SEARCH

SEARCH BY CITATION

Keywords:

  • body size;
  • hormone therapy;
  • breast cancer;
  • Asian–Americans

Abstract

  1. Top of page
  2. Abstract
  3. Subjects and methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. References

Historically, breast cancer rates have been low in Asia but rates have increased substantially in Asian–Americans for reasons that are not well understood. The authors conducted a population-based case–control study of breast cancer in Los Angeles County, which included 1,277 (450 Chinese, 352 Japanese, 475 Filipinos) women with incident, histologically confirmed breast cancer and 1,160 control subjects (486 Chinese, 311 Japanese, 363 Filipinos). A detailed in-person interview was conducted, which included questions on menopausal hormone therapy (HT) use, height, weight in each decade of life and reproductive factors. Breast cancer risk increased with increasing recent weight in postmenopausal women (p trend = 0.015). There was a significant 16% (95% CI = 2–35%) increase in risk per 10 kg of body weight in postmenopausal women. In both premenopausal and postmenopausal women, risk increased with increasing waist to hip ratio; this remained statistically significant after adjustment for recent weight in all subjects combined (p trend = 0.042). The increased risk associated with high recent weight in postmenopausal women was more apparent for women with high waist to hip ratio (p trend = 0.013). Use of HT was a significant risk factor; risk increased 26% per 5 years of current use of estrogen and progestin therapy (p trend = 0.017). The increased risk associated with high body weight was observed irrespective of HT use. Use of HT and high body size might have contributed to the rapid increase of breast cancer in Asian–Americans. © 2006 Wiley-Liss, Inc.

Reasons for the increase of breast cancer in Asian–American women are not well understood.1 In migrant studies conducted in the 1950s and 1960s, breast cancer rates increased only slightly in the migrating generation, but in more recent studies, including one we participated in,2 breast cancer rates increased substantially in the migrating generation. Such a rapid increase in risk may be due to changes in modifiable lifestyle factors.

High body weight and use of menopausal hormone therapy (HT) represent examples of such modifiable lifestyle factors, and both are important risk factors for postmenopausal breast cancer in Western populations, reflecting the critical etiological role of estrogen.3 The contribution of these two risk factors to the increasing breast cancer incidence in Asian–American women has been examined in a limited number of studies. In a study conducted in Hawaii between 1975 and 1980 that included a modest sample size of postmenopausal Japanese women (138 cases, 154 controls), recent weight was nonsignificantly but positively associated with risk.4 Long-term (73+ months) use of replacement estrogens was associated with an increased risk that was not statistically significant.5 In the multicenter case–control study of breast cancer in Asian–American (Chinese, Japanese and Filipino) women that we participated in during the late 1980s, high body weight and weight gain were significant risk factors; however, few postmenopausal women (114 cases, 243 controls) were included because this study was limited to women who were 55 years old or younger.6 The role of HT was not reported in this previous study conducted in Asian–Americans.6 In most studies conducted in Western populations, the harmful effects of HT were more pronounced in women with low body weight.7, 8, 9, 10 Conversely, the increased risk associated with high body weight was confined to nonusers of HT.7, 8, 9, 10 Given that the majority of Asian–American women are classified as thin or of normal weight according to the standards used in Western countries, we hypothesized that HT use may have a particularly deleterious effect in Asian–American women.

We describe later our findings on body weight, body mass index (BMI), weight gain, waist to hip ratio (WHR), use of HT and the interrelationships between body size and use of HT and breast cancer risk for Asian–American women of Los Angeles County.

Subjects and methods

  1. Top of page
  2. Abstract
  3. Subjects and methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. References

The methods of this case–control study have been described previously.11 Briefly, this population-based case–control study included women who were identified as Chinese, Japanese or Filipino, between the ages of 25 and 74 years at the time of diagnosis of an incident breast cancer on or after January 1995 through December 2001. Cases were identified through the Los Angeles County Cancer Surveillance Program, the population-based cancer registry that is a member of the National Cancer Institute's Surveillance, Epidemiology, and End Results program, and the statewide California Cancer Registry. Of the 2,221 (784 Chinese, 585 Japanese and 852 Filipino women) we contacted, 1,384 (492 Chinese, 384 Japanese, 508 Filipinos) were interviewed, 507 declined to be interviewed (193 Chinese, 143 Japanese, 171 Filipinos), 42 were deceased (8 Chinese, 15 Japanese, 19 Filipinos) and 288 could not be contacted (91 Chinese, 43 Japanese, 154 Filipinos).

One thousand two hundred and twenty-five controls (514 Chinese, 331 Japanese, 380 Filipino) were selected from the neighborhoods where cancer cases resided at the time of diagnosis using a well-established, standard algorithm to identify neighborhood controls that the University of Southern California Epidemiology Program has used in numerous case–control studies.12 This algorithm defined a specified sequence of houses to be visited in the neighborhoods where index cases lived at the time of diagnosis. We sought to interview as the control the first eligible resident in the sequence. If the first eligible control subject refused to participate, the second eligible one in the sequence was asked and so on. Letters were left when no one was home, and follow-up was by mail and telephone (if a number could be determined). Controls were sought to frequency-match to the cases on specific Asian-ethnicities and 5-year age groups. On average, a suitable control was identified after walking 64.6 houses. Of the controls interviewed, 62% were the first identified eligible control, 21% were the second identified eligible control and 17% were the third or later eligible control.

