)]	Exposed controls [n (%)]	Adjusted OR (95% CI)
Lymphoid leukemia	Aliphatic/alicyclic HCS	1,943	35,924	35 (1.8)	742 (2.1)	0.86 (0.61, 1.21)
	Aromatic HCS	1,943	35,924	151 (7.8)	2,943 (8.2)	0.91 (0.76, 1.08)
	Chlorinated HCS	1,943	35,924	136 (7.0)	2,496 (6.9)	0.99 (0.83, 1.19)
	Gasoline EEF	1,943	35,924	70 (3.6)	1,170 (3.3)	1.16 (0.91, 1.49)
	Diesel EEF	1,943	35,924	59 (3.0)	1,000 (2.8)	1.16 (0.88, 1.52)
Myeloid leukemiaa	Aliphatic/alicyclic HCS	338	5,952	<5	140 (2.4)	Not estimable
	Aromatic HCS	338	5,952	25 (7.4)	496 (8.3)	0.81 (0.53, 1.24)
	Chlorinated HCS	338	5,952	13 (3.8)	407 (6.8)	0.54 (0.31, 0.96)
	Gasoline EEF	338	5,952	12 (3.6)	194 (3.3)	1.23 (0.68, 2.24)
	Diesel EEF	338	5,952	12 (3.6)	161 (2.7)	1.53 (0.84, 2.81)
Hodgkin lymphoma	Aliphatic/alicyclic HCS	527	9,089	14 (2.7)	196 (2.2)	1.22 (0.70, 2.13)
	Aromatic HCS	527	9,089	63 (12.0)	917 (10.1)	1.15 (0.87, 1.52)
	Chlorinated HCS	527	9,089	48 (9.1)	744 (8.2)	1.06 (0.77, 1.45)
	Gasoline EEF	527	9,089	12 (2.3)	173 (1.9)	1.23 (0.68, 2.23)
	Diesel EEF	527	9,089	11 (2.1)	130 (1.4)	1.60 (0.85, 3.02)
Non-Hodgkin lymphoma	Aliphatic/alicyclic HCS	233	4,596	<5	78 (1.7)	Not estimable
	Aromatic HCS	233	4,596	25 (10.7)	335 (7.3)	1.64 (1.05, 2.58)
	Chlorinated HCS	233	4,596	18 (7.7)	308 (6.7)	1.24 (0.75, 2.06)
	Gasoline EEF	233	4,596	10 (4.3)	149 (3.2)	1.43 (0.74, 2.79)
	Diesel EEF	233	4,596	9 (3.9)	133 (2.9)	1.44 (0.71, 2.91)
Ependymoma	Aliphatic/alicyclic HCS	246	4,440	5 (2.0)	92 (2.1)	0.98 (0.39, 2.48)
	Aromatic HCS	246	4,440	19 (7.7)	344 (7.7)	1.05 (0.64, 1.72)
	Chlorinated HCS	246	4,440	14 (5.7)	271 (6.1)	1.00 (0.57, 1.76)
	Gasoline EEF	246	4,440	8 (3.3)	141 (3.2)	1.21 (0.58, 2.53)
	Diesel EEF	246	4,440	6 (2.4)	120 (2.7)	1.05 (0.45, 2.45)
Astrocytoma and other gliomas	Aliphatic/alicyclic HCS	1,694	30,357	31 (1.8)	627 (2.1)	0.91 (0.63, 1.32)
	Aromatic HCS	1,694	30,357	156 (9.2)	2,707 (8.9)	1.04 (0.87, 1.23)
	Chlorinated HCS	1,694	30,357	126 (7.4)	2,264 (7.5)	1.00 (0.83, 1.21)
	Gasoline EEF	1,694	30,357	55 (3.2)	829 (2.7)	1.29 (0.98, 1.71)
	Diesel EEF	1,694	30,357	49 (2.9)	702 (2.3)	1.40 (1.04, 1.88)
CNS embryonal tumors	Aliphatic/alicyclic HCS	510	9,004	9 (1.8)	187 (2.1)	0.89 (0.45, 1.75)
	Aromatic HCS	510	9,004	36 (7.1)	733 (8.1)	0.81 (0.57, 1.15)
	Chlorinated HCS	510	9,004	32 (6.3)	644 (7.2)	0.86 (0.59, 1.25)
	Gasoline EEF	510	9,004	14 (2.7)	283 (3.1)	0.90 (0.51, 1.56)
	Diesel EEF	510	9,004	11 (2.2)	251 (2.8)	0.81 (0.44, 1.51)
Neuroblastoma	Aliphatic/alicyclic HCS	352	6,859	<5	119 (1.7)	Not estimable
	Aromatic HCS	352	6,859	25 (7.1)	417 (6.1)	1.16 (0.75, 1.79)
