Abstract
Background:
Native Hawaiian (NH) women experience high breast cancer (BC) incidence and mortality. While obesity is a known risk factor, the role of ethnic mixture, defined as self-reported combinations of racial or ethnic backgrounds, remains understudied. This population-based study examined whether breast cancer risk varies by ethnic mixture among NH women and whether these associations differ across BMI categories.
Methods:
We analyzed data from 7,700 NH women in the Multiethnic Cohort Study. Participants were categorized into ethnic mixture groups: NH-only, NH/White, NH/Chinese, NH/White/Chinese, NH/Other Asian, and NH/Other. Age-adjusted incidence rates were calculated per 1,000 person-years. Cox proportional hazards models estimated hazard ratios (HRs) and 95% confidence intervals (CIs), adjusting for age, BMI, and other factors.
Results:
Incidence rates were highest among NH-only and NH/White/Chinese (5.7 and 5.9 per 1,000 person-years) and lowest among NH/Other (3.6). However, ethnic mixture was not significantly associated with risk in adjusted models. Stratified analyses showed modest, non-significant risk elevations among overweight NH/Chinese (HR=1.33, 95% CI: 0.85–2.10). Among obese women, risk was consistent across all groups (HRs: 0.78–0.96), highlighting the dominant role of BMI. Overall, the association between ethnic mixture and breast cancer risk did not differ by BMI category.
Conclusions:
Ethnic mixture was not independently associated with BC risk; BMI remained a consistent predictor. These findings underscore the need for research integrating ancestry, body composition, and social context to address cancer disparities.
Impact:
Obesity, not ethnic mixture, is the dominant factor influencing BC risk in NH women.
INTRODUCTION
Breast cancer is a complex disease influenced by genetic, environmental, and lifestyle factors. While advancements in detection and treatment have improved outcomes, disparities in breast cancer incidence and mortality persist across populations. Among women in Hawaii, Native Hawaiians (NH) women experience the highest breast cancer burden compared with other major racial and ethnic groups. According to the Hawaii Tumor Registry, breast cancer incidence in the state (139.6 per 100,000) exceeds the U.S. average (126.9 per 100,000) [1]. NH women have among the highest incidence rates (165.9 per 100,000), higher than non-Hispanic White women (141.1 per 100,000) [1]. NH women also experience the highest breast cancer mortality in the state (24.9 per 100,000), nearly double that of Japanese American women (13.7 per 100,000) [1]. These disparities suggest that NH women may face a unique combination of social, cultural, and metabolic factors that contribute to their disproportionate breast cancer burden. NH and other Pacific Islander women exhibit some of the highest obesity rates in the U.S. (51.7%) [2], adding to this complex risk profile.
Obesity has emerged as an important causal factor contributing to these disparities. Excess body fat is strongly linked to increased postmenopausal breast cancer risk through pathways involving estrogen production, chronic inflammation, and insulin resistance [3–6]. Importantly, obesity and related metabolic outcomes differ across populations not only because of biological factors but also because of lifestyle, cultural, and environmental influences associated with ethnic background. Studies have shown that individuals with mixed ethnic ancestry often experience distinctive obesity and metabolic profiles, reflecting complex interactions among diet, physical activity, cultural norms, and social environments [7–9]. These findings suggest that understanding obesity within the context of ethnic mixture may provide insight into how social and cultural diversity influences chronic disease risk. In this study, obesity was assessed using body mass index (BMI), calculated from self-reported height and weight.
In Hawaii, many individuals who identify as Native Hawaiian also report additional racial or ethnic ancestries, reflecting the state’s long history of cultural blending and intermarriage. Ethnic mixture, defined here as self-reported combinations of Native Hawaiian and other racial or ethnic backgrounds, provides a sociocultural lens for understanding variation in health among NH women. Although multiethnic individuals represent nearly 25% of Hawaii’s residents and over 10% of the U.S. population [10, 11], there are currently no published breast cancer incidence rates for NH-mixed groups (e.g., NH/Chinese or NH/White). This absence of data highlights an important knowledge gap regarding how ethnic mixture relates to cancer risk. Prior studies have shown that ethnic mixture is associated with obesity and metabolic outcomes, including higher diabetes risk in NH/Asian individuals and greater obesity prevalence among NH/White adults compared with Whites alone [7, 12].
Despite the relevance of ethnic mixture to obesity-related health outcomes, no studies have examined whether the association between ethnic mixture and breast cancer risk differs across BMI categories in NH women. This question is particularly important because the high prevalence of both obesity and multiethnic identity in this population may create unique risk profiles that are not captured by single-race or ancestry-only models. To address this gap, the present study examines whether breast cancer incidence among NH women varies by ethnic mixture and whether this association differs across BMI categories. We hypothesize that ethnic mixture may be associated with variation in breast cancer susceptibility, and that these differences may be more or less apparent depending on BMI category, given the strong metabolic influence of obesity.
METHODS
Study Population:
The Multiethnic Cohort (MEC) is a population-based cohort established in 1993 to study diet and cancer among different ethnic groups in Hawaii and California [13]. The MEC includes over 9,000 breast cancer cases from five racial/ethnic groups: African American, Japanese, Latino, NH and White. It also contains comprehensive data on risk factors (e.g., BMI, menstrual history, parity, birth control use, hormone medication, and family history), enabling models to adjust for well-known predictors of breast cancer risk. All women reporting NH ancestry within the MEC who were free of breast cancer at cohort entry were included.