Data collection

In-person interviews were conducted using a standardized, structured questionnaire that covered demographic characteristics and migration history, menstrual and reproductive history, body size, physical activity and diet history. The diet questionnaire was developed by Dr. Jean Hankin at the University of Hawaii and was modeled after the validated diet instrument used in the Multiethnic Cohort Study being conducted in Hawaii and Los Angeles.13 Dietary intake during the year prior to cancer diagnosis (for cases) or during the past year (for controls) was determined. We have previously reported that breast cancer risk was significantly reduced in association with intake of soy during adolescence and adulthood,11 intake of green tea14 and regular physical activity after 10 years of age15 using data from the first 501 cases and 594 controls who were interviewed; these subjects are included in the present analysis on body size characteristics and menopausal hormones. Subjects were asked about their height and weight history at age 18 years, at age 30 years and each decade thereafter. Trained interviewers measured the current weight, and circumferences of the waist and hip of study participants. Waist circumference was measured at the narrowest torso circumference and hip circumference was measured at the widest hip circumference. Calendars were used to chart major life events, reproductive histories and hormone use. Lifetime history of hormone use [oral contraceptives (OCs) and menopausal hormones] was obtained with the aid of an album with color photographs of all hormone preparations used in the United States. For each episode of hormone use, brand of hormone used, age at starting, age at stopping, side effects and reasons for stopping were asked.

Statistical analysis

After the publication of the results from the Women's Health Initiative, use of menopausal hormones changed dramatically in the United States.16 Because of reports of discontinued HT use as a result of the Women's Health Initiative results and that controls in this study were interviewed up to a year after their neighborhood cases interview, we counted HT use of the control up to the comparable calendar time of the case by assigning the case's reference date (i.e., diagnosis date) to the control subject identified from the same neighborhood of the case.

A woman was considered to be premenopausal if her menstrual periods had not ended or a menstrual period had occurred within 6 months of her reference date. For naturally menopausal women, age at menopause was estimated as follows: Natural menstruation was taken to mean menstruating and not using OCs or menopausal HT at the time. For women who started HT before their last menstrual period, we set age at menopause as the age at which they started HT use because we assumed HT was taken for menopausal symptoms; their menopausal status is classified as “HT menopause.” For women taking OCs, age at menopause was taken at the end of the period of OC use, if no “natural” menstruation occurred thereafter. For women who had a bilateral oophorectomy, their age at “surgical menopause” is the age at oophorectomy. For women (97 cases, 82 controls) who had undergone a “simple hysterectomy” (i.e., without a bilateral oophorectomy), their ages at menopause are not truly known. We conducted our analysis on HT with and without including these women (Table V).

We examined HT use around the time of menopause, separately for estrogen alone therapy (ET) and estrogen and progestin therapy combined (EPT). Recency of HT use (current or former) and duration of ET and EPT use were calculated for each postmenopausal subject. Subjects with a history of ET as well as EPT use were counted as users in both the ET and EPT analysis.

Quartile distributions for controls were used to categorize anthropometric variables including body weight at various ages and BMI [weight (in kilograms) divided by height (in meters) squared]. Cases and controls were compared in terms of height, weight (and BMI) at each decade of age, waist and hip circumferences and WHR. We also determined breast cancer risk by weight in recent decade (referred to hereafter as recent weight). In our analysis on waist and hip circumferences and WHR, we examined the results unadjusted for recent weight as well as adjusted for recent weight. Similarly, we examined the results on recent weight with and without adjustment for WHR. Results using weight (adjusted for height) were similar to those based on BMI; we presented results by weight adjusted for height categories as well as by BMI. We examined the effects of adult weight gain on risk and presented results on changes between recent weight and weight at age 18 as well as between recent weight and weight at age 30.

We used the stepwise regression method to identify the set of body size variables that independently and significantly predicted breast cancer risk in postmenopausal women when race, age, education, birthplace, years of residence in the United States and height were taken into consideration. It was determined that once recent weight was fitted in the model, the residual effects of WHR and all other body size indices were no longer statistically significant (all p values greater than 0.05). Hence, recent weight was included as a covariate in all regression models examining HT and breast cancer risk.

Results presented are based on 1,277 cases and 1,160 controls. We excluded 107 cases and 65 controls from the final analyses because of previous cancer (44 cases, 37 controls) or missing information on body size, menstrual or pregnancy history or one of the other adjustment covariates (63 cases, 28 controls). We calculated odds ratios (ORs; relative risk estimates), their corresponding 95% confidence intervals (95% CIs) and p values by conditional logistic regression methods, with matched sets defined jointly by reference age (≤39, 40–44, 45–49, 50–54, 55–59, 60–64, 65–69, 70+), and specific Asian ethnicity (Chinese, Japanese, Filipino). All basic regression models in this study included as covariates birthplace and years of residence in the United States (US born, 20+ years, 11–20 years, ≤10 years), education (less than high school, high school, some college, college graduate), age at menarche (<12, 12–13, 14+), parity (0, 1, 2, 3, 4+ births), marital status (ever vs. never married), recent body weight (in quartiles), height (in quartiles), total caloric intake (continuous), years of regular (i.e., at least 1 hr per week) physical activity (<5, 5–9, 10–19, 20+), menopausal status (premenopausal, natural menopause, “HT menopause,” bilateral oophorectomy, simple hysterectomy) and age at menopause (<45, 45–49, 50–54, 55+). A more elaborate regression model included intake of soy during adolescence and adulthood, intake of tea (none, black tea only, green tea only, black and green tea), alcohol intake and family history of breast cancer; results are essentially identical and we show the results based on the more elaborate model. For anthropometric exposures, tests for trend (p values) were performed by coding each variable as a grouped (quartile) linear variable (i.e., as 1, 2, 3 and 4). To examine the potential effect modification of the body size–breast cancer association by current HT use (or time since menopause), interaction terms for specific body size measures with HT use (or time since menopause) were tested. P values less than 5% are considered statistically significant and all P values quoted are 2-sided.

Results

  1. Top of page
  2. Abstract
  3. Subjects and methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. References

Table I shows the cases and controls in terms of Asian ethnicity, birthplace/length of residence in the United States and level of education. In Table II, risk associations with age at menarche, parity, family history of breast cancer in mother or sister, total years engaging in regular physical activity and regular intake of tea and soy are shown. As reported previously,11, 14 breast cancer risk was significantly higher for women with a family history of breast cancer and was significantly lower for parous women, those who were high consumers of soy during adolescence, those who drank green tea regularly and those who were physically active. These results were comparable in premenopausal and postmenopausal Asian–American women (Table II).

