Risk Factors for Breast Cancer in the Breast Cancer Risk Model study of Guam and Saipan

Abstract

Background

Chamorro Pacific Islanders in the Mariana Islands have breast cancer incidence rates similar to, but mortality rates higher than, those of U.S. women. As breast cancer risk factors of women of the Mariana Islands may be unique because of ethnic and cultural differences, we studied established and suspected risk factors for breast cancer in this unstudied population.

Methods

From 2010–2013, we conducted retrospective case-control study of female breast cancer (104 cases and 185 controls) among women in the Mariana Islands. Odds ratios (ORs) and 95% confidence intervals (CIs) were calculated for each of various lifestyle-related factors from logistic regression of breast cancer, in all women and in pre- and postmenopausal women separately. Tests for interaction of risk factors with ethnicity were based on the Wald statistics for cross-product terms.

Results

Of the medical and reproductive factors considered — age at menarche, breastfeeding, number of live births, age at first live birth, hormone use, and menopause — only age at first live birth was confirmed. Age at first live birth, among parous women, was higher among cases (mean 24.9 years) than controls (mean 23.2 years); with increased breast cancer risk (OR=2.53; 95% CI, 1.04–6.19 for age ≥ 30y compared to <20y, P for trend = 0.01). Of the lifestyle factors —body mass index, waist circumference, physical activity, alcohol and betel-nut intake, and education — only waist circumference (OR=1.65; 95% CI 0.87–3.14 for the highest tertile group compared to the lowest, P for trend=0.04) was significantly associated with breast cancer risk and only in Filipino women. The association with many other established risk factors, such as BMI, hormone use and physical activity, were in the expected direction but were not significant. Associations for family history of breast cancer and alcohol intake were not evident

Conclusions

The results provide a basis for cancer prevention guidance for women in the Mariana Islands.

1. INTRODUCTION

The Mariana Islands are located in the northwestern Pacific Ocean, approximately 3700 miles west of Hawaii, 6000 miles west of California, and 1300 miles southeast of Japan. They consist of 15 islands with a total land area of 389 square miles and are divided into two administrative units: Guam, a U.S. Territory; and the (U.S.) Commonwealth of the Northern Mariana Islands (CNMI), which includes the islands of Saipan, Tinian, and Rota. Guam is the largest and most populous island of the group. Chamorros are the original inhabitants of Guam and the Marianas Islands, and are typically grouped with other Pacific Islanders, however genetically, they are primarily Malayo-Polynesian maritime voyagers who first settled in Guam 3,500 years before. Gene flow from the Carolinians and, later, from Filipinos and Spanish after the 17^th century, have contributed a unique genetic mixture to the Chamorros (1–6). The current population of Guam is characterized by substantial ethnic variation (7): 37% Chamorro, 26% Filipino, 7% White, 7% other Asian, 12% other Pacific Islander, and 11% other ethnicity. CNMI is also diverse; its ethnic breakdown includes 32% Chamorro, 35% Filipino, 4% Carolinian, 13% other Pacific Islander, 14% other Asian, and 2% other ethnicity (8). This ethnic diversity evolved through centuries of colonization and migration that continues today (9). Spanish colonization of the islands between 1521 and 1898 and the subsequent U.S. possession since then have led to substantial ethnic and cultural mixing.

Breast cancer, previously considered a disease of Western industrialized nations (10), is now the most common malignancy among women worldwide (11, 12). Even among women from other Pacific regions, such as the Maori and Pacific islanders in New Zealand, breast cancer is the most important malignancy (13, 14). In the United States (U.S.), breast cancer is the second most common cause of cancer mortality among women, and it has also been the second leading cause of cancer mortality among women on Guam over the last three decades (15), second only to lung cancer. Data are not available for Saipan. For the period 1998–2002, Chamorro women had the highest breast cancer mortality rate on Guam, at 32 per 100,000 women; rates were 26 for Whites, 6 for Filipinos, and 16 for other Asians (age-adjusted to the U.S. 2000 standard population) (16). The overall U.S. mortality rate for that time period was 28 per 100,000. Incidence rates for breast cancer on Guam are lower than among U.S. women, although they may be underestimated, as availability of mammography has been limited. According to the most recent Guam Cancer Facts and Figures (2008–2012) (17), breast cancer accounts for 30% of new cancer cases among women on Guam and for 14% of cancer deaths. The age-adjusted incidence rate of breast cancer for 2008–2012 on Guam was 97 per 100,000 for Chamorro women and 77 for Filipino women (17). The breast cancer mortality rate for this period was 20.8 for Chamorro women and 21.9 for women in the U.S.

Many risk factors for breast cancer have been established. Its etiology involves both genetic and behavioral factors. Increasing age is a major risk factor. Risk increases with inherited genetic mutations, such as in the BRCA1 and BRCA2 genes, and a personal or family history of breast cancer (18). Other risk factors include never having children, greater height, benign breast disease, early menarche, late menopause (after age 54), ionizing-radiation exposure, postmenopausal obesity, physical inactivity, and alcohol intake (19, 20). Breastfeeding, moderate/vigorous physical activity, and maintaining a healthy body weight decrease breast cancer risk (17, 18). High breast tissue density (5) and high circulating levels of estrogens (21–24) have also been associated with higher breast cancer incidence. Abdominal adiposity, as measured by waist circumference and waist-to-height ratio, are also considered risk factors (25–28).

The reasons for the higher breast cancer incidence and mortality rates among Chamorro Pacific Islanders than other ethnic groups living in the Mariana Islands are poorly understood, as no clinical and epidemiological studies of breast cancer have been conducted on Guam or Saipan. Women of the Marianas may be subject to unique risk factors, as different breast cancer risk models have been found to be needed for adequate estimation of the risks of women of diverse ethnic backgrounds (28). To study the reasons for the disparate rates of breast cancer in different ethnic groups in the Mariana Islands (15, 16, 29), the BRISK (Breast Cancer Risk Model) Project was conducted to improve understanding of the risk factors for breast cancer in this region. Here, we discuss established and suspected risk factors for breast cancer in this population, including demographics, reproductive factors, body size, and alcohol intake.

2. MATERIALS & METHODS

2.1 Overview of Study Design

BRISK is a retrospective case-control study that was designed to assess breast cancer risk factors and to develop a model for predicting breast cancer risk in Pacific populations of Asians and Pacific Islanders living on Guam and Saipan. The study was conducted from 2010 to 2013 by researchers at the University of Guam and the University of Hawaii Cancer Center under the auspices of an NCI-sponsored partnership (U54-CA-143727) and was approved by Institutional Review Boards at both institutions. The University of Guam lead investigator and project staff members were of Chamorro heritage.

