Besides the home environment, married couples share dietary and other lifestyle habits,1, 2, 3, 4 and nutrition and alcohol affect the risk of breast and other cancers.5 Furthermore, a cancer diagnosis in the husband is a cause of emotional stress for the patient and also for his family,6 and stressful events have been suspected to affect breast cancer risk.7 Thus, an association between history of cancer in the husband and breast cancer in the wife is conceivable. Given the scant epidemiological evidence, we have investigated the issue using the data of a large multicentric Italian case-control study.8
Spouses share the home environment, and dietary and other lifestyle habits. Furthermore, a cancer diagnosis in the husband is a stressful event for the wife also. Thus, a history of cancer in the husband may be an indicator of breast cancer risk. We investigated the issue in a large Italian multicentric case-control study on 2,588 women with incident breast cancer and 2,569 female hospital controls, admitted for acute, non neoplastic diseases. The adjusted odds ratio (OR) was 1.0 (95% confidence interval, CI, 0.7–1.4) for a history of any type of cancer in the husband, 1.0 (95% 0.4–2.7) for stomach, 0.7 (95% 0.2–2.3) for intestinal (chiefly colorectal), 0.9 (95% CI 0.5–1.7) for lung, and 1.3 (95% CI 0.4–4.3) for prostate cancer. The OR was close to unity also when data were analyzed in separate strata of patient's or husband's age, patient's education, or vital status of the husband. This study suggests that women whose husband had a diagnosis of cancer are not at increased risk of breast cancer, although results for individual cancer sites should be interpreted with caution, due to small numbers. © 2006 Wiley-Liss, Inc.
Subjects and methods
A hospital-based case-control study was conducted in 4 Italian regions between 1991 and 1994 on 2,569 women with histologically confirmed breast cancer (median age 55 years, range 23–74) and 2,588 controls (median age 56 years, range 20–74) admitted to the same hospitals as cases for acute, nonneoplastic conditions (22% traumas, 33% nontraumatic orthopedic disorders, 15% surgical conditions and 30% miscellaneous other diseases).8 For this analysis, only ever married women were included, thus leaving a total of 2,341 cases and 2,355 controls. Interviews were conducted in hospital using a structured questionnaire. Information was collected on socio-demographic factors, anthropometric variables, tobacco, alcohol and other lifestyle habits, physical activity, a validated food frequency questionnaire, obstetric and gynecologic history and history of selected conditions and of use of female hormone preparations.
In the section on family history was recorded, for any husband, parent, sibling, child or grandparent with history of cancer, the site of the cancer, the age at diagnosis, the vital status and the current age or the age at death. We recorded only husband's cancers that occurred before the woman's breast cancer diagnosis for cases or hospitalization for control women. On account of recall and classification difficulties, some sites were combined (i.e. the whole intestinal tract, Hodgkin's and non-Hodgkin lymphomas). No verification of the cancer diagnosis in the husband was performed.
For a husband with cancer, we recorded age at diagnosis and current age or age at death. Thus, we were able to determine the time elapsed since the cancer diagnosis only for the 29 (16%) husbands with cancer who were still alive at the time of the study. For these, the median elapsed time was 3.5 years. For deceased husbands, most deaths (81%) occurred within 2 years from the cancer diagnosis.
We estimated odds ratios (OR) and 95% confidence intervals (CI) of breast cancer, according to history of cancer at selected sites in the husband, using unconditional multiple logistic regression models,9 including terms for age (in quinquennia and continuous), study centre, husband's education, and in the extended model, patient's education, age at menarche, menopausal status, parity and family history of breast cancer. We had 83% power to detect an OR ≥ 1.5 for history of any cancer in the husband.
Table I shows the OR of breast cancer, according to history of cancer at selected sites in the husband. No significant association was found. The OR was 1.0 (95% CI, 0.7–1.4) for a history of any type of cancer in the husband, 1.0 (95% CI 0.4–2.7) for stomach, 0.7 (95% CI 0.2–2.3) for intestinal (chiefly colorectal), 0.9 (95% CI 0.5–1.7) for lung and 1.3 (95% CI 0.4–4.3) for prostate cancer. The OR was close to unity when data were analyzed in separate strata of patient's or husband's age and patient's education (not shown). Exclusion of the few divorced women did not materially change the results. The OR for a history of any cancer in the husband was 1.0 (95% CI 0.5–2.1) for women whose husband was still alive, and 1.1 (95% CI 0.8–1.7) for widowed women.