Table I. Demographic Characteristics of Breast Cancer Cases and Controls in the Los Angeles County Asian–American Breast Cancer Study
 Cases (n = 1,277)Controls (n = 1,160)Premenopausal cases (n = 572)Premenopausal controls (n = 613)Postmenopausal cases (n = 705)Postmenopausal controls (n = 547)
By race/ethnicity
 Chinese450486227302223184
 Japanese352311125136227175
 Filipino475363220175255188
Mean reference age (s.d)52.9 (10.5)50.6 (10.6)44.1 (5.4)43.0 (5.8)60.0 (7.9)59.2 (8.0)
Birthplace
 US born30330995144208165
 Non-US born974851477469497382
 Migrants: Yrs in US      
 21+ yr454423163191291232
 11–2033928421118912895
 ≤10 yr181144103897855
Migrants
 Ave years in US (s.d)20.4 (11.5)20.9 (11.0)17.3 (9.1)18.4 (9.5)23.4 (12.7)23.9 (12.0)
 Ave age at migration (s.d)32.1 (12.4)32.0 (12.6)27.6 (8.4)27.6 (9.3)36.5 (13.9)37.4 (13.9)
Education
 <High school2302056773163132
 High school28227996135186144
 Some college554473289281265192
 >college2112031201249179
Table II. Selected Characteristics of Breast Cancer Cases and Controls
 CasesControlsAdjusted OR1PremenopausalPostmenopausal
CasesControlsAdjusted OR1CasesControlsAdjusted OR1
  • 1

    Conditional logistic regression models with matched sets defined jointly by age (≤39, 40–44, 45–49, 50–54, 55–59, 60–65, 65–69, 70+) and Asian ethnicity (Chinese, Japanese, Filipino) were employed. Years in the US, education, number of live births, age at menarche, recent weight (quartiles), height (quartile), caloric intake (continuous), family history of cancer, years of physical activity, menopausal status and age at menopause were included as covariates in all models. Intake of tea, soy intake during adolescence and soy intake during adult were mutually adjusted for each other. Menopausal status and age at menopause were not included in analyses restricted to premenopausal women.

  • 2

    Low/low is defined as ≤monthly soy intake during adolescence and ≤6.24 mg/1000 Kcal isoflavones/day during adult life; low/high is defined as ≤monthly soy intake during adolescence and >6.24 mg/1000 Kcal isoflavones/day during adult life; high/low is defined as ≥weekly soy intake during adolescence and ≤6.24 mg/1000 Kcal isoflavones/day during adult life and high/high is defined as ≥weekly soy intake during adolescence and >6.24 mg/1000 Kcal isoflavones/day during adult life.

Age at menarche
 ≤112251951.001251061.00100891.00
 12–136335920.89 (0.70–1.13)2983320.62 (0.44–0.87)3352601.12 (0.78–1.61)
 14+4193760.83 (0.66–1.13)1491780.55 (0.37–0.81)2701981.14 (0.78–1.69)
 p trend  0.29  0.03  0.67
Parity
 0 birth3041911.001561221.00148691.00
 12311810.94 (0.69–1.08)1281160.97 (0.64–1.46)103650.87 (0.53–1.42)
 24003940.72 (0.54–0.95)1942340.71 (0.50–1.05)2061600.67 (0.44–1.03)
 32102190.58 (0.43–0.80)741070.61 (0.39–0.97)1361120.55 (0.35–0.87)
 4+1321750.33 (0.23–0.43)20340.38 (0.19–0.75)1121410.29 (0.18–0.47)
 p trend  <0.001  <.001  <0.001
Family history of breast cancer in mother or sister
 No104910441.004895661.005604781.00
 Yes176891.91 (1.44–2.55)69352.44 (1.55–3.87)107541.64 (1.12–2.41)
 p value  <0.001  <0.001  0.011
Years of physical activity
 0–42181471.0096761.00122711.00
 5–93132480.94 (0.70–1.24)1431460.79 (0.53–1.19)1701021.08 (0.71–1.63)
 10–194224040.77 (0.59–1.01)2102270.76 (0.52–1.11)2121770.73 (0.50–1.08)
 20+3243610.57 (0.43–0.76)1231640.58 (0.39–0.88)2011970.56 (0.38–0.82)
 p trend  <0.001  0.021  <0.001
Intake of tea
 No black/green3132471.001351291.001781181.00
 Black only3642541.16 (0.90–1.49)1841471.37 (0.95–1.96)1801070.97 (0.67–1.41)
 Green only1942300.60 (0.44–0.80)611110.60 (0.39–0.94)1331190.58 (0.39–0.87)
 Black & green4064290.71 (0.55–0.92)1922260.89 (0.62–1.28)2142030.58 (0.40–0.83)
 P (3df)  <.001  0.002  0.004
Soy intake during adolescence and adult life2
 Low/low4783441.002311961.002471481.00
 Low/high1821510.88 (0.61–1.18)77820.83 (0.54–1.27)105690.85 (0.56–1.31)
 High/low1841730.72 (0.54–0.97)86800.96 (0.62–1.48)98930.53 (0.35–0.80)
 High/high4334920.61 (0.47–0.81)1782550.63 (0.42–0.92)2552370.57 (0.39–0.83)
 P (3df))  0.002  0.070  0.005

Table III shows height and weight patterns in cases and controls by menopausal status. In premenopausal women, breast cancer risk was not significantly related to height; weight at age 18, recent weight, recent BMI, weight gain since age 18, weight gain since age 30 or WHR.

