Reproductive factors and breast cancer risk in relation to hormone receptor and menopausal status in Japanese women

Abstract

The associations between menstrual and reproductive factors and breast cancer risk in relation to estrogen/progesterone receptor (ER/PgR) status have been unclear in Japanese women. This case‐control study evaluated these associations, overall and separately, by menopausal status. A total of 1092 breast cancer cases and 3160 controls were selected from among female patients aged 30 years and over admitted to a single hospital in Miyagi Prefecture between 1997 and 2009. The receptor status distribution among the cases (missing: 8.4%) was 571 ER+/PgR+, 133 ER+/PgR−, 24 ER−/PgR+ and 271 ER−/PgR−. Menstrual and reproductive factors were assessed using a self‐administered questionnaire. Polytomous logistic regression and tests for heterogeneity across ER+/PgR+ and ER−/PgR− were conducted. Later age at menarche was significantly associated with a decreased risk of both ER+/PgR+ and ER−/PgR− cancer among women overall (P _trend = 0.0016 for ER+/PgR+; P _trend = 0.015 for ER−/PgR−) and among postmenopausal women (P _trend = 0.012 for ER+/PgR+; P _trend = 0.0056 for ER−/PgR−). Nulliparity was associated with an increased risk of ER+/PgR+, but not ER−/PgR− cancer among women overall (P _{heterogeneity} = 0.019) and among postmenopausal women (odds ratio for ER+/PgR+ = 2.56, 95% confidence interval = 1.61–4.07; P _{heterogeneity} = 0.0095). A longer duration of breastfeeding tended to be associated with a decreased risk in all subtypes among women overall. Later age at menarche has a protective effect against both ER+/PgR+ and ER−/PgR− cancer. However, parity might impact differently on various subtypes of breast cancer. Further studies are needed to clarify the etiology of the rare ER+/PgR− and ER−/PgR+ cancer subtypes.

Over the past few decades, numerous epidemiologic studies of breast cancer have been conducted, based mainly on Caucasian populations. These studies show that menstrual and reproductive factors and menopausal status are associated with breast cancer risk.1, 2 In Japan, cohort studies,3, 4 case‐control studies5, 6, 7 and a meta‐analysis have revealed similar associations.8

Breast cancers are known to express the estrogen receptor (ER) or progesterone receptor (PgR). Tumor subtypes defined by these receptors represent biologically different entities.9 In Western countries, many studies have evaluated breast cancer risk according to hormone receptor status.10, 11, 12, 13 A meta‐analysis shows that nulliparity is associated with a higher risk of ER+ tumors, but not ER− tumors.13 Another meta‐analysis suggests that nulliparity is associated with an increased risk of ER+/PgR+ tumors, but not ER−/PgR− tumors. The protective effects of late age at menarche and longer duration of breastfeeding do not differ across ER/PgR status.12

Among studies conducted in the Asian region, a large‐scale case‐control study from China evaluates risk factors defined according to the four types of hormone receptor status and finds an association with parity history similar to that in the abovementioned meta‐analysis.14 Although a few epidemiologic studies focus on the hormone receptor status of breast cancer in Japan,15, 16, 17 their results are inconsistent. One study shows that parity, the number of births and age at menarche have different associations with the risk of breast cancer according to ER and PR status.15 Another study shows that only age at menarche is differently associated with the risk of breast cancer according to ER status.16 A third study shows no gradient in the risk associated with reproductive factors,17 including age at menarche, age at menopause, age at first birth, parity number and duration of breastfeeding. In most of the Japanese studies, however, hormone receptor data are incomplete, and, therefore, the percentage of breast cancer cases for which the hormone receptor status is unknown is relatively large. Consequently, the sample sizes might have been too small to allow comprehensive evaluation of breast cancer risk according to hormone receptor status. The inconsistencies among the results obtained in these Japanese studies are likely attributable to such limitations.

Therefore, we conducted a hospital‐based case‐control study to precisely evaluate the association between reproductive factors and breast cancer risk according to hormone receptor status. Data were obtained from women aged 30 years and over who were admitted to a single hospital in Miyagi Prefecture, Japan. Analyses were performed based on joint ER and PR status; that is, ER+/PgR+, ER+/PgR−, ER−/PgR+ and ER−/PgR−. In this study, data on hormone receptor status were available for over 90% of the breast cancer cases included.

Methods

Data collection

In January 1997, we began a questionnaire survey in connection with the present study. Information on lifestyle and personal history was collected from all patients at their first admission to the Miyagi Cancer Center Hospital (MCCH) using a self‐administered questionnaire. The questionnaire was distributed to patients on the day of their reservation for initial admission (i.e. 10–15 days before admission) and collected by nurses on the actual day of admission. The MCCH is located in Natori City, situated in the southern part of Miyagi Prefecture, and functions as a hospital for both cancer and benign disease. Details of the questionnaire survey have been described elsewhere.18, 19, 20, 21

The questionnaire covered demographic characteristics, personal and family histories of cancer and other diseases, including family history of breast cancer in mother or sisters, current height and weight, general lifestyle factors before the development of current symptoms, including cigarette smoking, alcohol drinking, physical activity, occupation, menstrual and reproductive histories, and history of oral contraceptives (OC) and other exogenous female hormone uses. Items related to the referral base were also included. The items related to menstrual and reproductive histories included age at menarche, menopausal status, age at menopause, parity history, parity number, age at first birth, history of breastfeeding, duration of breastfeeding and quantity of milk secretion. A question on the duration of breastfeeding was added after 2000. Between January 1997 and December 2009, the questionnaire was distributed to 23 531 first‐admitted patients, of whom 21 056 responded.

