Effect of lifetime lactation on breast cancer risk: A Korean Women's Cohort Study

Authors


Abstract

The objective of our study was to examine the effect of lifetime lactation on breast cancer risk among premenopausal women. The data were from a prospective cohort study with a follow-up period of 6 years in Korea (1995–2000). The cohort was composed of 110,604 premenopausal parous Korean women, aged 20 years and older, who received health insurance from the Korea Medical Insurance Corporation and who had medical evaluations in 1992 and 1994. Multivariate Cox proportional hazard models were tested, controlling for age, age at menarche, number of children, age at first pregnancy, oral contraceptive use, smoking, exercise and obesity. At baseline, 57,440 (51.9%) reported breastfeeding and 4,584 (4.1%) reported breastfeeding more than 24 months. From 1995–2000, 360 incident cases of breast cancer (61.8/100,000 person-years) occurred. Compared to parous women who had no history of lactation, a period of lactation of 13–24 months decreased the risk of breast cancer (RR, 0.7; 95% CI, 0.5–1.1), and this risk was decreased even further for those who breastfed for more than 24 months (RR, 0.6; 95% CI, 0.3–1.0). There was a clear trend of decreasing breast cancer risk with the duration of lactation (p for trend <0.001). In conclusion, our study of a large Korean cohort provides additional empirical evidence to current theoretical conjecture that lactation decreases the risk of breast cancer among premenopausal women. © 2003 Wiley-Liss, Inc.

Although the protective effect of lactation on breast cancer has been discussed among researchers for more than 20 years, no study ever found a detrimental effect of breastfeeding on risk of breast cancer. For example, an inverse association between lifetime lactation and breast cancer risk among premenopausal women has been found in some case-control studies,1, 2, 3, 4, 5, 6, 7, 8 whereas other studies show no association.9, 10, 11, 12, 13 It is noteworthy that studies from the countries in which the prevalence of prolonged lactation is high, such as China6, 7 and Japan,13, 14 have consistently shown an association between lactation and a reduced risk of breast cancer. On the contrary, studies of Western populations, in which the duration of lactation is relatively short, have provided inconsistent results regarding the relation between breastfeeding and the risk of breast cancer.10, 11, 15, 16

These findings from case-control studies are also evident in cohort studies: Six cohort studies17, 18, 19, 20, 21, 22 to date have attempted to clarify the association between lifetime lactation and a reduced risk of breast cancer. Neither the first cohort study for premenopausal women in Norway17 nor 2 studies from the Nurses' Health Study for U.S. women18, 19 showed the protective effect of lactation on breast cancer risk. However, the most recent nested case-control studies for an Iceland cohort20 and a Connecticut cohort21 reported a negative association between breastfeeding and breast cancer risk in both pre- and postmenopausal women. In particular, a significant reduction in breast cancer risk was observed among the women who had breastfed for more than 36 months in a Shanghai Textile Workers' cohort study, although a significant reduction in risk was not observed in women who had breastfed for fewer than 36 months.22

Despite the several previous researches that uncovered the relationships between various aspects of lactation and breast cancer risk, gaps and inconsistencies remain. Given the finding that lactation is a potentially modifiable risk factor for breast cancer, which is a disease with great public health concerns for women of all ages and races, rigorous investigations are warranted to clarify the inconsistent conclusions from previous studies of lactation and breast cancer.

Our study is the seventh reported cohort study to examine prospectively the relation between lactation and breast cancer. Our study used data gathering from 110,604 premenopausal Korean women to clarify the potential beneficial effect of lactation on the risk of breast cancer among premenopausal parous women.

MATERIAL AND METHODS

The Korea Medical Insurance Corporation (KMIC) is the largest Korean insurance group that provides health insurance to government employees, teachers and their family members. Of the entire Korean population in 1992 (approximately 44 million), 4,662,438 (10.7%) were insured by KMIC, including 1,297,833 workers and their 3,364,605 family members. All insured workers are encouraged to participate in biennial health maintenance physicals performed by the KMIC.23 In 1992 and 1994, 95% and 94% completed biennial medical examinations.

The Korean Women's Cohort (KWC) Study is an ongoing prospective cohort study designed to assess the effects of gender-related variables on chronic diseases in Korean women using the KMIC sample. The KWC Study population consists of 199,521 women aged 20 years and older who worked as government employees or teachers in 1994. Because only 5.7% (n = 14,166) of the women in this cohort were postmenopausal in 1994, we restricted our analyses to premenopausal women (n = 185,355) in our study. We excluded nulliparous women, who accounted for 73,900 (39.9%) of the 185,355; 383 women (0.2%) who had no information available concerning lactation; and 468 women (0.2%) who had any cancer diagnosis in 1994, yielding a final sample size of 110,604.