Recruitment required identification of women diagnosed with breast cancer (cases) and women of similar age and ethnicity without a diagnosis of breast cancer (controls) who were willing to participate. Breast cancer cases were identified through the Guam Cancer Registry, CNMI Department of Public Health, and health clinics on Guam. Eligibility criteria for cases were (1) primary, invasive breast cancer newly diagnosed between 2009 and 2012; (2) no prior history of cancer (other than skin cancer); (3) residence on Guam or Saipan for at least 5 years; (4) ability to provide consent for the study; and (5) age between 25 and 80 years. The majority of breast cancer cases were identified through the Guam Cancer Registry. A more in-depth description of the recruitment of breast cancer cases from this underserved minority population is given by Leon Guerrero et al. (30).

Control participants on both Guam and Saipan were recruited from among women undergoing screening mammography in local clinics/facilities and other community-based settings (health fairs, senior citizen centers). The research team also used posted printed media (e.g. flyers and posters) at selected sites, and broadcast mass media (television). Control participants were also encouraged to “spread the word” and tell their friends and family members about the study. Women interested in participating in the study as ‘control participants’ were encouraged to contact the research team in either Guam or Saipan to determine eligibility. Eligibility criteria for controls were (1) no prior history of cancer (other than skin cancer); (2) residence on Guam or Saipan for at least 5 years; (3) ability to provide consent for the study; and (4) age between 25 and 80 years. Controls were frequency-matched to cases on age, ethnicity, and location (Saipan or Guam).

Once willing and eligible women were identified as potential participants, a package containing an introduction letter, the consent form, an informational flyer, and study questionnaires was mailed to each participant. A trained interviewer then called the potential participant to assess eligibility further and to schedule an appointment for interview.

Of the 275 cases contacted, 38% agreed to participate, 21% were ineligible, and 41% refused; 48% of the eligible cases participated. The corresponding percentages for the 250 controls contacted were 74%, 20% and 6%; 93% of the eligible controls participated. The study included 104 breast cancer cases (83 from Guam and 21 from CNMI) and 185 controls (140 from Guam and 45 from CNMI) between 27 and 80 years of age. The main reasons for refusal were participant mistrust of research, cultural shame about having breast cancer, and unwillingness of many newly diagnosed breast cancer cases to participate because they were still overwhelmed with their diagnosis.

Participants were interviewed in face-to-face sessions, those on Guam at the University of Guam Cancer Research Center Office and those on Saipan at the CNMI Department of Public Health. During the interviews, participants were asked to complete a detailed questionnaire on exposure to risk factors before a reference date, which was defined as the diagnosis date for cases and as the interview date for controls. The questionnaire included demographic information such as age, date of birth, date of diagnosis (cases), education level, place of birth, and number of years living in the Mariana Islands. Participants were asked to identify their ethnicity and to provide a percentage ethnic breakdown of their parents (e.g., 50% Chamorro and 50% White), as well as birthplaces for both parents and grandparents. Participants were also asked about personal and family history of selected diseases and conditions such as cancers, Alzheimer’s, Parkinson’s, cardiovascular disease, and diabetes; tobacco and betel-nut use; menstrual and reproductive histories; pregnancy history; details about each pregnancy (including outcome, age at birth of child); breastfeeding history (number of months per child); use of oral contraceptives (type and duration); use of estrogen and progesterone/progestin hormone therapy (type and duration); breast biopsy history; recreational physical activity and alcohol consumption in various life periods; and acculturation based on a questionnaire used in a multiethnic study (31, 32).

In addition, alcohol intake and betel-nut consumption, along with a comprehensive diet history, during the year previous to the reference date were measured with the Marianas Food Frequency Questionnaire (33).

Current waist circumference, weight, height, and sitting height were measured by one of two anthropometrists who were trained by one of the study investigators (RN). A stadiometer (Seca, Germany) was used to measure height and sitting height, and a digital scale (ProFit Lifesource, Milpitas, CA) was used to measure weight. Body mass index (BMI) was calculated as kg/m². Waist circumference (WC) was measured to the nearest 0.1 centimeter with an inelastic tape measure (Seca Model 201; Chino, CA) maintained in a horizontal plane, with the subject standing comfortably with weight distributed evenly on both feet. The measurement was taken at the level of the umbilicus (34). Two independent readings were taken for WC, and the mean of the two was used in analysis. The waist-height ratio (WHtR) was calculated as WC in cm divided by height in cm. In addition, height, weight, BMI, and body-shape history of subjects was assessed by questionnaire at seven ages (excluding pregnancies): 15, 20, 30, 40, 50, and 60 years and the reference date.

2.2 Statistical Analysis

The goal of the present analysis was to examine established or suspected risk and protective factors for breast cancer in this population. We investigated each potential risk factor individually in unconditional logistic models, adjusting for ethnicity and age. Exposures were parameterized as indicator variables measuring presence/absence for a dichotomous factor, such as betel-nut chewing, or as indicator variables representing group membership, such as predefined range categories or quantiles of a continuous exposure, computed among both cases and controls. Trend variables were defined by the continuous values of the exposure variables.

Body size was parameterized in a variety of ways at different times of adult life. BMI was defined as underweight (<18.5), “healthy” (18.5–24.9), overweight (25.0–29.9), or obese (≥30), and healthy was used as the reference category, based on the WHO definitions. In addition to finer categorizations, WC was parameterized as ≤88 or >88 cm, and WHtR was parameterized as ≤0.5 and >0.5 on the basis of cutpoints for a reduction in cardiometabolic outcomes (35, 36). We examined the effect of weight gain from age 20 to 50 years. We also examined the association between health behaviors that may influence risk, such as physical activity and tobacco and alcohol use, as well as exposures unique to the population under study, such as betel-nut chewing. Physical activity was described as average hours of moderate and vigorous activity per day, hours of sedentary activity, as well as metabolic equivalents (METS) based on activity at mild, moderate, and vigorous levels (37).

A summary ethnicity variable was created on the basis of each participant’s reported ethnic breakdown of her mother’s and father’s ethnicities and a desire to provide information on as many reasonably homogeneous groups as the sample size permitted. In sequence, participants with any Chamorro background were characterized as “Any Chamorro,” then any remaining participants with any Filipino background were characterized as “Filipino,” then any remaining participants with any Micronesian background were characterized as “Micronesian,” and the rest of the participants were characterized as “Other.” For our study, we defined “Micronesian” as belonging to the indigenous peoples of Micronesia who were not Chamorro. We could not examine smaller race/ethnic categories (White, African-American, and Asian) separately because cell sizes were too small for statistical analysis. Participants’ “generation” in the Mariana Islands was determined by assessment of birthplace of participants, their mothers, and their grandmothers. Participants were considered “1^st generation” if they were born outside the Marianas, “2^nd generation” if they were born in the Marianas but their mothers were born elsewhere, and “3^rd generation” if they and their mothers were born in the Marianas but their grandmothers were born elsewhere. Percentage of lifetime lived in the Marianas was calculated as number of years reported living in the Marianas divided by total age at reference date.

