Background
Worldwide, breast cancer is the most common cancer among females, and its incidence is increasing continuously. Ovarian cancer is the third most common gynecological cancer worldwide, next to cervix uteri and corpus uteri cancers, and has the second highest mortality rate among gynecological cancers, following that for cervix uteri cancer. Globally in 2008, breast and ovarian cancers accounted for 26.6 % of all cancers among females [
1]. The main risk factors for breast cancer are reproductive factors such as age at menarche, number of births (parity), age at first birth, lactation (breastfeeding), and age at menopause [
2]. Ovarian cancer is also influenced by reproductive risk factors such as parity, breastfeeding, and oral contraceptive (OC) use [
3]. Each of these reproductive factors is associated with changes in circulating estrogen and progesterone levels and can be controlled by exogenous hormone treatment, such as OC use and menopausal hormone replacement therapy (HRT). Epidemiological studies regarding hormonal factors support the hypothesis that female hormones, particularly exogenous hormones such as those used for OC and HRT purposes, play an important role in the development of breast and ovarian cancers in women [
4,
5].
In Korea, breast and ovarian cancers account for 14.3 % and 2.0 %, respectively, of all female cancers. In Korea, breast cancer is the second most common cancer following thyroid cancer, while ovarian cancer is ranked as 10
th most common among females [
6]. Breast and ovarian cancer incidences have been continuously increasing over the past 20 years [
7], and the average annual percentage increases in breast and ovarian cancers are 6.3 % and 1.6 %, respectively [
6]. Increases in the prevalence of breast and ovarian cancers have been linked to rapid changes in reproductive factors, including age at menarche, menopause, parity, and birth-related characteristics (i.e., age at first birth, number of births, and breastfeeding), as well as a rapidly ageing population this country [
8‐
10]. Particularly in Korea, rapid development and economic growth since the 1950 Korean War have given rise to marked westernization, leading to rapid changes in the reproductive risk factors of cancers. In 2005, Korea implemented a nationwide breast cancer screening program, which may be a contributor to the increased prevalence of breast cancer cases observed in this country.
The epidemiological patterns of breast and ovarian cancers and their risk factors in Korean women may require the development of population-specific strategies for cancer prevention. The determination of population attributable risks (PARs), defined as the quantified contribution of each risk factor to a disease, can help policymakers establish appropriate public health interventions [
11].
The purpose of this study was to estimate the burden of reproductive risk factors on the prevalence of breast and ovarian cancer in Korea using Korean-specific risk estimates. Since breast and ovarian cancers are major female cancers in the Korean population, such population-specific prevalence and risk estimates should help in the development of cancer control plans in Korea.
Results
The RRs and prevalences for breast and ovarian cancer applied in the current study along with the data sources are summarized in Table
1. A later first pregnancy age showed a higher risk for breast cancer, and those whose total period of breastfeeding was ≤6 months showed an increased breast cancer risk (RR = 1.28 [95 % CI 1.07–1.53]) compared with females with a breastfeeding period of ≥7 months. The use of OC was associated with a 1.31-fold higher risk for breast cancer in the pooled analysis (95 % CI 1.04–1.64); however, HRT use was not significantly associated with breast cancer (RR = 1.16 [95 % CI 0.36–3.78]). Nulliparous women had a 1.42-fold higher risk for ovarian cancer (95 % CI 1.31–1.54). Women who did not have breastfeeding experience had an increased risk of ovarian cancer (RR = 1.17 [95 % CI 1.02–1.33]). Female who did not undergo tubal ligation had an increased risk of ovarian cancer (RR = 1.44 [95 % CI 1.33–1.56]) and those who did not have experience to take OC also had an increased risk (RR = 1.87 [95 % CI 0.89–3.94]).