Table III. Body Size Measures in Breast Cancer Cases and Controls Stratified by Menopausal Status
 CasesControlsAdjusted OR1PremenopausalPostmenopausal
CasesControlsAdjusted OR1CasesControlsAdjusted OR1
  • 1

    Conditional logistic regression models with matched sets defined jointly by age (≤39, 40–44, 45–49, 50–54, 55–59, 60–64, 65–69, 70+), and Asian ethnicity (Chinese, Japanese, Filipino). Duration of residence in the US (US born, 21+ yrs, 11–20 yr, ≤10 yr), education, age at menarche, number of live births, menopausal status (type) and age at menopause, intake of tea and soy during adolescence and adult life, and years of physical activity were included as covariates.

  • 2

    In analysis to examine the effect of weight at age 18 or recent weight, height was included in the model.

  • 3

    Height and recent weight were not included in the model.

Height (cm)
 Mean157.3157.6 158.1158.5 156.7156.6 
 ≤152.43503111.01381261.002121851.00
 >152.4–157.54623881.15 (0.93–1.43)1931961.10 (0.78–1.54)2691921.23 (0.91–1.65)
 >157.5–160.01951841.06 (0.80–1.40)921070.89 (0.60–1.32)103771.30 (0.88–1.92)
 >160.02802871.06 (0.83–1.36)1511880.99 (0.69–1.40)129991.14 (0.80–1.63)
 p trend  0.83  0.63  0.40
Weight, age 182
 Mean47.047.6 47.248.1 46.847.1 
 ≤43.12882251.001301111.001581141.00
 >43.1– 47.03583310.91 (0.71–1.16)1631710.92 (0.64–1.32)1951600.93 (0.66–1.32)
 >47.0–51.32762770.92 (0.70–1.21)1341500.99 (0.67–1.47)1421270.86 (0.59–1.26)
 >51.32872840.91 (0.68–1.20)1331720.78 (0.52–1.17)1541121.04 (0.69–1.55)
 p trend  0.56  0.30  0.97
Recent weight (kg)2
 Mean57.056.6 55.856.1 58.157.1 
 ≤49.93152931.001571611.001581321.00
 >49.9–55.43102891.07 (0.84–1.36)1561571.06 (0.75–1.50)1541321.10 (0.78–1.57)
 >55.4–61.33333210.99 (0.77–1.27)1461670.85 (0.59–1.21)1871541.12 (0.79–1.60)
 >61.33192571.14 (0.87–1.49)1131280.75 (0.50–1.12)2061291.62 (1.11–2.36)
 p trend  0.50  0.099  0.015
Recent BMI2
 Mean23.122.8 22.422.4 23.623.3 
 ≤20.433142861.001751751.001391111.00
 >20.43–22.322732950.81 (0.64–1.04)1351670.74 (0.53–1.04)1381280.94 (0.65–1.36)
 >22.32–24.603292890.96 (0.75–1.23)1421450.82 (0.58–1.17)1871441.13 (0.79–1.62)
 >24.603512900.99 (0.77–1.29)1201260.67 (0.46–0.98)2411641.35 (0.95–1.93)
 p trend  0.76  0.070  0.045
Weight gain (recent weight–weight at 18) (kg)2
 Mean9.99.0 8.88.0 11.110.1 
 ≤106776721.003584011.003192711.00
 >10 to ≤15 kg2442021.13 (0.89–1.43)106931.05 (0.74–1.48)1381091.24 (0.90–1.72)
 >15 to ≤20 kg1411360.88 (0.66–1.17)46600.64 (0.41–1.05)95761.10 (0.75–1.62)
 >20 kg1461021.23 (0.91–1.68)51470.87 (0.54–1.40)95551.66 (1.09–2.53)
 p trend  0.47  0.24  0.036
Weight gain (recent weight–weight at 30) (kg)2
 Mean5.54.6 4.03.6 6.75.9 
 ≤1010129711.004945441.005184271.00
 >10 to ≤15 kg1321011.12 (0.83–1.51)41440.68 (0.41–1.12)91571.51 (1.02–2.22)
 >15 to ≤20 kg61451.08 (0.71–1.66)17130.96 (0.43–2.16)44321.17 (0.70–1.96)
 >20 kg38171.79 (0.96–3.34)1171.48 (0.52–4.25)27102.23 (1.00–4.94)
 p trend  0.098  0.88  0.023
Waist/hip ratio3
 Mean0.820.81 0.800.80 0.830.82 
 ≤0.762502851.001411781.001091071.00
 >0.76–0.802992961.12 (0.87–1.44)1471641.10 (0.79–1.55)1511321.13 (0.77–1.66)
 >0.80–0.843312811.24 (0.96–1.60)1541471.13 (0.80–1.61)1771341.34 (0.92–1.96)
 >0.843722891.32 (1.02–1.72)1221141.20 (0.82–1.77)2501751.48 (1.02–2.15)
 p trend  0.027  0.35  0.035

In postmenopausal women, breast cancer risk was not significantly associated with height or weight at age 18 (Table III). However, risk increased significantly with increasing recent weight (p trend= 0.015); women in the highest quartile of recent weight (>61.3 kg) showed a significant 62% increased risk compared with women in the lowest quartile of weight (≤49.9 kg). We repeated the analysis using BMI instead of weight and found similar patterns of elevated risks. Breast cancer risk in postmenopausal women was associated with an increased risk of 1.13 (95% CI = 1.00–1.26) per 4 kg/m2 increment in BMI.

Risk of breast cancer increased significantly in association with weight gain in postmenopausal women (Table III). Women who gained >20 kg since age 18 showed a significant increased risk (OR = 1.66, 95% CI = 1.09–2.53) compared with women who gained less than 10 kg. Similar results were obtained when we compared women who gained 20 kg or more with those who did not gain any weight since age 18; the adjusted OR was 1.79 (95% CI = 1.04–3.08) (p trend = 0.02). Results remained essentially unchanged when weight at age 18 was included as a covariate in the multivariable logistic regression model; the adjusted OR was 1.77 (95% CI = 1.01–3.10) for weight gain of 20 kg or more versus no weight gain. An even larger increased risk (OR = 2.23, 95% CI = 1.00–4.94) was observed for women who gained >20 kg since 30; this result also remained after adjusting for weight at age 18.