Study subjects

Cases and controls were selected from among patients who responded to the above questionnaire survey. To identify incident cases of female breast cancer, a list of the patients was linked with the hospital‐based cancer registry files. The registry records all cancer cases confirmed by clinical, cytological and/or histopathological examination at the MCCH. Through linkage to the registry, 21 056 patients were classified into 1812 with a past history of cancer, 6848 male patients with cancer, 1096 female patients with breast cancer, 4171 female patients with other cancers, and 7129 non‐cancer patients (3708 male and 3421 female patients). Among the 1096 female patients with breast cancer, 1092 aged 30 years and over were included as the cases for the present study.

Controls were selected from among female non‐cancer patients. Patients with benign tumors were classified as non‐cancer patients for the present study. Accordingly, 3160 female non‐cancer patients aged 30 years and over were identified as controls. The diagnoses among the controls were as follows: benign tumor in 1824 (57.7%), cardiovascular disease in 116 (3.7%), digestive tract disease in 377 (11.9%), respiratory tract disease in 122 (3.9%), urologic‐gynecologic disease in 170 (5.4%), other benign disease in 302 (9.5%) and no abnormal findings in 249 patients (7.9%). The sites of benign tumors were the digestive tract in 637 subjects, gynecologic organs in 375, urologic organs in 17, breast in 36, bone or connective tissue in 545 and other sites in 214. The final response rate in the questionnaire survey was 94.1% for the case group and 89.8% for the control group.

This study was approved by the ethical review board of the Miyagi Cancer Center and was conducted in accordance with the principles specified in the Declaration of Helsinki. We considered the return of self‐administered questionnaires signed by the subjects to imply their consent to participate in the study.

Hormone receptor status

Information on hormone receptor status (i.e. expression of the ER and PgR in breast cancers) was extracted from medical records. In brief, enzyme immunoassays (EIA) were used in the early period of the study to determine hormone receptor status. After mid‐2003, immunohistochemistry (IHC) assays were conducted on tumor tissue samples. The cut‐off point for receptor positivity in the EIA was 14 fmol/mg for ER and 13 fmol/mg for PgR. In the IHC assay, a histology score (HSCORE) of ≥20 for ER and one of ≥6 for PgR were evaluated as positive.22 The concordance between the two assays was 94.3% for ER and 100% for PgR in the laboratory of the MCCH.22 Among the total of 1092 cases, data on joint ER/PgR status were available for 1000 (91.6%); 571 cases were ER+/PgR+, 133 were ER+/PgR−, 24 were ER−/PgR+ and 271 were ER−/PgR−.

Statistical analysis

We used multiple polytomous unconditional logistic regression analysis to estimate odds ratios (OR) and 95% confidence intervals (CI) for hormone receptor‐defined breast cancer risk in relation to menstrual and reproductive factors, family history of breast cancer, use of OC, and use of exogenous female hormones other than OC.

The exposure variables analyzed in the present study were menstrual and reproductive factors (age at menarche, menopausal status, age at menopause, parity, parity number, age at first birth, history of breastfeeding, total duration of breastfeeding and quantity of milk secretion), family history of breast cancer in mother or sisters (yes or no), history of OC use (ever or never) and use of exogenous female hormones other than OC (ever or never). For history of breastfeeding (formula only, mixed breastfeeding and formula, or breastfeeding only), use of formula only was recognized as no history of breastfeeding, and used as a reference. Breastfeeding only and mixed breastfeeding and formula were both regarded as a positive history of breastfeeding.

We considered the following variables to be potential confounders: age, referral base (from screening or other), area of residence (southern Miyagi Prefecture or other), year of recruitment, smoking (ever or never), alcohol drinking (ever or never), occupation (housewife or other), body mass index (BMI) and physical activity (more or less than 1 h per week). BMI was calculated as weight divided by squared height (kg/m²). In the analysis, menstrual and reproductive factors and history of breast cancer in mother or sisters were also adjusted for each other. Missing values for confounders were treated as an additional variable category, and were included in the model.

In the analysis, we stratified case subjects according to joint hormone receptor status. Stratification by menopausal status was also performed. Menopause was defined as the cessation of menstrual periods due to natural or other reasons, including surgery. With regard to menopause due to other reasons, we were unable to obtain any information about history of oophorectomy; therefore, case subjects aged 45–57 years and controls aged 43–57 years (defined as the mean age at natural menopause ±2 SD) were regarded as patients with unknown menopausal status. In the analysis stratified by menopausal status, case subjects who had ER+/PgR− or ER−/PgR+ tumors were too few to allow precise estimation of OR in comparison with subjects who had ER+/PgR+ or ER−/PgR− tumors; therefore, we excluded these subjects from the analysis according to menopausal status.

Dose‐response relationships were tested by treating each exposure category as a continuous variable. We conducted Wald tests for estimating the heterogeneity of breast cancer risk across ER+/PgR+ and ER−/PgR−. Values were considered significant if the two‐sided P were <0.05. All analyses were performed using sas version 9.3 (SAS Institute, Cary, NC, USA).

