Epidemiology of occupational and environmental risk related to ovarian cancer

Epidemiology of occupational and environmental risk factors related to ovarian cancer. Scand J Work Environ Health 1998;24(3):175-82. This paper reviews articles published during 1970-1997 from 48 epidemiologic studies on occupational and environmental risk factors of ovarian cancer. Culrent evidence is characterized by poorly focused data for occupational and environmental agents, vulnerability to biases, and an almost complete lack of quantitative exposure-response data. The moderate amount of data on nurses, teachers, professionals, dry cleaning employees, women in agriculture, the pharmaceutical industry, pharmacists, waitresses, and cooks show very little, if any, evidence of excess risk. Hairdressers, beauticians, and women employed in the printing industry may be at increased risk, but the data are insufficient for strong conclusions. Some case-referent studies suggest a modest-to-moderate excess in association with genital talc application. Few high-quality studies have been carried out, and no chemical agents have been studied extensively, with the exception of exposure to talc. Ovarian cancer may have occupational and environmental etiologies intertwined with cultural, behavioral, and life-style factors and genetic susceptibility, but current knowledge is insufficient to quantify occupational and environmental etiologies reliably. Well-designed analytic epidemiologic studies with sufficient power are needed.

helpful for clues because of inconsistencies. For example, in Brazil, Italy, and Colombia, higher social strata exhibit excess mortality or incidence of ovarian cancer, while northern European data suggest no consistent trends in social class (10).
There is a strong familial component to ovarian cancer (43,44). Ovarian cancers in some breast cancer families are due to a breast cancer susceptibility locus on chromosome 17 (45). The breast cancer susceptibility gene, BRCA1, may act as an inhibitor of cell proliferation. Locations of other genes for ovarian cancer susceptibility have been suggested (46)(47)(48). Thus the ovarian cancer risk due to a second major breast cancer susceptibility gene (BRCA2) on chromosome 13q is elevated over that of the general population, although less so than for BRCA1. BRCAl and BRCA2 together would explain about 6% of the ovarian cancers of northern European populations (Goldgar D, personal communication). Loss of heterozygosity on chromosome 13q occurs in 50-60% of sporadic ovarian cancers (49). However, no clearly diseasecausing somatic mutations have been observed for sporadic ovarian cancers.
The epidemiologic literature contains data on the risk of ovarian cancer for some 40 job titles. This review summarizes epidemiologic results published during the past 28 years on occupational and environmental risk factors.

Material and methods
Reports and articles published in 1970-1997 that provided data on the occupational and environmental etiology of ovarian cancer were identified by MEDLINE searches, from reference lists of papers on ovarian cancer published in major scientific journals, from the perusal of journals, and from an inspection of publications of the International Agency for Research on Cancer. When the ecological studies are excluded, a total of 48 studies were reviewed. There were 12 population-or hospital-based case-referent studies, 15 occupational cohort studies, 11 record-linkage studies, and 10 proportionate mortality studies. A total of 20 studies reported mortality data, while 28 studies were based on incident cases.
There were recurrent quality problems in the studies. None of the proportionate mortality, record-linkage, or occupational cohort studies controlled for potential confounding by reproductive determinants of ovarian cancer. Another feature was the almost exclusive use of job titles and industries that could lead to misclassification because of heterogeneity in exposure to chemical and physical agents. Even the validity of classifying subjects into job and industry categories, particularly in recordlinkage and proportionate mortality studies, can often be questioned, since the data usually derive from cross-sectional censuses or from death records. The probability of misclassification of job or industry is increased since the data sources usually provide poor surrogates for lifelong or usual job or industry.
Occupational hazards for ovarian cancer Nurses. Data on ovarian cancer risk for nurses were available in 1 occupational cohort study with a small number of cases (50), 2 record-linkage studies (5 1,52), and 4 proportionate mortality studies (18,(53)(54)(55) (table 1). The cohort study found a nonsignificant excess for Icelandic nurses with over 20 years of employment. The Finnish linkage study (51,52) found no excess for nurses, even after adjustment for social status. The Italian linkage and proportionate studies (52) had too few cases. No excess was found in the proportionate mortality studies from the United Kingdom (UK) (54) or the United States (US) (55). The UK study (54) reported an excess for physiotherapists. A proportionate mortality study in British Columbia (18,53) reported significantly elevated risks for breast or ovarian cancer among teachers, nurses, office clerks, and sales clerks. When homemakers were excluded from the reference population, however, most of the excesses for breast and ovarian cancers in the 4 occupational groups disappeared. The results therefore indicate confounding by reproductive factors.