	Chlorinated HCS	352	6,859	13 (3.7)	404 (5.9)	0.60 (0.34, 1.05)
	Gasoline EEF	352	6,859	17 (4.8)	283 (4.1)	1.21 (0.73, 2.01)
	Diesel EEF	352	6,859	13 (3.7)	253 (3.7)	1.02 (0.57, 1.81)
Retinoblastoma	Aliphatic/alicyclic HCS	196	3,493	5 (2.6)	60 (1.7)	1.23 (0.48, 3.16)
	Aromatic HCS	196	3,493	14 (7.1)	267 (7.6)	0.85 (0.48, 1.52)
	Chlorinated HCS	196	3,493	13 (6.6)	223 (6.4)	1.02 (0.56, 1.84)
	Gasoline EEF	196	3,493	6 (3.1)	114 (3.3)	1.03 (0.44, 2.41)
	Diesel EEF	196	3,493	5 (2.6)	98 (2.8)	1.01 (0.40, 2.54)
Renal tumors	Aliphatic/alicyclic HCS	494	8,927	9 (1.8)	163 (1.8)	0.96 (0.48, 1.90)
	Aromatic HCS	494	8,927	37 (7.5)	620 (6.9)	1.03 (0.72, 1.47)
	Chlorinated HCS	494	8,927	27 (5.5)	536 (6.0)	0.89 (0.60, 1.34)
	Gasoline EEF	494	8,927	16 (3.2)	321 (3.6)	0.97 (0.58, 1.63)
	Diesel EEF	494	8,927	14 (2.8)	283 (3.2)	0.96 (0.55, 1.68)
Hepatic tumors	Aliphatic/alicyclic HCS	119	2,182	10 (8.4)	155 (7.1)	1.06 (0.53, 2.11)
	Aromatic HCS	119	2,182	7 (5.9)	124 (5.7)	1.03 (0.46, 2.28)
	Chlorinated HCS	119	2,182	5 (4.2)	83 (3.8)	1.12 (0.43, 2.87)
	Gasoline EEF	119	2,182	5 (4.2)	73 (3.3)	1.29 (0.50, 3.33)
	Diesel EEF	119	2,182	10 (8.4)	155 (7.1)	1.06 (0.53, 2.11)
Bone tumors	Aliphatic/alicyclic HCS	424	7,428	8 (1.9)	174 (2.3)	0.81 (0.39, 1.68)
	Aromatic HCS	424	7,428	44 (10.4)	665 (9.0)	1.09 (0.79, 1.52)
	Chlorinated HCS	424	7,428	33 (7.8)	551 (7.4)	0.95 (0.65, 1.37)
	Gasoline EEF	424	7,428	8 (1.9)	180 (2.4)	0.83 (0.40, 1.71)
	Diesel EEF	424	7,428	7 (1.7)	154 (2.1)	0.84 (0.39, 1.82)
Soft tissue sarcomas	Aliphatic/alicyclic HCS	492	9,062	10 (2.0)	204 (2.3)	0.89 (0.46, 1.70)
	Aromatic HCS	492	9,062	44 (8.9)	713 (7.9)	1.09 (0.78, 1.51)
	Chlorinated HCS	492	9,062	40 (8.1)	650 (7.2)	1.11 (0.79, 1.55)
	Gasoline EEF	492	9,062	16 (3.3)	301 (3.3)	1.02 (0.61, 1.71)
	Diesel EEF	492	9,062	9 (1.8)	270 (3.0)	0.63 (0.32, 1.24)
Germ cell tumors	Aliphatic/alicyclic HCS	486	8,599	16 (3.3)	188 (2.2)	1.52 (0.89, 2.59)
	Aromatic HCS	486	8,599	50 (10.3)	787 (9.2)	1.12 (0.82, 1.52)
	Chlorinated HCS	486	8,599	41 (8.4)	621 (7.2)	1.17 (0.83, 1.64)
	Gasoline EEF	486	8,599	12 (2.5)	214 (2.5)	1.05 (0.58, 1.91)
	Diesel EEF	486	8,599	9 (1.9)	167 (1.9)	1.02 (0.51, 2.02)
Epithelial tumors	Aliphatic/alicyclic HCS	678	11,402	11 (1.6)	263 (2.3)	0.75 (0.41, 1.39)
	Aromatic HCS	678	11,402	60 (8.8)	1,082 (9.5)	0.91 (0.69, 1.20)
	Chlorinated HCS	678	11,402	55 (8.1)	869 (7.6)	1.06 (0.79, 1.42)
	Gasoline EEF	678	11,402	19 (2.8)	236 (2.1)	1.51 (0.93, 2.44)
	Diesel EEF	678	11,402	13 (1.9)	194 (1.7)	1.27 (0.71, 2.25)

