An Ecological Study of Organochlorine Pesticides and Breast Cancer in Rural Victoria, Australia
- First Online:
- Cite this article as:
- Khanjani, N., English, D.R. & Sim, M.R. Arch Environ Contam Toxicol (2006) 50: 452. doi:10.1007/s00244-004-7217-5
- 112 Views
A number of studies have suggested that environmental contamination with organochlorine pesticides may be related to risk of breast cancer. To investigate this association in a rural part of Australia, organochlorine contamination data from a breast milk organochlorine study conducted in the state of Victoria in 1993 were used. The state was divided into 11 statistical divisions. Standardized incidence ratios (SIRs) for the 11 regions were calculated using breast cancer incidence data from 1983 to 2002. During that time, 47,250 breast cancer cases occurred in Victoria, which had an average population of 2,147,409 women. The Ovens-Murray region, which was the region most contaminated with organochlorine pesticides, showed an elevated SIR of 1.10 (95%CI, 1.03–1.17), although two other regions with lower organochlorine contamination levels also had elevated SIRs. The rural part of the Ovens-Murray region, where the main pesticide use occurred, had the highest SIR, 1.15 (95%CI, 1.07–1.23). We did not find any significant correlation between organochlorine contamination and the age-standardized rate of breast cancer across all regions. But a positive dose-response relationship using an adjusted negative binomial model was detected for heptachlor epoxide. Our study may provide limited support for the role of environmental contamination with organochlorine pesticides in the development of breast cancer.
Pesticides are an integral and important part of agriculture. The use of pesticides has greatly increased world food security and standards of living (Radcliffe 2002). The most widely known is DDT. This insecticide had signal success in the control of agricultural pests and saved millions of human lives by controlling vector-borne diseases of man (Metcalf 1973). However, despite the widely recognized benefits, there has been continuing public concern worldwide about possible adverse effects of pesticides, especially the organochlorine group, on human health, including breast cancer (Radcliffe 2002). High average body burdens are associated with living on farms, being non-white, male, older, or below the poverty line (Mitra et al.2004) and non-vegetarian (Mathur et al.2002). In women, lactation is the single most effective pathway of organochlorine excretion, thereby providing a protective effect due to elimination (Mitra et al.2004).
Organochlorine pesticides used in Victoria and date of end of registration
End of registration
March 27, 1991
December 20, 1989
March 26, 1991
June 29, 1995
Atrazine, Simazine, Triazine, Cyanazine
Breast cancer is the second most frequent cancer in the world (Parkin 2001) and the commonest cause of cancer death in women worldwide (Key et al.2001). Rates vary about fivefold around the world, but are increasing in regions that until recently had low rates of the disease (Key et al.2001). Many of the established risk factors for breast cancer are linked to estrogen. Risk is increased by early menarche and late menopause. Childbearing reduces risk, with greater protection for early birth and a larger number of births. Breast feeding may have a protective effect (Key et al.2001).
Evidence that organochlorines increase the risk of breast cancer is mainly derived from case-control studies and is inconsistent. Some recent studies have reported higher concentrations of certain organochlorine pesticides in the blood or adipose tissue of breast cancer patients than in controls (Charlier et al.2003; Hoyer et al.2000, 2001; Romieu et al.2000); but other studies have reported similar levels in cases and controls (Laden et al.2001; Wolff et al.2000). The rationale for examining these exposures in relation to risk of breast cancer involves their persistence in the environment and animal tissues, with half lives of many years, carcinogenicity in animals (Buranatrevedh and Roy 2001; Wolff et al.1996), estrogenic activity and additive synergic effects (Payne et al.2001), and the ability to depress the number of natural killer cells, which are crucial in immunological defense against early stages of cancer (Mitra et al.2004).
The aim of this study was to investigate associations between breast cancer incidence and exposure to organochlorines using exposure data from a previous cross-sectional study in which organochlorine concentrations were measured in breast milk samples from mothers of different areas in the state of Victoria, Australia (Sim 1993).
Rationale for the Study
Studies about organochlorine pesticides and breast cancer have been conducted in different countries and many suffer from several methodological problems, such as not enrolling subjects from areas of high pesticide use. However, no such study has been done in Australia.
Materials and Methods
An ecological study was conducted to examine the association between the amount of organochlorine contamination and the incidence of female breast cancer in different parts of Victoria, Australia. The study protocol was approved by the Human Research Ethics Committees at The Cancer Council Victoria and Monash University.
