Burden of cancer attributable to never breastfeeding in Japan in 2015

Summary

We estimated the population attributable fraction (PAF) of cancers in Japan attributed to never breastfeeding in 2015. The data on breastfeeding in Japan were derived from two sources. Data on women aged younger than 40 were obtained from the Longitudinal Survey of Babies in the 21st Century (LSB21); those for women aged 40 to 69 was derived from the Japan Public Health Center-based Prospective Study for the Next Generation (JPHC-NEXT). For the present study, the optimal frequency of breastfeeding was a history of ever breastfeeding. The PAF was calculated for each age group using a standard formula. Overall, 0.3% of total cancer incidence and 0.1% of total cancer mortality in Japanese women were attributable to never breastfeeding. The high prevalence of women who breastfed children may explain the lower fraction of cancer attributable among Japanese women compared to studies from other parts of the world.

Introduction

The Continuous Update Report (CUP) on Food, Nutrition, Physical Activity and Prevention of Cancer by the World Cancer Research Fund (WCRF) and American Institute for Cancer Research (AICR) concluded that there is probable evidence for an association between breastfeeding and decreased risk of breast cancer (1). In 2002, the Collaborative Group on Hormonal Factors in Breast Cancer showed that for every year of breastfeeding, the risk of breast cancer decreases by 4.3% (2). On the other hand, the association between breastfeeding and cancer risk of female reproductive organs is unknown. There is a limited or suggestive evidence for an association between breastfeeding and increased risk for ovarian cancer (1). While that report also discussed evidence for endometrial cancer, it was considered too limited to allow the drawing of a conclusion (1).

Here, we estimated the population-attributable fractions (PAF) of cancer incidence and mortality in Japan in 2015 attributed to never breastfeeding.

Materials and Methods

Cancers associated with breastfeeding

We defined the PAF of cancers associated with breastfeeding as the proportion of cancers diagnosed in a certain period in a population that could possibly have been prevented if everyone in the population ever breastfed children. In this study, we included cancers that were discussed in the WCRF/AICR CUP project (1). The cancer sites included in this estimate were breast, endometrium, and ovary.

Theoretical minimum risk exposure level

The optimum level of breastfeeding was considered as a history of ever breastfeeding.

Prevalence of breastfeeding

The latent period between cancer onset and breastfeeding is unknown. Therefore, for this study, we decided the optimal exposure to breastfeeding is a history of ever breastfeeding. Since 2001, the Japanese Ministry of Health, Labour and Welfare (MHLW) has been implementing a survey, called the Longitudinal Survey of Babies in the 21^st Century (LSB21), with the aim of developing strategies to combat the declining fertility rate in Japan (3). The LSB21 surveyed families from all over Japan with newborns delivered between the 10^th and 17^th of January or July of 2001. The baseline questionnaire was sent to all of the families six or seven months after the baby was born. Follow-up questionnaires were then sent annually. Each participating child's birth records were linked with the Japanese vital statistics, which includes information such as birth length and weight, gestational age, sex and parental age.

For the history of breastfeeding, LSB21 obtained this information from the baseline questionnaire. The question asked whether the mother breast fed, formula fed, or both, in addition to the duration (ranging from zero to seven months). For people aged 40 and over, we used data from the Japan Public Health Center-based Prospective Study for the Next Generation (JPHC-NEXT), launched in 2011 (4). JPHC-NEXT is population-based cohort study being conducted in seven prefectural areas all over Japan. Self-administered questionnaires about lifestyle were provided to all residents aged 40 to 74 at the time of the baseline survey. The questionnaire asked female participants if they had ever breastfed their child, and if they answered yes, the duration. Table 1 shows the proportion of women who were breastfeeding in 2005.

Table 1. Staining Intensity of ALM and SSM lesions.

Age at exposure (2005)	Proportion of ever breastfeeding (%)
	Ever (%)	Never (%)
0 - 4	0	100
5 - 9	0	100
10 - 14	0	100
15 - 19	0	100
20 - 24	94	6
25 - 29	94	6
30 - 34	94	6
35 - 39	94	6
40 - 44	94.4	5.6
45 - 49	93.5	6.5
50 - 54	91.2	8.8
55 - 59	90.4	9.6
60 - 64	90.6	9.4
65 - 69	92.1	7.9
70 - 74	92.1	7.9
≥ 75	92.1	7.9

Cancer incidence and mortality in 2015

Cancer incidence in 2015 was estimated using the annual estimate of cancer incidence in 2013 by the Monitoring of Cancer Incidence in Japan (5). Estimation was done using an age and period spline model, a type of analysis which is used for short-term projection of cancer incidence in Japan (6). The sex-and age-specific incidence data for target cancers were coded in accordance with the International Statistical Classification of Diseases and Related Health Problems, 10^th edition (ICD-10), using the morphology code of the International Classification of Disease for Oncology, 3^rd edition (ICD-O-3).

The data on cancer mortality from 2015 were based on the vital statistics of Japan (7). We obtained sex-and age-specific mortality data by cause of death from an available data source from the Health, Labour, and Welfare Statistics Association (8). Similar to the cancer incidence data, 4-digit ICD-10 codes we used to classify the cause of death.