Note: ORs and corresponding 95% CIs estimated using conditional logistic regression models inherently adjusted for matching variables child’s birth year (continuous) and sex (male, female) and further adjusted for child’s region of residence at diagnosis (Stockholm-Gotland, Uppsala-Örebro, West, South, Southeast, North), birth order (first, subsequent birth), maternal country of birth (Sweden, other) and age (continuous), education (, 10–12, , missing), and history of cancer at the child’s birth (yes, no). CI, confidence interval; EEF, engine exhaust fumes; HCS, hydrocarbon solvents; OR, odds ratio.

Includes myelodysplastic and myeloproliferative syndromes.

Table 3

Odds ratios of childhood cancer subtypes associated with paternal occupational exposure to hydrocarbon solvents or engine exhaust fumes around the time of the child’s birth.

Cancer	Paternal occupational exposure	Cases (n)	Controls (n)	Exposed cases [n (%)]	Exposed controls [n (%)]	Adjusted OR (95% CI)
Lymphoid leukemia	Aliphatic/alicyclic HCS	2,346	51,054	228 (9.7)	5,498 (10.8)	0.90 (0.78, 1.04)
	Aromatic HCS	2,346	51,054	397 (16.9)	9,368 (18.3)	0.92 (0.82, 1.03)
	Chlorinated HCS	2,346	51,054	291 (12.4)	6,602 (12.9)	0.97 (0.85, 1.10)
	Gasoline EEF	2,346	51,054	484 (20.6)	10,402 (20.4)	1.02 (0.92, 1.13)
	Diesel EEF	2,346	51,054	323 (13.8)	7,256 (14.2)	0.96 (0.85, 1.08)
Myeloid leukemiaa	Aliphatic/alicyclic HCS	450	9,903	43 (9.6)	1,001 (10.1)	0.90 (0.65, 1.25)
	Aromatic HCS	450	9,903	93 (20.7)	1,781 (18.0)	1.16 (0.91, 1.47)
	Chlorinated HCS	450	9,903	54 (12.0)	1,196 (12.1)	0.96 (0.71, 1.29)
	Gasoline EEF	450	9,903	81 (18.0)	1969 (19.9)	0.87 (0.68, 1.11)
	Diesel EEF	450	9,903	52 (11.6)	1,248 (12.6)	0.89 (0.66, 1.21)
Hodgkin lymphoma	Aliphatic/alicyclic HCS	739	16,257	68 (9.2)	1,671 (10.3)	0.87 (0.67, 1.13)
	Aromatic HCS	739	16,257	137 (18.5)	3,131 (19.3)	0.95 (0.79, 1.16)
	Chlorinated HCS	739	16,257	85 (11.5)	2030 (12.5)	0.89 (0.71, 1.13)
	Gasoline EEF	739	16,257	167 (22.6)	3,195 (19.7)	1.21 (1.01, 1.44)
	Diesel EEF	739	16,257	98 (13.3)	1,831 (11.3)	1.20 (0.96, 1.50)
Non-Hodgkin lymphoma	Aliphatic/alicyclic HCS	239	5,225	26 (10.9)	600 (11.5)	0.93 (0.61, 1.42)
	Aromatic HCS	239	5,225	39 (16.3)	948 (18.1)	0.86 (0.60, 1.23)
	Chlorinated HCS	239	5,225	40 (16.7)	720 (13.8)	1.27 (0.89, 1.82)
	Gasoline EEF	239	5,225	45 (18.8)	1,026 (19.6)	0.95 (0.68, 1.33)
	Diesel EEF	239	5,225	37 (15.5)	758 (14.5)	1.09 (0.75, 1.57)
Ependymoma	Aliphatic/alicyclic HCS	311	6,795	29 (9.3)	734 (10.8)	0.84 (0.56, 1.24)
	Aromatic HCS	311	6,795	59 (19.0)	1,269 (18.7)	1.01 (0.75, 1.37)
	Chlorinated HCS	311	6,795	35 (11.3)	863 (12.7)	0.86 (0.60, 1.24)
	Gasoline EEF	311	6,795	57 (18.3)	1,336 (19.7)	0.90 (0.67, 1.21)
	Diesel EEF	311	6,795	28 (9.0)	887 (13.1)	0.65 (0.43, 0.97)
Astrocytoma and other gliomas	Aliphatic/alicyclic HCS	2,254	49,588	254 (11.3)	5,203 (10.5)	1.09 (0.95, 1.25)
	Aromatic HCS	2,254	49,588	434 (19.3)	9,205 (18.6)	1.05 (0.94, 1.17)
	Chlorinated HCS	2,254	49,588	282 (12.5)	6,229 (12.6)	1.01 (0.88, 1.15)
	Gasoline EEF	2,254	49,588	443 (19.7)	9,850 (19.9)	0.98 (0.88, 1.09)
	Diesel EEF	2,254	49,588	303 (13.4)	6,176 (12.5)	1.08 (0.95, 1.23)