For monitoring chemical exposure, the use of biological exposure indices has several advantages over environmental sampling. Biological exposure indices provide a measure of the internal dose rather than the exposed dose; they account for all routes of absorption and they allow the inclusion of a variety of sources of exposure (Sim and McNeil 1992). Breast milk fat concentrations of organochlorines are highly correlated with concentrations in fat stores within the body (Kanja et al. 1992; Skaare et al.1988). Therefore, human milk is a good indicator for monitoring human beings and their environment for organochlorine contamination (Skaare et al.1988).
In the Victorian Breast Milk Study done in 1993, a representative sample of 797 lactating mothers was recruited from different parts of the state (Sim et al.1998). This was done by using a two-stage stratified random selection process. Subjects were recruited from 850 randomly selected maternal and child health centers throughout Melbourne and country Victoria. These centers have a wide geographical spread, are community based, and are used by about 96% of women who give birth in the state (Sim 1993). Mothers who had lived in Victoria since at least 1987 and who were fully breastfeeding their first baby, who was between six and 12 weeks old at the time of data collection, were eligible to participate. The samples were collected from 1991 to 1993. There may have been a small decline in the measurements over time, but this decrease would have happened in all parts of the state regions as these chemicals were registered for use and banned at the same time all over the state.
The number of samples taken from each statistical division and the female population of each region based on the Victorian 1991 census
Number of samples
Ovens and Murray
Each woman donated a 75-ml breast milk sample and the concentrations of the chemicals DDE, DDD, DDT, dieldrin, heptachlor epoxide, HCB (hexachlorobenzene), oxychlordane, aldrin, and lindane were measured in the samples. DDE and DDD are metabolites of DDT. Oxychlordane, heptachlor epoxide, and dieldrin are, respectively, metabolites of chlordane, heptachlor, and aldrin. However, dieldrin has also been used as a separate pesticide (Ahlborg et al.1995). Chemicals with shorter half lives like endosulfan were not measured. But, even if it were, its short half life (about 6 months) means that current levels cannot predict past exposure. From these data, median concentrations of each organochlorine chemical in breast milk were calculated for the 11 statistical divisions. As the data were skewed, the median was used to represent the contamination level of each region (Sim 1993).
Different regions of Victoria and their contamination level with organochlorine pesticides measured in mg/kg of maternal milk
Total detectable samples
The organochlorine insecticides DDT, dieldrin, and aldrin were widely used in the Ovens and King Valley particularly on tobacco farms. Organochlorine use commenced prior to the 1950s (Sim 1993). Aldrin was used to control wire worm/black beetle and dieldrin to control grasshoppers (Environment Australia 1997). There were approximately 2000 hectares of tobacco grown in this region and more than 3000 hectares in the state. Due to concerns about the persistence of these chemicals in the environment and their bioaccumulation in grazing cattle, their use in the Ovens and King Valley was reduced during the 1970s. Concerns about the impact of contamination were so high between 1987 and 1991 that 35 cattle herds grazing on ex-tobacco properties were quarantined because of concerns over the levels of organochlorine residues in export cattle (Environment Australia 1997; Sim 1993).
Data on occurrence of female breast cancer was obtained from the Victorian Cancer Registry, which is a population-based registry of all cancers diagnosed in Victoria. Place of residence of patients was documented by the postcode of residence at the time of diagnosis. De-identified, aggregated data on breast cancer occurrence were obtained for the years 1983 to 2002, stratified by age at diagnosis (five-year groups), period of diagnosis (four-year groups), and postcode. Postcodes were matched to the 11 statistical divisions by using a postcode map. In the few cases where postcodes straddled the borders of two or more statistical divisions, they were allocated to the division with the biggest area.
Estimate of the female population in the state of Victoria, stratified by statistical division and age, were obtained from censuses conducted by the Australian Bureau of Statistics (ABS) in 1986, 1991, 1996, and 2001.
People in Victoria were exposed to organochlorine pesticides from the 1950s and usage of these chemicals continued until the 1990s (Table 1). The exposure measurements done in 1991–1993 reflect body burdens from exposures over the previous years, as these chemicals have long half lives. The breast cancer incidence data that we used were from 1983 to 2002, which allows sufficient time for these chemicals to accumulate in human tissues and show their possible carcinogenic effects. Thus there is a suitable window of time from first exposure to the occurrence of breast cancer. Also, 1983 was the first date of complete, population-based cancer registration data in Victoria.