Results

The background characteristics of the study subjects are presented in Table 1. Among the case subjects included in the analysis (n = 1000), 416 were premenopausal, 555 were postmenopausal and 29 were undefined. Among the premenopausal subjects, 260 (62.5%) were ER+/PgR+, 44 (10.6%) were ER+/PgR−, 12 (2.9%) were ER−/PgR+ and 100 (24.0%) were ER−/PgR−. Among the postmenopausal subjects, 300 (54.1%) were ER+/PgR+, 87 (15.7%) were ER+/PgR−, 11 (2.0%) were ER−/PgR+ and 157 (28.3%) were ER−/PgR−. Among the control subjects (n = 3160), 1081 (34.2%) were premenopausal, 1963 (62.1%) were postmenopausal and 116 (3.7%) were undefined. Cases with ER+/PgR+ tumors tended to be heavier, and were more likely to be referred from screening, to engage in physical activity and to be drinkers. Cases with ER−/PgR− tumors tended to be lighter, and were less likely to be referred from screening, to engage in physical activity, and to be smokers. Cases with unknown ER/PgR status were less likely to be referred from screening in comparison with the other subtypes.

Table 1.

Background characteristics in cases and controls

	All
	Cases					Controls
	Hormone receptor status
	ER+/PgR+	ER+/PgR−	ER−/PgR+	ER−/PgR−	Missing
Total (n)	572	133	24	271	92	3160
Menopausal status (n)a
Premenopausal	260	44	12	100	22	1081
Postmenopausal	300	87	11	157	43	1963
Unknown menopausal status	12	2	1	14	27	116
Age group (years old) (%)
30–39	6.1	3.8	4.2	5.9	10.9	8.4
40–49	25.0	18.8	41.7	23.2	13.0	17.6
50–59	28.5	28.6	20.8	30.6	34.8	22.1
60–69	23.4	28.6	12.5	22.9	20.7	25.3
≥70	17.0	20.3	20.8	17.3	20.7	26.5
Average	57.2	59.2	56.3	57.2	57.7	59.6
SD	12.6	11.7	14.0	12.1	12.9	13.7
BMI (%)
<18.5	4.9	6.0	—	4.8	9.8	5.8
18.5–25	59.4	62.4	62.5	65.3	60.9	63.4
25–30	27.6	26.3	37.5	24.0	22.8	26.0
≥30	8.0	5.3	—	5.5	3.3	4.2
Missing	—	—	—	0.4	3.3	0.7
Average	24.1	23.6	23.9	23.5	23.0	23.5
SD	3.8	3.6	3.1	3.8	3.6	3.6
Year of recruitment (%)
1997–2002	24.7	39.8	58.3	45.4	50.0	54.7
2003–2009	75.3	60.2	41.7	54.6	50.0	45.3
Area of residence (%)
Southern Miyagi Prefecture	82.7	85.0	87.5	83.4	78.3	88.4
Other	17.3	15.0	12.5	16.6	21.7	11.6
Referral base (%)
From screening	21.2	20.3	16.7	13.3	8.7	18.1
Other	78.8	79.7	83.3	86.7	91.3	81.9
Occupation (%)
Housewife	20.1	21.8	25.0	20.7	31.5	21.4
Other	68.2	68.4	54.2	66.4	54.3	61.7
Missing	11.7	9.8	20.8	12.9	14.1	16.9
Physical activity (%)
More than 1 h per week	43.9	43.6	41.7	40.2	41.3	44.9
<1 h per week	50.2	50.4	54.2	51.7	50.0	47.4
Missing	5.9	6.0	4.2	8.1	8.7	7.7
Smoking (%)
Never	79.9	79.7	66.7	81.5	80.4	80.0
Ever	17.7	17.3	20.8	16.2	17.4	15.6
Missing	2.4	3.0	12.5	2.2	2.2	4.4
Alcohol drinking (%)
Never	68.4	78.9	70.8	69.7	75.0	71.3
Ever	28.7	20.3	12.5	26.9	19.6	23.3
Missing	3.0	0.8	16.7	3.3	5.4	5.3

^aMenopause was defined as the cessation of menstrual periods due to natural or other reasons including surgery. BMI, body mass index; ER, estrogen receptor; PgR, progesterone receptor.

Table 2 shows the OR and 95% CI for menstrual and reproductive factors, family history of breast cancer, and exogenous female hormone use according to the four hormone receptor subtypes. A later age at menarche is significantly associated with a decreased risk of ER+/PgR+ (P _trend = 0.0016; OR = 0.61, 95% CI 0.45–0.83 for ≥15 years) and ER−/PgR− (P _trend = 0.015; OR = 0.57, 95% CI 0.38–0.86 for ≥15 years) cancer. Natural menopause (OR = 0.64, 95% CI 0.49–0.84) and menopause due to other reasons (OR = 0.53, 95% CI 0.35–0.80) are associated with a lower risk of ER+/PgR+ cancer in comparison with premenopause. Nulliparity is associated with a higher risk of ER+/PgR+ cancer (OR = 1.30, 95% CI 0.96–1.78; P = 0.094), but not ER−/PgR− cancer (P _{heterogeneity} = 0.019). An older age at first birth is significantly associated with an increased risk of ER+/PgR+ cancer (P _trend = 0.0086; OR = 1.26, 95% CI 1.00–1.59 for ≥25–≤29 years; OR = 1.57, 95% CI 1.08–2.30 for ≥30 years) and ER−/PgR+ cancer (P _trend = 0.009; OR = 9.04, 95% CI 1.92–42.68 for ≥25–≤29 years; OR = 7.80, 95% CI 1.13–54.07 for ≥30 years). Multiparity is associated with a decreased risk of ER−/PgR− cancer (P _trend = 0.045). Breastfeeding only and a good quantity of breast milk secretion are associated with a decreased risk of cancers for all hormone receptor subtypes, but not statistically significantly. Data on duration of breastfeeding were available for 2222 subjects (52.3%). A longer period of breastfeeding is associated with a lower risk of cancers of all subtypes, although the result for ER+/PgR− is not statistically significant (P _trend = 0.013 for ER+/PgR+, P _trend = 0.082 for ER+/PgR−, P _trend = 0.04 for ER−/PgR+ and P _trend = 0.023 for ER−/PgR−). A family history of breast cancer in mother or sisters is significantly associated with an increased risk of all subtypes. The heterogeneity test for a family history of breast cancer reveals a significant difference in risk across ER+/PgR+ and ER−/PgR− tumors (P _{heterogeneity} = 0.044). The use of OC and exogenous female hormones other than OC is not significantly associated with breast cancer risk for any of the subtypes.

