Women who have mutations in either the BRCA1 or BRCA2 gene have a very high lifetime risk of breast cancer. It has been estimated that the breast cancer risk associated with a mutation in either gene is as high as 80% by the time the woman reaches age 70.1 It is of interest to identify nongenetic factors that modify the inherited risk of cancer in these women so that preventive programs can be designed. In 1998, we reported that BRCA carriers who smoked had a reduced lifetime risk of breast cancer and proposed that the protective effect of smoking on breast cancer could be due to the anti-estrogenic effect of smoking.2 In that case-control study, we reported that smoking more than 4 pack-years reduced the risk of breast cancer by approximately 50% in carriers (OR = 0.46) compared to mutation carriers who had never smoked. Our study was based on a relatively small number of subjects (186 matched sets). Over the past 5 years, our international study group has continued to identify BRCA carriers with and without breast cancer and to collect lifestyle information on these. Here we use this extended data set to reanalyze the effect of smoking on breast cancer risk in BRCA1 and BRCA2 carriers.
The effect of cigarette smoking on the risk of breast cancer is controversial, although most studies show little or no effect. It has been suggested that smoking may reduce the risk of developing hereditary breast cancer. We completed a case-control study on 1,097 women with breast cancer who were BRCA1 or BRCA2 mutation carriers and 1,097 age-matched controls with a mutation in the same gene but without breast cancer. There were no statistically significant differences between the cases and controls in terms of the number of current and ex-smokers (41.2% and 40.4%, respectively) or the age at smoking commencement (18.2 years and 18.5 years, respectively). There were no statistically significant differences between cases and controls regarding beginning smoking within 5 years of menarche (OR = 1.03; 95% CI 0.83 to l.28) or before the first pregnancy (OR = 1.09; 95% CI = 0.90 to 1.33). In conclusion, contrary to our previous report, smoking does not appear to be a risk factor for breast cancer among carriers of BRCA mutations. © 2004 Wiley-Liss, Inc.
Eligible study subjects were women who carried a pathogenic mutation in either the BRCA1 or BRCA2 gene. Information on study subjects was submitted from 52 participating centers in 11 countries. These women participated in clinical and research protocols at the host institutions. All study subjects received genetic counseling and provided written consent for genetic testing. In most cases, testing was offered initially to women who had been affected with breast cancer or ovarian cancer. When a BRCA1 or BRCA2 mutation was identified in an affected proband, genetic testing was then offered to other unaffected and affected women in the family. In a few families (<10% of the total) only unaffected carriers were identified. Mutation detection was performed with a range of techniques, but all abnormal nucleotide sequences were confirmed by direct sequencing of DNA. A subject was eligible for the current study when the molecular analysis established that she was a carrier of a pathogenic mutation. The great majority (>95%) of the mutations identified in the study were either nonsense mutations, deletions, insertions or small frameshifts and resulted in a truncated protein.
Information was submitted to the data center on a total of 5,025 mutation carriers. Forty-seven women were excluded because they had ovarian cancer prior to the diagnosis of breast cancer; 878 subjects had incompletely recorded their smoking history and were excluded. Information was also missing from 933 women regarding their cancer status, parity, age of first birth, use of oral contraceptives or date of birth and these were excluded. After these exclusions, there was a total of 3,167 eligible women, including 1,556 women with breast cancer (potential cases) and 1,611 women without breast cancer (potential controls).
An attempt was made to identify a single matched control for each of the 1,556 eligible breast cancer cases. Control subjects had never had breast cancer but were known to be carriers of a mutation in the BRCA1 or BRCA2 gene. The control was matched to the case according to the type of mutation (BRCA1 or BRCA2), year of birth (within 1 year) and place of residence (USA, Quebec, other Canada, Europe and Israel). A subject was not eligible to be a control for a given case if she had a bilateral prophylactic mastectomy prior to the age of diagnosis of breast cancer of the matched case. To be matched to a given case, the control had to be at least as old as the case at the time of diagnosis of breast cancer in the case (i.e., the age of the control at the time she filled out the questionnaire had to be greater or equal to the age of the breast cancer diagnosis in the case). We did not match by family because in most cases a suitable control from within the family could not be found.
A total of 1,097 matched case-control pairs were generated for the analysis, including 806 pairs with BRCA1 mutations and 291 pairs with BRCA2 mutations. Some of the study subjects were included in the 186 matched sets in our original report,2 but the matching was not retained from the previous study. Case and control subjects completed a questionnaire that asked about their medical and reproductive histories, and selected lifestyle factors including smoking. In most cases the questionnaire was completed when blood was drawn for genetic testing or within a year of receiving the test result. They were asked if they ever smoked, if they currently smoked, at what age they first began smoking, at what age they stopped smoking and average smoking consumption. A pack-year index was created that was determined as the total number of years smoked times the average consumption (in packs per day) over the smoking period. When considering smoking related to pregnancies, we considered only births that occurred prior to the age-of-onset of breast cancer in the case. Only smoking exposures that occurred prior to the date of cancer diagnosis were considered (for the case and her matched control).