Standardized incidence ratios (SIRs) were calculated for each statistical division and the Ovens and King Valley separately by dividing the number of observed cancers in each region by the expected number for the same region, with stratification by age (five-year groups) and year of diagnosis (four-year groups). The Melbourne age- and period-specific rates of female breast cancer were used to calculate expected numbers of female breast cancer in each region. As the aim of the study was to evaluate the possible effects of pesticides used in agriculture and these chemicals are mainly used in rural areas, the urban population of Melbourne is a suitable comparison group. Population estimates between censuses were calculated by linear interpolation.
For each chemical, scatter plots of breast cancer SIRs and median organochlorine breast milk concentrations were drawn and the correlation was estimated.
The dose-response relationship between the incidence rates of breast cancer in the 11 regions and regions’ median contamination levels was modelled using negative binomial regression. Each pesticide was analysed separately. To control for possible confounding, the adjusted model included 10-year age groups, 4-year diagnosis periods, and regions’ median family income (as an indicator of socio-economic status). Cases diagnosed before the age of 30 were not included in the model, because the incidence was very low. The regression coefficients and their 95% upper and lower limits were divided by 100, changing the unit of measurement to 0.01 mg/kg lipid contamination in breast milk, which is more sensible, and then exponentiated to estimate the relative increase in the incidence rate (i.e., the incidence rate ratio or IRR) for an increment of 0.01 mg/kg lipid contamination in breast milk.
Organochlorine pesticides have been shown to have additive or synergistic estrogenic effects. Although it is interesting to determine the magnitude of the organochlorine load of each region by using concepts such as estrogen equivalents and to model these equivalents against the incidence rates, we did not think it was possible due to several reasons. These potency measures vary among different estrogenicity assays and hormone receptor specificity is also unclear for some chemicals. Also non-additive and complicated interactions have been reported for different combinations of these chemicals (Borgert et al.2003; Safe 1998). Thus, there is no validated way of combining the organochlorine load taking into account all the organochlorines tested.
Standard incidence ratios of female breast cancer for the 11 statistical divisions of Victoria
Breast Cancer Cases
Ovens and King Valleya
Incidence rate ratio of breast cancer for each organochlorine pesticide, crude and adjusted for 10-year age groups, 4-year diagnosis periods, and categorized median family income of each region
Crude rate ratioa (95% confidence interval)
Adjusted rate ratio (95% confidence interval)
1.07 (0.94, 1.22)
1.06 (1.02, 1.11)*
0.97 (0.85, 1.12)
0.96 (0.91, 1.00)
1.04 (0.73, 1.48)
1.03 (0.91, 1.15)
1.00 (0.99, 1.01)
1.00 (0.99, 1.00)
0.96 (0.88, 1.05)
0.96 (0.93, 0.99)*
1.02 (0.94, 1.11)
1.01 (0.98, 1.04)
The findings of this study show that three statistical divisions within Victoria had higher rates of female breast cancer than expected when compared with rates in the metropolitan area of Melbourne. The division with the highest contamination of organochlorine pesticides in the state had one of the highest SIRs among all 11 statistical divisions, and the rural part of this division, where most pesticide use occurred, had an even higher SIR. However, we found no evidence of a dose-response relationship between level of contamination and incidence rate of breast cancer, except for heptachlor epoxide.
The ecological nature of the study means that the exposure level of each individual was not measured and we do not know whether the women who developed breast cancer were really those who had higher exposure. However, there is no reason to believe that the mothers who donated breast milk samples were different from the general female population in regard to exposure. Breast milk monitoring is a non-invasive way of measuring organochlorine contamination in human populations. Non-differential misclassification of contamination levels is more likely to have biased our results towards the null.
The place of residence of the breast cancer patients at the time of diagnosis was used to estimate exposure, which may not be representative of their exposure during the etiologically relevant period in a mobile population such as in Victoria. However, we did not have relative population movement data for adjusting.
We were able to classify people into rural and urban locations, but we had no information on possible confounding factors such as the age at menarche, menopause, and birth of first child, lactation or ethnic origin, which can play an important role in the development of breast cancer.
Our exposure measure was based on the work of Sim and colleagues in 1993; (Sim 1993; Sim et al.1998). This study included 815 women (Table 2). There is a possibility of imprecision due to the small sample size. However, we believe that measuring contamination in human samples is more accurate and relevant than other methods such as measuring contamination in the environment, measuring pesticide sales in the past, or self-report (Sim and McNeil 1992). We considered that the data from the Victorian breast milk study was the best way we had for measuring human contamination with organochlorine pesticides in the state and comparing urban and rural areas.