CNS embryonal tumors	Aliphatic/alicyclic HCS	673	14,750	71 (10.5)	1,508 (10.2)	1.05 (0.81, 1.36)
	Aromatic HCS	673	14,750	122 (18.1)	2,723 (18.5)	0.99 (0.81, 1.22)
	Chlorinated HCS	673	14,750	80 (11.9)	1,786 (12.1)	0.99 (0.78, 1.27)
	Gasoline EEF	673	14,750	131 (19.5)	3,000 (20.3)	0.95 (0.78, 1.15)
	Diesel EEF	673	14,750	88 (13.1)	1946 (13.2)	0.98 (0.78, 1.24)
Neuroblastoma	Aliphatic/alicyclic HCS	388	8,389	39 (10.1)	888 (10.6)	0.98 (0.69, 1.38)
	Aromatic HCS	388	8,389	68 (17.5)	1,462 (17.4)	1.05 (0.79, 1.38)
	Chlorinated HCS	388	8,389	49 (12.6)	1,070 (12.8)	1.02 (0.75, 1.40)
	Gasoline EEF	388	8,389	82 (21.1)	1,748 (20.8)	1.02 (0.79, 1.31)
	Diesel EEF	388	8,389	59 (15.2)	1,362 (16.2)	0.91 (0.68, 1.22)
Retinoblastoma	Aliphatic/alicyclic HCS	253	5,547	24 (9.5)	534 (9.6)	1.02 (0.66, 1.59)
	Aromatic HCS	253	5,547	41 (16.2)	988 (17.8)	0.94 (0.66, 1.34)
	Chlorinated HCS	253	5,547	30 (11.9)	641 (11.6)	1.10 (0.74, 1.64)
	Gasoline EEF	253	5,547	52 (20.6)	1,144 (20.6)	1.02 (0.75, 1.40)
	Diesel EEF	253	5,547	41 (16.2)	734 (13.2)	1.27 (0.89, 1.82)
Renal tumors	Aliphatic/alicyclic HCS	611	13,369	66 (10.8)	1,343 (10.0)	1.08 (0.83, 1.41)
	Aromatic HCS	611	13,369	117 (19.1)	2,326 (17.4)	1.13 (0.91, 1.40)
	Chlorinated HCS	611	13,369	79 (12.9)	1,658 (12.4)	1.06 (0.83, 1.35)
	Gasoline EEF	611	13,369	121 (19.8)	2,716 (20.3)	0.97 (0.79, 1.18)
	Diesel EEF	611	13,369	80 (13.1)	1,894 (14.2)	0.90 (0.71, 1.15)
Hepatic tumors	Aliphatic/alicyclic HCS	132	2,818	14 (10.6)	304 (10.8)	1.06 (0.59, 1.90)
	Aromatic HCS	132	2,818	22 (16.7)	515 (18.3)	0.94 (0.58, 1.53)
	Chlorinated HCS	132	2,818	15 (11.4)	344 (12.2)	0.95 (0.54, 1.67)
	Gasoline EEF	132	2,818	30 (22.7)	567 (20.1)	1.15 (0.75, 1.75)
	Diesel EEF	132	2,818	20 (15.2)	414 (14.7)	0.96 (0.58, 1.59)
Bone tumors	Aliphatic/alicyclic HCS	565	12,490	61 (10.8)	1,286 (10.3)	1.06 (0.81, 1.40)
	Aromatic HCS	565	12,490	93 (16.5)	2,306 (18.5)	0.87 (0.69, 1.09)
	Chlorinated HCS	565	12,490	58 (10.3)	1,585 (12.7)	0.77 (0.59, 1.03)
	Gasoline EEF	565	12,490	103 (18.2)	2,449 (19.6)	0.92 (0.74, 1.15)
	Diesel EEF	565	12,490	72 (12.7)	1,470 (11.8)	1.09 (0.84, 1.41)
Soft tissue sarcomas	Aliphatic/alicyclic HCS	608	13,281	63 (10.4)	1,437 (10.8)	0.98 (0.75, 1.29)
	Aromatic HCS	608	13,281	104 (17.1)	2,451 (18.5)	0.94 (0.76, 1.18)
	Chlorinated HCS	608	13,281	76 (12.5)	1,720 (13.0)	0.98 (0.77, 1.26)
	Gasoline EEF	608	13,281	140 (23.0)	2,655 (20.0)	1.22 (1.00, 1.48)
	Diesel EEF	608	13,281	101 (16.6)	1,875 (14.1)	1.22 (0.97, 1.52)
Germ cell tumors	Aliphatic/alicyclic HCS	630	13,849	70 (11.1)	1,519 (11.0)	1.06 (0.82, 1.38)
	Aromatic HCS	630	13,849	114 (18.1)	2,660 (19.2)	0.98 (0.79, 1.21)
	Chlorinated HCS	630	13,849	87 (13.8)	1,776 (12.8)	1.14 (0.90, 1.44)
	Gasoline EEF	630	13,849	121 (19.2)	2,637 (19.0)	1.04 (0.84, 1.27)
	Diesel EEF	630	13,849	76 (12.1)	1,575 (11.4)	1.07 (0.83, 1.37)
Epithelial tumors	Aliphatic/alicyclic HCS	948	20,835	91 (9.6)	2,168 (10.4)	0.94 (0.75, 1.17)
	Aromatic HCS	948	20,835	174 (18.4)	3,953 (19.0)	0.99 (0.84, 1.18)
	Chlorinated HCS	948	20,835	108 (11.4)	2,565 (12.3)	0.93 (0.75, 1.14)
	Gasoline EEF	948	20,835	173 (18.2)	4,085 (19.6)	0.95 (0.80, 1.13)
	Diesel EEF	948	20,835	94 (9.9)	2,275 (10.9)	0.90 (0.72, 1.12)

