Night work and breast cancer: A population-based case–control study in France (the CECILE study)
Night work involving disruption of circadian rhythm was suggested as a possible cause of breast cancer. We examined the role of night work in a large population-based case-control study carried out in France between 2005 and 2008. Lifetime occupational history including work schedules of each night work period was elicited in 1,232 cases of breast cancer and 1,317 population controls. Thirteen percent of the cases and 11% of the controls had ever worked on night shifts (OR = 1.27 [95% confidence interval = 0.99–1.64]). Odds ratios were 1.35 [1.01–1.80] in women who worked on overnight shifts, 1.40 [1.01–1.92] in women who had worked at night for 4.5 or more years, and 1.43 [1.01–2.03] in those who worked less than three nights per week on average. The odds ratio was 1.95 [1.13–3.35] in women employed in night work for >4 years before their first full-term pregnancy, a period where mammary gland cells are incompletely differentiated and possibly more susceptible to circadian disruption effects. Our results support the hypothesis that night work plays a role in breast cancer, particularly in women who started working at night before first full-term pregnancy.
Breast cancer is the most common cancer in women worldwide with an annual incidence of ∼100 cases per 100,000 in developed countries. It is estimated that over 1,300,000 women are diagnosed with breast cancer each year around the world,1 and 53,000 in France.2
Recognized risk factors for breast cancer include genetic mutations, family history of breast cancer, and several aspects of reproductive history, but lifestyle, environmental, or occupational causes of breast cancer are incompletely identified.3 Following the publication of studies indicating a possible role of night shift work in breast cancer, the International Agency for Research on Cancer (IARC) in 2007 classified shift work that involves circadian disruption as probably carcinogenic to humans, on the basis of sufficient evidence in experimental animals and limited evidence of carcinogenicity in humans.4 Whether night work is implicated in breast cancer etiology is of major importance for public health because of the increasing number of women working on a nonstandard day schedule in modern societies. In 2005, for example, 11% of European women were working on shifts that included night work.5
Overall, among 12 epidemiological studies conducted so far to investigate the association between night work and breast cancer,6–17 eight reported positive associations,6–10, 14, 15, 17 of which six were cohort studies of nurses8–10, 14, 15 or radio and telegraph operators17 enrolled in shift work, and two were population-based studies where night work was assessed in a wide range of occupations.6, 7 Other studies did not report an association with breast cancer.11–13, 16 Although the body of evidence generally points to a role of night work in breast cancer occurrence, there is a need of additional studies to better identify the characteristics of night work that may lead to an increased risk.18
Several mechanistic hypotheses for how shift work may be related to cancer have been reviewed recently.19 They include exposure to light at night that suppresses the nocturnal peak of melatonin and its associated anticarcinogenic effects; disruption of the circadian rhythm regulated by several “clock” genes controlling cell proliferation and apoptosis; repeated phase shifting leading to internal desynchronization and defects in the regulation of the circadian cell cycle; and sleep deprivation that alters the immune function.
The human breast undergoes several stages of maturation throughout life, and may be particularly susceptible to carcinogens when exposure occurs during periods of mammary gland development and differentiation such as puberty or pregnancy.20 Although measuring environmental exposures during critical periods of breast development in a woman's lifetime is a key issue for identifying exposures that may lead to breast cancer later in life, the role of night work during these critical exposure windows has not been specifically investigated in epidemiological studies. Full differentiation of the mammary gland occurs during first childbirth and lactation.21–23 It can thus be hypothesized that the risk of breast cancer is particularly elevated when night work involving circadian disruption occurs in the period of life before first full-term pregnancy, i.e., when mammary gland cells are incompletely differentiated.
We conducted a large population-based case-control study in France (CECILE) to investigate the role of environmental and genetic factors in breast cancer that included data on lifelong occupational history. In the present article, we analyzed the role of type, duration and frequency of night work in breast cancer. We also focused on night work in the period before first full-term pregnancy as a possible critical window of exposure.
Material and Methods
Eligible cases were women aged 25–75 years, newly diagnosed for breast cancer between 2005 and 2007 and residing in the French départements of “Côte d'Or” or “Ille-et-Vilaine” (administrative areas) at the time of diagnosis. Patients were recruited in the main cancer hospital of each area, as well as from smaller public and private hospitals that also recruited breast cancer patients, by specifically trained investigators. All breast cancer diagnoses were confirmed histologically. Among the 1,553 eligible cases identified during the study period, 163 refused to participate, 151 women could not be contacted and 7 died before the interview. Finally, 1,232 (79%) incident breast cancer cases were included in the study.