In another ecological study on this issue, the state of Kentucky in the United States was divided into counties of low, medium, and high exposure for the triazine herbicide. Exposure to triazines was estimated by use of water contamination data, corn crop production, and pesticide use data. The incidence of breast cancer was compared between these regions. The study suggested a moderate relationship between exposure to triazine herbicides and increased breast cancer risk (Kettles et al.1997). However, another study 3 years later, which was an expansion of the original study, found no association between atrazine and breast cancer across all exposure indices, both by county and by area development districts (Hopenhayn-Rich et al.2002).
In a study done by Duell et al.(2000), residence or work on farms was associated with a reduced risk of breast cancer. Their results also suggested a possible increased risk of breast cancer among a subgroup of farming women who were most likely to be exposed to pesticides (Duell et al.2000).
In Duell’s study, the women who lived or worked on farms had a healthier risk profile in terms of traditional breast cancer risk factors such as a later age at menarche, earlier age at first birth, higher mean number of live births, lower nulliparity, and lower socioeconomic status. Farming women also had more physical activity and were less likely to smoke cigarettes or drink alcohol than non-farming women (Duell et al.2000). Similar to Duell et al.’s study, we found that the incidence of breast cancer in rural Victoria was significantly less than the urban population in 4 regions and non-significantly less in 2 other regions. However, the incidence was significantly higher than the urban region in 3 rural regions and one of them was a region with a history of agriculture and extensive pesticide usage.
In Israel for at least 10 years until 1978, milk and dairy products were contaminated by very high levels of three pesticides: α-BHC (benzene hexachloride), γ-BHC (lindane), and DDT (Westin and Richter 1990). It has been argued that the marked decrease in breast cancer mortality rates in women under 65 years in Israel in the late 1980s may be a result of the pesticide ban (Westin and Richter 1990).
Allen et al.(1997) have suggested that the increase in the breast cancer incidence rate in Hawaii over the past few decades may be related to agricultural chemicals. Agricultural pesticides, like chlordane, heptachlor, and 1,2-dibromo-3-chloropropane (DBCP), have been intensively used in Hawaii’s island ecosystem over the past 40 years leaching into ground water and leading to unusually widespread occupational and general population exposures. Chlordane, heptachlor, and DBCP had been measured at levels that sometimes exceeded federal standards by several orders of magnitude (Allen et al.1997).
The results of the methodologically stronger case-control and nested case-control studies are conflicting. The latest meta-analysis of 22 reports relating DDE (the main metabolite of DDT) to risk of breast cancer resulted in an odds ratio of 0.97 (95% CI, 0.87–1.09) and does not support the hypothesis of a causal relationship (Lopez-Cervantes et al.2004). Although the recent meta-analysis shows no relationship, this meta-analysis was done for only one chemical, DDE, a metabolite from the main pesticide DDT. A meta-analysis for other chemicals has not been published yet. In fact, up to the time of writing there were only two case-control studies published on heptachlor and breast cancer risk (Dello Iacovo et al.1999; Mathur et al.2002), and both show positive associations. Therefore, we believe our report provides valuable new information especially in relation to organochlorines other than DDT.
Some authors have hypothesized that exposure to xenoestrogens such as DDT, in particular during critical periods of human growth and development (e.g., during the development of breast tissue in adolescence), may have irreversible health effects such as increasing the risk of future breast cancer (Ardies and Dees 1998; Buranatrevedh and Roy 2001; Lopez-Cervantes et al.2004). Chronic, long-term exposure to pesticides that disrupt endocrine function after maturity may also present a health risk (Buranatrevedh and Roy 2001). These hypotheses can be areas for future research.
The results of this study may provide limited evidence of an association between breast cancer and the organochlorine pesticides for which we had contamination data, in particular cyclodienes. Although the use of organochlorine chemicals has been limited in many countries mainly due to the belief that they are carcinogens, their carcinogenicity in humans is still uncertain. Further study is needed to clarify the relation between breast cancer and organochlorine pesticides, especially for those chemicals that are still currently used. Developing better ways of measuring chronic exposure to organochlorine pesticides, especially during critical periods of human development and in specific genetic and population subtypes, is essential.
The authors thank Dr. Rory Wolfe and Dr. James Cui for statistical advice. N. Khanjani also acknowledges the Iranian Ministry of Health and Medical Education, which sponsored her to undertake this research.