A participant was considered postmenopausal if her most recent menstrual period was more than 12 months before the reference date (natural menopause) or if she had undergone surgical menopause prior to the reference date. Women were considered premenopausal if, at the reference date, they were still menstruating or perimenopausal if menstrual bleeding had become irregular. Perimenopausal women were grouped with premenopausal women for analyses (see Table 1).

Table 1.

Characteristics^† of breast-cancer cases and controls in the BRISK study of Guam and Saipan. Reference was diagnosis date for cases, interview date for controls.

Characteristic	Cases (n = 104)	Controls (n = 185)	p-value
Age at reference, years (mean ± SD)	55.5 ± 10.4	54.1 ± 10.4	0.26
<40	7 (6.7)	13 (7.0)	0.88
40–49	25 (24.0)	54 (29.2)
50–59	38 (36.5)	64 (34.6)
60–69	23 (22.2)	39 (21.1)
≥70	11 (10.6)	15 (8.1)
Ethnicity
Chamorro	53 (51.0)	112 (60.5)	0.32
Filipino	34 (32.7)	46 (24.9)
Micronesian	8 (7.7)	9 (4.9)
Other	9 (8.7)	18 (9.7)
Highest education level completed
High school diploma or less	48 (46.2)	77 (41.6)	0.94
Some college	28 (26.9)	58 (31.4)
College degree or more	28 (26.9)	50 (27)
Percentage of lifetime lived in Mariana Islands	72.2 ± 28.8	72.4 ± 26.6	0.96
Age at menarche, yearsˆ
<12	24 (23.1)	51 (27.9)	0.65
12–13	44 (42.3)	75 (41.0)
≥14	36 (34.6)	57 (31.2)
Ever been pregnant	95 (91.4)	170 (91.9)	0.87
Total number of pregnancies
0	9 (8.7)	15 (8.1)	0.71
1–2	32 (30.8)	46 (24.9)
3–4	37 (35.6)	70 (37.8)
5 or more	26 (25.0)	54 (29.2)
Number of live births
Nulliparous	11 (10.6)	19 (10.3)	0.21
1–2	42 (40.4)	55 (29.7)
3–4	30 (28.8)	74 (40.0)
5 or more	21 (20.2)	37 (20.0)
Age at first live birth, years, parous women only (mean ± SD)ˆ	24.9 ± 5.7	23.2 ± 5.2	0.01
<20	23 (24.7)	51 (31.1)	0.06
20–24	26 (28.0)	63 (38.4)
25–29	27 (29.0)	33 (20.1)
≥30	17 (18.3)	17 (10.4)
Ever breastfed, parous women only
No	24 (25.8)	46 (27.7)	0.74
Yes	69 (74.2)	120 (72.3)
Number of first-degree relatives with breast cancer
0	92 (88.5)	155 (83.8)	0.13
1	8 (7.7)	27 (14.6)
2	4 (3.8)	3 (1.6)
Hormone useˆ
Never used estrogen or progesterone	88 (87.1)	156 (84.8)	0.91
Yes, previously	11 (10.9)	23 (12.5)
Yes, currently	2 (2.0)	5 (2.7)
Menopausal status
Pre	26 (25.0)	55 (29.7)	0.22
Peri	6 (5.8)	19 (10.3)
Post	72 (69.2)	111 (60.0)
Age at menopause, years, postmenopausal women only
<46	21 (29.2)	33 (29.7)	0.66
46–53	37 (51.4)	62 (55.9)
≥54	14 (19.4)	16 (14.4)
Type of Menopause, postmenopausal women only
Natural	52 (72.2)	87 (78.43)	0.07
Medication	3 (4.2)	0
Surgical	16 (22.2)	24 (21.6)
Unknown	1 (1.4)	0
Height, cm (mean ± SD)	156.7 ± 6.5	156.1 ± 6.9	0.50
Sitting height, cm (mean ± SD)ˆ	82.5 ± 3.4	82.4 ± 5.8	0.89
Body mass Index, kg/m2 (mean ± SD)	30.4 ± 7.1	30.2 ± 7.4	0.79
<18	0	1 (0.5)	0.11
18–24.9	20 (19.2)	52 (28.1)
25–29.9	37 (35.6)	52 (28.1)
30–34.9	27 (26.0)	32 (17.3)
≥35	20 (19.2)	48 (26.0)
Waist Circumference, cm (mean ± SD)	98.5 ± 15.0	94.5 ± 15.0	0.03
Tertile 1 (≤89)*	27 (27.8)	63 (36.6)	0.26
Tertile 2 (89.1–99.5)	32 (33.0)	56 (32.6)
Tertile 3 (>99.5)	38 (39.2)	53 (30.8)
Waist/Height Ratio (mean ± SD)ˆ	0.63 ± 0.09	0.61 ± 0.10	0.07
Quartile 1 (≤0.54)*	17 (17.5)	51 (29.7)	0.18
Quartile 2 (0.55–0.62)	27 (27.8)	39 (22.7)
Quartile 3 (0.62–0.67)	26 (26.8)	41 (23.8)
Quartile 4 (>0.67)	27 (27.8)	41 (23.8)
Recreational physical activity, Current MET-hr/week (mean ± SD)	19.6 ± 36.6	21.5 ± 37.5	0.68
None	44 (42.3)	71 (38.4)	0.63
<15.0	24 (23.1)	39 (21.0)
15.0–24.9	11 (10.6)	27 (14.6)
25.0–34.9	9 (8.7)	11 (5.9)
35.0–44.9	4 (3.8)	13 (7.0)
≥45.0	12 (11.5)	24 (13.0)
Moderate–vigorous physical activity, hr/d (mean ± SD)ˆ	0.99 ± 1.5	1.0 ± 1.4	0.92
None	45 (43.3)	73 (39.5)	0.91
Quartile 1 (≤ 0.286)*	19 (18.3)	31 (16.8)
Quartile 2 (0.287–0.589)	12 (11.5)	23 (12.4)
Quartile 3 (0.590–1.071)	15 (14.4)	28 (15.1)
Quartile 4 (>1.071)	13 (12.5)	30 (16.2)
Alcohol intake, drinks/week (mean ± SD)ˆ	1.7 ± 7.4	1.3 ± 3.3	0.54
None	55 (72.4)	100 (60.2)	0.07
Any alcohol reported	21 (27.6)	66 (39.8)
Betel-Nut useˆ
No	52 (65.8)	95 (59.4)	0.34
Yes	27 (34.2)	65 (40.6)
Smoked daily for > 6 monthsˆ
No	73 (70.9)	110 (59.5)	0.05
Yes	30 (29.1)	75 (40.5)
Weight change (age 20 to present)ˆ
≤ 5 lb gain	4 (5.6)	12 (10.7)	0.31
>5–35 lb gain	29 (40.3)	52 (36.8)
>35 lb gain	39 (54.2)	74 (52.5)

^†% is based on non-missing data and may not add up to 100 due to rounding

^ˆMissing values were excluded: 2 cases and 2 controls for percentage of lifetime lived in Marianas, 2 controls for age at menarche, 2 controls for age at first live births, 3 cases and 1 control for hormone use, 24 cases and 24 controls for sitting height, 7 cases and 13 controls for waist/height ratio, 28 cases and 19 controls for ethanol intake, 25 cases and 25 controls for betel-nut use, 1 case for smoked daily for >6 months, 32 cases and 44 controls for weight change (age 20 to present).