Note: ORs and corresponding 95% CIs estimated using conditional logistic regression models inherently adjusted for matching variables child’s birth year (continuous) and sex (male, female) and further adjusted for child’s region of residence at diagnosis (Stockholm-Gotland, Uppsala-Örebro, West, South, Southeast, North), birth order (first, subsequent birth, missing), paternal country of birth (Sweden, other) and age (continuous), education (, 10–12, , missing), and history of cancer at the child’s birth (yes, no). CI, confidence interval; EEF, engine exhaust fumes; HCS, hydrocarbon solvents; OR, odds ratio.

Includes myelodysplastic and myeloproliferative syndromes.

ORs of childhood cancers associated with maternal occupational exposures are shown in Table 2 and Figure 2. We found a 64% increased risk of non-Hodgkin lymphoma associated with maternal exposure to aromatic hydrocarbons (; 95% CI: 1.05, 2.58). Numbers were too small to examine risks of non-Hodgkin lymphoma from maternal exposure to benzene or PAH separately (Table S10). Although risks of Hodgkin lymphoma associated with aromatic HCS exposure was small (; 95% CI: 0.87, 1.52), we found markedly increased risks of this cancer in children of mothers exposed to benzene (; 95% CI: 1.14, 3.33), PAH (; 95% CI: 1.07, 3.35), and gasoline (; 95% CI: 1.20, 6.06) (Table S10). The latter association was based on only seven cases. In addition, there was an indication of an association between maternal exposure to aliphatic/alicyclic hydrocarbons and germ cell tumors (; 95% CI: 0.89, 2.59).