Additional analyses revealed that the divergence in weight between cases and controls became increasingly clear with increasing age. In analyses restricted to women in their 50s, cases (n = 366) and controls (n = 289) did not differ significantly in their weight around age 18 (OR was 1.13 for the highest versus lowest quartile of weight, p trend = 0.62), or weight in their 30s (OR was 1.51 for the highest versus lowest quartile of weight, p trend = 0.15), or weight in their 40s (OR was 1.39 for the highest vs. lowest quartile of weight, p trend = 0.24). However, cases were significantly heavier than controls in their 50s (OR was 2.00 for highest versus lowest quartile of weight, p trend = 0.031). Similar patterns of increasing risk associated with weight at increasing ages were observed in analyses restricted to women in their 60s or older (data not shown).

WHR was a significant risk factor in all subjects combined; this remained statistically significant even after adjusting for recent weight and other covariates (p trend = 0.042) (data not shown). No significant associations with waist or hip circumferences were found in premenopausal or postmenopausal women (data not shown). In postmenopausal women, breast cancer risk increased significantly with increasing WHR (p trend = 0.035) (Table III); the effects of WHR and recent weight diminished when both body size factors were adjusted for simultaneously (Table IV). Our data show a modest effect of recent weight in women with below median WHR (≤0.80) (p trend = 0.88) but a clear and significant trend of increasing risk with increasing recent weight in women with above median WHR (>0.80) (p trend = 0.013). Postmenopausal women in the highest quartile of recent weight and WHR >0.80 showed a significant increased risk (OR = 1.67, 95% CI = 1.10–2.55) compared with women in the lowest quartile of recent weight and WHR ≤0.80. Results were similar when we examined recent BMI and WHR. For women with above median WHR, the respective OR by quartile of recent BMI was 1.00, 1.00, 1.59, 1.73 (p trend = 0.012); risk was unrelated to recent BMI for women with below median WHR (p trend = 0.42).

Table IV. Odds Ratio (95% Confidence Interval)1 of Breast Cancer in Association With Recent Weight and Waist/Hip Ratio (WHR) Among Postmenopausal Women
WHRRecent Weight (Kg)p trendWHR Adj for recent weight
≤49.9>49.9–55.4>55.4–61.3>61.3
  • 1

    Conditional logistic regression models with matched sets defined jointly by age and Asian ethnicity. Birthplace and duration of residence in the US, education, age at menarche, number of live births, menopausal status and age at menopause, intake of tea and soy during adolescence and adult life and years of physical activity were included as covariates.

≤0.801.000.97 (0.59–1.57)0.89 (0.54–1.45)1.38 (0.75–2.53)0.881.00
>0.800.95 (0.56–1.65)1.19 (0.75–1.90)1.28 (0.84–1.96)1.67 (1.10–2.55)0.0131.31 (0.89–1.95) p = 0.14
Recent weight adjusted for WHR1.001.01 (0.70–1.47)1.03 (0.71–1.49)1.43 (0.95–2.15)0.097 

The increased risk associated with recent weight in postmenopausal women was strongest for the most recent migrants. The respective ORs by quartile of recent weight after adjusting for WHR were 1.00, 1.16, 1.23 and 2.33, respectively (p trend = 0.039) for Asian migrants who had lived in the United States for 20 years or fewer; 1.00, 1.02, 1.24 and 1.55 (p trend = 0.11) for Asian migrants who had lived in the United States for 21+ years; and 1.00, 1.48, 1.05 and 1.42 (p trend = 0.65) for US born Asians. Results were similar using recent BMI (data not shown). Likewise, the magnitude of risk for weight gain (recent weight–weight at age 18) and WHR was highest in recent immigrants. Relative to no weight gain, the OR associated with weight gain of 20 kg or more was 1.28 (95% CI = 0.46–3.56) in US born women, 1.95 (95% CI = 0.74–4.99) in long-term immigrants and 3.46 (95% CI = 0.94–12.80) in recent immigrants. Relative to a WHR value of 0.75 or lower, the OR associated with WHR of >0.84 was 1.55 (95% CI = 0.69–3.49) in US born Asians, 1.14 (95% CI = 0.58–2.21) in long-term immigrants and 3.11 (95% CI = 1.05–9.20) in recent immigrants. However, these differences in ORs by birthplace/duration of residence in the United States were not statistically significant.

Recent weight was highly correlated with weight at age 18 (and at other ages), waist and hip circumferences and WHR. The corresponding Pearson correlation coefficients between recent weight and weight at age 18, waist circumference, hip circumference and WHR were 0.44, 0.71, 0.75 and 0.32. Stepwise regression identified recent weight as the strongest predictor of breast cancer risk while the effects of the other body size variables were small when height, age, Asian ethnicity, education, nativity and duration of residence in the United States were already included in the model. Thus, recent weight was adjusted for in subsequent analysis on HT and risk.

Table V shows relative risk patterns according to menopausal HT use. We found a significant increased risk in relation to current use of EPT but not with ET use; the adjusted OR was 1.26 per 5 years of EPT use (p trend = 0.017) and 0.99 per 5 years of ET use (p trend = 0.95). Duration of past ET and past EPT was not significantly associated with risk. Results were similar when we restricted the analysis to women who had a natural menopause or a bilateral oophorectomy; the adjusted OR was 1.28 per 5 years of use for current EPT users (p trend = 0.026) and 1.12 per 5 years of use for current ET users (p trend = 0.33). We also investigated the HT-breast cancer association according to recent weight. The risk estimates per 5 years of use for current EPT and ET users were not noticeably different for women with lower (e.g., below median recent weight or lowest three quartiles of recent weight) versus those with higher body weight (e.g., above median recent weight or highest quartile of recent weight). All p values for the HT-body size interactions were greater than 0.05 (data not shown).