Teachers.
Only proportionate mortality studies were available for teachers. No unusual pattern of risk was found for teachers in 3 European (51, 52) studies and 1 Canadian (53) study. The latter reported a 70% excess, but the estimate was unstable. In the UK study (54), a significant 50% excess was reported.
Professionals. Professional occupations is a large category and is likely to be a heterogeneous entity in terms of work, exposures, and life-style. A Chinese case-referent study (17) reported a 40% nonsignificant excess for professionals and scientists. Another Chinese study (56) found an increased risk of ovarian cancer for professionals and scientists [standardized incidence ratio (SIR) 1.3, P<0.05], but occupational information for the cases and the comparison population came from different sources and therefore the result may have been biased in an unknown direction. Proportionate mortality data from the United Kingdom (54, 57) suggested elevated risks for Table 1. Summary of studies of occupation and ovarian cancer. (C-R = case-referent study, 95% CI = 95% confidence interval, COH = occupational cohort study, IR = incidence ratio, MOR = mortality odds ratio, NEC = not elsewhere classified, NS = not significant (P>0.05), OPCS = Office of Population Censuses and Surveys, OR = odds ratio, PCMR = proportionate cancer mortality ratio, PMR = proportionate mortality ratio and proportionate mortality study, PRR = proportionate registration ratio, R-L = record-linkage study, SIR = standardized incidence ratio, SMR = standardized mortality ratio) Rubber manufacturing. The only source of data for rubber manufacturing was the Finnish census-cancer incidence linkage study (51). The risk of ovarian cancer for employees in the rubber industry, adjusted for social class, was elevated.
Telephone industry. The results of 1 proportionate mortality study in the telephone industry (74), conducted in 24 US states, were available. Excesses were found for both white and black women, and for engineers, technicians, office workers, telephone operators, mechanics, and repairers.

Pharmaceutical industry, pharmacists, chemists.
Three occupational cohort studies were available for the pharmaceutical industry, pharmacists and chemists, all from Nordic countries. A 40% excess in ovarian cancer risk was reported for employees in a Danish pha~maceutical company (73, while no excess was found for Danish pharmaceutical technicians (76) or for employees in a Swedish pharmaceutical company (77). The Finnish record-linkage study (51) found no excess for pharmacists. On the other hand, a community-based case-referent study in China (21) found an excess for women in the pharmaceutical industry, and a US proportionate mortality study (78) found an excess for chemists.

Miscellaneous industrial workers. No clear excess
ovarian cancer risk was found in an occupational cohort study of manual workers in different industries in Iceland in 1970-1986 (79). In an Italian proportionate mortality study (52), significantly increased risks were found for metal workers and clothing workers. The UK proportionate mortality study (53) found low risks for sewers and embroiderers, spinners and winders, machine tool operators, and packers and sorters.
Printing industry. Positive associations between ovarian cancer and work in the printing industry have been reported in 1 large record-linkage study from Finland (5 1) and a UK proportionate mortality study (57). In the linkage study (51), significantly positive associations were found for the printing industry and for a number of specific occupational categories within the printing industry. Similar results have been obtained in Italy (52), but, as there were only 3 cases, the estimate was highly unstable. No associations were found for women working in the Swedish graphics industry (81), and there were very few cases in the Danish study (82).