Data collection

KMIC biennial health maintenance physical exams are conducted in a standardized fashion by medical staff at the designated local hospitals. In the 1994 data collection, a standardized, structured questionnaire was used to obtain information on past and present illness and menstrual and reproductive factors. Respondents were asked in the questionnaire about their reproductive history: whether they had ever been pregnant, how many live births they had had, at what age they had given birth to each child, whether they had breastfed any or all of their children, and whether they had ever taken oral contraceptives. To examine the duration of breastfeeding, we used self-reported questionnaires, which included lactation records of a maximum of 5 children per mother. Duration of breastfeeding was from summation of lactation duration for each child per mother. Lifestyle-related factors for smoking habits, number of cigarettes smoked per day, duration of cigarette smoking in years, amount and frequency of alcohol consumed and exercise were also addressed. Completed questionnaires were reviewed and trained staff entered the data.

The 1992 and 1994 biennial physical examination included obtaining weight and height measurements. Body mass index (BMI) was calculated as weight/height2 (kg/m2). In our analysis, BMI was the average of measurements collected in 1992 and 1994.

The principal outcome variable was morbidity and mortality from breast cancer (International Classification of Disease, 10th Revision [ICD-10] codes C50). The follow-up period, which lasted 6 years, was from January 1, 1995–December 31, 2000. Outcomes were ascertained from diagnoses on hospital discharge summaries, from the National Cancer Registry and from causes of death on death certificates. In Korea, the data of diagnosis, which are documented by certified personnel in a standardized fashion using a discharge summary note for the inpatients, are automatically gathered into KMIC. In a random sample (n = 54) of breast cancers that occurred in KMIC enrollees, we were able to confirm 96% of the breast cancer diagnoses by extensive medical chart review.24 Computerized searches of the National Cancer Registry data from National Cancer Center and death certificate data from the National Statistical Office in Korea were performed on each of the KMIC-insured women. The system allowed near-perfect availability of follow-up mortality data for participants in our study.

Statistical analysis

In bivariate analyses, we examined the relation between lactation and traditional breast cancer risk factors, adjusting for age. In these bivariate analyses, we tested for trends across categories of duration of lactation, using parous women who never breastfed as the reference. In Cox proportional hazard models, age, age at menarche, number of children, age at first pregnancy, oral contraceptives, smoking, exercise and obesity were controlled to delineate the independent effects of lifetime lactation. In all analyses, a 2-sided alpha level of 0.05 was considered statistically significant.

RESULTS

At baseline, the mean (S.D.) age of the study participants was 35.2 (± 6.0) years. Among our cohort of 110,604 participants, 57,440 (51.9%) of these women reported breastfeeding. Seventy-nine percent of 57,440 women who breastfed reported breastfeeding from 1–12 months, 13% from 13–24 months and 8% more than 24 months.

During the 6 years of follow-up (582,352 person-years), the 360 incident breast cancers were developed from 1995–2000. Among 360 incident breast cancer cases, there were 24 deaths. Characteristics of parous women in the cohort according to their duration of lifetime lactation are presented in Table I. After adjustment for age, duration of lactation was associated with BMI (p < 0.001), number of children (p < 0.001), age at first pregnancy (p < 0.001), age at menarche (p = 0.043), a rate of taking oral contraceptives (p = 0.015) and smoking (p < 0.001).

Table I. Baseline Characteristics of 110,604 Parous Women According to Breastfeeding History1: The Korean Women's Cohort Study, 1992–2000
CharacteristicDuration of lactationp for trend
None (n = 53,164)1-12 months (n = 45,591)13-24 months (n = 7,265)>24 months (n = 4,584)
  • 1

    Data are expressed as mean (S.D.) unless otherwise indicated. Except for age, all values were age-adjusted

  • 2

    Menarche at age <14 years;

  • 3

    Consumption of any alcohol

  • 4

    For exercise, dichotomized as yes or no.

Age33.4 (5.4)34.9 (5.7)40.7 (5.7)43.6 (5.8)<0.001
Body mass index, kg/m221.1 (2.2)21.2 (2.1)22.3 (2.4)23.0 (2.6)<0.001
No. of children1.4 (0.7)1.7 (0.6)2.1 (0.5)2.7 (0.8)<0.001
Age at first pregnancy27.9 (2.4)27.5 (2.2)26.3 (2.3)25.3 (2.6)<0.001
Conditions (%)     
 Earlier menarche246.844.635.825.70.043
 Oral contraceptive use19.921.128.333.70.015
 Ever smoke0.40.40.50.9<0.001
 Alcohol use310.59.710.211.30.130
 Exercise (yes)49.812.318.719.20.114

Compared to parous women who had no history of lactation, periods of lactation of 1–12 months or 13–24 months were associated with decreased risk of breast cancer (RR, 0.8; 95% CI, 0.7–1.0 or RR, 0.7; 95% CI, 0.5–1.1, respectively), and this risk decreased even further for those with who breastfed for more than 24 months (RR, 0.6; 95% CI, 0.3–1.0) (Table II).

Table II. Risk of Breast Cancer Among Parous Women: The Korean Women's Cohort Study, 1992–2000
 Person-yearsBreast cancerAge-adjusted modelMultivariate model1
RR95% CIp-valueRR95% CIp-value
  • 1

    Relative risk (RR) and 95% confidence interval (CI) from multivariate Cox proportional hazard models. All factors were simultaneously included in one model.