Estimates of relative risk of history of breastfeeding

Table 2 shows a summary of relative risk estimates used in the present estimate, in which relative risk (RR) was measured compared to women who had no history of breastfeeding. The estimates, derived from the studies listed, were adjusted for potential major confounders. For breast (9) and ovarian cancer (10), the RRs were derived from a Japanese-based cohort study among women who were parous. For the risk of endometrial cancer, the RR was derived from a meta-analysis consisting of 15 studies, including three Japanese studies (11).

Table 2. Summary of risk estimate of site-specific cancers associated with breastfeeding for the present analysis.

Factors	Cancer type	Studies	Reference group	Decrease in risk
Breastfeeding	Breast	Iwasaki et al. (2007) (9)	Never	0.86 (0.65 - 1.15)
	Ovarian	Weiderpass et al. (2012) (10)	Never	1.00 (0.5 - 1.90)
	Endometrium	Zhan et al. (2015) (11)	Never	0.88 (0.72 - 1.06)

Since RRs for ever breastfeeding compared with never were estimated for the target cancers, we calculated the reciprocal of each RR to obtain the RR for never breastfeeding versus ever.

Estimation of population attributable fractions (PAFs)

PAF was calculated using the standard formula (12):

Where P refers to the prevalence of never breastfeeding by age. The numbers of attributable cancers were then totalled across age categories, in order to show a percentage of the total number of all cancer incidence and mortality in Japan in 2015.

Results and Discussion

Table 1 shows the proportion of Japanese women with ever and never experience of breastfeeding in 2005 by age group. Among women aged 20 and over, over 90% have breastfed at least once. In comparison to women who had no history of breastfeeding women who breastfed had the following RR: breast 0.86 (0.65-1.15) (9), ovarian 1.00 (0.5-1.90) (10), and 0.88 (0.72-1.06) (11) (Table 2).

The estimated PAF of cancer incidence and mortality in 2015 attributed to never breastfeeding in Japan is summarized in Table 3. Detailed results for each cancer, sex, and age-group are shown for in Tables S1-S2 (online data, https://www.ghmopen.com/site/supplementaldata.html?ID=38). In the Japanese setting, the only breast cancer incidence and mortalities attributable to never breastfeeding were of breast and endometrium, with about 1% of incidence and mortality for each (breast: 1,066 attributable cases, 176 attributable deaths; endometrium: 165 attributable cases, 22 attributable deaths). Overall, in 2015, there were 1,231 cases and 202 deaths attributable to a history of never breastfeeding in 2005. Accordingly, the overall PAF for never breastfeeding was 0.3% for cancer incidence and 0.1% for cancer mortality in Japanese women.

Table 3. Proportion (%) of cancer in 2015 attributable to never breastfeeding in Japan.

Factors	Incidence		Mortality
	Women	Both sexes	Women	Both sexes
Breast (C50)	1.3		1.3
Endometrium (C54)	1.1		1.1
Ovary (C56)	0.0		0.0
Total cancer (C00-C96)	0.3	0.1	0.1	0.1

In this report, we used RRs calculated for cancers in relation to never breastfeeding on the number of cancer incidence and mortality among Japanese women in 2015. We estimated that 1,231 cancer incident cases (0.30% of total cancer incidence) and 202 cancer deaths (0.13% of total cancer mortality) in 2015 could be attributed to never breastfeeding. These estimates are relatively low compared to results from the United Kingdom (UK) (1.5%) (13) but similar to the results from Australia (0.5%) (14).

With regard to breast cancer risk, these UK and Australian studies (13,14) found that 4.7% and 1.7% of cases were attributable to breastfeeding, respectively. These numbers are higher than our present finding (1.3%). This could be attributable to the difference in the proportion of women who have breastfed, as well as in the duration of breastfeeding used for calculation. In the UK (13) and Australia (14), 52-66% and 83% of women had initiated breastfeeding upon the birth of a child in 2000 and 2001, respectively. Compared to these two Western studies, over 90% of all Japanese women included in the study had breastfed. Further, the definition used as a history of breastfeeding differed - while the UK and Australia studies had clear definitions of what was considered breastfeeding, our study defined a history of breastfeeding as a history of ever breastfeeding. The result from Australian study suggested that breastfeeding for less than 12 months is associated with an increased risk of cancer. Given these methodological differences, it is difficult to make direct comparisons between these previous and our present studies.

The protective effect of breastfeeding on breast cancer is not direct, but likely an indirect cause, such as due to lactation. According to the CUP project, the reduction in breast cancer may result from the hormonal influence of the associated period of amenorrhea and infertility (1). A longer lactation period leads to a reduced number of menstrual cycles throughout life, altering the cumulative exposure to sex hormones, which are known risk factor for post-menopausal breast cancer. The exfoliation of breast tissues during lactation, as well as the apoptosis of epithelial cells after the breastfeeding period is over, might eliminate cells with DNA damage and mutations, leading to lower breast cancer risk (1).

A more accurate history of lactation history estimates for risk would allow a better understanding of the impact of possible protective factors of breastfeeding on cancer burden among Japanese women.

Conclusion

Our estimate found an overall 0.3% of total cancer incidence and 0.1% of total cancer mortality in Japanese women was attributable to never breastfeeding. The high prevalence of women who breastfed may explain the lower fraction of cancer attributable among Japanese women compared to studies from other parts of the world.

Funding

This study was supported by JSPS KAKENHI Grant Number 16H05244.

Conflict of Interest

The authors have no conflicts of interest to disclose.