^*Quartile and tertile values were based on cases and control

The OR and 95% CI for breast cancer from the logistic regression models were the primary measures of association. ORs were calculated for each of the following 19 lifestyle-related factors in all women combined and in pre- and post-menopausal women separately: education, age at menarche, number of live births, age at first live birth, ever breastfed, first-degree family history of breast cancer, hormone use, type of menopause, age at menopause, BMI, WC, WHtR, physical activity, daily energy intake, alcohol intake, betel-nut use, smoking tobacco, height, and weight change. These factors were selected among lifestyle-related variables because they have been reported to be risk factors for breast cancer (18, 37) or were unique exposures to Guam and Saipan (e.g., betel nut). Models included an individual risk/protective factor, as well as age at reference date and ethnicity as adjustment variables. Additionally, a multivariate model was fit including all significant risk factors; a joint model with all investigated risk factors was unstable. Tests for interaction of risk factors with ethnicity were based on the Wald statistics for cross-product terms. We used SAS version 9.2 (SAS Institute, Cary, NC) for all analyses. P values of <0.05 were considered statistically significant, whereas P values of 0.05 to 0.10 were described as borderline significant.

3. RESULTS

3.1 Demographics

Average age of BRISK study participants was about 55 years on the reference date; by design, cases and controls did not differ by design (Table 1). The largest age groups were 50–59 years and 40–49 years, constituting 35% and 27% of participants, respectively. Most participants were Chamorro (57%); 28% were Filipino, 9% were “Other” (2% were Other Asian, 6% were White, and 1% were African American), and 6% were Other Micronesian. By design, cases and controls did not differ significantly in ethnic distribution. Approximately one-third of the sample were high-school graduates, and another third had some college. About 19% graduated from college, 8% had graduate degrees, and about 12% did not complete high school. Cases and controls did not differ on education. Both cases and controls had lived 72% of their lives on either Guam or Saipan.

3.2 Participant Characteristics

About 42% of study participants had menarche between 12 and 13 years of age, about 33% at age 14 or later, and about 26% at age 11 or earlier. Cases and controls were similar in age at menarche, but differences were observed in the number of pregnancies. About 38% of controls had had 3–4 pregnancies, about 29% 5 or more, 25% only 1–2, and 8% had never been pregnant. In contrast, the number of pregnancies was somewhat lower in cases with only 36% reporting 3–4 pregnancies and 25% reporting with 5 or more pregnancies. Age at first live birth, among parous women, was higher among cases (mean 24.9 years) than controls (mean 23.2 years), and more cases (47%) had their first child after age 25 than did controls (31%). Over 84% percent of cases and controls never used hormones. Twenty-eight percent of study participants were premenopausal, about 9% were perimenopausal, and the remainder were postmenopausal. Cases and controls were similar in menopausal status, probably because of age matching. Although only borderline significant (p=0.07), control participants were more likely to experience natural menopause than breast cancer cases (78.4% versus 72.2%). About 21% of postmenopausal women had undergone surgical menopause, and this percentage was comparable for cases and controls. About 86% of participants had no first-degree relatives with cancer; cases and controls were similar.

Mean height was 156.0 cm, sitting height was 82.5 cm, and mean BMI was 30.2 kg/m² (obese). Cases and controls did not differ substantially in the means, but waist circumference was greater among cases (mean 98.5 cm) than among controls (mean 94.5 cm). A high proportion of both cases (94%) and controls (86%) were classified “at risk” for cardiometabolic disease on the basis of the WHtR cutoff (>0.5).

Women, both cases and controls, engaged in about 20.8 MET hours/week of physical activity. Although only borderline significant, breast cancer cases tended to consume more alcohol per week than control participants (1.7 versus 1.3 drinks per week).

When asked if they ever had a breast biopsy, 16.8% (n = 31) of controls responded “yes,” and of those 22.6% (n = 7) had undergone two or more biopsies (data not shown). Cases were not asked for their biopsy history.

The associations between breast cancer risk and established risk factors are displayed in Table 2, overall and for pre- and post-menopausal women. No association was found between breast cancer incidence and level of education or age at menarche. The association was in the expected decreasing direction for number of live births and breastfeeding and increasing direction for age at menopause, but none of these relationships were statistically significant. Age at first live birth was associated with increased breast cancer risk (P for trend = 0.01), and ORs were over 2 for women whose first live birth was at age 30 or older. Hormone use was not common in this population, but hormone use was associated with increased risk of premenopausal breast cancer (P for trend = 0.03). Family history of breast cancer was not related to a higher risk of breast cancer in this population. Greater BMI was associated with increased risk of postmenopausal breast cancer. The OR for weight loss or maintenance, defined as a change of < 5 pounds from age 20 to reference age, was one-half ot that for weight gain, but not significant and showed no evidence of a linear trend. Increasing WC was associated with breast cancer risk (P for trend = 0.04); ORs were above 1.6 for women with waist measurements >99.5 cm. Weak evidence supported an association of the highest levels of physical activity with lower postmenopausal breast cancer risk. The OR for the highest to the lowest categories was 0.6 (0.2–1.6) for recreational MET-hours per week; the corresponding OR for hours per day of moderate-vigorous physical activity was of borderline significance at 0.38 (0.14–1.04). No associations were observed for alcohol intake or use of betel nut or tobacco. After running a multivariate analysis that included age, ethnicity, menopausal status, age at first live birth, and waist circumference in one model for all women, the OR (95% CI) for waist circumference for the second and third highest tertile groups compared to the first group were 1.47 (0.76, 2.86) and 1.65 (0.85, 3.22), respectively, similar to those adjusted for age and ethnicity in Table 2. The estimates for age at first live birth were strengthened: OR (95% CI) for age 20–24, age 25–29, and age ≥30 compared to age<20 were 0.89 (0.43, 1.87), 2.09 (0.95, 4.62), 3.19 (1.24, 8.20), respectively (data not shown).

Table 2.

The association of established and suspected risk factors for breast cancer with breast-cancer risk, BRISK study, 2010–2013.