Environmental and occupational agents
Solvents and related compounds. Exposures to organic solvents are found across a wide spectrum of industries and processes, such as chemical manufacturing, the rubber industry, rayon viscose production, varnishing, insulation, printing, painting and paint manufacturing, the metal industry, technical services, other services and cleaning industries, including dry cleaning. In a US casereferent study (83) of ovarian cancer, years of occupational exposure to unspecified solvents was not associated with risk of ovarian cancer. A Chinese case-referent study (17)  not find an elevated risk for women exposed to folmaldehyde. A Finnish cohort study of workers occupationally exposed to halogenated solvents (85) found no excess; however the number of exposed cases was small. Chlorinated naphthalenes were key agents used in a US cable manufacturing plant during World War 11. A mortality follow-up of employees at this plant (86) failed to find any evidence for an excess risk of ovarian cancer.
Pesticides. An association between risk of ovarian neoplasms and triazine herbicides has been reported in an Italian case-referent study (87). This study may have suffered from the usual problem of confounding from other pesticidal agents used by exposed cases.
Asbestos and talc. In 1960, Keal(88) reported that women exposed to asbestos appeared to have a higher frequency of ovarian neoplasms than those without such exposure. Newhouse et a1 (89) found excess deaths from ovarian cancer among women employed at an asbestos factory. This study was based on 9 cases.
Women can be exposed to talc through occupational contact, in mining, talc milling, or the talc product manufacturing industry, but, more commonly, in daily activities through the use of cosmetic talc, baby powder, facial powder, sanitary napkins, and contraceptive devices, the application of talc powder in the perineal area, and talc powder in towels. Table 2. Epidemiologic data on chemical agents. (95% CI = 95% confidence interval, C-R = case-referent study, COH = occupational cohort study, R-L = record-linkage study, SIR = standardized incidence ratio, SMR = standardized mortality ratio)  (98,95,90) and the United Kingdom (90). Each found a 50-90% excess. The results of the smaller studies were more heterogeneous, the point estimates of the odds ratios exceeding unity in each.

Discussion and concluding remarks
Considerable international variation in the incidence rates for ovdan cancer, together with results from migrant studies, suggest that either life-style or environmental factors or both may affect risk. The published data are not extensive and suffer from various limitations. Much of the evidence on occupational and environmental risk factors is derived from record-linkage and proportionate mortality studies. Some of the studies had jab titles reported crosssectionally. Data on potential confounders such as reproductive history, use of oral contraceptives, hormone replacement therapy, and dlugs that induce ovulation were missing in most of the occupational cohort, record-linkage, and proportionate studies. A further limitation of the published studies was the possibility of reporting and publication bias. Given the relative rarity of ovarian cancer, it is possible that increased risks found by chance only have been preferentially reported and published in both cohort and case-referent studies, This problem is particularly relevant to occupations, industries, and agents for which results from only a few studies are available.
In conclusion, the evidence for risks of ovarian cancer in association with job titles, industries, and occupational and environmental agents is not convincing because of the lack of high-quality data. The aggregated evidence for nurses, teachers, professionals, dry cleaners, farmers and farm workers, employees in the pharmaceutical industry and pharmacists, waitresses, and cooks is negative or inconsistent. Hairdressers, beauticians, and women employed in the printing industry may have an increased risk, but the data are insufficient. For the rest of the occupations and industries studied, not enough studies exist on which to base conclusions. There are very few studies on chemical agents in the occupational or general environment, and the evidence is grossly insufficient. The only exception is genital talc application, with 9 interviewbased case-referent studies that provided rather consistent suggestions for a modest-to-moderate excess.
Elucidation of occupational and environmental risk factors of ovarian cancer calls for well-focused, high-quality epidemiologic studies with sufficient study size and valid individual data on histories of exposures to occupational, environmental and life-style factors.