Table 1
Summary of relative risks and prevalence (%) of exposure to modifiable reproductive factors in Korean women and data source
Breast cancer | | | | | | |
Pregnancy/age at first birth | Nulliparous | 1.06 (0.82-1.36) | Case–control study | SeBCS | 3 | KNHANES, 2005 |
| ≤23 years | 1.00 | | | 36 | |
| 24 – 30 year | 1.13 (0.98-1.32) | | | 57 | |
| ≥31 year | 1.27 (0.98-1.66) | | | 4 | |
Total period of breast feeding | Never | 1.03 (0.87-1.21) | Case–control study | SeBCS | 17 (18b) | KNHANES, 2005 |
| ≤6 months | 1.28 (1.07-1.53) | | | 9 (10b) | |
| ≥7 months | 1.00 | | | 70 (72b) | |
Oral contraceptive use | Never | 1.00 | Meta-anlaysis | | 82 | KNHANES, 2005 |
| Ever | 1.31 (1.04-1.64) | | | 18 | |
Hormone replacement therapy use | Never | 1.00 | | | 30 (95c) | KNHANES, 2005 |
| Ever | 1.16 (0.36-3.78) | Meta-anlaysis | | 2 (5c) | |
Ovarian cancer | | | | | | |
Pregnancy | Nulliparous | 1.42 (1.31-1.54) | Meta-anlaysis | | 3 | KNHANES, 2005 |
| Parous | 1.00 | | | 97 | |
Breast feeding experience | Never | 1.17 (1.02-1.33) | Meta-anlaysis | | 17 (18b) | KNHANES, 2005 |
| Ever | 1.00 | | | 79(82b) | |
Tubal ligation | No | 1.44 (1.33-1.56) | Meta-anlaysis | | 75 (77b) | Ko-EVEd, 2009-2011 |
| Yes | 1.00 | | | 22 (23b) | |
Oral contraceptive use | Never | 1.87 (0.89-3.94) | Meta-anlaysis | | 82 | KNHANES, 2005 |
| Ever | 1.00 | | | 18 | |
The PARs and numbers of breast and ovarian cancer incidences due to modifiable reproductive factors among females aged 20 years and older for the year 2010 in Korea are presented in Table
2. Pregnancy/age at first birth was the most important modifiable reproductive factor for breast cancer (PAR = 8.0 %), followed by OC use, total period of breastfeeding, and HRT use, which were attributed to 5.3 %, 3.1 %, and 0.3 %, respectively, of breast cancer incidences. The PAR for the selected modifiable reproductive factors was 16.7 % (95 % CI 15.8–17.6) for breast cancer, and those factors were responsible for 2,404 (95 % CI 2,283–2,535) breast cancer cases in the year of 2010.
Table 2
The population attributable risks and estimated number of new cancer cases in Korean women caused by modifiable reproductive factors in the year 2010
Breast | Pregnancy/age at first birth | 8.0 (7.4-8.6) | 1152 (1066–1246) |
| (Nullipara or age at first birth ≥ 24 years) |
| Total period of breast feedinga | 3.1 (3.1-3.2) | 453 (441–464) |
(Duration of breast feeding ≤ 6 months) |
| Oral contraceptive use (Ever) | 5.3 (5.1-5.5) | 763 (732–795) |
| Hormone replacement therapy use (Ever)b | 0.3 (0.2-0.3) | 37 (29–44) |
| Total | 16.7 (15.8-17.6) | 2404 (2283–2535) |
Ovary | Pregnancy (Nulliparous) | 1.2 (1.2-1.2) | 24 (24–24) |
| Breast feeding experience (Never)a | 2.9 (2.8-2.9) | 55 (55–56) |
| Tubal ligation (No)a | 24.5 (23.0-26.2) | 473 (444–504) |
| Oral contraceptive use (Never) | 53.3 (27.9-100.0d) | 1027 (537–1927) |
| Total | 81.9 (55.0-100.0 d) | 1579 (1059–1927) |
| | | Attributed % of cases |
| % of breast cancer in women | | 16.7 (15.8-17.6) |
| % of ovary cancer in women | | 81.9 (55.0-100.0d) |
For ovarian cancer, OC use and tubal ligation were the most important modifiable reproductive factors (PAR = 53.3 % [95 % CI 27.9-100.0] and PAR = 24.5 % [95 % CI 23.0–26.2], respectively), whereas breast feeding and pregnancy were attributed to 2.9 % and 1.2 %, respectively. The PAR for the selected modifiable reproductive factors was 81.9 % (95 % CI 55.0–100.0) of ovarian cancer cases and they were responsible for 1579 (95 % CI 1059–1927) ovarian cancer incidences in the year of 2010 in Korea.