Figure 2.

Plots show adjusted odds ratios (ORs) and corresponding 95% confidence intervals (CIs; error bars) of childhood cancer subtypes associated with maternal occupational exposure to hydrocarbon solvents (HCS) or engine exhaust fumes (EEF). ORs were estimated using conditional logistic regression models inherently adjusted for matching variables child’s birth year (continuous) and sex (male, female) and further adjusted for child’s region of residence at diagnosis (Stockholm-Gotland, Uppsala-Örebro, West, South, Southeast, North), birth order (first, subsequent birth), maternal country of birth (Sweden, other) and age (continuous), education (, 10–12, , missing), and history of cancer at the child’s birth (yes, no). Point estimates and confidence intervals are available in Table 2. ORs were not estimated when only less that 5 exposed cases or controls were available for the analysis. Myeloid leukemia includes myelodysplastic and myeloproliferative syndromes. Note: CNS, central nervous system.

We also observed a 40% higher risk of astrocytoma and other gliomas associated with diesel EEF (; 95% CI: 1.04, 1.88) and 29% increased risk owing to gasoline EEF exposure (; 95% CI: 0.98, 1.71) (Table 2 and Figure 2). Maternal occupational exposure to carbon monoxide could not explain these findings (Table S10). In addition to CNS gliomas, exposure to diesel EEF was likely associated with higher risks of myeloid leukemia (; 95% CI: 0.84, 2.81]), Hodgkin lymphoma (; 95% CI: 0.85, 3.02), and non-Hodgkin lymphoma (; 95% CI: 0.71, 2.91), and gasoline EEF with epithelial tumors (; 95% CI: 0.93, 2.44; Table 2).

Data on ORs of childhood cancer associated with paternal occupational exposures are displayed in Table 3 and Figure 3. Exposure to gasoline and diesel EEF was associated with a 1.2-fold higher risk of Hodgkin lymphoma [ (95% CI: 1.01, 1.44) for gasoline EEF and (95% CI: 0.96, 1.50) for diesel EEF], and soft tissue sarcomas [ (95% CI: 1.00, 1.48) for gasoline EEF and (95% CI: 0.97, 1.52) for diesel EEF]. These findings could not be explained by exposure to carbon monoxide as ORs were close to null, except for Hodgkin lymphoma (; 95% CI: 1.00, 1.40) (Table S10).

Figure 3.

Plots show adjusted odds ratios (ORs) and corresponding 95% confidence intervals (CIs; error bars) of childhood cancer subtypes associated with paternal occupational exposure to hydrocarbon solvents (HCS) or engine exhaust fumes (EEF). ORs were estimated using conditional logistic regression models inherently adjusted for matching variables child’s birth year (continuous) and sex (male, female) and further adjusted for child’s region of residence at diagnosis (Stockholm-Gotland, Uppsala-Örebro, West, South, Southeast, North), birth order (first, subsequent birth, missing), paternal country of birth (Sweden, other) and age (continuous), education (, 10–12, , missing), and history of cancer at the child’s birth (yes, no). Point estimates and confidence intervals are available in Table 3. Myeloid leukemia Includes myelodysplastic and myeloproliferative syndromes. Note: CNS, central nervous system.

We did not identify notable differences in ORs of childhood cancer when we compared higher and lower maternal or paternal occupational exposure to HCS or EEF (Figures 4 and 5; Table S11).

Figure 4.

Plots show adjusted odds ratios (ORs) and corresponding 95% confidence intervals (CIs; error bars) of childhood cancer subtypes associated with higher and lower maternal occupational exposure to hydrocarbon solvents (HCS) or engine exhaust fumes (EEF). Occupations with higher exposure had a period specific product of probability and level of exposure greater or equal to the median among all exposed mothers of controls. Occupations with lower exposure had a nonzero probability level below the median. ORs were estimated using conditional logistic regression models inherently adjusted for matching variables child’s birth year (continuous) and sex (male, female) and further adjusted for child’s region of residence at diagnosis (Stockholm-Gotland, Uppsala-Örebro, West, South, Southeast, North), birth order (first, subsequent birth), maternal country of birth (Sweden, other) and age (continuous), education (, 10–12, , missing), and history of cancer at the child’s birth (yes, no). Point estimates and confidence intervals are available in Table S11. ORs were not estimated when only less that 5 exposed cases or controls were available for the analysis. Myeloid leukemia includes myelodysplastic and myeloproliferative syndromes. Note: CNS, central nervous system.