Participants completed a 26-page, self-administered mailed questionnaire that assessed demographic background, medical conditions, anthropometric measures, and diet and lifestyle factors. Receipt of a completed, mailed baseline questionnaire was considered written informed consent. Study procedures were conducted in accordance with the principles of the Declaration of Helsinki and the U.S. Common Rule. The study protocol was reviewed and approved by the Institutional Review Boards (IRBs) of the University of Hawaii and the University of Southern California. Of the 7,700 total NHs in the MEC, 922 were diagnosed with incident invasive breast cancer.
Exposure:
The primary exposure in this analysis was self-reported ethnic mixture. At baseline, participants were asked to mark all applicable racial or ethnic backgrounds (African American, Chinese, Filipino, Japanese, Korean, Latino, Native Hawaiian, White, or Other). All women reporting any Native Hawaiian ancestry were included in this study. To ensure adequate statistical power while maintaining meaningful distinctions, we grouped participants into six ethnic mixture categories reflecting the most common ancestry combinations in the cohort: Native Hawaiian (NH) only (reference group), NH/White, NH/Chinese, NH/White/Chinese, NH/Other Asian (Japanese, Filipino, or Korean), and NH/Other (Latino, African American, or Native American). These categories capture self-identified sociocultural and ancestral backgrounds rather than genetically determined ancestry. The NH-only group was used as the reference category in all models to assess whether ethnic mixture conferred additional or reduced breast cancer risk compared with women who identified solely as NH.
Invasive breast cancer ascertainment:
Incident invasive breast cancer cases in the MEC were identified by linkage to the statewide Surveillance, Epidemiology and End Results (SEER) cancer registries of Hawaii and California. Vital status was identified via linkages with the National Death Index and death certificate files for Hawaii and California. Cancer and death ascertainment was completed through December 31, 2019.
Statistical analysis:
Obesity was assessed using body mass index (BMI), calculated as weight in kilograms divided by height in meters squared (kg/m2). BMI was derived from self-reported height and weight on the baseline questionnaire. For analysis, BMI was categorized as <25 kg/m2 (healthy/underweight), 25–<30 kg/m2 (overweight), and ≥30 kg/m2 (obese), consistent with World Health Organization definitions [14]. The age-adjusted breast cancer incidence rates per 1,000 person-years, truncated to ages 45–85 years, weighted by the age distribution of the 2000 U.S. standard million population, was used to examine differences in incidence. They were computed as the sum of the weighted weights: number of newly diagnosed breast cancer cases divided by the total person-years of follow-up for each age group, weighted by the proportion of the U.S. 2000 standard in each age group. Cox proportional regression model using SAS version 9.4 (RRID:SCR_008567) (SAS Institute, Inc, Cary, NC) estimated breast cancer risk as hazard ratios (HR) with 95% confidence intervals (CI). The proportional hazards assumption was tested using Schoenfeld residuals [15], all exposures met assumption. The interval from age at cohort entry to age at diagnosis of breast cancer, death or December 31, 2019 (closure date) was used as the time metric. Time to event was defined as the interval from age at cohort entry to age at invasive breast cancer diagnosis, death, loss to follow-up or December 31, 2019 (end of follow-up), whichever occurred first. Participants without a diagnosis of invasive breast cancer by that date were censored at their last known follow-up or death. Given the stratification of NH women into multiple ethnic mixture groups, we assessed whether the study had sufficient statistical power to detect meaningful differences in breast cancer risk. A post-hoc power analysis indicated that to detect a hazard ratio of 1.33 with 80% power (α=0.05), approximately 193 cases per group would be required. This suggests that some subgroup analyses may have been underpowered to detect moderate associations. The NH-only group was used as our reference group because it would allow us to compare if ethnic mixture provided additional risk or protection in comparison to full NHs. Ethnic mixture was modeled as a categorical exposure variable (NH-only as reference) with hazard ratios estimated for each ethnic mixture group, but the models were not stratified by ethnic mixture. To examine whether the association between ethnic mixture and breast cancer risk differs across BMI categories, models were stratified by BMI (<25, 25–<30, ≥30 kg/m2).
Three models with progressive levels of adjustment were computed to assess the contribution of various covariates and determine whether additional adjustments improved model performance. In model 1, we estimated HRs and 95% CIs adjusted only for age at cohort entry. Model 2 additionally adjusted for BMI (<25, 25-<30, and ≥30 kg/m2). Model 3, the fully adjusted model, further adjusted for all covariates previously associated with breast cancer: history of breast cancer (yes, no), age at menarche (≤12, 13–14, >14 years, missing), age at first live birth (<20, 21–30, >30 years, missing), age and type of menopause (premenopausal, natural <45, 45–49, 50–54, ≥55 years, oophorectomy (<45, ≥45 years), hysterectomy (<45, ≥45 years; missing), number of children (0, 1, 2–3, 4, missing), smoking status (never, former, current, missing), hormone replacement therapy (HRT) (no estrogen use, past estrogen use, current estrogen use only, and current estrogen with past or current progestin use, missing), alcohol intake (<1/month, ≥1/month-<1/day, ≥1/day), and physical activity (<30 min/day, ≥30 min/day). To examine whether the association between ethnic mixture and breast cancer risk differed across BMI categories, models were stratified by BMI categories (<25, 25-<30, and ≥30 kg/m2). To examine differences in the ethnic mixture breast cancer association across BMI categories, we also included an interaction term between ethnic mixture and BMI to formally test whether BMI modified the ethnic mixture-breast cancer association. Ethnic mixture was treated as the primary exposure, while BMI was included as a covariate and stratification variable to test for potential effect modification.