  • 2

    Due to missing values, the sum of person-years does not add up to the total 582,352 person-years.

Age in 5 years (1-year increase)  1.11.1–1.10.00011.11.1–1.10.0001
Breastfeeding (months)        
 Never263,4721611.0  1.0  
 1–12256,1991490.80.7–1.00.09610.80.7–1.00.1184
 13–2439,125320.70.4–1.00.03710.70.5–1.10.1042
 >2423,556180.50.3–0.80.00340.60.3–1.00.0438
 p for trend    <0.001  <0.001
Age at menarche (years)        
 <14465,340264   1.0  
 ≥14117,01296   0.80.7–1.00.0596
No. of children        
 1227,95395   1.0  
 2309,641224   1.31.0–1.60.0883
 ≥344,75841   1.10.7–1.70.7812
Age at first pregnancy (years)        
 <2699,96855   1.0  
 26–28312,409170   1.20.8–1.60.3858
 ≥29169,975135   1.61.1–2.20.0092
Oral contraceptive use2        
 No458,179286   1.0  
 Yes49,90631   0.80.6–1.00.0651
Smoking        
 Nonsmoker580,244357   1.0  
 Ever smoker2,1083   2.20.7–6.80.1773
Exercise        
 No89,086292   1.0  
 Yes12,11056   1.00.8–1.50.5667
Body mass index (kg/m2)        
 <23258,088168   1.0  
 ≥23324,264192   1.00.8–1.30.7575

There was strong indication that lifetime duration of breastfeeding was inversely associated with the risk of breast cancer among parous women (p for trend < 0.001). In further analysis, lifetime duration of breastfeeding was also inversely associated with the risk of breast cancer after excluding younger women aged less than 30 years.

Later menarche aged 14 years or older was also inversely associated with the risk of breast cancer (RR, 0.8; 95% CI, 0.7–1.0). Late age at first birth increased the risk of breast cancer. When the first birth occurred after age 29 years, women had an approximately 60% increase in the risk of breast cancer compared to women who had their first child at age 26 years or younger (RR, 1.6; 95% CI, 1.1–2.2).

DISCUSSION

In this prospective, observational cohort study of a large sample of Korean women, we confirmed that lactation is associated with a significant reduced risk of breast cancer. The results of our study suggest a significant and clear inverse relation between lifetime lactation and breast cancer risk among premenopausal parous Korean women.

The important finding of our study is that breastfeeding for about 2 years is associated with a significantly reduced risk of breast cancer. This has been observed in some other cohort studies of lactation.20, 21, 22 It has also generally been accepted that the major factor contributing to the inconsistent findings of earlier studies could be the relatively shorter duration of lactation in Western populations.16, 21, 25 The majority of study subjects from Western populations had 2 or fewer live births, and the average duration of lactation was 4 months per child.25 For example, only 17% of the women study subjects from Western populations breastfed for at least 12 months.19 In a Chinese study population, on the other hand, half of the women breastfed for at least 3 years.25 In the case of our study population, the average number of births was 1.7, and the average duration of lactation was 8.6 months over the course of a lifetime. Among those reporting at least 1 live birth, 24% had breastfed for at least 12 months. In other words, the overall figures for reproductive statistics are closer to those for Western populations than those previously reported in the Japanese or Chinese studies. Therefore, it is noteworthy that our study found a significant inverse association between lactation and breast cancer, despite the fact that the average duration of lactation of our study participants was much shorter than that of study subjects from Japan26 or China.6, 7, 13

The findings on other variables such as age at menarche and age at first pregnancy are consistent with previous reports, which also point out the quality of our study. In our study, women whose menarche started before age 14 years had an increased risk of breast cancer when compared to women whose menarche started at or after 14 years of age. Our study also showed that early age at first birth is a protective factor, which is consistent with previous studies.27

In our study, the average age of participants was 35.3 years. Because some of these women are still in their childbearing years, there is potential risk of measurement errors associated with underestimation of lifetime lactation period. However, this error probably only attenuated the strength but not the direction of the association because data was collected in the same fashion for the whole cohort. We also conducted the multivariate analysis using only the subsample of women older than 30 years who were less likely to give additional births than the women in their 20s. The same result was obtained (data not shown), which supports that the effect of underestimation is negligible. Another limitation of our study was the relatively brief duration of follow-up. We believe, however, that the large size of the cohort (>580,000 person-years) provided sufficient statistical power to overcome the limitation of a short-term follow-up period.

One of the strengths of our cohort study was a large sample (>110,000) that allowed us to have enough statistical power to examine the effect of lactation on breast cancer after controlling for all other potential covariates. Computerized searches of death certificate data from the National Statistical Office and admission data from the National Health Insurance Corporation in Korea were performed on each of the KMIC-insured women. The system allowed near-perfect availability of follow-up mortality and admission data for participants of our study.

In summary, our prospective study provides strong empirical evidence that lactation reduces the risk of breast cancer.

Acknowledgements

The authors appreciate the assistance of the National Health Insurance Corporation, which provided the data for our study. We thank Dr. S.H. Jee from Yonsei University for his expert advice during statistical modeling.

Ancillary