We investigated whether timing of menopause modified the recent weight–breast cancer association in women who achieved menopause naturally or by bilateral oophorectomy (Table VI). Increasing recent weight was a significant risk factor for women who have been menopausal for less than 15 years as well as for those who have been menopausal for 15 years or more. The risk ratios for all categories of recent weight were uniformly larger for women who achieved menopause in the more distant past. When current HT users were excluded from the analysis, the larger ORs in women with a longer history of menopause became more apparent. For women who had been menopausal for 15+ years, those who weighed >61.3 kg experienced a significant 4.08 (95% CI = 1.46–11.43) fold increase in risk relative to those who weighed under 49.9 kg (p trend=0.005). The corresponding OR for women with a shorter duration of menopause was 1.68 (95% CI = 0.82–3.44) (p for interaction = 0.16). Timing of menopause influenced the recent BMI–breast cancer and the WHR–breast cancer association in a similar manner. The adjusted OR for the highest versus the lowest quartile of recent BMI was 3.59 (95% CI = 1.37–9.41) for women who had been menopausal for 15+ years while the comparable OR for women with a shorter duration of menopause was 1.09 (95% CI = 0.56–2.13). The adjusted OR for the highest versus the lowest quartile of WHR was 2.63 (95% CI = 0.92–7.49) for women who had been menopausal for 15+ years while the comparable OR for women with a shorter duration of menopause was 1.25 (95% CI = 0.60–2.62).

Table V. Menopausal Hormone use in Postmenopausal Breast Cancer Cases and Controls
 All postmenopausal womenNatural menopause or bilateral oophorectomy
Cases (n = 722)Controls (n = 555)Adjusted OR1Adjusted OR2 (95 % CI)Cases (n = 563)Controls (n = 436)Adjusted OR1Adjusted OR2 (95 % CI)
  • 1

    See footnote 1 in Table IV.

  • 2

    Recent weight and height are included in the model.

  • 3

    For all postmenopausal women, 87 cases and 73 controls used HT for <1 yr and they were combined with never users (336 cases, 270 controls) in the analyses. For women who had natural menopause or bilateral oophorectomy, 65 cases and 59 controls used HT for <1 yr and they were combined with never users (291 cases, 237 controls) in the analyses.

Never/short term34233431.001.003562961.001.00
Current hormone therapy
 ≥1 to <5yr87651.191.22 (0.81–1.82)67431.421.46 (0.91–2.34)
 ≥5 to <10 yr77631.091.07 (0.70–1.62)51401.301.30 (0.78–2.14)
 ≥10 yr89531.411.44 (0.91–2.22)56341.551.57 (0.91–2.72)
 Per 5 yr  1.101.10 (0.96–1.26)  1.161.17 (0.99–1.37)
Current estrogen alone
 ≥1 to <5yr31231.311.31 (0.69–2.48)21151.471.46 (0.66–3.25)
 ≥5 to <10 yr34310.960.92 (0.51–1.66)20181.231.21 (0.55–2.66)
 ≥10 yr43340.970.98 (0.55–1.74)30191.411.42 (0.67–3.03)
 Per 5 yr 1.000.99 (0.83–1.19)  1.121.12 (0.89–1.42)
Current estrogen + progestin
 ≥1 to <5yr62491.021.06 (0.67–1.67)49341.141.18 (0.70–1.99)
 ≥5 to <10 yr50331.421.41 (0.83–2.37)34221.681.67 (0.90–3.10)
 ≥10 yr33112.552.59 (1.20–5.60)2082.102.15 (0.87–5.30)
 Per 5 yr  1.251.26 (1.04–1.52)  1.271.28 (1.03–1.58)
Former hormone therapy
 ≥1 to <5yr22181.151.15 (0.58–2.29)15141.191.19 (0.53–2.67)
 ≥524131.511.51 (0.69–3.31)1892.022.02 (0.78–5.22)
 Per 5 yr  1.211.21 (0.86–1.72)  1.351.35 (0.89–2.05)
Table VI. Odds Ratio (95% Confidence Interval)1 of Breast Cancer in Postmenopausal Women and Body Size Stratified by Number of Years Between Age at Menopause And Reference Date
 Natural menopause + bilateral oophorectomyNatural menopause + bilateral oophorectomy excluded current HT users
<15 yr since menopause (346 cases/287 controls)15+ yr since menopause (200 cases/141 controls)<15 yr since menopause (221 cases/207 controls)15+ yr since menopause (151 cases/105 controls)
  • 1

    Conditional logistic regression models with matched sets defined jointly by age and Asian ethnicity. Birthplace and duration of residence in the US, education, age at menarche, number of live births, intake of tea and soy during adolescence and adult life, years of physical activity and height were included as covariates.

Recent weight
 ≤49.91.001.001.001.00
 >49.9–55.41.13 (0.67–1.91)1.28 (0.57–2.84)0.86 (0.43–1.72)2.32 (0.87–6.14)
 >55.4–61.31.11 (0.65–1.89)2.61 (1.20–5.66)0.94 (0.47–1.87)5.52 (1.97–15.51)
 >61.31.92 (1.11–3.32)2.38 (1.04–5.44)1.68 (0.82–3.44)4.08 (1.46–11.43)
 P trend0.0230.0120.140.005

Table VII shows the relationship between recent weight and breast cancer risk separately for non-HT users, current ET alone users and current EPT users. The effects of recent weight did not differ significantly between current ET users, current EPT users and noncurrent users (p for interaction effect = 0.75). Similar results were observed when we restricted the analyses to women who had a natural menopause or a complete hysterectomy. Likewise, there was no evidence that HT use modified the associations between breast cancer risk and either recent BMI or WHR (data not shown).