Data availability
The data used in this manuscript may be available upon request to the MEC research Committee (https://www.uhcancercenter.org/for-researchers/mec-data-sharing).
RESULTS
Participant Characteristics:
After excluding participants with prevalent breast cancer at cohort entry, 7,700 women who reported any NH ancestry were part of this analysis (Table 1). Among the 7,700 NH women, the distribution across ethnic-mixture categories was as follows: NH-only = 1,182 (15.4%), NH/White = 1,765 (22.9%), NH/Chinese = 1,001 (13.0%), NH/White/Chinese = 1,662 (21.6%), NH/Other Asian = 1,493 (19.4%), and NH/Other = 597 (7.7%). The overall cohort of women with NH ancestry had a mean age of 56 (±8.6) years at cohort entry in 1993–1996. BMI varied by ethnic mixture category, with NH-only women having the highest rate of obesity (45%), while those in the NH/Chinese (30%) and NH/Other Asian (30%) had comparatively lower rates of obesity.
Table 1.
Characteristics of participants by Native Hawaiian ethnic mixture in the Multiethnic Cohort Study (N=7,700)
| NH only | NH White | NH Chinese | NH White Chinese | NH Asian | NH Other | |
|---|---|---|---|---|---|---|
| N | 1,182 | 1,766 | 1,002 | 1,662 | 1,493 | 595 |
| Number of breast cancer cases | 137 | 212 | 120 | 218 | 169 | 66 |
| Follow-up years, mean (SD) | 17.49 (8.0) | 19.06 (7.8) | 18.96 (7.9) | 20.04 (7.4) | 20.59 (7.2) | 20.90 (6.8) |
| Person years total | 20668 | 33668 | 18995 | 33303 | 30741 | 12435 |
| Age at cohort entry, mean (SD) | 57.45 (8.6) | 57.52 (8.8) | 58.65 (8.8) | 55.03 (8.5) | 53.75 (7.6) | 52.95 (7.5) |
| Age at diagnosis, mean (SD) | 67.82 (9.3) | 69.46 (8.9) | 68.70 (8.9) | 66.38 (8.9) | 65.74 (8.8) | 64.42 (7.6) |
| Years of education, n (%) | ||||||
| ≤12 high school or less | 797 (67.4) | 1,017 (57.6) | 589 (58.8) | 835 (50.2) | 796 (53.3) | 268 (45.0) |
| 13–15 some college | 240 (20.3) | 513 (29.1) | 251 (25.1) | 528 (31.8) | 429 (28.7) | 220 (37.0) |
| ≥16 college graduate + | 145 (12.3) | 236 (13.4) | 162 (16.2) | 299 (18.0) | 268 (18.0) | 107 (18.0) |
| Alcohol intake, no. of drinks, n (%) | ||||||
| <1/month | 730 (61.8) | 1,050 (59.5) | 695 (69.4) | 1,008 (60.6) | 1,008 (67.5) | 364 (61.2) |
| ≥1/month –<1/day | 268 (22.7) | 446 (25.2) | 195 (19.5) | 454 (27.3) | 333 (22.3) | 154 (25.9) |
| ≥1/day | 81 (6.8) | 197 (11.2) | 58 (5.8) | 145 (8.7) | 92 (6.2) | 54 (9.1) |
| Physical activity, n (%) | ||||||
| <30 min/d | 476 (40.3) | 588 (33.3) | 400 (39.9) | 504 (30.3) | 508 (34.0) | 176 (29.6) |
| ≥ 30 min/d | 669 (56.6) | 1,152 (65.2) | 585 (58.4) | 1,140 (68.6) | 967 (64.8) | 412 (69.2) |
| Smoking status, n (%) | ||||||
| Never | 458 (38.8) | 732 (41.5) | 550 (54.9) | 726 (43.7) | 682 (45.7) | 237 (39.8) |
| Former | 389 (32.9) | 588 (33.3) | 258 (25.8) | 534 (32.1) | 446 (29.9) | 219 (36.8) |
| Current | 319 (27.0) | 432 (24.5) | 175 (17.5) | 388 (23.4) | 349 (23.4) | 135 (22.7) |
| BMI, kg/m2, n (%) | ||||||
| <20 | 33 (2.8) | 61 (3.5) | 44 (4.4) | 43 (2.6) | 69 (4.6) | 12 (2.0) |
| 20-<25 | 266 (22.5) | 449 (25.4) | 341 (34.0) | 477 (28.7) | 475 (31.8) | 152 (25.6) |
| 25-<30 | 351 (29.7) | 574 (32.5) | 315 (31.4) | 549 (33.0) | 500 (33.5) | 212 (35.6) |
| 30+ | 532 (45.0) | 682 (38.6) | 302 (30.1) | 593 (35.7) | 449 (30.1) | 219 (36.8) |
| Family history of breast cancer, n (%) | 160 (13.5) | 248 (14.0) | 148 (14.8) | 196 (11.8) | 165 (11.1) | 58 (9.8) |
| Ever had a mammography, n(%) | ||||||
| No | 188 (16.6) | 210 (12.2) | 146 (15.0) | 196 (12.1) | 196 (13.4) | 58 (9.9) |
| Yes | 943 (83.4) | 1,514 (87.8) | 825 (85.0) | 1,430 (88.0) | 1,268 (86.6) | 525 (90.0) |
| Age at Menarche, n (%) | ||||||
| <13 years | 667 (56.4) | 923 (52.3) | 562 (56.1) | 930 (56.0) | 887 (59.4) | 360 (60.5) |
| 13–14 years | 343 (29.0) | 615 (34.8) | 303 (30.2) | 535 (32.2) | 453 (30.3) | 181 (30.4) |
| >15 years | 148 (12.5) | 200 (11.3) | 120 (12.0) | 171 (10.3) | 131 (8.8) | 46 (7.7) |
| Age at first child, n (%) | ||||||
