Article Text
Abstract
Background Increased mammographic density is one of the strongest risk factors for breast cancer. Night shiftwork and its related factors, which include light at night, phase shift and sleep disruption, are believed to increase breast cancer risk however, their effects on mammographic density have barely been studied.
Methods This study included 1821 women enrolled in the Breast Cancer Environment and Employment Study between 2009 and 2011. Mammographic density was measured using the Cumulus software program. The association of night shiftwork factors with square root transformed absolute dense area (DA) and percentage dense area (PDA) were modelled using linear regression adjusted for confounders.
Results Ever doing graveyard shiftwork (between 24:00 and 05:00 hours) was not associated with PDA (β=−0.10; 95% CI −0.27 to 0.08)) and DA (β=−0.12; 95% CI −0.33 to 0.09)). No association was found between night shiftwork related factors (light at night, phase shift and sleep disturbance) with PDA or DA.
Conclusions Shiftwork and its related factors are not associated with mammographic density. Using high-quality, comprehensive shiftwork data from a large population-based breast cancer case–control study, this study suggests that mammographic density does not play a role in the relationship between shiftwork and breast cancer risk.
- shift work
- cancer
Statistics from Altmetric.com
Key messages
What is already known about this subject?
Shiftwork and its related factors, which include light at night, phase shift and sleep disruption, are believed to increase breast cancer risk.
Few studies have investigated the effect of shiftwork on mammographic density.
What are the new findings?
No evidence of an association between shiftwork and mammographic density was observed in this study.
How might this impact on policy or clinical practice in the foreseeable future?
This study suggests that mammographic density might not play a role in the relationship between shiftwork and breast cancer risk.
Introduction
Mammographic density is one of the strongest known risk factors for breast cancer1 and a strong intermediate marker for prevention and intervention.2 It represents the white/opaque radiographic appearance of epithelial and stromal tissue on a mammogram, as opposed to fatty tissue which appears radiologically lucent/dark. Mammographic density is modifiable through adjuvant and preventative breast cancer endocrine therapies2 and it has been shown that reducing density reduces breast cancer risk.2 The large variation of mammographic density measures observed in screen-aged women can largely be explained by genetic factors.3 However, some breast cancer risk factors such as reproductive and lifestyle factors are also associated with breast density.4
In 2019, the International Agency for Research on Cancer classified night shiftwork that involves circadian disruption has as ‘probably carcinogenic to humans’ on the basis of sufficient evidence in experimental animals and limited evidence of breast cancer in humans.5 There are a number of possible pathways by which night shiftwork might result in breast cancer, including phase shift, light at night and sleep disruptions. Exposure to these biological mechanisms may influence circadian patterns of melatonin production, and thereby change the hormonal profiles that play a role in cancer-related pathways.6 Melatonin is a hormone integral to the sleep–wake cycle and may also have anticancer activity. Multiple consecutive nights of shiftwork may result in phase shift, where the central sleep–wake cycle becomes adjusted to being awake at night, and potentially causing the slower-adapting peripheral rhythms, including cell division, to lag behind.7
The associations between night shiftwork and mammographic density have been assessed in only two studies, one reporting higher breast density with shiftwork and the other reporting no association.8 9 It has been hypothesised that prolonged night shiftwork, through melatonin-related mechanisms or other mechanisms, may possibly influence mammographic density. Studies investigating the association between mammographic density and 6-sulfatoxymelatonin, the primary metabolite of melatonin in urine that serves as biomarker of circadian disruption, have shown either an inverse association in premenopausal women or no association at all.9 10 Given the limited epidemiological evidence currently available and the inconsistent findings, we used high-quality, comprehensive shiftwork data from a population-based breast cancer case–control study to investigate the association between shiftwork factors and mammographic density.
Methods
Study population and design
The Breast Cancer Environment and Employment Study (BCEES) was a large, population-based case–control study of women aged between 18 and 80 years.7 The methods used for this study have been described previously and are summarised briefly here. Informed consent was obtained for all participants.
Cases with invasive breast cancer (n=1205; response fraction 57.8%) were identified and recruited via the Western Australia Cancer Registry between 2009 and 2011.7 Controls (n=1789; response fraction 41.1%) were randomly selected from the Western Australian electoral roll and frequency age matched to the expected distribution of cases. Participants completed a questionnaire that included questions on demographic, lifestyle and reproductive factors and on all jobs held for at least 6 months and shiftwork. Women who reported working graveyard shift (regular working between midnight and 05:00 hours) were asked additional questions regarding the times the shifts started and finished, the roster pattern, and for how long they had done the shiftwork. In addition, they were asked questions about exposure to light at night, phase shift and sleep disturbance based on an a priori framework established to assess possible health effects of shiftwork using biologically plausible mechanisms.6 ,7 Light at night was defined as being exposed bright or medium light at night at work and/or light in the bedroom when sleeping during the day. Phase shift was defined as working two or more nights of forward rotation or three or more nights of backward rotation consecutively. Sleep disturbance was defined as experiencing decreases in both quantity and quality of sleep when working night shifts.7
Mammographic data
From the 2994 BCEES participants, we obtained digitised mammographic images from craniocaudal film mammograms for 1903 (668 cases and 1235 controls) women who attended BreastScreen Western Australia (WA). This represents 64% of BCEES participants. For breast cancer cases, prediagnosis mammograms were used. For controls, the earliest mammogram taken between 2009 and 2011 was used—to match the years the cases were diagnosed. The Cumulus software (Sunnybrook Health Sciences Centre, Toronto, Canada) was used to estimate absolute dense area (DA), percentage dense area (PDA) and non-dense area (NDA).