Table 2.

OR (95% CI) of breast cancer risk by hormone receptor status associated with risk factors

	Control	ER+/PgR+ (n = 572)				ER+/PgR− (n = 133)				ER−/PgR+ (n = 24)				ER−/PgR− (n = 271)				P heterogeneity ER+/PgR+ vs ER−/PgR−
		Case	OR	95% CI	P	Case	OR	95% CI	P	Case	OR	95% CI	P	Case	OR	95% CI	P
Age at menarche
≤12	691	187	1.00 (reference)a			27	1.00 (reference)a			7	1.00 (reference)a			81	1.00 (reference)a
13	600	133	0.93	0.71–1.21		25	1.10	0.62–1.95		5	0.95	0.29–3.16		60	0.85	0.59–1.22
14	574	105	0.83	0.62–1.11		36	1.82	1.05–3.15		4	0.83	0.22–3.13		57	0.89	0.61–1.31
≥15	1021	128	0.61	0.45–0.83		41	1.28	0.71–2.32		7	0.73	0.20–2.74		61	0.57	0.38–0.86
P for trend					0.0016				0.23				0.62				0.015	0.93
Menopausal status
Premenopause	1081	260	1.00 (reference)b			44	1.00 (reference)b			12	1.00 (reference)b			100	1.00 (reference)b
Natural menopause	1424	241	0.64	0.49–0.84		74	1.20	0.72–2.02		11	0.59	0.19–1.89		128	1.22	0.83–1.80
Menopause due to other reason	539	59	0.53	0.35–0.80		13	0.71	0.32–1.58		0	–	–		29	0.95	0.53–1.69
Parity
Parous	2590	460	1.00 (reference)c			112	1.00 (reference)c			21	1.00 (reference)c			234	1.00 (reference)c
Nulliparous	235	69	1.30	0.96–1.78	0.094	10	0.94	0.47–1.85	0.85	1	0.48	0.06–3.72	0.48	16	0.65	0.38–1.11	0.12	0.019
Age at first birthi
≤24	1242	179	1.00 (reference)d			49	1.00 (reference)d			2	1.00 (reference)d			102	1.00 (reference)d
25–29	1071	213	1.26	1.00–1.59		47	1.09	0.71–1.66		15	9.04	1.92–42.68		99	0.97	0.72–1.32
≥30	211	57	1.57	1.08–2.30		15	1.77	0.91–3.44		3	7.80	1.13–54.07		30	1.31	0.81–2.11
P for trend					0.0086				0.17				0.009				0.48	0.26
Parity numberi
1	273	58	1.00 (reference)e			13	1.00 (reference)e			4	1.00 (reference)e			36	1.00 (reference)e
2	1243	250	1.03	0.73–1.45		57	1.07	0.55–2.05		13	0.67	0.19–2.30		120	0.77	0.51–1.17
3	773	115	0.91	0.62–1.35		35	1.22	0.59–2.50		1	0.12	0.01–1.21		64	0.71	0.44–1.15
4	214	29	1.02	0.60–1.73		4	0.58	0.18–1.90		1	0.43	0.04–4.71		11	0.52	0.25–1.08
≥5	87	8	0.87	0.38–1.99		3	1.23	0.32–4.77		2	2.28	0.29–18.15		3	0.39	0.11–1.36
P for trend					0.59				0.94				0.64				0.045	0.17
Breastfeedingi
Formula only	410	89	1.00 (reference)f			26	1.00 (reference)f			3	1.00 (reference)f			41	1.00 (reference)f
Mixed breastfeeding and formula	1268	262	0.99	0.75–1.32		58	0.78	0.48–1.27		13	1.70	0.45–6.44		134	1.10	0.75–1.60
Breastfeeding only	891	107	0.73	0.53–1.02		28	0.60	0.33–1.08		4	0.72	0.14–3.74		59	0.88	0.57–1.37
Total month of breastfeedingi
0–3	394	143	1.00 (reference)f			29	1.00 (reference)f			7	1.00 (reference)f			56	1.00 (reference)f
3–12	302	74	0.70	0.50–0.97		17	0.79	0.42–1.49		3	0.39	0.08–1.97		37	0.84	0.53–1.32
12–24	396	89	0.65	0.47–0.89		18	0.61	0.32–1.13		3	0.47	0.11–2.10		31	0.57	0.35–0.93
>24	478	94	0.68	0.48–0.97		21	0.59	0.30–1.16		1	0.07	0.004–0.99		36	0.61	0.36–1.03
P for trend					0.013				0.082				0.04				0.023	0.58
Quantity of breast milk secretioni
Poor or no	761	168	1.00 (reference)f			43	1.00 (reference)f			11	1.00 (reference)f			74	1.00 (reference)f
Fair	876	141	0.82	0.64–1.06		28	0.61	0.37–1.01		6	0.44	0.14–1.37		82	1.08	0.77–1.52
Good	885	141	0.80	0.62–1.04		38	0.82	0.52–1.31		3	0.30	0.08–1.16		67	0.90	0.63–1.29
Family history of breast cancer in mother or sisters
No	3037	524	1.00 (reference)g			116	1.00 (reference)g			21	1.00 (reference)g			238	1.00 (reference)g
Yes	123	48	2.14	1.49–3.08	<.0001	17	3.52	2.03–6.09	<.0001	3	4.06	1.15–14.31	0.029	33	3.51	2.32–5.31	<.0001	0.044
Oral contraceptives use
Never	2604	504	1.00 (reference)h			115	1.00 (reference)h			22	1.00 (reference)h			241	1.00 (reference)h
Ever	158	30	0.90	0.59–1.37	0.62	8	1.22	0.57–2.59	0.61	0	–	–	–	16	1.03	0.60–1.78	0.91	0.68
Use of exogenous female hormones other than oral contraceptives
Never	2588	498	1.00 (reference)h			112	1.00 (reference)h			21	1.00 (reference)h			241	1.00 (reference)h
Ever	134	26	0.86	0.55–1.36	0.52	9	1.56	0.76–3.19	0.23	1	0.79	0.10–6.17	0.82	11	0.79	0.42–1.50	0.47	0.82