| <20 years | 553 (46.8) | 785 (44.5) | 378 (37.7) | 628 (37.8) | 663 (44.4) | 262 (44.0) |
| 21–30 years | 462 (39.1) | 737 (41.7) | 461 (46.0) | 789 (47.5) | 616 (41.3) | 245 (41.2) |
| >31 years | 29 (2.5) | 44 (2.5) | 40 (4.0) | 54 (3.2) | 48 (3.2) | 22 (3.7) |
| Number of children, n (%) | ||||||
| None | 84 (7.1) | 142 (8.0) | 85 (8.5) | 144 (8.7) | 120 (8.0) | 48 (8.1) |
| 1 | 71 (6.0) | 104 (5.9) | 62 (6.2) | 136 (8.2) | 115 (7.7) | 42 (7.1) |
| 2–3 | 318 (26.9) | 631 (35.7) | 319 (31.8) | 657 (39.5) | 581 (38.9) | 240 (40.3) |
| >4 | 692 (58.5) | 872 (49.4) | 522 (52.1) | 708 (42.6) | 664 (44.5) | 262 (44.0) |
| Menopausal status, n (%) | ||||||
| Premenopausal | 216 (18.3) | 314 (17.8) | 163 (16.3) | 416 (25.0) | 449 (30.1) | 189 (31.8) |
| <45 years at natural menopause | 133 (11.2) | 163 (9.2) | 89 (8.9) | 117 (7.0) | 112 (7.5) | 54 (9.1) |
| 45–49 years at natural menopause | 181 (15.3) | 249 (14.1) | 130 (13.0) | 217 (13.1) | 183 (12.3) | 74 (12.4) |
| 50–54 years at natural menopause | 171 (14.5) | 272 (15.4) | 168 (16.8) | 244 (14.7) | 210 (14.1) | 77 (12.9) |
| >55 years at natural menopause | 54 (4.6) | 86 (4.9) | 62 (6.2) | 67 (4.0) | 48 (3.2) | 15 (2.5) |
| Surgical menopause | 287 (24.3) | 528 (29.9) | 266 (26.6) | 460 (27.7) | 371 (24.9) | 157 (26.4) |
| HRT use, n (%) | ||||||
| Never (estrogen and progesterone) | 800 (67.7) | 991 (56.1) | 583 (58.2) | 981 (59.0) | 923 (61.8) | 354 (59.5) |
| Past (estrogen with/without progesterone) | 149 (12.6) | 307 (17.4) | 141 (14.1) | 204 (12.3) | 157 (10.5) | 85 (14.3) |
| Current (estrogen only) | 91 (7.7) | 199 (11.3) | 111 (11.1) | 210 (12.6) | 158 (10.6) | 70 (11.8) |
| Current (estrogen and progesterone) | 80 (6.8) | 207 (11.7) | 125 (12.5) | 213 (12.8) | 214 (14.3) | 73 (12.3) |
| History of Diabetes, n (%) | 213 (18.0) | 230 (13.0) | 157 (15.7) | 185 (11.1) | 224 (15.0) | 77 (12.9) |
Abbreviations: NH, Native Hawaiian; BMI, body mass index; HRT, hormone replacement therapy
Breast Cancer Incidence Rates:
During follow-up of 19.5 (±7.7) years (Table 2), 922 breast cancer cases were identified among NH women. The age-adjusted breast cancer incidence rates differed by ethnic mixture category. Breast cancer incidence was lowest among NH/Other women (3.6 per 1,000 pyears) and highest among NH-only (5.7 per 1,000 pyears) and NH/White/Chinese women (5.9 per 1.000 pyears).
Table 2.
Breast cancer incidence rates by Native Hawaiian (NH) ethnic mixture across body mass index (BMI) categories
| Variable/ Ethnic Mixture Group | NH only | NH/ White | NH/ Chinese | NH/ White/ Chinese | NH/ Asian | NH/ Other | All |
|---|---|---|---|---|---|---|---|
| Number of participants | 1182 | 1766 | 1002 | 1662 | 1493 | 595 | 7700 |
| Incident cancer cases | 137 | 212 | 120 | 218 | 169 | 66 | 922 |
| Breast Cancer Incidence (1,000 pyears) a | |||||||
| All | 5.7 | 5.5 | 5.6 | 5.9 | 5.0 | 3.6 | 6.0 |
| Healthy weight (<25 kg/m2) | 5.7 | 4.2 | 4.5 | 5.5 | 3.9 | 3.2 | 4.5 |
| Overweight (25-<30 kg/m2) | 4.7 | 5.5 | 6.3 | 5.8 | 5.2 | 3.5 | 5.4 |
| Obese (≥30 kg/m2) | 6.8 | 6.7 | 6.6 | 6.3 | 6.7 | 4.5 | 6.5 |
Incidence rates for breast cancer (age adjusted to U.S. 2000, truncated at age 45–85)
Abbreviations: NH, Native Hawaiian; BMI, body mass index;
Breast Cancer Incidence by Ethnic Mixture Group:
As seen in Table 3, the HRs for breast cancer incidence across the ethnic mixture groups did not differ significantly from the reference group (NH-only), and the corresponding 95% CIs consistently included 1. In model 1, which adjusted for age only, the HRs for groups such as NH-alone, NH/White, NH/Chinese, and NH/White/Chinese were all close to 1, suggesting no significant deviation in breast cancer incidence in ethnic mixture groups compared to the NH-only individuals. The NH/Other Asian and NH/Other groups had slightly lower HRs (HR=0.84, 95% CI: 0.67–1.05 and HR=0.8, 95% CI: 0.61–1.09, respectively) relative to NH-only women, though the estimates were not statistically significant.
