Female Breast Cancer Trends in Disaggregated Asian American Populations: Analysis of 2003-2017 U.S. Mortality Data

cancer mortality rates in Asians may improve culturally-tailored outreach, prevention, and treatment strategies to reduce cancer deaths from this critical disease


Introduction
In the United States, 1 in 8 women (12%) will be diagnosed with invasive breast cancer in their lifetime and 1 in 39 women (3%) will die due to breast cancer 1 .Efforts such as emerging treatments and more aggressive screening practices have signi cantly reduced breast cancer mortality by 40% from 1989 to 2017 1 .Despite this progress, breast cancer is still predicted to be the leading form of new cancer diagnosed in 2020 and will account for 30% of all predicted cancer cases in all American females 2 .
When aggregated, women with Asian/Paci c Islander ancestry appear to have the lowest breast cancer incidence (93.7 per 100,000) and death (14 per 100,000) rates of all American racial groups 1 , but the aggregate Asian American (AA) breast cancer incidence has been increasing by 1.5% yearly 3,4 .With over 17.3 million AAs in the U.S. in 2010, this rate of increase in breast cancer indicates a signi cant healthcare burden 5,6 .
The distinct ethnic subgroups comprising the AA population are extremely diverse in country of origin, length of US residence, acculturation, socioeconomic status, language, culture, religion, and more [7][8][9][10] .By aggregating all AAs into a single group, health care providers and policy makers may miss critical heterogeneity around breast cancer healthcare and mortality 11 .For example, Thompson et al. examined   disaggregated death certi cate data of AAs (Asian Indian, Korean, Japanese, Chinese, Vietnamese, and Filipino) from 2003 to 2011 and found that female Filipino Americans had the highest comparative rates of breast cancer amongst the six AA ethnic groups, even if the aggregated AA death rates trended downward 12 .However, this study did not nd any signi cant breast cancer mortality trends for any of the AA groups.
In this study, we present updated trends in female breast cancer mortality rates from 2012 to 2017 among six disaggregated AA ethnicities in comparison with Non-Hispanic White (NHW) females using the Centers for Disease Control and Prevention (CDC) National Vital Statistics System (NVSS) mortality dataset.We also compare current breast cancer mortality trends to previous trends from 2003 to 2011.

Data
U.S. mortality records between 2003 to 2017 for all fty states were obtained under an IRB data-use agreement from the CDC NVSS mortality dataset (Protocol # 53429).Death certi cates were completed by clinicians and/or medical coroners at time of death and include information about age, sex, race/ethnicity, and cause of death.Underlying causes of death were classi ed by the International Classi cation of Diseases (ICD), 10th revision.ICD-10 code C50 (malignant neoplasm of the breast) identi ed breast cancer as the primary underlying cause of death.Other variables of interest included sex and age.

Participants
Female NHWs and AAs within six ethnic groups were included: Asian Indians, Chinese, Filipinos, Japanese, Koreans, Vietnamese.Hispanics, African American or Black, Paci c Islanders, and any AAs not in the six selected subgroups were excluded.Males were also excluded from the study.Incomplete death certi cates (decedents with missing data for age and primary cause of death variables) and individuals with mixed racial identity or unknown age were excluded.Data analysis was conducted for 473,927 NHW females and 11,388 AA females (1425 Asian Indians, 2898 Chinese, 3865 Filipina, 1549 Japanese, 949 Korean, 702 Vietnamese) (Fig. 1).

Statistical Analysis
Three different numerical measures were calculated: age-adjusted mortality rates (AAMR), standardized mortality ratios (SMR), and proportionate mortality.AAMR was calculated as the deaths per 100,000 people.Annual percentage change (APC) was calculated as the average percentage change in the fteenyear period.SMR was calculated as a ratio of the AAMRs with NHW as the reference.Proportionate mortality was calculated as the proportion of breast cancer deaths compared to all cancer deaths.
Population counts were extrapolated based on the 2003 and 2010 US Census.To account for the rolling adoption of the 2003 death certi cate by state, population adjustment was conducted by enrolling descendants from each state as disaggregated information became available 13 .Death and population data from each state were included in the numerator and denominator, respectively, when that state adopted the 2003 death certi cate.For AAMRs from 2003-2017, age adjustment was calculated using the 2010 US standard age distribution of the population.The same standard age distribution was applied to each group's age-speci c mortality rates in order to compare groups.Nativity population data for each of the six AA subgroups was gathered from the American Community Survey (ACS) 14 .We did not suppress data with < 10 deaths per cell.A linear model, and not a joinpoint analysis, to analyze the overall trend.

Results for Aggregate AAs
We identi ed 624,221 female AA decedents and 30,269,449 female NHW decedents over the fteen-year study period, of whom 612,833 female AAs and 30,222,057 female NHWs were excluded and 11,388 female AAs and 473,927 female NHWs died of breast cancer (Supplementary Fig. 1).Overall, 13.7% of aggregate female AA and 14.4% female NHW cancer deaths were from breast cancer.AAMRs within disaggregated female AAs ranged from 3.5 per 100,000 (Vietnamese) to 21.6 per 100,000 (Filipina) (Fig. 2 and Supplementary Fig. 2).
Proportion of breast cancer deaths: Although proportionate mortality of breast cancer deaths compared to all cancer deaths for aggregate AAs (13.7%) and NHWs (14.4%) appear comparable, the disaggregated AA ethnicities show great variability (Fig. 4).Asian Indians (19.8%) and Filipinas (18.6%) had the highest proportion of breast cancer deaths, as well as NHWs (14.4%).On the other end, Korean (10.0%),Japanese (10.7%), and Vietnamese women (10.5%) had the lowest proportion of breast cancer mortality.