All models were adjusted by age, BMI (<18.5, 18.5–25, 25–30, ≥30), smoke (never, current or past), alcohol (never, current or past), occupation (housewife, other), physical activity (<1 h per week, more than 1 h per week), year of recruitment (continuous), area (southern Miyagi Prefecture, other) and reference (from screening, other). ^aAdditionally adjusted by family history of breast cancer (yes, no), parity number (0, 1, 2, 3, 4, ≥5). ^bAdditionally adjusted by family history of breast cancer, age at menarche (≤12, 13, 14, ≥15), parity number (0, 1, 2, 3, 4, ≥5). ^cAdditionally adjusted by family history of breast cancer, age at menarche. ^dAdditionally adjusted by family history of breast cancer, age at menarche, parity number (1, 2, 3, 4, ≥5). ^eAdditionally adjusted by family history of breast cancer, age at menarche, age at first birth (≤24, 25–29, ≥30). ^fAdditionally adjusted by family history of breast cancer, age at menarche, age at first birth, parity number (1, 2, 3, 4, ≥5). ^gAdditionally adjusted by parity number (0, 1, 2, 3, 4, ≥5). ^hAdditionally adjusted by family history of breast cancer, age at menarche, parity number (0, 1, 2, 3, 4, ≥5). ⁱFor parous women only. BMI, body mass index; CI, confidence interval; ER, estrogen receptor; OR, odds ratio; PgR, progesterone receptor.

Table 3 shows the results according to ER+/PgR+ and ER−/PgR− status among premenopausal women. A later age at menarche is marginally associated with a decreased risk of ER+/PgR+ cancer (P _trend = 0.056). An older age at first birth is significantly associated with an increased risk of ER+/PgR+ cancer (P _trend = 0.027). However, tests of heterogeneity between the risks of ER+/PgR+ and ER−/PgR− cancer show non‐significance for these factors. A family history of breast cancer is positively associated with the risk of both ER+/PgR+ and ER−/PgR− cancer.

Table 3.

OR (95% CI) of breast cancer risk by hormone receptor status among premenopausal women