Table 3.
Breast cancer risk by Native Hawaiian ethnic mixture across body mass index (BMI) categories. (Ref=NH only)
| Variable/ Ethnic Mixture Group | NH only | NH/ White | NH/ Chinese | NH/ White/ Chinese | NH/ Asian | NH/ Other |
|---|---|---|---|---|---|---|
| Breast Cancer Risk, HR (95% CI) | ||||||
| Model 11 | 1 | 0.95 | 0.96 | 1.00 | 0.84 | 0.81 |
| (0.77–1.18) | (0.75–1.22) | (0.81–1.24) | (0.67–1.05) | (0.61–1.09) | ||
| Model 22 | 1 | 0.97 | 0.99 | 1.03 | 0.87 | 0.83 |
| (0.78–1.20) | (0.78–1.27) | (0.83–1.27) | (0.70–1.10) | (0.62–1.12) | ||
| Model 33 | 1 | 0.95 | 0.98 | 1.01 | 0.85 | 0.82 |
| (0.76–1.18) | (0.77–1.26) | (0.81–1.25) | (0.68–1.07) | (0.61–1.11) | ||
| By BMI category, HR (95% CI) | ||||||
| Healthy weight (<25 kg/m2) | 1 | 0.78 | 0.76 | 0.92 | 0.69 | 0.60 |
| (0.49–1.23) | (0.47–1.23) | (0.59–1.42) | (0.44–1.09) | (0.31–1.13) | ||
| Overweight (25-<30 kg/m2) | 1 | 1.12 | 1.33 | 1.19 | 1.06 | 0.78 |
| (0.74–1.69) | (0.85–2.10) | (0.79–1.80) | (0.69–1.62) | (0.45–1.36) | ||
| Obese (≥30 kg/m2) | 1 | 0.92 | 0.91 | 0.90 | 0.78 | 0.96 |
| (0.67–1.26) | (0.62–1.34) | (0.65–1.25) | (0.55–1.12) | (0.63–1.47) | ||
Hazard ratios and 95% confidence intervals were obtained by Cox proportional hazard models, stratified by age at cohort entry and adjusted for age at cohort entry
Additionally, adjusted for BMI ((<25 kg/m2, 25-<30 kg/m2, (≥30 kg/m2)
Further adjusted for BMI, history of breast cancer, age at menarche, age at first live, age and type of menopause, number of children, smoking status, hormone replacement therapy (HRT), alcohol, and physical activity
Abbreviations: NH, Native Hawaiian; BMI, body mass index
When BMI was added in model 2, HRs across ethnic mixture groups remained largely unchanged, indicating that differences in BMI did not account for variation in risk. In the fully adjusted model (Table 4), overweight (HR=1.19, 95% CI: 1.00–1.41), obese (HR=1.36, 95% CI: 1.15–1.61), smoking former (HR=1.20, 95% CI: 1.03–1.39), smoking current (HR=1.22, 95% CI: 1.02–1.45) and family history (HR=1.34, 95% CI: 1.12–1.60) were the most significant contributors to breast cancer incidence. The NH/Other Asian and NH/Other groups had HRs of 0.85 (95% CI: 0.68–1.07) and 0.82 (95% CI: 0.61–1.11), respectively, suggesting a slightly lower risk compared to NH-only women, but CIs overlapped 1.
Table 4.