Discussion
In Korea in the year 2010, 16.7 % of breast cancer and 81.9 % of ovarian cancer cases in women were attributable to modifiable reproductive factors. The modifiable reproductive factors included pregnancy/age at first birth (8.0 %), total period of breastfeeding (3.1 %), OC use (5.3 %), and HRT use (0.3 %) for breast cancer and included pregnancy (1.2 %), breastfeeding (2.9 %), tubal ligation (24.5 %), and OC never use (53.3 %) for ovarian cancer.
Several recent studies have reported PARs of reproductive factors for breast cancer. In one study, a combination of parity number and age at first birth explained 17.9 % of breast cancers [
48], a percentage higher than that in our results for age at first birth only (8.4 %). However, that study’s overall PAR of breast cancer attributable to reproductive factors was similar to ours. Barnes et al. included the most reported risk factors for breast cancer including modifiable and non-modifiable factors in their PAR calculation and showed that about 50 % of breast cancers in post-menopausal women were attributed to hormone and reproductive factors such as age at menarche (7.7 %), age at menopause (12.0 %), parity (10.9 %), and HRT use (19.4 %) [
49]. Parkin et al. considered only breastfeeding and attributed it to 3.2 % of female breast cancers [
50]. A study from China showed that 6.7 % of breast cancers in women aged 15–49 years were attributed to reproductive factors, which included parity, number of children, age at first birth, and breastfeeding, and 0.7 % and 0.3 % were attributed to OC use and HRT use, respectively [
28].
Among the assessed modifiable reproductive factors, large differences in PARs between studies were observed in HRT use. In French women, HRT use was associated with 12.7 % of breast cancer cases and 10.2 % of breast cancer deaths [
27], whereas HRT use was attributed to 19.4 % [
49], 3.2 % [
50], 0.3 % [
28], 4.4 % [
51], and 2.4 % [
52] of female breast cancer in Germany, the United Kingdom, China, the United States, and Japan, respectively. In the Korean population, the PAR of HRT was very low (0.3 %), similar to that reported for China [
28]. The low level in Korea may be because HRT use is not common in Korea (5 % in post-menopausal women) and its RR is low.
Regarding the classification of risk factors, parity was considered a non-modifiable factor in the work of Barnes et al., but was assessed as a modifiable factor in this study. Barnes et al. restricted their PAR calculation to post-menopausal women, and parity-related factors could thus be considered as non-modifiable [
49]. In contrast, the present study included all women aged 20 or older, and parity, pregnancy/age at first birth, and total period of breastfeeding were thus classified as modifiable factors. The International Agency for Research on Cancer and France working group estimated PAR changes from 1980 to 2000 by assessing changes in reproductive factors including parity (nulliparous vs. parous), mean number of children, age at first birth, and breastfeeding duration, and showed that changes in reproductive factors over those 20 years were associated with 6.7 % and 0.38 % of breast and ovarian cancers increases, respectively [
27]. Those results indicate that such factors can produce temporal changes in cancer incidence. Korea has an experience in decreasing fertility rate fast (from 6.0 births per woman in 1960 to 1.08 in 2005 and to 1.23 in 2010) through national birth control program as part of a population control policies began in 1958. Although the decreasing fertility rate is common phenomenon worldwide, the speed of decrement is one of the fast and now fertility rate in Korea is the lowest in the world [
22]. So, Korean government has changed family planning policy to childbirth encouragement and tries to increase the birth rate intensively with many benefits to families. Considering that the PAR can help policy makers establish appropriate public health interventions and efforts to control the birth rate in Korea for about past 60 years, including pregnancy/age at first birth as modifiable factors would helpful for policymakers not only in Korea but also other countries which have family planning policy to support their recommendation about childbirth. Thus, under this new policy, the encouragement of childbirth, particularly at an early age, might reduce breast and ovarian cancer incidences and deaths.