Statistical analysis
Multivariable linear regression was used to examine the associations of the mammographic measures (dependant variables: DA and PDA) with independent variables, graveyard shiftwork and three mechanistic variables (light at night, phase shift, sleep disturbances). The mammographic measures were square root transformed to improve normality of the regression residuals. A brief description of how to interpret regression coefficients for square root transformed mammographic measures can be found in the footnote of table 1. Examination of potential confounders included all covariates listed in online supplementary table 1. Women who had missing data on these covariates were excluded (n=82) leaving a final sample size of 1821 women. Out of the 1821 women, body mass index (BMI) was missing for 40 women, and therefore, we used NDA as a proxy for BMI for all subjects. We conducted two sensitivity analyses: the first adjusted for BMI as a continuous variable (instead of NDA) to assess the suitability of NDA as a proxy; the second analysed breast cancer cases and controls separately. χ2 and t-tests were performed to assess difference in the covariates summarised in online supplementary table 1 between women with and without mammograms from which breast density was measured and available from the publicly funded BreastScreen program. All analyses were performed using Stata V.14.1 (StataCorp).
Supplemental material
Results
Characteristics of subjects and the mean of DA and PDA by these characteristics are presented in online supplementary table 1. Compared with women in the BCEES study for whom mammograms were not available, women included in this study were more likely to be older, not diagnosed with breast cancer, have higher social economic status, less educated, born overseas, overweight or obese, non-smoker, non-alcohol consumer, have had at least one child, have used hormone replacement therapy, and premenopausal (data not shown).
There was no evidence of associations of either mammographic measure with graveyard, phase shift, light at night and sleep disruption or with duration and dose of exposure with shift work variables (table 1). Adjustment for BMI instead of NDA or analysing breast cancer cases and controls separately did not meaningfully alter the results (data not shown).
Discussion
We did not observe any evidence of association between mammographic density measures and shiftwork and its associated factors. Two studies have investigated these associations previously, with one reporting higher mammographic density in women with 15 years of cumulative exposure to night shiftwork,8 and the other one finding no association with rotating night shiftwork in a population of nurses and midwives.9 The discrepancy between studies could be related to differences in the definitions of what constitutes night shiftwork (eg, hours of start and finish and whether the shifts are rotating or not). Measurement of cumulative exposure to shiftwork might better predict changes in mammographic density over time through melatonin-related mechanisms, similar to hormonal changes as proposed by Pike’s ‘breast tissue ageing’ model.11 Thus far, the only interventions that have been shown to clinically significant changes in breast density have been adjuvant or preventive breast cancer endocrine therapies.2 Other possible sources of discrepancy include sample size, study population, exposure assessment and type of statistical analysis.
Strengths of this study include its size and data quality. However, shiftwork information was self-reported and collected retrospectively and, therefore, vulnerable to recall bias although there is evidence that recall bias was not a major concern in the BCEES.12 Mammograms were obtained from 64% of BCEES participants which is higher than the BreastScreen WA participation rate for women in the targeted screening age. While differences in demographic, lifestyle and reproductive factors between women with and without mammograms were observed, women are unaware of their mammographic density prior to screening and therefore would not impact a woman’s intention to screen (ie, no selection bias). Finally, although the sample size was large, this study may have lacked statistical power to detect small effect sizes.
In summary, our results did not support an association between shiftwork factors and mammographic density.
Acknowledgments
The authors would like to thank all the study participants, as well as the members of the BCEES team (Allyson Thomson, Ann D’Orsogna, Terry Slevin, Jennifer Girschik, Pierra Rogers, Jane Deborah Glass, Troy Sadkowsky).
References
Footnotes
Twitter @drterryboyle
Contributors LF, JH, CS, EW and JS designed and carried out the original study. SE-Z conducted the data analysis. SE-Z drafted the manuscript, and all authors critically revised it and approved the final version.
Funding The Breast Cancer Environment and Employment Study was funded by a National Health and Medical Research Council Australia (NHMRC) Project Grant #572530. SE-Z was supported by a Cancer Council Western Australia Cancer Epidemiology Initiative grant. JS is funded by the National Breast Cancer Foundation as a principal research fellow.
Competing interests None declared.
Patient consent for publication Not required.
Ethics approval Study approval was obtained from the The University of Western Australia’s Human Research Ethics Committee and the Western Australian Department of Health.
Provenance and peer review Not commissioned; externally peer reviewed.
Data availability statement No data are available.