The association between risk factors and breast cancer risk in NH women
| Risk Factora | HR | 95% CI Lower | 95% CI Upper | p value |
|---|---|---|---|---|
| Family History of BC (yes) | 1.34 | 1.12 | 1.60 | <0.001 |
| Menarche (13–14y) | 0.93 | 0.80 | 1.08 | 0.34 |
| Menarche (15+ y) | 1.00 | 0.81 | 1.24 | 0.99 |
| Age at first child (=<20) | 0.50 | 0.09 | 2.72 | 0.43 |
| Age at first child (21–30) | 0.51 | 0.10 | 2.77 | 0.44 |
| Age at first child (31+) | 0.59 | 0.11 | 3.28 | 0.55 |
| <45 years at natural menopause | 0.95 | 0.72 | 1.26 | 0.73 |
| 45–49 years at natural menopause | 0.95 | 0.74 | 1.20 | 0.65 |
| 50–54 years at natural menopause | 1.10 | 0.87 | 1.38 | 0.42 |
| >55 years at natural menopause | 1.00 | 0.69 | 1.44 | 0.99 |
| Surgical menopause | 0.82 | 0.65 | 1.03 | 0.09 |
| 1 child | 1.51 | 0.27 | 8.38 | 0.64 |
| 2–3 children | 1.71 | 0.31 | 9.33 | 0.54 |
| >4 children | 1.46 | 0.27 | 7.98 | 0.66 |
| Smoking (Former) | 1.20 | 1.03 | 1.39 | 0.02 |
| Smoking (Current) | 1.22 | 1.02 | 1.45 | 0.03 |
| Hormone Therapy Past (estrogen with/without progesterone) | 1.01 | 0.81 | 1.25 | 0.94 |
| Hormone Therapy Current (estrogen only) | 0.94 | 0.73 | 1.21 | 0.62 |
| Hormone Therapy Current (estrogen and progesterone) | 1.14 | 0.92 | 1.40 | 0.23 |
| Physical Activity (≥ 30 min/d) | 0.89 | 0.78 | 1.02 | 0.11 |
| Physical Activity (missing) | 0.83 | 0.44 | 1.58 | 0.57 |
| Alcohol use (≥1/month –<1/day) | 0.92 | 0.78 | 1.07 | 0.29 |
| Alcohol use (≥1/day) | 0.86 | 0.66 | 1.11 | 0.24 |
| Education (13–15 some college) | 0.97 | 0.83 | 1.14 | 0.73 |
| Education (≥16 college graduate +) | 0.94 | 0.77 | 1.14 | 0.51 |
| Overweight | 1.19 | 1.00 | 1.41 | 0.05 |
| Obese | 1.36 | 1.15 | 1.61 | <0.001 |
All hazard ratios (HRs) were estimated using a Cox proportional hazards model and adjusted for age at cohort entry. Significance p<0.05. Reference group in order of risk factor grouping: NH only, Family of BC (none), Menarche (<13y), Age at first child (no child), Premenopausal, No kids, Smoking (Never), Hormone Therapy (Never), Physical Activity (<30min/d), Alcohol use (never), Education (<=12 high school or less), Normal Weight (BMI<25 kg/m2),
Further stratification by BMI categories revealed that among healthy-weight women, NH/White, NH/Chinese, and NH/White/Chinese had lower HRs of 0.76 (95% CI: 0.47–1.23) to 0.92 (95% CI: 0.59–1.42) relative to the NH-only group, while NH/Other Asian and NH/Other had even lower HRs (0.69, 95% CI:0.33–1.09 and 0.60, 95% CI: 0.31–1.13 respectively). However, the wide CIs indicate considerable uncertainty in these estimates. For the overweight group, NH/White and NH/White/Chinese had HRs close to 1.12 (95% CI:0.74–1.69) and 1.19 (95% CI: 0.79–1.80) in comparison to the NH-only women, respectively, while NH/Chinese had the highest HR (1.33, 95% CI: 0.85–2.10), but the findings were not statistically significant. For the obese group, HRs across all ethnic mixture groups were close to 1, ranging from 0.78 to 0.96, indicating minimal differences in breast cancer risk as compared to NH-only individuals. No significant interactions were observed, indicating that ethnic mixture did not modify the association between obesity (BMI) and breast cancer risk (Table S1).
DISCUSSION
In this population-based cohort of 7,700 NH women, we found that differences in breast cancer risk by ethnic mixture were largely explained by variation in BMI status. After adjusting for BMI and other known risk factors, ethnic mixture was not independently associated with breast cancer incidence. Although age-adjusted incidence rates were highest among NH-only and NH/White/Chinese women (5.7 and 5.9 per 1,000 person-years, respectively) and lowest among NH/Other (3.6 per 1,000 person-years). These differences did not translate into statistically significant risk estimates. In contrast, overweight (HR=1.19, 95% CI: 1.00–1.41) and obesity (HR=1.36, 95% CI: 1.15–1.61) were significantly associated with increased breast cancer risk. These findings indicate that BMI, rather than ethnic mixture, is the primary driver of breast cancer risk in this population. Consistent with this, the association between ethnic mixture and breast cancer risk did not differ by BMI.
Stratification by BMI revealed additional patterns that may be hypothesis-generating. Among normal-weight women, NH-only and NH/White/Chinese had the highest relative risks as compared to NH-only women, while NH/Other Asian and NH/Other had lower HRs (0.69 and 0.60, respectively).Among women with overweight, NH/Chinese had the highest risk, followed by NH/White/Chinese and NH/White. These patterns may reflect underlying biological, socioeconomic, or environmental differences that merit further investigation. Among women with obesity, breast cancer risk appeared uniformly high across NH ethnic mixture groups, supporting the notion that obesity itself may be a stronger determinant of breast cancer risk than ancestry alone.
Despite the largely null findings, this study contributes valuable insights into the multifactorial nature of breast cancer risk among NH women. The consistent lack of significant differences across ethnic mixture groups—even after stratification—underscores the dominant influence of BMI over ethnic background. In women with obesity, breast cancer risk appeared relatively uniform across all ethnic mixture categories, suggesting that excess adiposity may overshadow more subtle genetic or sociocultural risk factors.