Variables	All women		Premenopausal		Postmenopausal
	No. of cases/controls	Adjusted OR (95% CI)†	No. of cases/controls	Adjusted OR (95% CI)†	No. of cases/controls	Adjusted OR (95% CI)†	P-value for interactionˆ
Education							P = 0.41
High school diploma or less	48/77	1.0	12/33	1.00	36/44	1.00
Some college	28/58	0.78 (0.43, 1.41)	7/22	0.89 (0.30, 2.65)	21/36	0.72 (0.34, 1.51)
College degree or more	28/50	0.83 (0.43, 1.60)	13/19	1.76 (0.61, 5.07)	15/31	0.54 (0.22, 1.28)
P-value for trend		0.69		0.45		0.36
Age at menarche, years							P = 0.86
<12	24/51	1.00	5/22	1.00	19/29	1.00
12–13	44/75	1.22 (0.65, 2.27)	18/32	2.36 (0.74, 7.49)	26/43	0.81 (0.37, 1.77)
≥14	36/57	1.18 (0.61, 2.28)	9/20	2.10 (0.57, 7.69)	27/37	0.91 (0.41, 2.02)
P-value for trend		0.39		0.42		0.54
Number of live births							P = 0.30
Nulliparous	11/19	1.00	3/8	1.00	8/11	1.00
1–2	42/55	1.36 (0.58, 3.18)	20/29	1.65 (0.37, 7.29)	22/26	1.13 (0.38, 3.38)
3–4	30/74	0.68 (0.29, 1.61)	8/27	0.73 (0.15, 3.56)	22/47	0.64 (0.22, 1.85)
5 or more	21/37	0.87 (0.33, 2.30)	1/10	0.28 (0.02, 3.39)	20/27	0.99 (0.31, 3.19)
P-value for trend		0.38		0.21		0.61
Age at first live birth (parous women only)							P = 0.78
<20	23/51	1.00	7/22	1.00	16/29	1.00
20–24	26/63	0.92 (0.46, 1.83)	6/22	1.02 (0.29, 3.63)	20/41	1.03 (0.44, 2.42)
25–29	27/33	2.00 (0.93, 4.27)	9/13	2.12 (0.58, 7.75)	18/20	2.05 (0.78, 5.37)
≥30	17/17	2.53 (1.04, 6.19)	7/9	2.82 (0.66, 12.1)	10/8	2.95 (0.87, 10.0)
P-value for trend		0.01		0.18		0.01
Ever breastfed (parous women only)							P = 0.67
No	24/46	1.00	19/30	1.00	19/31	1.00
Yes	69/120	1.06 (0.58, 1.93)	45/69	1.33 (0.42, 4.24)	45/69	0.96 (0.47, 1.95)
P-value		0.65		0.84		0.90
Family history of breast cancer (1st degree relatives)							P = 0.94
No	92/155	1.00	27/59	1.00	65/96	1.00
Yes	12/30	0.74 (0.36, 1.54)	5/15	0.77 (0.25, 2.41)	7/15	0.76 (0.29, 2.02)
P-value		0.43		0.65		0.58
Hormone use (estrogen or progesterone)							P = 0.03
No	88/156	1.00	28/71	1.00	60/85	1.00
Ever	13/28	0.80 (0.39, 1.65)	4/3	3.08 (0.64, 14.8)	9/25	0.53 (0.22, 1.23)
P-value		0.55		0.16		0.14
Type of menopause (excluding medication)
Natural					52/88	1.00
Surgical					16/23	1.35 (0.64, 2.84)
P-value						0.44
Age at Menopause, y
<46					21/33	1.00
46–53					37/62	0.88 (0.44, 1.77)
≥54					14/16	1.27 (0.50, 3.24)
P-value for trend						0.33
Body Mass Index, kg/m2*							P = 0.36
18–24.9	20/52	1.00	9/20	1.00	11/33	1.00
25–29.9	37/52	2.01 (1.01, 3.99)	8/18	1.17 (0.36, 3.84)	29/34	2.90 (1.20, 7.0)
30–34.9	27/32	2.40 (1.12, 5.13)	9/11	2.62 (0.69, 9.97)	18/21	2.43 (0.92, 6.45)
≥35	20/48	1.34 (0.61, 2.97)	6/25	0.76 (0.21, 2.84)	14/22	2.08 (0.74, 5.88)
P-value for trend		0.55		0.84		0.40
Waist circumference, cm							P = 0.48
Tertile 1 (≤89)&	27/63	1.00	10/27	1.00	17/36	1.00
Tertile 2 (89.1–99.5)	32/56	1.28 (0.67, 2.43)	13/19	2.39 (0.82, 6.99)	19/37	0.99 (0.43, 2.25)
Tertile 3 (>99.5)	38/53	1.65 (0.87, 3.14)	7/22	1.05 (0.31, 3.56)	31/31	1.96 (0.90, 4.30)
P-value for trend		0.04		0.38		0.06
Waist/height ratio							P = 0.42
Quartile 1 (≤0.54)	17/51	1.00	6/25	1.00	11/26	1.00
Quartile 2 (0.55–0.61)	27/39	2.05 (0.96, 4.39)	10/12	5.01 (1.32, 19.0)	17/27	1.42 (0.54, 3.71)
Quartile 3 (0.62—0.67)	26/41	1.84 (0.87, 3.92)	10/14	4.25 (1.16, 15.5)	16/27	1.20 (0.45, 3.16)
Quartile 4 (>0.67)	27/41	1.94 (0.90, 4.16)	4/17	1.33 (0.29, 5.99)	23/24	2.04 (0.81, 5.20)
P-value for trend		0.10		0.58		0.11
Recreational physical activity, current MET-hr/wk							P = 0.14
None	44/71	1.00	9/29	1.00	35/42	1.00
<15.0	24/39	0.97 (0.51, 1.85)	8/16	1.31 (0.40, 4.26)	16/23	0.89 (0.40, 1.99)
15.0–24.9	11/27	0.65 (0.29, 1.46)	4/8	1.62 (0.36, 7.24)	7/19	0.51 (0.18, 1.42)
25.0–34.9	9/11	1.38 (0.51, 3.70)	5/6	2.38 (0.54, 10.5)	4/5	0.96 (0.22, 4.11)
35.0–44.9	4/13	0.46 (0.14, 1.53)	1/6	0.48 (0.05, 4.80)	3/7	0.45 (0.10, 1.98)
≥45.0	12/24	0.82 (0.37, 1.84)	5/9	1.60 (0.41, 6.21)	7/15	0.56 (0.20, 1.58)
P-value for trend		0.73		0.53		0.25
Moderate–vigorous physical activity, hr/d							P = 0.39
None	45/73	1.00	9/30	1.00	36/43	1.00
Quartile 1 (≤ 0.286)&	19/31	1.00 (0.50, 2.00)	7/14	1.41 (0.42, 4.75)	12/17	0.89 (0.37, 2.16)
Quartile 2 (0.287–0.589)	12/23	0.78 (0.34, 1.76)	4/9	1.37 (0.32, 6.00)	8/14	0.71 (0.26, 1.98)
Quartile 3 (0.590–1.071)	15/28	0.89 (0.42, 1.87)	6/12	1.54 (0.44, 5.39)	9/16	0.70 (0.27, 1.86)
Quartile 4 (>1.071)	13/30	0.69 (0.32, 1.48)	6/9	2.02 (0.55, 7.38)	7/21	0.38 (0.14, 1.04)
P-value for trend		0.94		0.78		0.36
Alcohol intake in past year							P = 0.17
No	55/100	1.00	17/35	1.00	38/65	1.00
Yes	21/66	0.60 (0.32, 1.10)	10/29	0.76 (0.29, 2.02)	11/37	0.49 (0.21, 1.13)
P-value for trend		0.41		0.26		0.23
Betel-nut use							P = 0.85
No	52/95	1.00	18/39	1.00	34/56	1.00
Yes	27/66	0.82 (0.43, 1.56)	9/23	0.83 (0.26, 2.69)	18/42	0.82 (0.36, 1.87)
P-value		0.56		0.99		0.20
Smoked daily for > 6 months							P = 0.94
No	73/110	1.00	21/41	1.00	52/69	1.00
Yes	30/75	0.66 (0.38, 1.15)	10/33	0.58 (0.22, 1.55)	20/42	0.71 (0.36, 1.41)
P-value		0.25		0.87		0.18
Weight change (age 20 to present)							P = 0.54
≤ 5 lb gain	4/15	0.54 (0.16, 1.82)	2/6	0.57 (0.09, 3.61)	2/9	0.48 (0.09, 2.54)
>5–35 lb gain	29/52	1.00	12/19	1.00	17/33	1.00
>35 lb gain	39/74	1.05 (0.57, 1.95)	13/29	0.83 (0.28, 2.45)	26/45	1.14 (0.52, 2.48)
P-value for trend		0.55		0.83		0.58