Table VII. Odds Ratio (95% Confidence Interval)1 of Breast Cancer in Postmenopausal Women and Recent Weight Stratified by Menopausal Hormone Use
 Use of menopausal hormones
Non-current HT user (never + Ex)Current ET alone usersCurrent EPT users
  • 1

    Conditional logistic regression models with matched sets defined jointly by age and Asian ethnicity. Birthplace and duration of residence in the US, education, age at menarche, number of live births, intake of tea and soy during adolescence and adult life, years of physical activity and height were included as covariates.

  • 2

    Number of cases and controls in each category of recent weight and menopausal hormone use.

Recent weight
 ≤49.91.00 [101/82]20.90 (0.44–1.84) [26/24]1.10 (0.57–2.12) [31/26]
 >49.9–55.40.97 (0.62–1.51) [98/91]1.09 (0.52–2.29) [26/21]1.79 (0.90–3.56) [33/20]
 >55.4–61.31.21 (0.78–1.87) [120/96]0.81 (0.41–1.60) [28/31]1.28 (0.67–2.46) [40/27]
 >61.31.41 (0.90–2.20) [138/98]2.83 (1.22–6.56) [27/12]2.18 (1.09–4.35) [41/19]
 Mean weight (case/control)58.25/57.5257.39/56.0257.95/56.24

Discussion

  1. Top of page
  2. Abstract
  3. Subjects and methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. References

Historically, breast cancer incidence rates have been low in Asia. However, in recent years, breast cancer incidence rates have increased rapidly in Asian countries including Japan and Singapore. They have also risen sharply in Asian–Americans.1, 17 Between 1993 and 1997, the annual rate of increase in invasive breast cancer for women over 50 years of age was 4 times higher in Asian women (6.3% per year) than in non-Hispanic White women (1.5% per year) in Los Angeles County.1 Similar increases have been observed using incidence data from the California Cancer Registry and the Surveillance, Epidemiology and End Results Program.

This population-based case–control study of breast cancer for Asian–American women in Los Angeles County was designed to investigate reasons for the large increase in breast cancer incidence in Asian–American women. Consistent with reports by others,18, 19, 20 we found a trend of decreasing risk with increasing recent weight (and recent BMI) in premenopausal women. In contrast, high recent body weight (and recent BMI) was a significant, positive risk factor in postmenopausal Asian–American women, especially for those with high WHR. Weight gain over the lifetime was also associated with a significant increased risk in postmenopausal Asian–American women.21, 22, 23 Current use of replacement hormones, especially the combined use of estrogen and progestin, was another significant risk factor for postmenopausal breast cancer in Asian–American women.

A pooled analysis of 7 prospective cohort studies including 4,385 incident breast cancers identified in over 330,000 Caucasian women reported a combined RR of 1.06 (95% CI = 1.03–1.10) per 10 kg increment in body weight and 1.07 (95% CI = 1.02–1.11) per 4 kg/m2 increase in BMI.20 In this study of Asian–American women, recent weight was strongly associated with an increased risk of postmenopausal breast cancer (Table III). We observed a 16% increase in risk per 10 kg increase in recent weight and a 13% increase in risk per 4 kg/m2 increase in BMI for postmenopausal Asian–American women in Los Angeles County. The stronger effects of recent weight (and recent BMI) may be related, in part, to Asians having a higher percentage of body fat at the same BMI level as Whites. In a study conducted in Whites and Asians in New York City, Wang et al. found that the percent body fat was 3–5% higher in Asians than Whites of comparable BMI. In terms of percent upper body fat (abdomen and umbilicus), Asians were 16% higher than their White counterparts.24

There are supportive data indicating that Asian–American women are heavier today than those of a generation ago.25 In a case–control study of breast cancer conducted in the mid to late 1970s in Japanese–American women of Hawaii,4 the mean of self-reported current weights of control subjects (average age of 58.3 years) was 53.5 kg. The means of their self-reported weights at ages 20 and 30 were 48.0 kg and 50.2 kg, respectively. In our Japanese control subjects who were postmenopausal, the mean of their self-reported current weights was 57.0 kg while the means of their self-reported weights at ages 18 and 30 were 48.8 kg and 51.7 kg, respectively.

In postmenopausal women, the effect of recent weight on risk was stronger for women with an above versus below median WHR (Table IV). The suggested interaction of recent weight and WHR on risk may be explained by the deleterious effects of both high recent weight and WHR on circulating levels of sex hormones and that high WHR additionally influences the insulin pathway. There is a consistent body of literature showing that in postmenopausal women high BMI is positively associated with higher estradiol level and is inversely associated with levels of sex-hormone binding globulins (SHBG), resulting in higher free estradiol levels.26 High WHR (or waist circumference) is also negatively associated with levels of SHBG although the effects of WHR on circulating estrogen and androgen levels are less consistently reported.27, 28, 29 High abdominal adiposity, decreased levels of SHBG and increased levels of androgens have been associated with hyperinsulinemia and diabetes30 and both conditions have been found to be independent risk factors for breast cancer in some studies.31, 32

There is increasing recognition that Asians are especially prone to developing hyperinsulinemia even at relatively low body size (i.e., BMI <25 kg/m2) and that they may be more susceptible to a spectrum of chronic diseases such as diabetes and heart disease even at relatively low or average BMI.33 Reasons for this are not understood but there is suggestive evidence that Asians may have higher circulating levels of androgens compared with Whites and other ethnicities. In a cross-sectional study of postmenopausal Japanese–American, Whites, African–Americans and Hispanics in Hawaii and Los Angeles County, we found levels of androstenedione levels were highest in Japanese–American women.34 In the Study of Women's Health Across the Nation, perimenopausal Chinese and Japanese women had higher levels of dehydroepiandrosterone sulfate relative to Whites, Hispanics and African–American women.35 Further studies are warranted to determine whether the higher levels of total circulating androgens in Asians may help explain, in part, their susceptibility to hyperinsulinemia despite their relatively low/normal weight. There is also accumulating evidence that high androgen levels have direct deleterious effects on breast cancer risk.3, 36