Figure 5.

Plots show adjusted odds ratios (ORs) and corresponding 95% confidence intervals (CIs; error bars) of childhood cancer subtypes associated with higher and lower paternal occupational exposure to hydrocarbon solvents (HCS) or engine exhaust fumes (EEF). Occupations with higher exposure had a period specific product of probability and level of exposure greater or equal to the median among all exposed fathers of controls. Occupations with lower exposure had a nonzero probability level below the median. ORs were estimated using conditional logistic regression models inherently adjusted for matching variables child’s birth year (continuous) and sex (male, female) and further adjusted for child’s region of residence at diagnosis (Stockholm-Gotland, Uppsala-Örebro, West, South, Southeast, North), birth order (first, subsequent birth, missing), paternal country of birth (Sweden, other) and age (continuous), education (, 10–12, , missing), and history of cancer at the child’s birth (yes, no). Point estimates and confidence intervals are available in Table S11. Myeloid leukemia Includes myelodysplastic and myeloproliferative syndromes. Note: CNS, central nervous system.

In sensitivity analyses, associations for maternal exposures were slightly stronger when we restricted to children born close to a census or in 2000 and onward when employment data were available annually and in children of blue-collar mothers (Table S12). Setting a 5% threshold to exposure probabilities and considering only children whose parents’ employment could be ascertained before birth resulted in fewer exposed mothers and attenuated some ORs. Controlling for maternal first trimester smoking or including parents whose employment could not be ascertained as unexposed did not change any of our results from the main analyses. In terms of paternal exposures, the results of the sensitivity analyses were similar to those in the main analyses (Table S13).

Discussion

This large register-based case–control study, which included all childhood cancer cases diagnosed in Sweden since 1960, suggested a small increased risk of cancer associated with maternal and paternal exposure to gasoline and diesel EEFs around the time of a child’s birth. Specifically, analyses by cancer subtype revealed increased risks of astrocytoma and other gliomas, myeloid leukemia, Hodgkin and non-Hodgkin lymphomas, and epithelial tumors in children of mothers exposed to EEF. Some of these associations were based on a small number of cases, leading to wide CIs. Paternal occupational exposure to EEF was associated with Hodgkin lymphomas and soft tissue sarcomas. Paternal occupational exposure to HCS was not associated with any increased risk of childhood cancer, whereas maternal exposure to aromatic HCS was associated with higher risks of non-Hodgkin lymphomas and aliphatic/alicyclic HCS with germ cell tumors in the offspring of exposed mothers. Of note, there was no indication of an exposure–response relation between any of the examined exposures and childhood cancer risk, likely due to small numbers in most analyses.

This is the first study to our knowledge to report an increased risk of lymphoma, particularly non-Hodgkin lymphomas in children of mothers occupationally exposed to aromatic HCS. ORs did not vary by the degree of exposure (higher/lower). In line with a small Swiss study,[7] neither benzene nor PAH, two potent aromatic hydrocarbon carcinogens in the index individual,[37] appeared to be responsible for the observed association for non-Hodgkin lymphomas in the offspring. On the other hand, benzene and PAH were strongly associated with a higher risk of Hodgkin lymphoma. For other cancer types, including leukemia and CNS tumors, findings from previous studies are conflicting, with some suggesting considerable risks for these cancers,[5,6,12,13,38] whereas others did not.[7-11] In line with the latter, we found no indication of higher risks of leukemia or CNS tumors or their subtypes owing to HCS exposure in either mothers or fathers, suggesting that parental exposure to these agents plays a limited role in the occurrence of these cancers.

Few studies examined risks of childhood cancer related to maternal occupational exposure to EEF around birth, and their findings were largely inconsistent. Our observation of a moderately increased risk of astrocytoma and other gliomas was supported by a study from Denmark that used employment data from a register and a JEM in which authors found a 50% increased risk of astrocytoma in mothers exposed to diesel EEF.[15] These results were not replicated, however, in a pooled analysis of several European case–control studies with self-reported information on parental occupation, which found no association between maternal exposure to diesel EEF and CNS tumors.[16]