With regard to the PAR in ovarian cancer, Granstrom et al. reported that the PARs for family history, parity/age at first birth, and residential area were 2.6 %, 22.3 %, and 7.2 %, respectively [
53]. Parazzini et al. included more risk factors, and the PARs were 4 % for a family history of breast and ovarian cancer, 8 % for age at menopause, 5 % for parity, 12 % for OC use, 7 % for high fat intake score, and 24 % for low green vegetable intake [
54]. Parkin et al. considered only HRT use as a reproductive risk factor for ovarian cancer and attributed 0.7 % of the ovarian cancer incidence to HRT. The OC use is a protective factor for ovarian cancer [
55] and in Korea the PAR of OC was higher than other studies because prevalence OC use was very low (18 %). As a contraceptive method, tubal ligation was associated with a lower risk of ovarian cancer and, during the era of encouraging birth control in Korea, the most common artificial sterilization method was tubal ligation. Therefore, the PAR of tubal ligation for ovarian cancer was also higher.
There are several study limitations to be considered. Although the standardized population used in this study was from 1990, prevalence of most reproductive factors in 2005 was used because of the lack of representative data for 1990. Considering the increased age at first birth and the decreased fertility rate between 1990 and 2005, our results might underestimate the PAR for reproductive factors in Korea. In addition, the prevalence of tubal ligation showed limited representativeness because it was estimated by using control subjects from a hospital-based case–control study (Ko-Eve). Thus, we compared our estimates of the prevalence of tubal ligation with results from the National Survey on Fertility, Family Health & Welfare in Korea, a nationwide representative survey for females of childbearing age (15–44 years) [
56], and the results were comparable to ours (18.3 % vs. 23 %, respectively). Moreover, our estimates of RRs were based on a limited number of studies, which may have introduced uncertainty in the pooled RR estimate and, hence, uncertainty in the calculated PARs. Although several review articles and meta-analysis studies have reported stable results for RR estimates of breast and ovarian cancer risk factors, in this study, we used Korean-specific results. We considered ethnicity- or country-specific risk estimates, distributions, and their effects on PAF estimates. In addition, while we did not consider the quality of each study included in our RR estimation, Korean studies included in the breast/ovarian cancer RR estimation were community based case–control studies. For ovarian cancer, we pooled results from previous international studies with those from a Ko-Eve. Therefore, the RR values might be less appropriate for the Korean female population. However, due to the lack of Korean data and followed by instable results for ovarian cancer, pooled estimates with previous international studies would be unavoidable as Shin et al. did in the previous study [
26]. The cut-off points were arbitrary and identified for convenience in our meta-analysis.
Despite these limitations, this study has several strengths. First, we used nationwide cancer incidence data that are representative of nearly the entire population. Thus, we had access to precise numbers of cancer cases for inclusion in our PAR estimation. Second, the estimated prevalence of exposure to each risk factor in 1990 was used in the consideration of a 20-year lag period between exposure to a risk factor and subsequent cancer development. Although we did not measure the quality of each study, the studies included in the meta-analysis for breast cancer were conducted within the Korean population, thus providing Korean-specific results.
Acknowledgments
The study is part of a systematic analysis of attributable causes of cancer in Korea conducted by working group experts in collaboration with the National Cancer Center, Korea and the International Agency for Research on Cancer. This study was supported by a research grant from the National Cancer Center, Korea (NCC-0710160) and a grant from the National R&D Program for Cancer Control, Ministry for Health, Welfare and Family affairs, Republic of Korea (1420190).
We thank Mathieu Boniol from the International Prevention Research Institute, Lyon, France and Paolo Boffetta from the Tisch Cancer Institute, Mount Sinai School of Medicine, New York, United States of America for their help while they were working at the International Agency for Research on Cancer, Lyon, France.
Competing interests
The authors declare that they have no competing interests.
Authors’ contributions
B P analyzed data and has been involved in drafting manuscript. S P, Y Y and J-Y C has been involved in analyzing the data. B-G K and Y-M K has been involved in drafting manuscript. S P, H-R S, A S, K-W J and SK P have made substantial contributions to conception and design of the study and to interpretation of the data from the public sector’s perspectives. D-Y N, S-H A, DK, and K-Y Y have made substantial contributions to acquisition and interpretation of the breast cancer related data from breast cancer surgeons’ and epidemiologists’ perspectives. JWK, SK, JHK, and TJK have made substantial contributions to acquisition and interpretation of the ovarian cancer related data from obstetricians’ and gynecologists’ perspectives. SKP has been involved in drafting manuscript and given final approval of the version to be published. All authors read and approve the final manuscript.