Controls were selected among general population women free of cancer and resident in the study areas at the time of the cases' diagnoses. For including controls, quotas by age were established as a preliminary to yield the control group similar to the case group in terms of age to achieve frequency-matching (10-year age group). Quotas by socio-economic status (SES) were also set a priori to control for potential selection bias arising from differential participation rates across SES categories. These quotas by SES were calculated from the census data available in each study area, to obtain a distribution by SES among controls identical to the SES distribution among general population women, conditionally to age. The recruitment of controls was conducted as follows: phone numbers of private homes were selected at random from the telephone directory of each study area where unlisted numbers had first been recreated. A phone number was dialed up to 15 times at different times of the day and different days of the week until contact could be established with the residents. When a woman was living in the residence reached by phone, she was invited to participate to the study, as long as the predefined quota corresponding to her age group and socioeconomic status (SES) was not completed. When the quota was exceeded, the woman was excluded. To obtain the desired number of controls within the limits of age and SES categories, ∼30,000 phone numbers were dialed for identifying 1,731 eligible controls. Among these, 1,317 (76%) accepted to participate to an in-person interview and were included in the study.
The study was approved by the French Ethic Committee (Jan 2005), the National Data Protection Commission (Dec 2004) and the Advisory Committee on the Treatment of Health Research Information (Apr 2004). All participants signed informed consent before inclusion.
A standardized questionnaire was administered during in-person interviews by trained interviewers, to obtain information on demographic and socioeconomic characteristics, reproduction, medical history, family history of cancer, diet, lifestyle factors, residential and occupational history over the lifetime. A blood sample was also collected during interview.
For each job held for at least 6 consecutive months, we obtained a description of the work tasks, work places, occupational exposures and work schedules. Women were asked whether they had worked for at least 1 hr between 11:00 pm and 5:00 am during all or part of each job. We characterized any night work period with the month and the year of beginning and ending, the usual number of nights per week, and the hour when the night shift started and ended. Any night work period was categorized as overnight (night shift of 6 consecutive work hours or more spanning the time period 11:00 pm–5:00 am), late evening (night shift ending between 11:00 pm and 3:00 am), or early morning (night shift starting between 3:00 and 5:00 am).
Unconditional logistic regression models were used to estimate odds ratios (OR) and their 95% confidence intervals (CI) using women who had never worked at night as the reference group. Analyses were systematically adjusted for the original matching variables, i.e., age (5-year period) and study area, and for well-established risk factors for breast cancer categorized as follows: age at menarche (<12, 12:reference, 13, 14, 15 years and more), age at first full-term pregnancy (<22, 22–24:reference, 25–27, >27 years), parity categorized (nulliparous:reference, 1, 2, 3, 4+ children), current use of menopausal hormone therapy (Yes, No), family history of breast cancer in first-degree relatives (Yes, No), body mass index according to the WHO categories (<18.5, 18.5–24:reference, 25–30, >30), alcohol consumption (≤3 drinks/week:reference, 4–7 drinks/week, 8–14 drinks/week, >14 drinks/week), and tobacco consumption (Never smokers: reference, former smokers, current smokers). Duration of night work was categorized into two groups according to the median value among controls (<4.5, ≥4.5 years or ≤4, >4 years for the analysis of night work before first full-term pregnancy). The average number of nights per week was categorized into two groups according to the median of the distribution among controls (<3, ≥3). Analyses were also conducted after stratification by age group (<55 years, ≥55 years) used as a proxy of menopausal status.
We also conducted analysis according to estrogen- or progesterone-receptor status (ER-positive or ER-negative, PR-positive or PR-negative), and histological subtypes of breast cancer using polytomous logistic regression models, but results were not modified and are not shown.
Analyses were performed using SAS software (version 9.2, Cary, NC).
We included 1,232 breast cancer cases and 1,317 controls. The distributions by age, study area and socioeconomic characteristics are shown in Table 1. Cases and controls were similarly distributed in terms of age and study area (stratification variables). Cases were more frequently single and had higher education levels than the controls.
Table 1. Tumor characteristics of breast cancer and sociodemographic characteristics of cases and controls in the CECILE study
Consistently with the literature, we found that the following variables were associated with breast cancer (Table 2): early age at menarche, late age at first full-term pregnancy, low parity, current use of menopausal hormone therapy, low body mass index in premenopausal women, lack of physical activity and family history of breast cancer in first-degree relatives. No association was apparent between breast cancer and alcohol consumption, or high BMI in postmenopausal women.