	Control	Premenopausal								P heterogeneity
		ER+/PgR+ (n = 260)				ER−/PgR− (n = 100)
		Case	OR	95% CI	P	Case	OR	95% CI	P
Age at menarche
≤12	465	129	1.00 (reference)a			49	1.00 (reference)a
13	270	67	0.94	0.66–1.35		22	0.77	0.45–1.32
14	183	41	0.91	0.60–1.40		13	0.70	0.36–1.36
≥15	150	21	0.48	0.26–0.87		16	1.08	0.54–2.18
P for trend					0.056				0.72	0.37
Parity
Parous	891	215	1.00 (reference)b			88	1.00 (reference)b
Nulliparous	144	33	0.82	0.52–1.27	0.37	9	0.58	0.28–1.21	0.15	0.41
Age at first birthh
≤24	401	75	1.00 (reference)c			34	1.00 (reference)c
25–29	391	104	1.28	0.90–1.84		41	1.04	0.63–1.72
≥30	89	34	1.85	1.07–3.19		13	1.57	0.74–3.34
P for trend					0.027				0.35	0.57
Parity numberh
1	135	29	1.00 (reference)d			11	1.00 (reference)d
2	445	125	1.38	0.84–2.29		55	1.62	0.79–3.35
3	250	51	1.28	0.72–2.28		18	0.97	0.41–2.29
4	47	8	1.32	0.52–3.34		3	0.93	0.23–3.76
≥5	14	2	0.88	0.17–4.51		1	0.97	0.10–9.39
P for trend					0.78				0.44	0.4
Breastfeedingh
Formula only	160	42	1.00 (reference)e			13	1.00 (reference)e
Mixed breastfeeding and formula	542	136	0.92	0.60–1.41		57	1.28	0.67–2.44
Breastfeeding only	188	36	0.63	0.36–1.09		18	1.43	0.65–3.13
Total month of breastfeedingh
0–3	155	77	1.00 (reference)e			16	1.00 (reference)e
3–12	120	37	0.63	0.39–1.03		13	1.35	0.59–3.10
12–24	130	41	0.60	0.37–0.98		12	0.98	0.41–2.36
>24	108	34	0.70	0.40–1.22		10	1.15	0.44–2.96
P for trend							0.091		0.88	0.28
Quantity of breast milk secretionh
Poor or no	317	86	1.00 (reference)e			26	1.00 (reference)e
Fair	299	62	0.72	0.49–1.07		32	1.38	0.78–2.42
Good	257	64	0.80	0.54–1.19		26	1.45	0.80–2.64
Family history of breast cancer in mother or sisters
No	1046	239	1.00 (reference)f			89	1.00 (reference)f
Yes	35	21	2.86	1.56–5.23	0.0007	11	4.34	2.07–9.07	<.0001	0.31
Oral contraceptives use
Never	942	227	1.00 (reference)g			91	1.00 (reference)g
Ever	99	25	1.22	0.74–2.01	0.44	8	0.91	0.42–1.97	0.8	0.5
Use of exgenous female hormones other than oral contraceptives
Never	943	228	1.00 (reference)g			90	1.00 (reference)g
Ever	73	17	0.95	0.53–1.72	0.87	6	1.00	0.41–2.42	0.99	0.93

All models were adjusted by age, BMI (<18.5, 18.5–25, 25–30, ≥30), smoke (never, current or past), alcohol (never, current or past), occupation (housewife, other), physical activity (<1 h per week, more than 1 h per week), year of recruitment (continuous), area (Southern Miyagi Prefecture, other), reference (from screening, other). ^aAdditionally adjusted by family history of breast cancer (yes, no), parity number (0, 1, 2, 3, 4, ≥5). ^bAdditionally adjusted by family history of breast cancer, age at menarche (≤12, 13, 14, ≥15). ^cAdditionally adjusted by family history of breast cancer, age at menarche, parity number (1, 2, 3, 4, ≥5). ^dAdditionally adjusted by family history of breast cancer, age at menarche, age at first birth (≤24, 25–29, ≥30). ^eAdditionally adjusted by family history of breast cancer, age at menarche, age at first birth, parity number (1, 2, 3, 4, ≥5). ^fAdditionally adjusted by parity number (0, 1, 2, 3, 4, ≥5). ^gAdditionally adjusted by family history of breast cancer, age at menarche, parity number (0, 1, 2, 3, 4, ≥5). ^hFor parous women only. BMI, body mass index; CI, confidence interval; ER, estrogen receptor; OR, odds ratio; PgR, progesterone receptor.

Table 4 shows the results for postmenopausal women. A later age at menarche is associated with a decreased risk of both ER+/PgR+ and ER−/PgR− cancer (P _trend = 0.012 and 0.0056, respectively). Nulliparity is positively associated with a risk of ER+/PgR+ cancer, but not ER−/PgR− cancer (P _{heterogeneity} = 0.0095). Among parous women, no dose‐response relationship with parity number is observed for either of the receptors. A longer period of breastfeeding is associated with a lower risk of both ER+/PgR+ and ER−/PgR− cancer; however, this is not statistically significant (P _trend = 0.062 and 0.076, respectively). A family history of breast cancer is associated with an increased risk of both ER+/PgR+ and ER−/PgR− cancer; the magnitude of the risk appears to be greater for ER−/PgR− cancer (P _{heterogeneity} = 0.052; OR = 3.23, 95% CI 1.86–5.62).

Table 4.

OR (95% CI) of breast cancer risk by hormone receptor status among postmenopausal women