Only living women of known mutation status were approached to participate in our study. This is because mutation testing can only be done if a blood sample is available and because of the difficulty in obtaining accurate information about oral contraceptives from surrogate sources. The questionnaires were completed between 1977 and 2002. For the cases, an average of 8.9 years had elapsed from the date of diagnosis of breast cancer until the date the questionnaire was completed.
In an attempt to be consistent with our previous study we performed the statistical analysis as described by Brunet et al.2 Smoking histories were compared between case subjects and control subjects. The odds ratio for breast cancer associated with smoking was estimated by use of conditional logistic regression for matched sets. Odds ratios were adjusted for reproductive variables (parity, age of first birth and age at last birth) and for oral contraceptive use (ever/never). Odds ratios were estimated first by univariate analysis and then by multivariate analysis, adjusting for the reproductive risk factors.
A total of 1,097 matched cases and controls was studied. Approximately 1/2 of the subjects resided in the United States and 30% resided in Canada (Table I). The remainder were from Europe and Israel. The mean age at diagnosis of breast cancer in the cases was 40.0 years (range 19 to 74 years). Cases were well-matched for age and were similar to controls in terms of age of first birth, and parity (Table I). The mean age at menarche was slightly younger for the cases (12.8 years) than for the controls (13.0 years) (p = 0.05).
|Variable||Cases N (%)||Controls N (%)||p-Value|
|−30||15 (1.4%)||22 (2.0%)|
|30–39||212 (19.3%)||244 (22.2%)|
|40–49||442 (40.3%)||407 (37.1%)|
|50+||428 (39.0%)||424 (38.6%)|
|Place of residence|
|Europe||159 (14.5%)||159 (14.5%)|
|Israel||71 (6.4%)||71 (6.4%)|
|Quebec||93 (8.5%)||93 (8.5%)|
|Canada (other than Quebec)||252 (23.0%)||252 (23.0%)|
|USA||522 (47.6%)||522 (47.6%)|
|BRCA1||806 (73.5%)||806 (73.5%)|
|BRCA2||291 (26.5%)||291 (26.5%)|
|Mean age at menarche||12.8||13.0||0.05|
|Mean number of children||2.1||2.1||0.51|
|Mean age at 1st birth||29.2||29.0||0.41|
|Oral contraceptive (ever)||748 (68.2%)||735 (67.0%)|
Forty-one percent of the cases and 40% of the controls reported ever having smoked regularly (OR = 1.05; 95% CI 0.88 to 1.25). The results were similar for BRCA1 carriers (multivariate OR = 1.09; 95% CI 0.87 to 1.33) and for BRCA2 carriers (OR = 0.97; 95% CI 0.68 to 1.38).
There was no trend observed between the risk of breast cancer and the lifetime amount smoked. Women who smoked less than 20 packs-years had a slightly increased risk of breast cancer (OR = l.ll; 95% CI 0.92 to 1.33) compared to nonsmokers, but women who smoked more than 20 pack-years had a slightly lower risk of breast cancer than women who never smoked (OR = 0.81; 95% CI = 0.57 to 1.15). Neither of these associations was statistically significant (Table II).
|Smoking (cigarette)||Cases||Controls||OR; 95% CI||p-value|
|Pack per year:|
|Mean age started||18.2||18.5||0.31|
|Mean age last smoked||34.4||34.4||0.94|
|Mean year smoked||16.2||15.9||0.60|
Among smokers, the cases smoked for an average of 16.2 years compared to 15.9 years for the controls (p = 0.60). Cases started smoking at a slightly younger age than controls (mean age of 18.2 years vs. 18.5 years) but the difference was not statistically significant (p = 0.31). We also considered smoking prior to first pregnancy and smoking shortly after menarche (Table III). There was no significant increase in the breast cancer risk associated with ever-smoking prior to first pregnancy (OR = 1.09; 95%; CI 0.90 to 1.33) or with commencement of smoking within 5 years of menarche (OR 1.03; 95%; CI 0.83 to 1.28). There were no statistically significant differences in the risks associated with smoking and breast cancer among BRCA gene carriers from different regions (Table IV).