Table 2. Odds ratios associated with family history of breast cancer, reproductive factors, lifestyle factors and body mass index in the CECILE study
Overall, 311 women (12%) had ever worked during night shifts (Table 3). Overnight work was the most frequent type of night work schedule (n = 222) followed by late evening work (n = 80) and early morning work (n = 21). Night work was more common among cases than among controls (OR = 1.27 [CI = 0.99–1.64]) (Table 3). The odds ratio for overnight work (OR = 1.35 [CI = 1.01–1.80]) was slightly higher than the odds ratio for late evening work (OR = 1.25 [CI = 0.79–1.98]). Early morning shift was not associated with breast cancer.
Table 3. Odds ratios for breast cancer associated with duration, frequency and type of night work among women of the CECILE study
Duration of night work of 4.5 or more years was associated with an OR of 1.40 [CI = 1.01–1.92]. Working overnight for 4.5 or more years yielded a similar odds ratio of 1.40 [CI = 0.96–2.04].
The odds ratio in women who worked at night, less than three nights per week on average was 1.43 [CI = 1.01–2.03], whereas it was only 1.14 [CI = 0.82–1.59] in women who worked at night ≥3 nights per week. Similarly, the ORs for overnight shifts less than three nights per week and ≥3 nights per week were 1.61 [CI = 1.07–2.42] and 1.13 [CI = 0.76–1.68], respectively.
In the analyses combining the duration of night work and the average number of nights per week, the association with breast cancer was particularly apparent among night workers of long duration (≥4.5 years) working less than three nights per week on average (OR = 1.83 [CI = 1.15–2.93]). This association was more pronounced when only overnight work was considered (OR = 2.09 [CI = 1.26–3.45]).
In analyses restricted to parous women, we investigated the effect of night work before or after first full-term pregnancy (FFTP) (Table 4). The odds ratio for breast cancer in women who ever worked at night before FFTP was 1.47 [CI = 1.02–2.12] as compared to never night workers, whereas it was 1.09 [CI = 0.77–1.55] in women who started working at night after FFTP (Table 4). The odds ratio for night work before FFTP was stronger among women who had been working at night for >4 years before FFTP (OR = 1.95 [CI = 1.13–3.35]), and in those who worked at night less than three nights per week on average during this period of life (OR = 2.24 [CI = 1.35–3.71]). Combining duration and frequency of night work before FFTP yielded an OR of 3.03 [CI = 1.41–6.50] for night work >4 years and <3 nights per week.
Table 4. Odds ratios for breast cancer associated with night work after or before first full-term pregnancy (FFTP), and according to night work characteristics before FFTP among parous women of the CECILE study
Analyses stratified by age group (<55, ≥55 years) are presented in Table 5. Night work was more common among younger (14.5%) than among older women (8.1%). Ever working at night, or working at night before FFTP, was associated with breast cancer in women below 55 years. Among older women, there was some evidence of an increased risk of breast cancer in women with long duration of night work and working less than three nights per week.
Table 5. Odds ratios for breast cancer associated with night work characteristics by age group (<55 or ≥ 55 years) among women of the CECILE study
In this study, we have shown that breast cancer risk is associated with characteristics of night work, and provided new evidence that night work may play a role in the occurrence of the disease. The association of night work with breast cancer was mainly observed in women working during overnight shifts, those who worked at night for 4.5 or more years and less than three nights per week on average. The association was stronger in women who worked at night before their first full-term pregnancy than in women who started working at night later in life.
Overall, based on the available epidemiologic literature,8, 9, 12, 13, 24 the evidence of an association between night work and breast cancer has been seen as limited. However, cohort studies of nurses involved in night shift work8–10, 14, 15 have been more consistent than population-based studies.6, 7 In these population-based studies, the large number of occupational groups with various night work patterns, and the lack of standardization of exposure assessment may explain some of the inconsistencies, because the categories defining type, duration and frequency of night work were based on cut-off points that varied across studies, and may represent different degrees of circadian disruption. It has been suggested that several key domains should be used to better capture circadian disruption based on detailed information on night work in epidemiological studies.18 Among those domains were the rotating type of night-shift work, direction and rate of rotation, and the number of consecutive nights at work. It was also suggested to collect data on sleep habits, and on subject chronotype.
In the present population-based study, our questionnaire did not allow to go as deeply in the description of the night shifts as recommended in this article, due to large differences between night shift systems across occupations. Nevertheless, we were able to categorize the type of night work using time schedule data (late evening, overnight, early morning), and examined the duration of night work in years as well as the average number of nights per week.