Our findings on the association between HT use and breast cancer risk in Asian–American women are compatible with previous studies, in particular, an increase in risk associated with EPT use.37 In a recent overview on EPT and breast cancer risk that included results from randomized trials as well as observational studies, one of us (MCP) estimated an increase of 5.2% in the risk of breast cancer per year of use of EPT in US women (or a 29% increase in risk per 5 years of use).37 Thus, our finding of a 26% increase in breast cancer risk per 5 years of use in current users of EPT is compatible with the literature's overall estimate in US women. The increases associated with HT use for Asian–American women are also similar to those observed for Japanese–Americans in the Multi-ethnic Cohort Study who experienced a 33% increase in breast cancer risk per 5 years of use of EPT and a 18% increase in risk per 5 years of use of ET.38

The prevalence of ever HT use (including HT use of <1 year) in our control subjects was 52%, identical to the prevalence of use reported in Los Angeles County Whites and African–Americans during the late 1980s and early 1990s.39 Prevalence of HT use was highest in US-born Asians (69%), intermediate (55%) in long-term migrants (in the United States for 21+ years) and lowest (31%) in recent migrants (in the United States for 20 years or fewer). Our subjects indicated a much higher prevalence of HT use relative to women in Japan (∼6%),40 Singapore (∼5%)41 and Shanghai, China (∼3%),19 and Asian–Americans surveyed in the late 1970s.5 Nomura et al.5 reported that 51.3% of Japanese–American control women in Hawaii in the late 1970s were ever users of HT, 16.7% used HT for 1–12 months, 20.6% for 13–72 months and 14.0% for 73+ months. For our Japanese–American control subjects, 66.3% were ever users of HT, 16.6% used HT for 1–12 months, 25.7% for 13–72 months and 24.0% for 73+ months.

Some studies suggest that the effects of obesity on postmenopausal breast cancer may be strongest 10–15 years after menopause.42, 43 These observations have been interpreted to support the notion that prolonged exposure to the proliferative effects of elevated circulating estrogens from adipose tissue is needed. Results from our study are supportive of these findings. A stronger effect of recent weight, recent BMI and WHR was seen for women who have been menopausal for 15+ years, especially when current HT users were excluded from the analysis (Table VI).

In this population of Asian–American women, the pattern of increasing risk with increasing recent weight was observed in current HT users as well as in non-hormone users (Table VII). In most studies conducted in western countries, stronger effects of body size for non-HT users and stronger effects of HT for leaner women are usually found.8, 9, 42, 43, 44 However, in some studies, increases in risk associated with HT use are still evident in women with very high (>30+) BMI.38 A recent cohort in France reported a weight effect that was actually more apparent in HT users than in non-HT users. The women in this French cohort also were relatively thin.45 Our results may be explained by the fact that virtually the entire spectrum of absolute weight measurements across our study subjects fall within the “lean” range according to western standards and that the average weight of non-current HT users and HT users were quite comparable. Therefore, it should not be surprising that the effect of exogenous hormone on breast cancer risk is still apparent against the background effect of endogenous estrogens that is derived from peripheral fat.

Strengths of this study include a large population-based study including three Asian ethnic groups in Los Angeles County. To assess anthropometry, we collected lifetime information on body size, including body weight for each decade of age and obtained measurements of height, weight and waist and hip circumferences using standardized procedures for all study participants. This allowed us to examine the effects on risk of decade-specific weights and to evaluate the relative importance of early versus more recent weights. Our questions on HT use included dates, brands and dosages and reasons for prescribing and stopping use of each regimen. Because of the comparable methods that were used, we could compare our risk estimates on HT-breast cancer with results conducted in other populations.39

A limitation of this case–control study is that we interviewed 63% of the cases we contacted. The largest loss was due to refusal by the cases or that we failed to contact them. We compared cases who refused or could not be contacted with cases we interviewed in terms of age, social class (based on census tract of residence), birthplace and tumor stage at diagnosis. The cases we did not interview were somewhat older (mean age was 54.3) than cases we interviewed (mean age was 52.9) but they were similar in terms of birthplace. Birthplace information on cases was available in 81% of the cases we identified from the CSP. For cases with known birthplace, 20% of those we interviewed were US born compared with 17% of subjects who were not interviewed. We examined the social class of cases interviewed and those not interviewed based on census tract of residence. Of the cases interviewed, 29% resided in census tracts that were classified as high social class (census 1) compared with 26% of subjects we did not interview. Finally, cases interviewed and those not interviewed did not differ in terms of stage of disease at diagnosis; 31% of those interviewed and 31% of those not interviewed had regional/metastatic cancer at the time of diagnosis.

In summary, in this population-based case–control study of breast cancer in Asian–American women, recent weight, especially for women with high waist and hip ratios, and use of replacement hormones, especially the combined use of estrogen and progestin, were significantly and positively related to postmenopausal breast cancer risk. The magnitude of associations associated with high body weight (BMI) and with EPT use in Asian–Americans is similar to those observed in western populations. Asian–Americans, especially postmenopausal women, have steadily become heavier in the last 30 years. Prevalence of HT use and long-term HT use also have increased steadily in Asian–Americans. These modifiable lifestyle factors might have contributed substantially to the increase of postmenopausal breast cancer in Asian–American women.

Acknowledgements

  1. Top of page
  2. Abstract
  3. Subjects and methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. References

The authors thank all the study participants for their contributions and support. The authors also thank the entire data collection team, especially Annie Fung, Lydia Tran, June Yashiki and Sushma Jain. Incident breast cancer cases for this study were collected by the USC Cancer Surveillance Program (CSP), which is supported under subcontract by the California Department of Health. The CSP is also part of the National Cancer Institute's Division of Cancer Prevention and Control Surveillance, Epidemiology, and End Results Program, under contract number N01CN25403.

References

  1. Top of page
  2. Abstract
  3. Subjects and methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. References