|Cases||Controls||OR; 95% CI||p-value*|
|Pregnancy and smoking|
|No pregnancy, smoker||49||56||0.93 (0.67–1.11)||0.73|
|Started before 1st pregnancy||343||319||1.09 (0.90–1.33)||0.36|
|Started after 1st pregnancy||60||65||0.92 (0.63–1.35)||0.67|
|Menarche and smoking|
|Started smoking within 5 years of menarche||256||253||1.03 (0.83–1.28)||0.77|
|Started smoking after 5 years from menarche||196||187||1.07 (0.85–1.35)||0.56|
|Regions||Cases (Smokers)||Controls (Smokers)||OR; 95% CI||p-value*|
|Age ≤50||975 (396)||975 (389)||1.03 (0.86–1.25)||0.73|
|Age >50||122 (56)||122 (51)||1.16 (0.68–1.97)||0.58|
|Age ≤50||770 (327)||770 (313)||1.07 (0.87–1.32)||0.50|
|Age >50||97 (45)||97 (44)||0.99 (0.54–1.80)||0.97|
|Europe and Israel|
|Age ≤50||205 (69)||205 (76)||0.84 (0.55–1.30)||0.44|
|Age >50||25 (11)||25 (7)||2.29 (0.62–8.44)||0.21|
This case-control study failed to confirm our previous report of an association between smoking and risk of breast cancer in BRCA1 or BRCA2 carriers. The previous study of smoking in BRCA1 and BRCA2 carriers, which included only 186 matched pairs,2 suggested that smoking might reduce the risk of breast cancer in carriers of BRCA1 or BRCA2 gene mutations. In the current study we included a total of 1,097 women with breast cancer who were BRCA1 or BRCA2 gene carriers and 1,097 age-matched women with a mutation in the same gene, but without breast cancer. We found neither an increased or decreased risk of developing breast cancer associated with smoking and no association with early initiation of smoking. Because the methods and analyses for the present study and that of Brunet et al.2 were similar, we believe that the positive result of the first study may have been spurious. The present study is much larger and the confidence limits are narrow. Also, the first study was restricted to subjects from North America, and a significant proportion (approximately 20%) of the current subjects are from Europe. However, there appears to be no protective effect when we restrict our study to subjects from North America (Table IV). The main difference in the date between the present and the previous that the proportion of ever-smokers among controls has declined from 52 to 40%. Couch et al.3 studied smoking as a risk factor for breast cancer in 132 high-risk families. Each family contained at least 3 related women with breast or ovarian cancer, but the BRCA status of the study subjects was not known. They reported that among sisters and daughters of the index cases, ever-smoking was associated with a 2.4-fold increased risk of breast cancer (95% CI: 1.2–5.1), compared to nonsmokers. Cigarette smoking is an established cause of a variety of cancers, but its role in breast cancer etiology is not clear. The mammary gland is exposed to metabolites of tobacco constituents that have been modified in the process of absorption and transportation to the breast.4 Anti-estrogenic effects of smoking has been described, including early menopause4, 5, 6 and increased risk of osteoporosis.7 Smokers are also at decreased risk of endometrial cancer8 as well as decreased risk of breast fibroadenoma.9 It has been reported that women who smoke have reduced urinary concentration of estrogen during the luteal phase of the menstrual cycle.10 The plausibility of an anti-estrogenic effect of smoking is discussed by Baron et al.11
In our study, we included 2,194 out of a possible 5,025 study subjects. This selection procedure was done in order to include only women who had complete information on smoking and the relevant covariates. A matched analysis was undertaken in order that the cases and controls be as similar to each other as possible and resulted in a groups of cases and controls that were similar with respect to age, residence, parity and oral contraceptive use (Table I). We studied prevalent cases and an average of 8.9 years had elapsed since the diagnosis of breast cancer and completion of the questionnaire. If smoking were to negatively influence the prognosis of women with breast cancer, then our estimate of the fraction of smoking cases would be an underestimate, and a deleterious effect of smoking on breast cancer risk would be obscured.
There have been several studies regarding smoking and breast cancer in the general population, which are contradictory. Some investigators have reported that smokers have a lower risk of developing breast cancer,12, 13 whereas other studies have reported a significantly increased risk.6, 14, 15, 16, 17 In the latter group of studies (except reference15), passive as well as active smoking was included, and women were considered to be unexposed if they had no history of either. We had no details about passive smoking in our study, and it is not clear if including this information would have modified our results. In the Nurses Health Study,18 the RR of breast cancer for smokers was only 1.04 (95% CI: 0.94–1.15), but in that study there was a modest increase in risk among smokers who began smoking before the age of 17 (RR = 1.19, 95% CI (1.03–1.37). In a case-control study of smoking and breast cancer during adolescence, Marcus et al.19 showed that women who began to smoke cigarettes between the age of 10 and 14 years have an elevated risk of developing breast cancer (OR = 1.5; 95% CI: 0.9 to 2.5). A recent study20 found that risk of breast cancer was significantly increased (OR = 1.69; 95% CI: 1.13 to 2.51) in women who had had at least 1 pregnancy and had started to smoke within 5 years of menarche. We did not observed an increased risk for breast cancer among BRCA1 and BRCA2 gene carriers, regarding smoking prior to first pregnancy or smoking near the age of menarche. In conclusion, it seems there is no association between the risk of breast cancer and lifetime smoking among BRCA1 and BRCA2 gene carriers.
We thank M. Daly, J. Garber, Hakan Olsson, O. Olopade, H. Saal, K. Huelsman, G. Evans, E. Warner, M. Osborne, D. Fishman, B. Karlan, S. Merajver, D. Gilchrist, A. Chudley, A. Eisen, P. Moller, K. Heimdal, W. McKinnon, M. Wood, C. Eng, E. Lemire, M.K. Dabney and J. McLennan for contributing data to this article. S.L.N. is supported by NCI ROICA74415.