	Control	Postmenopausal								P heterogeneity
		ER+/PgR+ (n = 300)				ER−/PgR− (n = 157)
		Case	OR	95% CI	P	Case	OR	95% CI	P
Age at menarche
≤12	194	54	1.00 (reference)a			28	1.00 (reference)a
13	304	64	0.87	0.56–1.35		35	0.79	0.46–1.37
14	375	61	0.71	0.46–1.10		41	0.80	0.47–1.37
≥15	855	105	0.61	0.40–0.93		43	0.46	0.26–0.80
P for trend					0.012				0.0056	0.48
Age at natural menopause
≤47	259	37	1.00 (reference)b			19	1.00 (reference)b
48–50	546	98	1.31	0.85–2.03		42	1.00	0.56–1.79
51–53	398	62	0.93	0.58–1.50		42	1.23	0.68–2.21
≥54	167	42	1.64	0.97–2.76		22	1.47	0.75–2.87
P for trend					0.38				0.17	0.56
Parity
Parous	1620	238	1.00 (reference)c			136	1.00 (reference)c
Nulliparous	85	33	2.56	1.61–4.07	<.0001	6	0.76	0.32–1.81	0.54	0.0095
Age at first birthi
≤24	800	101	1.00 (reference)d			64	1.00 (reference)d
25–29	648	107	1.26	0.92–1.74		55	0.90	0.61–1.35
≥30	117	22	1.16	0.65–2.09		14	1.11	0.56–2.22
P for trend					0.26				0.96	0.43
Parity numberi
1	131	28	1.00 (reference)e			21	1.00 (reference)e
2	757	122	0.73	0.44–1.21		62	0.47	0.27–0.84
3	497	61	0.66	0.38–1.15		43	0.57	0.31–1.06
4	163	21	0.86	0.43–1.70		8	0.44	0.18–1.09
≥5	72	6	0.64	0.23–1.78		2	0.30	0.06–1.41
P for trend					0.51				0.18	0.48
Breastfeedingi
Formula only	234	45	1.00 (reference)f			25	1.00 (reference)f
Mixed breastfeeding and formula	683	123	1.06	0.71–1.58		71	1.06	0.64–1.76
Breastfeeding only	683	69	0.76	0.48–1.20		40	0.92	0.51–1.64
Total month of breastfeedingi
0–3	226	65	1.00 (reference)f			38	1.00 (reference)f
3–12	173	37	0.74	0.46–1.19		19	0.56	0.30–1.04
12–24	248	48	0.70	0.45–1.09		19	0.54	0.29–1.00
>24	361	58	0.63	0.38–1.04		26	0.60	0.31–1.17
P for trend					0.062				0.076	0.73
Quantity of breast milk secretioni
Poor or no	420	79	1.00 (reference)f			44	1.00 (reference)f
Fair	550	79	0.93	0.65–1.35		45	0.95	0.60–1.50
Good	602	73	0.78	0.54–1.13		40	0.82	0.51–1.32
Family history of breast cancer in mother or sisters
No	1879	274	1.00 (reference)g			138	1.00 (reference)g
Yes	84	26	1.67	1.01–2.76	0.044	19	3.23	1.86–5.62	<.0001	0.052
Oral contraceptives use
Never	1586	267	1.00 (reference)h			139	1.00 (reference)h
Ever	52	5	0.49	0.19–1.30	0.15	7	1.39	0.59–3.28	0.46	0.095
Use of exgenous female hormones other than oral contraceptives
Never	1571	262	1.00 (reference)h			141	1.00 (reference)h
Ever	56	7	0.68	0.29–1.60	0.38	4	0.64	0.22–1.85	0.41	0.93

All models were adjusted by age, BMI (<18.5, 18.5–25, 25–30, ≥30), smoke (never, current or past, missing), alcohol (never, current or past), occupation (housewife, other), physical activity (<1 h per week, more than 1 h per week), menopausal status (natural menopause, menopause due to other reason), age at menopause (≤47, 48–50, 51–53, ≥54), year of recruitment (continuous), area (southern Miyagi Prefecture, other), reference (from screening, other). ^aAdditionally adjusted by family history of breast cancer (yes, no), parity number (0, 1, 2, 3, 4, ≥5). ^bAdditionally adjusted by family history of breast cancer, age at menarche (≤12, 13, 14, ≥15), parity number (0, 1, 2, 3, 4, ≥5). ^cAdditionally adjusted by family history of breast cancer, age at menarche. ^dAdditionally adjusted by family history of breast cancer, age at menarche, parity number (1, 2, 3, 4, ≥5). ^eAdditionally adjusted by family history of breast cancer, age at menarche, age at first birth (≤24, 25–29, ≥30). ^fAdditionally adjusted by family history of breast cancer, age at menarche, age at first birth, parity number (1, 2, 3, 4, ≥5). ^gAdditionally adjusted by parity number (0, 1, 2, 3, 4, ≥5). ^hAdditionally adjusted by family history of breast cancer, age at menarche, parity number (0, 1, 2, 3, 4, ≥5). ⁱFor parous women only. BMI, body mass index; CI, confidence interval; ER, estrogen receptor; OR, odds ratio; PgR, progesterone receptor.

Discussion

This hospital‐based case‐control study revealed the associations between menstrual and reproductive factors and breast cancer risk in terms of joint hormone receptor status. A few epidemiologic studies conducted in Japan have focused on tumor subtypes.15, 16, 17 However, it has been difficult to determine whether the associations among Japanese women differ from those among Western women. It is known that the proportion of tumor subtypes differs across menopausal status23 and the breast cancer patients' survival rates are reported to be different according to tumor subtypes.24 Etiology and biology of subtypes might be different from each other. Therefore, the present study is important for clarifying the impact of menstrual and reproductive factors on breast cancer risk in relation to tumor subtypes and menopausal status among Japanese women.

Regarding menstrual factors, a meta‐analysis showed that a late age at menarche is associated with a decreased risk of ER+/PgR+ and ER−/PgR− cancers.12 A recent study from China demonstrates a similar association.14 In the present study, a later age at menarche is associated with a decreased risk of ER+/PgR+ and ER−/PgR− cancers among both women overall and postmenopausal women. It has been hypothesized that a later age at menarche might result in a shorter proliferation of mammary gland cells, which might be more susceptible to carcinogenesis.25 This hypothesis might explain the association between a later age at menarche and a lower risk for both ER+/PgR+ and ER−/PgR− cancer.