Several hypotheses may explain the modest or null associations observed. One possibility is that self-reported ethnic mixture, while culturally meaningful, may not adequately reflect underlying genetic or ancestral variation that could influence breast cancer risk. Prior genomic research among Native Hawaiians has shown that higher proportions of Polynesian ancestry are associated with increased BMI and metabolic disease risk, suggesting potential biological pathways that warrant further investigation [16, 17]. However, since our study relied on self-reported ethnic mixture rather than genomic data, we cannot directly assess these relationships. Future studies integrating both sociocultural and genomic measures of ancestry may help clarify how inherited and environmental factors interact to shape breast cancer risk in Native Hawaiian women.
Another explanation lies in the overwhelming effect of obesity as a risk factor [18, 19], for all women, particularly after menopause. Adipose tissue serves as the primary source of estrogen in postmenopausal women, and elevated estrogen levels are strongly linked to increased breast cancer risk [3, 6]. The consistent HRs observed among obese women across all ethnic-mixture categories support the interpretation that excess adiposity, through hormonal and metabolic mechanisms, may be the dominant driver of breast cancer risk in this population.
Importantly, these findings should not be interpreted as evidence that ethnic background is irrelevant to breast cancer risk. Rather, they emphasize the complex and multifactorial nature of risk, shaped by interactions among genetics, environment, body composition, and social determinants of health. For instance, the elevated, but non-significant risk observed in overweight NH/Chinese women may point toward specific gene-environment interactions that warrant further investigation in larger studies with more precise measures of ancestry and biological factors. Moreover, as recent work by Coots et al. (2025) [20] highlights, the inclusion of race and ethnicity in disease risk models, while sometimes yielding only modest gains in clinical utility, can still offer important value, particularly when assessing risk in historically marginalized populations. Their findings show that race-aware predictions can reduce miscalibration of risk models and improve targeting of preventive care, even when the overall impact on screening decisions appears small. In the context of NH women, who have a unique ethnic mixture profile and a disproportionately high burden of breast cancer, our study underscores the need for risk models that are both accurate and contextually informed [20].
Beyond biological mechanisms, socioeconomic and environmental context plays a substantial role in shaping obesity risk among NH women. National data shows that nearly one in five Native Hawaiian and Pacific Islander youth live in food-insecure households, and lower household income is significantly associated with higher BMI [21]. These findings underscore how economic instability can influence diet quality and obesity trajectories from an early age. Similarly, community-based research in Hawaii has demonstrated that socioeconomic and acculturative stressors, including the high cost of healthy foods, limited financial resources, and constrained access to appropriate health services, can hinder sustained weight loss and obesity prevention efforts in NH adults [22]. Addressing these structural inequities, including affordability of healthy foods and access to supportive community environments, is therefore central to any intervention aimed at reducing obesity and related cancer disparities among NH women.
This study has several strengths, including the use of a large, diverse cohort with extensive follow-up, and the ability to control for a wide range of known breast cancer risk factors. The stratified analysis by BMI adds a critical layer of nuance, revealing patterns that would have been obscured in unstratified models. However, limitations must be acknowledged. The statistical power to detect moderate differences was limited in some ethnic mixture and BMI subgroups. The NH/Chinese overweight group had only 120 cases, yielding approximately 60% power. Additionally, unmeasured factors such as healthcare access or socioeconomic status cannot be ruled out. This study focuses exclusively on NH women living in Hawaii, findings may not be generalizable to other NH or Pacific Islander populations living on the continental U.S., where environmental and healthcare contexts may differ. Not knowing the percentages for each ethnic mixture limited our ability to investigate if increasing the percentage of NH ancestry may have been linked to breast cancer incidence. Also, we had limited information on socioeconomic factors and only assessed lifestyle factors once. Finally, BMI was used as an indirect measure of adiposity and may not accurately capture variation in body composition or fat distribution across individuals [23]. Despite this limitation, BMI remains the most widely used and comparable indicator of obesity in large population-based studies such as the MEC.
From a public health perspective, these findings underscore the central importance of obesity as a modifiable risk factor for breast cancer among Native Hawaiian women, regardless of ethnic mixture background. Given the high prevalence of obesity in this population, interventions targeting weight management may have a significant impact on reducing breast cancer. Nonetheless, continued attention to ethnic background remains important as more precise genetic, social and environmental data become available to contextualize cancer disparities
This is the first population-based study to investigate how self-reported ethnic mixture and adiposity jointly shape breast cancer risk in Native Hawaiian women, an intersection that remains virtually unexplored in the cancer disparities literature. This type of analysis is only possible within the Multiethnic Cohort study, which uniquely captures both the racial complexity, large sample size and comprehensive breast cancer risk factors required to explore these questions. These findings represent a crucial early step in understanding how ancestry and adiposity interact to shape cancer risk in this understudied population.
Supplementary Material
ACKNOWLEDGMENTS
Funding:
This research was supported by the National Cancer Institute, under grant number (U01CA164973, Multiethnic Cohort Study) awarded to L. Le Marchand, C.A. Haiman, and L.R. Wilkens and R01CA263491 awarded to J.A. Shepherd. Additional support was provided by a Diversity Supplement to R01CA263491 (D.Valdez) and by a National Cancer Institute training grant (T32CA229110) to J. Abe. The funding agency had no roles in the design of the study, data collection and analysis, decision to publish, or preparation of the manuscript.
Footnotes
Conflict of Interest: The authors declare no conflicts of interest related to this work.
References
- 1.Hawaii Tumor Registry, University of Hawaii Cancer Center. Cancer at a Glance 2014–2018. [cited 2026 Jan 30]. Available from: https://www.uhcancercenter.org/pdf/htr/Cancer%20at%20a%20Glance%202014-2018.pdf.