|Site of cancerin the husband||No. of cases||No. of controls||OR2 (95% CI)3||OR4 (95% CI)3|
|Oral cavity||4||6||0.8 (0.2–2.8)||0.8 (0.2–2.8)|
|Esophagus||1||2||0.4 (0.0–4.6)||0.4 (0.0–4.6)|
|Stomach||8||10||1.0 (0.4–2.4)||1.0 (0.4–2.7)|
|Intestine||5||8||0.7 (0.2–2.1)||0.7 (0.2–2.3)|
|Liver||4||7||0.6 (0.2–1.9)||0.6 (0.2–2.1)|
|Pancreas||5||6||0.8 (0.3–2.8)||0.9 (0.3–3.1)|
|Larynx||10||7||1.7 (0.6–4.5)||1.9 (0.7–5.1)|
|Lung||20||25||0.9 (0.5–1.6)||0.9 (0.5–1.7)|
|Prostate||6||5||1.3 (0.4–4.5)||1.3 (0.4–4.3)|
|Bladder||3||6||0.6 (0.1–2.3)||0.5 (0.1–2.1)|
|Brain||6||5||1.3 (0.4–4.3)||1.2 (0.4–4.1)|
|Leukemias||1||1||0.9 (0.1–14.2)||0.9 (0.0–16.4)|
|All lymphomas||1||1||0.8 (0.1–13.2)||0.4 (0.0–7.6)|
|All sites||86||95||1.0 (0.7–1.3)||1.0 (0.7–1.4)|
In this large case-control study, a history of cancer in the husband was not associated with risk of breast cancer in the wife. Thus, neither the home environment and the husband's lifestyle habits that may have led to a cancer diagnosis, nor the emotional stress following the diagnosis of cancer in the husband, appear to be important determinants of breast cancer.
When specific sites are considered, we found a nonsignificant 30% increase in breast cancer risk in women married to a man diagnosed with prostate cancer. In 3 other studies, the risks ranged between 0.8 and 1.2,10, 11, 12 thus suggesting that a prostate cancer in the husband does not appreciably increase the risk of breast cancer.
Although it has been noted that breast and colon/colorectal cancer may share important etiological factors,13 including dietary factors and alcohol drinking, a non significant inverse association between these 2 neoplasms in spouses was found in this study. Similarly, a relative risk of breast cancer of 0.77 was found in a cohort of wives of men with colon cancer,14 and no cases of colon cancer were found in a cohort of wives of men with breast cancer, as compared to 1.82 expected.15 In contrast, a study of the proportional frequency of combinations of sites in 195 married couples found a relative risk of 2.9 (95% CI 1.9–4.1) for a husband with colon and a wife with breast cancer, when compared to the general population, but the risk was 0.96 (0.6–1.4) when estimated using a permutation test on the sample distribution.11
In this study, only 1 woman with breast cancer, and no control, reported a history of breast cancer in the husband. Thus, our data were too scanty to investigate the relation between breast cancer in both spouses. The cohort of wives of men with breast cancer had a risk of breast cancer of 0.97 (0.31–2.26),15 while in the Swedish Cancer Family Database the risk of breast cancer was 1.29 (95% CI 0.89–1.77) in women with a husband diagnosed with breast cancer.16
In a Swedish study on wives of men with breast cancer,15 the only significant association was with brain cancer (standardized mortality ratio = 4.37, 95% CI 0.90–12.78), based however on 3 cases only. In this analysis, we did not find a significant association between brain cancer in the husband and breast cancer in the wife, although brain cancer was one of the few sites where the point estimate of the OR was slightly above unity (OR = 1.2), the other 2 being the larynx and the prostate.
This study has all the strengths and weaknesses of a case-control study with hospital controls,9 which have been discussed elsewhere.8 Moreover, the history of cancer in the husband was reported by the woman, and we did not have any verification of the diagnosis. Several studies have examined the validity of self-reported information on history of cancer in relatives finding satisfactory results for some common cancer sites, like breast and colon, and a lower accuracy for some other sites.17 We are not aware of studies investigating the validity of reported cancer in spouses. However, a long and burdensome period of medical management generally follows a cancer diagnosis, and thus most women should be aware if such an event happens to their husband.
Although our estimates were reasonably precise for history of any cancer, the number of women reporting a cancer at a specific site in the husband was small, and thus the OR estimates for specific sites were subject to considerable random variation. For the 2 sites more frequently reported, i.e. lung and stomach cancer, the ORs were 1.0 and 0.9, respectively, thus providing reassuring results.
Although we adjusted for major breast cancer risk factors, we did not have information on risk factors in the husband, besides years of education and social class. In particular, we had no information on age of the husband. Age in the spouses is correlated, and thus the age of the wife was also a proxi of the age of the husband. Furthermore, we adjusted for husband's education, which is a correlate of cancer risk.
Our questionnaire was not designed to investigate stressful events, and thus we lacked information on several other stressful events. Among the variables used to investigate the relation between stress and breast cancer risk, disease and death in the husband have been used in several studies.7 Moreover, it has been shown that a cancer diagnosis in a member is a source of stress for the whole family, and particularly, for the spouse.6 Thus considering cancer in the husband as an indicator of stress appears justified.
In conclusion, this study suggests that women whose husband had a diagnosis of cancer are not at increased risk of breast cancer.
The authors thank Ms MP Bonifacino for editorial assistance.