Our data also suggested increased risks for Hodgkin and non-Hodgkin lymphoma in children of mothers occupationally exposed to gasoline/diesel EEF and Hodgkin lymphoma in the offspring of fathers exposed to EEF. Although few data are available for maternal exposures to compare our findings regarding higher risks of lymphoma and epithelial tumors, higher risks for lymphoma in children of exposed fathers was replicated in a smaller case–control study in which occupation was self-reported.[11] We also observed higher risks of soft tissue sarcoma in children of fathers occupationally exposed to EEF. In support to our findings, Kendall et al. have recently reported higher risks of rhabdomyosarcoma (and marginally higher for total tissue sarcomas) in children of fathers exposed to EEF in the United Kingdom.[39]

In terms of parental EEF exposure and leukemias, we found some modestly increased risks of lymphoid and especially myeloid leukemias associated with both gasoline and diesel EEFs, although numbers were too small to conclude with certainty. Paternal exposure to EEFs was not associated with leukemias even in fathers with higher exposure. Indications of higher risks of leukemia or subtypes from maternal or parental exposure to EEFs were found in some previous studies,[11,17,19] whereas others reported no such associations.[5,15]

In our study, most mothers and fathers exposed to EEFs worked as machine mechanics, engineering technicians, chemical/industrial engineers, and transportation clerks in approximately similar proportions between cancer cases and controls. It is largely unknown through which mechanisms carcinogenic substances included in EEFs, such as aromatic HCS including PAH, gasoline, and others, might cause cancer in the offspring of exposed parents.[3] Extrapolating results of animal studies,[40-42] it is believed that parental exposure to these cancerogenic agents, especially through inhalation and uptake from the lungs, might cause mutations also in the germ cells, including DNA damage and chromosomal aberrations resulting in tumorigenesis in the affected child. Passage via the placenta to the fetus and early exposure of the newborn are also likely mechanisms for carcinogenesis in the offspring.[3,15]

A limitation of our study was the potential misclassification of the exposures to which several factors may have contributed. True exposure likely differed across specific occupations or was influenced by individual tasks or modified by protective equipment. In addition, parental leave of absence could not be evaluated in our data. Similarly, and although limited,[43] we could not account for changes in parental occupation during the period when information was obtained from censuses. As a result, some parents may have been incorrectly classified as exposed, or less likely unexposed, resulting in an underestimation of some true effects as these factors were not influenced by a child’s cancer status.

We also could not ensure that the timing of the occupational exposure matched the period around pregnancy for mothers and the preconception period for fathers because employment data from censuses were available at best every 5 y during the period 1960–1996. Further, a slightly larger proportion of cases than controls were excluded from the analyses, an exclusion that may lead to selection bias, assuming this exclusion was associated with any of the parental exposures studied here. Nevertheless, a sensitivity analysis that included parents whose employment could not be ascertained as unexposed yielded results similar to those of the main analysis.[33] Last, an almost perfect correlation between gasoline and diesel EEF prevented us from separating risks between the two exposures or their components, except for gasoline and carbon monoxide for which no increased ORs were observed.

This investigation benefited greatly from the use of Swedish registers and SWEJEM. Using these sources, we were able to assess several exposures in both mothers and fathers ensuring that a cancer diagnosis in the child did not influence reporting of parental occupation or selection into the study. Further, we could account for changes in exposure probability and level during the long study period and adjust for several confounders. Our findings are generalizable to other contexts where occupations and exposure patterns are comparable to the Swedish context.

To conclude, findings from this large register-based study suggested that occupational exposure to hydrocarbon solvents and engine exhaust fumes of the mother and father around their child’s birth was associated with cancer occurrence in the offspring. Maternal exposure to aromatic hydrocarbons was associated with higher risks of non-Hodgkin lymphoma and alicyclic/aliphatic hydrocarbon solvents with germ cell tumors. Occupational exposure of the mother to diesel or gasoline engine exhaust fumes including PAH and benzene was associated with higher risks of gliomas, lymphomas, myeloid leukemias, and epithelial tumors. Some of these associations, particularly for rarer cancers, were based on a few exposed cases, leading to higher uncertainty. Paternal occupational exposure to hydrocarbon solvents and engine exhaust fumes was not associated with higher risks of childhood cancers except for Hodgkin lymphoma and soft tissue sarcomas owing to EEF exposure. Although the moderately increased relative risks from maternal and paternal exposure add to the growing body of literature on the harms of solvents and engine exhaust fumes, more data are needed to confirm these findings.

Click here for additional data file.

Click here for additional data file.