- 2.Centers for Disease Control and Prevention. Summary Health Statistics: National Health Interview Survey 2016. Table A-15a. [cited 2026 Jan 30]. Available from: https://ftp.cdc.gov/pub/Health_Statistics/NCHS/NHIS/SHS/2016_SHS_Table_A-15.pdf.
- 3.Picon-Ruiz M, Morata-Tarifa C, Valle-Goffin JJ, Friedman ER, and Slingerland JM, Obesity and adverse breast cancer risk and outcome: Mechanistic insights and strategies for intervention. CA: A Cancer Journal for Clinicians, 2017. 67(5): p. 378–397. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Glassman I, Le N, Asif A, Goulding A, Alcantara CA, Vu A, et al. , The Role of Obesity in Breast Cancer Pathogenesis. Cells, 2023. 12(16): p. 2061. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Parekh N, Chandran U, and Bandera EV, Obesity in cancer survival. Annu Rev Nutr, 2012. 32: p. 311–42. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Cleary MP and Grossmann ME, Minireview: Obesity and breast cancer: the estrogen connection. Endocrinology, 2009. 150(6): p. 2537–42. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Bacong AM, Gibbs SL, Rosales AG, Frankland TB, Li J, Daida YG, et al. , Obesity Disparities Among Adult Single-Race and Multiracial Asian and Pacific Islander Populations. JAMA Network Open, 2024. 7(3): p. e240734. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Cardel M, Higgins PB, Willig AL, Keita AD, Casazza K, Gower BA, et al. , African genetic admixture is associated with body composition and fat distribution in a cross-sectional study of children. Int J Obes (Lond), 2011. 35(1): p. 60–5. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Albright CL, Steffen AD, Wilkens LR, Henderson BE, and Kolonel LN, The prevalence of obesity in ethnic admixture adults. Obesity (Silver Spring), 2008. 16(5): p. 1138–43. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.U.S. Census Bureau. 2010 Census Shows Multiple-Race Population Grew Faster Than Single-Race Population. [cited 2026 Jan 30]. Available from: https://www.census.gov/newsroom/releases/archives/race/cb12-182.html.
- 11.U.S. Census Bureau. Quick Facts Hawaii. [cited 2026 Jan 30]. Available from: https://www.census.gov/quickfacts/fact/table/HI/PST045224.
- 12.Maskarinec G, Morimoto Y, Jacobs S, Grandinetti A, Mau MK, and Kolonel LN, Ethnic admixture affects diabetes risk in native Hawaiians: the Multiethnic Cohort. Eur J Clin Nutr, 2016. 70(9): p. 1022–7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Kolonel LN, Henderson BE, Hankin JH, Nomura AM, Wilkens LR, Pike MC, et al. , A multiethnic cohort in Hawaii and Los Angeles: baseline characteristics. Am J Epidemiol, 2000. 151(4): p. 346–57. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.World Health Organization. Body mass index (BMI). [cited 2026 Jan 30]. Available from: https://www.who.int/data/gho/data/themes/topics/topic-details/GHO/body-mass-index.
- 15.SCHOENFELD D, Partial residuals for the proportional hazards regression model. Biometrika, 1982. 69(1): p. 239–241. [Google Scholar]
- 16.Ioannidis AG, Blanco-Portillo J, Sandoval K, Hagelberg E, Barberena-Jonas C, Hill AVS, et al. , Paths and timings of the peopling of Polynesia inferred from genomic networks. Nature, 2021. 597(7877): p. 522–526. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Sun H, Lin M, Russell EM, Minster RL, Chan TF, Dinh BL, et al. , The impact of global and local Polynesian genetic ancestry on complex traits in Native Hawaiians. PLOS Genetics, 2021. 17(2): p. e1009273. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Dehesh T, Fadaghi S, Seyedi M, Abolhadi E, Ilaghi M, Shams P, et al. , The relation between obesity and breast cancer risk in women by considering menstruation status and geographical variations: a systematic review and meta-analysis. BMC Womens Health, 2023. 23(1): p. 392. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Tzenios N, Tazanios ME, and Chahine M, The impact of BMI on breast cancer – an updated systematic review and meta-analysis. Medicine, 2024. 103(5). [Google Scholar]
- 20.Coots M, Saghafian S, Kent DM, and Goel S, A Framework for Considering the Value of Race and Ethnicity in Estimating Disease Risk. Ann Intern Med, 2025. 178(1): p. 98–107. [DOI] [PubMed] [Google Scholar]
- 21.Mokiao RH, Carlin K, Spencer MS, Young BA, and Fretts AM, The social drivers of health for Native Hawaiian and Pacific Islander youth in the United States. Preventive Medicine Reports, 2024. 39: p. 102658. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Kaholokula JK, Townsend CK, Ige A, Sinclair K, Mau MK, Leake A, et al. , Sociodemographic, behavioral, and biological variables related to weight loss in native Hawaiians and other Pacific Islanders. Obesity (Silver Spring), 2013. 21(3): p. E196–203. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.White KK, Park SY, Kolonel LN, Henderson BE, and Wilkens LR, Body size and breast cancer risk: the Multiethnic Cohort. Int J Cancer, 2012. 131(5): p. E705–16. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data Availability Statement
The data used in this manuscript may be available upon request to the MEC research Committee (https://www.uhcancercenter.org/for-researchers/mec-data-sharing).