^†Adjusted for age at reference and ethnicity; OR, odds ratio; CI, confidence interval; BMI, body mass index.

^ˆInteraction between the trend variable and menopausal groups was tested by the Wald test with 1 degree of freedom.

^*Excludes one control with BMI < 18.

^&Quartile and tertile values were based on cases and controls

To explore further the relationship between anthropometry and breast cancer risk, we calculated ORs separately for Chamorros and Filipinos, the two largest ethnic groups in the study (Table 3). Although the ethnicity-specific ORs were somewhat unstable, the associations appeared to be stronger in Filipinos than in Chamorros. The ORs for Chamorros for a BMI of 30+ were not significant (1.4 for BMI of 30–34 and 0.9 for BMI 35+ compared to the 18–24 BMI group), whereas the OR for 30–34 was strong and significant in Filipinos (8.0 (1.7–38.6)). Although WC of cases and controls did not significantly differ among Chamorros, the OR for waist circumference above 99.5 cm was 7.03 (1.66–29.8) among Filipinos, and this difference between ethnic groups was significant (p for interaction=0.03). We also explored characteristics of Chamorro and Filipino control participants (Table 4). When compared to Chamorro control participants, Filipino control participants were significantly more educated, and had spent significantly less time during their lifetime living in Guam and/or Saipan. Although only borderline significant (p=0.09), more Chamorro control participants had a family history of breast cancer than Filipino control participants (19.6% versus 8.7%). Chamorro control participants tended to have a significantly larger body size, as measured by BMI, waist circumference, and WHtR, and were significantly more likely to smoke cigarettes and chew betel-nut; all of which put them at high risk for chronic disease such as breast cancer. However, age at first live birth, among parous women, was lower among Chamorro controls (mean 21.5 years) than among Filipino controls (mean 26.2 years), and more Filipino controls (56%) had their first child after age 25 than did Chamorro controls (16%).

Table 3.

The relationship between body size and the risk for breast cancer among Chamorro and Filipino women, BRISK study, 2010–2013.

	Chamorros		Filipinos
Variable	No. of cases/controls	Adjusted OR (95% CI)†	No. of cases/controls	Adjusted OR (95% CI)†	P-value for interactionˆ
BMI categories, kg/m2					P = 0.15
18–24.9	8/21	1.00	10/23	1.00
25–29.9	22/32	1.74 (0.65, 4.65)	13/17	1.75 (0.62, 4.95)
30–34.9	11/22	1.35 (0.45, 4.03)	9/3	8.02 (1.66, 38.6)
≥35	12/37	0.89 (0.31, 2.54)	2/3	1.57 (0.23, 11.0)
P-value for trend		0.54		0.18
Waist/Height Ratio					P = 0.03
Quartile 1 (≤0.54)	7/23	1.00	7/18	1.00
Quartile 2 (0.55–0.61)	19/22	2.73 (0.96, 7.81)	6/13	1.31 (0.34, 4.99)
Quartile 3 (0.62—0.67)	13/24	1.73 (0.59, 5.13)	9/11	2.23 (0.63, 7.86)
Quartile 4 (>0.67)	12/31	1.25 (0.42, 3.67)	7/4	5.24 (1.06, 25.9)
P-value for trend		0.83		0.01
Waist Circumference, cm					P = 0.03
Tertile 1 (≤89)*	14/29	1.00	10/24	1.00
Tertile 2 (89.1–99.5)	19/33	1.14 (0.48, 2.69)	9/18	1.26 (0.42, 3.77)
Tertile 3 (> 99.5)	18/38	0.98 (0.42, 2.29)	10/4	7.03 (1.66, 29.8)
P-value for trend		0.86		0.01
Waist (cm) and BMI category (kg/m2)					P = 0.07
Waist ≤88 cm	12/28	1.00	10/24	1.00
Waist >88 cm, and BMI ≤ 30	18/25	1.58 (0.63, 3.98)	8/16	1.24 (0.40, 3.86)
Waist >88 cm, and BMI > 30	21/47	1.07 (0.46, 2.51)	11/6	5.02 (1.36, 18.5)

^†Adjusted for age at reference (continuous); OR, odds ratio; CI, confidence interval; BMI, body mass index.

^ˆInteraction between the trend variable and ethnic groups was tested by the Wald test with 1 degree of freedom.

^*Quartile and tertile values were based on cases and controls.

Table 4.

Characteristics^† of Chamorro and Filipino breast-cancer controls in the BRISK study of Guam and Saipan. Reference was diagnosis date for cases, interview date for controls.

