The World Health Organization (WHO) seeks to strike a balance between an adequate protection of public health and the control of environmental contaminants, among them polychlorinated biphenyls (PCBs). PCBs are organochlorines with different degrees and positions of chlorination, which determine their persistence and toxicity. They are lipophilic and difficult to metabolize and any environmental exposure of living organisms to them results in their accumulation and persistence in fat tissues; in humans this is biomagnified via food (Angulo et al., 1999). The toxic action of these compounds, possibly due to enzymatic induction, manifests itself in the form of dermal, hepatic, gastroenteric, nervous and immunological disorders. In addition, they are powerful endocrine disrupters (Colman and Dawson, 1996) and cancer inducers (Carrasco, 1995). The contamination by and accumulation of individual PCB congeners in the adipose tissue of women's breasts reflects the degree of exposure to which they are subjected. This exposure was identified as a possible risk factor for cancer of the breast (Carrasco, 1995;Angulo et al., 1999).
The purpose of this study was to examine the possible relationship between polychlorinated biphenyls (PCBs) and breast cancer.
Materials and methods
All women who received surgical treatment in the form of excision biopsy because of breast lumps in Reina Sofia University Hospital of Cordoba, Spain between February and November 1997 were included in our study. Women were administered an interview questionnaire which included data on history of known breast cancer risk factors and body measurements for calculating the body mass index. All the excised lumps were histopathologically examined together with chemical estimation of the PCB congener levels in breast fat. The individual PCB congeners determined were numbers 28, 52, 101, 118, 138, 153, 170, 180, 183, 187 and 188. We used gas chromatography to determine the levels of PCBs congeners.
A new variable was constructed called ‘fertility period’ as a well-known risk factor (Kelsey and Bernstein, 1996); for menopausal women we used the difference between the age of menopause and of menarche, and for non-menopausal women the difference between the age of diagnosis of the breast lesion and of menarche.
The collected data were analysed statistically (with initial comparisons between patients with benign and malignant breast lumps) using Student's t -test for continuous variables and Pearson's chi-squared test for categorical variables.
The adjusted risk factors for breast cancer were obtained using multivariable logistic regression. The model evaluated the postulated possible associations between PCBs and breast cancer, considering all the possible known confounding factors. Smoking habit was included in the model because the organochlorate levels are known to be higher in smokers (Duarte-Davidson et al., 1994). The calibration of the final model was assessed using the Hosmer and Lemeshow goodness-of-fit test, and the area under the receiver operator characteristic (ROC) curve (Hosmer and Lemeshow, 1989) assessed its discrimination. All statistical analyses were performed using the Statistical Package for Social Science (SPSS) version 9.0.
The cohort studied consisted of 65 (48.5%) women with benign lesions and 69 (51.5%) with malignant lesions. The average age of the studied women was 51.3 ± 16.1 years. PCB congener numbers 138, 153, 170, 180 and 187 were identified in all the studied subjects.
The variables most highly associated with malignant lesions on univariate analysis were age, lactation periods in months, overweight, PCB n-28 and PCB n-52. The risk factors possibly associated with malignant lesions in the final multivariable analysis model are presented in Table 1
. The most important risk factor identified by the model was PCB n-28 (OR 9.59).
The variable representing lactation period in months was included in the model despite its lack of statistical significance because it was identified as a confounder: when the changes in the coefficients of the models with and without this variable were evaluated, significant changes were observed in the coefficients of number of children, therefore, warranting the presence of lactation period in months in our final model to adjust for this confounding (Hosmer and Lemeshow, 1989).
Breast cancer is a major public health problem in many countries, especially in Western industrialized nations (MMWR, 1996), and our results offer evidence that polychlorinated biphenyls as industrial contaminants, specifically PCB n-28, are associated with increased risk for this cancer. The magnitude of association between PCB n-28 and breast cancer (OR 9.6) together with the fact that its exposure proceeded the effect show the internal validity. The identified large 95% confidence interval (3.7–24.4) of PCB n-28 as a risk factor could be explained by the small sample size; however, its lower level is quite satisfactory. In agreement with Kelsey and Bernstein (1996) we identified age, alcohol, low parity and overweight as major breast cancer risk factors, but our regression model has shown that although there were many risk factors for breast cancer the most crucial one was PCB n-28.
Previous studies (Huisman et al., 1995;Lawrence and Friedman, 1995;Anwar, 1997;Angulo et al., 1999) showed that the level of organochlorine in the mother's milk (via PCBs excretion) is positively correlated with the lactation period in months, and in turn this offers more possible risk for this type of cancer. In this study we could not demonstrate this, but it could be explained by the identification of lactation period in months as a confounding factor associated with the number of children, a variable well known to have a protective effect against breast cancer (Kelsey and Bernstein, 1996).
Because of the difficulty of obtaining samples from healthy women, our control subjects had benign breast lesions; this had the drawback of hospital control where the exposure risk can be different from the general population (bias of Berkson) (Knapp and Clinton, 1992). However, our controls were representative of the population base from which the cases arrived.
We think that there is a great need for more studies of the relationship between cancer, especially breast cancer and PCBs. The future studies should be designed with controls matched for age and other possible confounding factors, and if possible, they should be non-hospital controls.
1. Angulo R , Martinez P , Jodral ML ( 1999 ). PCB congeners transferred by human milk, with an estimate of their daily intake . Food Chem Toxicol 37 : 11081 – 11088 .
2. Anwar WA ( 1997 ). Biomarkers of human exposure to pesticides’ . Environ Health Perspect 105 : 801 – 806 .
3. Carrasco R (1995). Amigos o enemigos de las consechas. Ecología Crítica Enero13
4. Colman A, Dawson J (1996). Un estudio realizado en el Reino Unido revela que las hormonas naturals en los afluentes de agues residuals afectan a los peces. Boletin Moduladores Endocrinos
. Crupo de dirección de Moduladores Endocrinos (CEFIC).
5. Duarte-Davidson R , Wilson SC , Jones KC ( 1994 ). PCBs and other organochlorines in human tissues, samples from the Welsh population: II-milk . Environ Pollut 84 : 79 – 87 .
6. Hosmer DW , Lemeshow SA ( 1989 ). Applied Logistic Regression John Wiley , New York .
7. Huisman M , Eerenstein SEJ , Koopman-Esseboom C et al. ( 1995 ). Perinatal exposure to polychlorinated biophenyls and dioxins through dietary intake . Chemosphere 31 : 4273 – 4287 .
8. Kelsey JL , Bernstein L ( 1996 ). Epidemiology and prevention of breast cancer
. Annu Rev Public Health 17 : 47 – 67 .
9. Knapp RG , Clinton M ( 1992 ). Risk and causality . In Clinical Epidemiology and Biostatistics (eds Knapp RG , Clinton M ); Harwal Publishing Company , Baltimore, Hong Kong, London, Sydney , pp. 109 – 130 .
10. Lawrence RA , Friedman LR ( 1995 ). Contaminants in milk: environmental substances in milk . In In Handbook of Milk Composition (ed Jensen RG ); Academic Press , London , pp. 873 – 874 .
11. MMWR. ( 1996 ). Breast caner incidence and mortality – United States, 1992 . Morbidity Mortality Weekly Rep 45 : 833 – 877 .