Parity, parity number and age at first birth have been recognized as factors affecting breast cancer risk in Japan and other countries.12, 14, 15, 17 A meta‐analysis has shown that nulliparity13 and a higher age at first birth12, 13 are associated with an increased risk of ER+/PgR+ cancer, indicating that the effects differ between ER+/PgR+ and ER−/PgR− status.12 In the present study, nulliparity was associated with an increased risk of ER+/PgR+ cancer among both women overall and postmenopausal women, and higher age at first birth was associated with an increased risk of ER+/PgR+ cancer among women overall. A significant association with age at first birth was also observed for ER−/PgR+ cancer. However, as the confidence interval was wide, the result for ER−/PgR+ cancer is questionable. Our overall analysis also showed that multiparity was associated with a decreased risk of ER−/PgR− cancer. Nulliparity showed a decreased risk, but not statistically significant. Although the precise mechanism is unknown, it has been reported that successive multiparity induces a protective effect through sequential differentiation of mammary gland stem cells;26 such cells are thought to be associated with ER−/PgR− cancer.27

A meta‐analysis has shown that breastfeeding is associated with a decreased risk of ER+/PgR+ and ER−/PgR− cancer.12 A previous study from the Asian region demonstrates an association between a longer duration of breastfeeding and a decreased risk of ER+/PgR+ cancer, but not ER−/PgR− cancer.14 Meanwhile, studies in Japan have indicated no association between breastfeeding and the risk of either receptor‐positive or negative breast cancer.16, 17 Although our data for the risk of ER−/PgR+ cancer were based on a small sample size, our findings suggest that a longer period of breastfeeding might protect against all subtypes of breast cancer, being almost consistent with the findings of the abovementioned meta‐analysis.

With regard to the risk associated with a family history of breast cancer, a meta‐analysis reveals a positive association between a history of breast cancer in mother or sisters and breast cancer risk among both premenopausal and postmenopausal women.28 Our previous study conducted in Miyagi prefecture also demonstrates such an association.3 In the present study, the increased risk posed by a family history of breast cancer is consistently observed for all subtypes. In addition, there is a variation in the magnitude of risk among the subtypes. A higher OR for family history is found for ER−/PgR− cancer (P _{heterogeneity} = 0.044). The mechanism might include genetic mutation, such as BRCA1,29 which has been associated with a positive family history of breast cancer, and might confer susceptibility to ER−/PgR− cancer.30

The present study had both strengths and limitations. First, we considered comparability between cases and controls. We selected the controls from among patients admitted to the same hospital as the cases. The participation rates were high for both cases and controls. However, the distribution of risk factors for breast cancer among control subjects may have differed from that in the general population. To improve comparability between the cases and controls, statistical analyses were appropriately controlled for background characteristics, such as area of residence and referral patterns. Although persistent bias might exist, it is likely that any problems with comparability have been weakened. Second, the problem of limited statistical power must be considered in the analysis of ER−/PgR+ cancer; the results for this subtype might be inconclusive because of the small number of cases. To confirm the risk for ER−/PgR+ cancer, further studies are needed. Third, we must evaluate the possibility of information bias. Self‐reported information on exposure might have been vulnerable to misclassification. However, any such misclassification in reproductive factors would have been non‐differential.31 This bias is unlikely to have distorted our present results. Fourth, it is possible that the inclusion of patients with benign tumors in the control group influenced the results, because patients with benign tumors of the gynecologic organs or breast might have a background similar to that of patients with breast cancer. Therefore, we performed additional analyses by excluding patients with benign tumors of the gynecologic organs (n = 375) or breast (n = 36) from the controls. However, the exclusion of these patients had no effect on the OR (data not shown).

One of the strengths of our study was the stability of menopausal status. In any prospective study, some of the premenopausal women in the original cohort may become postmenopausal by the end of follow up.3, 4, 15 In contrast, any case‐control study like the present one has information on menopausal status at the time of diagnosis. Another strength of our study was the low rate of missing data (8.4%) for hormone receptor status. Although missing cases were less likely to have been referred from screening, the distribution of the hormone receptor statuses in our study was roughly the same as that in a large previous study in Japan.32 Compared with our present study, the rates of missing data in previous studies, including cohort studies, which ranged from 9% to 61%, were relatively high.10, 11, 14, 15, 17 Cancer incidence in Japanese cohort studies has been evaluated based on population‐based cancer registries.3, 33, 34 However, data on hormone receptor status in population‐based cancer registries are incomplete.3, 33, 34 Therefore, hospital‐based studies would be more suitable for assessing the risk of breast cancer by hormone receptor status.18 From this viewpoint, the present study is considered to represent one of the most accurately conducted assessments of breast cancer risk in terms of hormone receptor status.

In conclusion, this hospital‐based case‐control study has clarified risk factor profiles according to breast cancer subtypes stratified by joint hormone receptor status and menopausal status. A later age at menarche is associated with a decreased risk of both ER+/PgR+ and ER−/PgR− among women overall and postmenopausal women. Nulliparity is associated with an increased risk of ER+/PgR+, but not ER−/PgR−, among postmenopausal women and women overall. A longer duration of breastfeeding is associated with a decreased risk of all subtypes among women overall. These results indicate that a later age at menarche has a protective effect against both ER+/PgR+ and ER−/PgR− cancer, but that parity might impact differently on various subtypes of breast cancer. A longer duration of breastfeeding might protect against breast cancer, irrespective of receptor type. Further studies are needed to clarify the etiology of the rare ER+/PgR− and ER−/PgR+ cancer subtypes among Japanese women.

Disclosure Statement

The authors have no conflict of interest to declare.

(Cancer Sci, doi: 10.1111/j.1349‐7006.2012.02379.x, 2012)