Characteristic	Chamorro Controls (n = 112)	Filipino Controls (n = 46)	p-value
Age at reference, years (mean ± SD)	53.4 ± 10.3	54.8 ± 11.5	0.46
<40	8 (7.1)	4 (8.7)	0.59
40–49	33 (29.5)	14 (30.4)
50–59	42 (37.5)	12 (26.1)
60–69	21 (18.8)	10 (21.7)
≥70	8 (7.1)	6 (13.0)
Highest education level completed
High school diploma or less	57 (50.9)	16 (34.8)	<0.001
Some college	39 (34.8)	7 (15.2)
College degree or more	16 (14.3)	23 (50.1)
Percentage of lifetime lived in Mariana Islands (mean ± SD)	46.2 ± 14.4	28.3 ± 12.1	<0.001
Age at menarche, yearsˆ
<12	37 (33.6)	8 (17.4)	0.10
12–13	45 (40.9)	21 (45.7)
≥14	28 (25.5)	17 (37.0)
Ever been pregnant	103 (92.0)	42 (91.3)	0.99
Total number of pregnancies
0	9 (8.0)	4 (8.7)	0.34
1–2	30 (26.8)	9 (19.6)
3–4	37 (33.0)	22 (47.8)
5 or more	36 (32.1)	11 (23.9)
Number of live births
Nulliparous	12 (10.7)	5 (10.9)	0.29
1–2	33 (29.5)	12 (26.1)
3–4	40 (35.7)	23 (50.0)
5 or more	27 (24.1)	6 (13.0)
Age at first live birth, years, parous women only (mean ± SD)ˆ	21.5 ± 4.4	26.2 ± 5.5	<0.001
<20	43 (43.9)	5 (12.2)	<0.001
20–24	39 (39.8)	13 (31.7)
25–29	10 (10.2)	16 (39.0)
≥30	6 (6.1)	7 (17.1)
Ever breastfed, parous women only
No	32 (32.0)	10 (24.4)	0.37
Yes	68 (68.0)	31 (75.6)
Family history of breast cancer (1st degree)
No	90 (80.4)	42 (91.3)	0.09
Yes	22 (19.6)	4 (8.7)
Number of first-degree relatives with breast cancer
0	90 (80.4)	42 (91.3)	0.24
1	19 (17.0)	4 (8.7)
2	3 (2.7)	0
Hormone useˆ
Never used estrogen or progesterone	97 (86.6)	36 (80.0)	0.28
Yes, previously	14 (12.5)	7 (15.6)
Yes, currently	1 (0.9)	2 (4.4)
Menopausal status
Pre	30 (26.8)	18 (39.1)	0.23
Peri	14 (12.5)	3 (6.5)
Post	68 (60.7)	25 (54.4)
Age at menopause, years, postmenopausal women only
<46	23 (33.8)	5 (20.0)	0.43
46–53	37 (54.4)	16 (64.0)
≥54	8 (11.8)	4 (16.0)
Type of Menopause, postmenopausal women only
Natural	51 (75.0)	21 (84.0)	0.36
Medication	0	0
Surgical	17 (25.0)	4 (16.0)
Unknown	0	0
Height, cm (mean ± SD)	156.0 ± 7.2	153.6 ± 5.1	0.05
Sitting height, cm (mean ± SD)ˆ	82.4 ± 6.4	80.5 ± 5.1	0.15
Body mass Index, kg/m2 (mean ± SD)	32.0 ± 7.6	26.2 ± 4.9	<0.001
<18	0	0	<0.001
18–24.9	21 (18.8)	23 (50.0)
25–29.9	32 (28.6)	17 (37.0)
30–34.9	22 (19.6)	3 (6.5)
≥35	37 (33.0)	3 (6.5)
Waist Circumference, cm (mean ± SD)	97.5 ± 15.8	88.0 ± 10.2	<0.001
Tertile 1 (≤89)*	41 (36.6)	24 (52.2)
Tertile 2 (89.1–99.5)	33 (29.5)	18 (39.1)
Tertile 3 (>99.5)	38 (33.9)	4 (8.7)
Waist/Height Ratio (mean ± SD)ˆ	0.63 ± 0.11	0.57 ± 0.07	0.002
Quartile 1 (≤0.54)*	35 (31.3)	18 (39.1)	0.07
Quartile 2 (0.55–0.61)	22(19.6)	13 (28.3)
Quartile 3 (0.61–0.67)	24 (21.4)	11 (23.9)
Quartile 4 (>0.67)	31 (27.7)	4 (8.7)
Recreational physical activity, Current MET-hr/week (mean ± SD)	20.7 ± 36.9	18.1 ± 21.1	0.65
None	47 (42.0)	14 (30.4)	0.51
<15.0	24 (21.4)	11 (23.9)
15.0–24.9	13 (11.6)	7 (15.2)
25.0–34.9	7 (6.25)	3 (6.5)
35.0–44.9	6 (5.4)	6 (13)
≥45.0	15 (13.4)	5 (10.9)
Moderate–vigorous physical activity, hr/d (mean ± SD)ˆ	1.00 ± 1.34	0.74 ± 0.59	0.30
None	49 (43.8)	14 (30.4)	0.23
Quartile 1 (≤ 0.286)*	20 (17.9)	9 (19.6)
Quartile 2 (0.287–0.589)	10 (8.9)	8 (17.4)
Quartile 3 (0.590–1.071)	18 (16.1)	5 (10.9)
Quartile 4 (>1.071)	15 (13.4)	10 (21.7)
Alcohol intake, drinks/week (mean ± SD)ˆ	1.2 ± 3.1	0.62 ± 1.79	0.26
None	60 (60.6)	27 (64.3)	0.71
Any alcohol reported	39 (39.4)	15 (35.7)
Betel-Nut useˆ
No	42 (44.2)	35 (87.5)	<0.001
Yes	53 (55.8)	5 (12.5)
Smoked daily for > 6 monthsˆ
No	53 (47.3)	39 (84.8)	<0.001
Yes	59 (52.7)	7 (15.2)
Weight change (age 20 to present)ˆ
≤ 5 lb gain	9 (10.1)	2 (6.9)	0.16
>5–35 lb gain	28 (31.5)	15 (51.7)
>35 lb gain	52 (58.4)	12 (41.4)

^†% is based on non-missing data and may not add up to 100 due to rounding

^ˆMissing values were excluded: 1 Chamorro control for percentage of lifetime lived in Marianas, 2 Chamorro controls for age at menarche, 2 Chamorro controls for age at live birth, 1 Filipino control for hormones, 13 Chamorro controls and 4 Filipino controls for ethanol intake, 17 Chamorro controls and 6 Filipino controls for betel nut use, 23 Chamorro controls and 17 Filipinos for weight change.

^*Quartile and tertile values were based on cases and control

4. DISCUSSION

BRISK is the first retrospective case-control study to explore the association of health-related risk factors including diet, physical activity, and obesity with breast cancer risk among Asian-Pacific Islander women living in Guam and the Northern Mariana Islands. Most breast cancer studies among Pacific Islanders have focused on exploring knowledge, attitudes, and screening behaviors (39–47) and support systems after cancer diagnosis (48). The few that have concentrated on health-related risk factors were limited to Chamorros in San Diego, California (47, 49–51), and residents of Hawaii (52–54).