Nativity Analysis
An analysis of breast cancer mortality trends from 2003-2017 for each of the six Asian groups showed no signi cant differences by place of birth, except for Japanese women (Fig. 5).US-born Japanese women had decreasing rates of breast cancer deaths (APC − 1.0; CI -1.7 -2.4; p 0.0001), while breast cancer deaths increased in foreign-born Japanese (APC 11.4; CI 0.8 6.12; p 0.02) (Fig. 6 and Supplementary Fig. 3).

Discussion
Over the past 15 years, we found that non-Hispanic white women have exhibited steadily decreasing breast cancer mortality trends (about 2.1% annually), while the aggregate Asian Americans breast cancer deaths have remained stable.However, when AAs groups are disaggregated, they demonstrate considerable heterogeneity.Speci cally, we found that Filipina, Korean and Chinese women have about a 2% annual increase in breast cancer deaths.Moreover, Filipina women had the highest age-adjusted breast cancer rates which were about two times higher than aggregate AAs.By contrast, Vietnamese, Asian Indians, and aggregated AAs exhibited a stable breast cancer mortality trend.While aggregated proportions of cancer deaths were the same for NHW and AA (about 14%), when disaggregated, AA groups demonstrated broad variability in mortality.The large variability in breast cancer mortality rates by AA groups corresponds to results from earlier studies.Thompson et al., found that Filipinas had the highest rate of breast cancer, contributing to 19.5% of all Filipina cancer deaths 12 .They also found 19.8% of all Asian Indian cancer deaths were from breast cancer, now the leading cause of all Asian Indian cancer deaths.These ndings might be in uenced by cancer type, hormone therapy, access to care, screening and lifestyle behaviors amongst women within each AA group.
High but stable breast cancer mortality in Asian Indian women and low but stable mortality in Vietnamese women may be in uenced by cancer type.Asian Indians are more likely than NHWs to have triple-negative breast cancer [15][16][17] , which is relatively uncommon and, until recently, without curative targeted therapies [18][19] .Conversely, Vietnamese women are more likely to have HER2-positive breast cancer 20 , which can have a poor prognosis if not treated optimally with targeted therapies 21 .The stable breast cancer mortality among Asian Indian and Vietnamese women may also be in uenced by access to care 22 .Increasing AA breast cancer mortality may be driven by post-menopausal hormone replacement therapy (HRT) 23 .Breast cancer risk increased 26% per 5 years of current use of estrogen and progestin therapy (p = 0.017) in a study of 1,277 Chinese, Japanese, and Filipinas 24 .Asian and Paci c Islander postmenopausal women may be at higher risk for developing breast cancer (OR 1.58; 95% CI of 1.18-2.11)than NHW women (OR 1.21; 95% CI 1.14-1.28)after using HRT 25 .
Increasing breast cancer mortality in Filipina, Chinese, and Korean women may be due to lack of preventative public health measures, including screening 26-28 and treatment 29 .AA cancer screening rates are lower than those of NHWs for all cancer types 30 and AA women (65%) tend to have lower breast cancer screening rates than NHWs (68%) 31 .In a study comparing mammography rates among AA subgroups, South Asian and Korean women had the lowest mammography rates while Japanese women had the highest rates, second only to NHW women 7 .By contrast, Korean American women had the lowest rates of breast cancer screening 32 .Acculturation, healthcare access, and low levels of health literacy have been posited as potential reasons for reduced breast cancer screening in AAs 30,33−35 .Breast cancer screening rates also vary between the AA subgroups 26 .
Lower levels of screening, perhaps coupled with an increase in risk factors associated with a Western lifestyle (e.g., obesity, alcohol, tobacco, unopposed continuous estrogen from reduced and/or delayed parity) may contribute to increasing breast cancer mortality trends within Korean women.In addition, screening behaviors vary by Asian American group.A meta-analysis found that only 34-65% of Korean women aged 50-74 reported obtaining a mammogram within the last two years, compared to 73% of white women and 72% of other Asian women 36 .Mammography screening among Korean American women in the past two decades has also signi cantly increased, but these rates are still much lower than NHWs, especially since Korean American women tended to be uninsured and sought out professionals from South Korea where preventative health measures are unusual 26 .Consistent with their lower utilization of mammography screening, Korean women were twice as likely to be diagnosed with a tumor size > 1 cm as compared to White women (OR 2.38; CI 1.49 3.80) 37 .
Amongst Asians, Japanese women had the lowest rates of breast cancer mortality.Not only have Japanese American women been shown to have increased mammography rates compared to other Asians 38 , Japanese American women with breast cancer also had a lower risk of having advanced-stage disease than NHW women or those in other Asian subgroups 39 .Increased adoption of breast screening among Japanese may be driven by a larger proportion of acculturated, highly educated, and wealthier Japanese American women as compared by other AA ethnicities 8,40 .
Our study has several limitations.While we captured all deaths in the United States, ethnic/racial groups may have been misclassi ed unintentionally by the healthcare provider or funeral director completing the death certi cate, especially if next-of-kin or family members were not present 41 .However, ethnicity misclassi cation has been diminishing over time, and are estimated at 3% for Asian Paci c Islanders 41 .States' adoption of the 2003 death certi cate Asian subgroup classi cation occurred over time, although the states with the largest Asian populations were among the earlier adopters 13 .Finally, as death certi cate data are not linked to other medical information, no information is available about other socioeconomic data that might inform further mortality-related analysis.

Conclusion
Breast cancer mortality may be improved with healthy lifestyles, early diagnosis via mammographic screening, tailored treatment, and ongoing follow-up.The high and often increasing rates of breast cancer mortality among speci c Asian American groups point to the need to disaggregate AA subgroups during research studies, to ensure that potential disparities in access to effective screening, treatment and follow-up care do not go unnoticed and unresolved.

Figure 5 Total
Figure 5