We found that breast cancer risk increased for duration of night work >4.5 years, a much shorter period than in cohort studies of nurses where the risk of breast cancer increased for long durations (≥20 years) of rotating night shift work.8, 10, 14, 15
Intriguingly, we also found that breast cancer incidence was inversely related to the average number of working nights per week. This finding requires clarification. It is not consistent with a previous study reporting that the risk of breast cancer increased with three or more working nights per week in the 10 years before diagnosis.6 It has been postulated however that the fewer nonday shifts in succession, the less adaptation can occur.18 Our results of a smaller number of nights per week associated with a higher risk of breast cancer may thus reflect more frequent changes between night and day schedules, conferring a higher degree of circadian disruption. It is also possible that the number of nights per week captured different types of rotating shift work patterns that may be associated differently with breast cancer. The number of consecutive nights has also been seen as a potentially important characteristic of night work, as breast cancer incidence was associated with rotating shifts of at least five consecutive nights per month during at least 5 years among nurses in one study.9 Unfortunately, we were not able to assess the number of consecutive nights at work in our study.
We reported that breast cancer risk was higher in women who started working at night before first full-term pregnancy, particularly if the duration of night work before FFTP was >4 years, and if the number of nights per week was less than 3. Pesch et al. reported an odds ratio of breast cancer of 1.51 [CI = 0.80–2.83] in women starting night shift work between 20 and 29 years of age, a period where most women give birth for the first time.12 Our finding of a higher risk of breast cancer related to night work exposure before FFTP is of particular interest as it is compatible with the early life etiological model for breast cancer, indicating that terminal differentiation of the mammary gland cells occurs at first childbirth and lactation.21–23, 25, 26 This finding supports the hypothesis that incompletely differentiated mammary gland cells may be more susceptible to the potentially carcinogenic effects of circadian disruption during this period of life.19
Limits and strengths of the study
Our findings are based on a large carefully designed population-based case-control study conducted to assess the role of environmental, occupational and genetic factors in breast cancer. The study power enabled to detect odds ratios of 1.5 or above assuming a prevalence of exposure among controls of 10 percent, consistent with the proportion of night workers among French27 and European women.5
Cases were women living in well-defined geographic areas diagnosed with a breast cancer in 2005–2007. To minimize selection bias, we aimed at recruiting all incident cases during the study period in the study areas, by identifying breast cancer patients in the main cancer hospital of each area (Côte d'Or and Ille-et-Vilaine), as well as from smaller public and private hospitals that also recruited patients. To select the controls from general population women in the same areas, quotas by socioeconomic status (SES) were established to yield the control group similar to the general population of women of the same age in terms of SES. After the selection process, we were able to compare the distribution by SES between controls and the female general population in each study area, and found no significant difference, indicating that no major selection bias by SES had occurred. In addition, the proportion of night workers among controls was similar to that expected among women in France.27 Women reporting night work in our study were employed in industries where night work is common, i.e., health and social work, hotels and restaurants, transportation and communication, manufacture of chemicals, rubber and plastic products, manufacture of motor vehicles, manufacture of food products and beverages.18, 27 The population-based design of the study also provides some reassurance that the association between breast cancer and night work is not restricted to a few occupations with frequent night shifts such as nurses.
Although recall bias cannot be totally excluded, it was minimized by the use of standardized questionnaires and the similar interviewing conditions for cases and controls. Moreover, the study was conducted in 2005–2008, a period where the carcinogenic potentials of night work was not a major public concern in France. In addition, the mean number of jobs and the total duration of employment reported by cases and controls were similar.28
To address the possibility of confounding by well-established risk factors for breast cancer, all models examining the association between breast cancer and night work were closely adjusted for potential confounders. Night work was weakly associated with tobacco smoking, body mass index or alcohol consumption, but adjusting for these variables in the models did not change the results.
In conclusion, our results support a possible role of night work in breast cancer, particularly if night work occurs before first full-term pregnancy, and may reflect the link between circadian disruption and mammary carcinogenesis. To go further in the understanding of night work and circadian disruption in cancer etiology, epidemiological studies should better standardize night work definition across studies. Besides collecting data on work schedules over lifetime work history, future epidemiological studies should assess the subjects' chronotype to better characterize circadian disruption. Investigations on genetic polymorphisms and/or epigenetic changes in genes involved in circadian rhythm are also of interest. Given the increasing prevalence of night work among women in modern societies, scrutinizing the relationship between night work and breast cancer constitutes a major issue for public health and may have an impact on work policy.