Two established risk factors for breast cancer were confirmed in the BRISK population: older age at first live birth and greater size as measured by WC. Sedentary behavior was suggested to be a risk factor in this population, but several established risk factors were not found to contribute, including education, age at menarche and menopause, and family history of breast cancer. Although associations for number of live births and breastfeeding were in the expected directions, they were weak and not statistically significant, probably because of the small sample size. The lack of an association with cigarette smoking is consistent with the discordant results reported in the literature (53). Betel-nut use was not related to breast cancer risk.

Our failure to observe some expected associations may be due to their lack of variability within ethnic-age groups and to small sample size. The population of the Mariana Islands is very heterogeneous, both in ethnicity and in its mixture of traditional and Western cultures. The diversity has come from immigration and acculturation, which tend to be found unevenly across subgroups of the population. The variability of exposure variables with matched sets of cases and controls matched on ethnicity and age group may have been relatively small, reducing the power to detect differences.

Anthropometric factors were strong risk factors for breast cancer among Mariana Island women. Obesity has become a major health problem in the U.S. Affiliated Pacific Island communities, including Guam and CNMI (56). Data from the 2012 Behavior Risk Factor Survey indicated that 61.5% of adults on Guam were overweight or obese (BMI ≥ 25), a prevalence similar to that for the United States as a whole. The prevalence of obesity and overweight was significantly higher among Chamorros than other ethnic groups on Guam (57). In BRISK, we found that a large proportion of breast cancer cases and controls were obese (>40%), as indicated by BMI’s of 30 and above, and they were at risk for cardiometabolic disease, as indicated by high WC measures and WHtR calculations.

On the basis of their relatively large waist circumferences, breast cancer cases are more likely to be at high risk for cardiometabolic disease, especially among the Filipino women. Interestingly, Chamorro women had higher BMI’s and larger waist circumferences than the Filipino women, but the association of breast cancer with BMI was stronger in the Filipino women, and the association with waist circumference was limited to them. Studies have suggested that disease risk occurs at lower levels of adiposity among Asian women (58). Differential cutpoints by ethnicity were considered, but the common WHO reference was used to ease interpretation, especially in this highly mixed populations.

According to the most recent Guam Cancer Facts and Figures (2008–2012) (17), incidence of breast cancer increased in Filipino women on Guam from 60.7 to 76.7 between 1998 and 2012. Breast cancer risk is rapidly increasing among Filipino women on Guam, as is clear from the increase in incidence and mortality rates, possibly fueled by changes to a more Westernized life style, more extensive use of mammography screening, and increased obesity. A similar transition has occurred between 1975 and 2005 among Filipino women in Hawaii, where breast cancer incidence rates increased from 44 to 88, and among Native Hawaiian women, where the rates increased from 119 to 148. The transitional period may be especially risky (59–69). Incidence rates of breast cancer among Asian-American, Native Hawaiian, and Pacific Islander women increased between 2003 and 2012 as compared to non-Hispanic white women, and the reasons for this increase are thought to be related to changes in reproductive patterns and increased body weight that occur with acculturation after immigration (58).

As further evidence that the ethnic groups included in our study are in transition from a more traditional to more Westernized life styles, Leon Guerrero and colleagues (57) recently observed a greater prevalence of overweight and obesity among Chamorros and Filipinos than were reported in an earlier study on Guam by Pinhey and colleagues (70). The greatest increase in BMI between the two studies was among Chamorro women, whose mean BMI increased from 25.7 in 1994 to 30.6 in 2008. In the most recent survey of adults on Guam (57), only 20.6% of Chamorro women were considered of healthy weight (BMI < 25.0); the remaining women were either overweight (38.2%) or obese (41.2%). A smaller increase in BMI was observed among Filipinos. Most Filipinos in both studies were “1^st generation” and had not spent a lifetime being exposed to the same environment, diet, and life-style patterns as Chamorros, but Filipino men and women appear to be experiencing a gradual upward shift in their body size. Compared to Whites, Asians tend to have higher amounts of visceral adipose tissue (71–74) and to be more likely to exhibit central obesity than other ethnic groups. Obesity, especially abdominal adiposity, is a risk factor for postmenopausal breast cancer (22, 23). Studies have shown that other ethnic groups (Japanese, Koreans, Chinese, Hispanics) experienced changes in diet and weight parameters as they migrated to the U.S. and became more acculturated into a Westernized life style (59, 60, 62, 65, 66, 68, 69) and that these changes affected their risk of cancer and other chronic diseases (61, 63, 64, 67).

Age at first live birth above 30 years is a well-known risk factor for breast cancer (6), and women in the present study who had a higher age at first birth had a significantly greater breast cancer risk. This population of women residing in the Mariana Islands has a relatively low age at first live birth. In the U.S., the age at first birth has risen to 26.0 years (75). In the BRISK study, the mean age at first birth was 24. If these women are representative of the women living in the Mariana Islands, this difference may explain why the overall rates of breast cancer on Guam are lower than those among U.S. women. The protection of low age at first live birth may disappear as women in the Mariana Islands wait to have their first child, as do those in the U.S. (75) and other Westernized countries.

Some limitations may influence the interpretation of these results. The small sample size of approximately 290 women has made statistical significance for many of the known risk factors difficult to attain. Recall bias for both cases and controls may have affected the associations for dietary, reproductive, and physical-activity factors. The interview was conducted after the cases had been diagnosed, although within 24 months after diagnosis. We measured weight and height after cancer diagnosis and treatment, and, for breast cancer cases, body-size associations with risk may have been biased because the body size we recorded may not have reflected body size before diagnosis, although the results for self-reported body size before diagnosis were similar. Control subjects were a convenience sample of volunteers and may not be representative of the population from which they were recruited, although the participation rate was good among controls at 74%. Despite these limitations, this retrospective case-control study provides the first assessment of breast cancer risk factors in the multiethnic populations of Guam and Saipan. The majority of breast cancer cases on both islands were overweight or obese and at increased risk for chronic disease as indicated by a high waist circumference. In particular, Filipino women with high waist circumference were at a higher risk of developing breast cancer. Results of our study provide information for cancer-prevention guidance for women in the Mariana Islands.

Highlights.

• A retrospective case-control study was conducted looking female breast cancer (104 cases and 185 controls) among women in the Mariana Islands.

• OR and 95% CI were calculated for each of various lifestyle-related factors from logistic regression of breast cancer, in all women and in pre- and postmenopausal women separately.

• Of the medical and reproductive-factors considered, only age at first live birth (was significantly associated with an increase in breast cancer risk.

• Of the lifestyle factors studied, only waist circumference was significantly associated with breast cancer risk and only in Filipino women

• The results provide a basis for cancer prevention guidance for women in the Mariana Islands.