Variation in Breast Cancer Subtype Incidence and Distribution by Race/Ethnicity in the United States From 2010 to 2015

This cohort study describes the incidence rates of 4 breast cancer molecular subtypes stratified by race/ethnicity among women in the US.


Introduction
Breast cancer is the most commonly diagnosed cancer and the leading cause of death in women worldwide. 1,2 Despite recent advances in treatment and prognosis of this disease, the racial/ethnic gap in incidence and death rates has widened. In the United States, the incidence of breast cancer is higher among Black and White individuals younger than 45 years, whereas the rate in White women between the ages of 60 and 84 years is substantially higher than in Black women. 3 Although the incidence in Black women is lower, the mortality rate is higher in Black women than in White women. 4,5 A propensity-matched study using the National Cancer Data Base reported that, among female patients with hormone receptor (HR)-positive breast cancer, Black women had twice the risk of death compared with White women. 6 Miller et al 7 reported that, in a study of more than 1.3 million women in the US who were diagnosed with breast cancer between 2001 and 2009, the survival rate for Black women was consistently 10 percentage points lower than for White women, and this disparity persisted over time.
The racial/ethnic disparity in breast cancer is complex, comprising biological heterogeneity; the presence of comorbid conditions; and the multifaceted interactions of behavioral and social factors as well as access to quality health care. 8,9 Dismantling the biological and access factors associated with the death rate in breast cancer is essential for the development of successful measures.
Emerson et al 10 used data from the Carolina Breast Cancer Study 3 involving 2998 women with invasive breast cancer to evaluate the interaction between race/ethnicity and tumor characteristics; the study oversampled younger women (<50 years) and Black women. The authors used latent class analysis to reduce the size of the data and to capture complex patterns, which varied across people from different racial/ethnic groups. As a result, compared with White women, Black women were found to have lower socioeconomic status, higher cure barriers, and multiple aggressive aggregation of tumors. 10 Warner et al 11 examined 20 025 female patients with breast cancer at the National Comprehensive Cancer Network (NCCN) centers to identify the race/ethnicity-specific survival rates based on subtypes and the mediating factors in breast cancer incidence and survival. Although this study by Warner et al 11 was well done, it analyzed limited data and subtype dimensions. Apart from the molecular subtype, race/ethnicity-based disparities in breast cancer might also be associated The racial/ethnic disparity in the incidence and distribution of breast cancer is still largely unknown. Analysis of this problem with a wider perspective and a larger sample size is necessary. In this cohort study, we used the Surveillance, Epidemiology, and End Results (SEER) Program to identify rates of breast cancer in the United States by molecular subtypes, histological grades, pathological variations, tumor size, TNM stages, and tumor locations among patients from different race/ethnic groups. We aimed to examine the racial/ethnic patterns associated with the incidence of the subtypes of breast cancer and distribution of patients across clinicopathological variables.

Methods
This cohort study was approved by SEER, and the requirement for informed patient consent was waived based on the Common Rule. Each study procedure complied with the Declaration of Helsinki 12,13 or similar standards of ethics. We followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline.

Study Population
Data were obtained from the SEER database, which collected data from 18  approximately 28% of the US population. 14 Because the relative breast cancer survival rate was more than 90%, data on the long-term follow-up of patients with breast cancer were collected. In addition, we included follow-up information for patients diagnosed in 2010 because recording of ERBB2 (formerly known as HER2) status began in 2010.
We used the following criteria for inclusion in this analysis: (1) female patients with primary unilateral breast cancer who underwent surgical treatment; (2) record of estrogen receptor, progesterone receptor, and ERBB2 status; (3) record of medical history and histological subtype of the specified tumor location; and (4) data on patient race/ethnicity, lateral tumor position, tumor size, tumor TNM stage (per the AJCC seventh edition), and number of tumors. We excluded patients who did not meet these criteria. The SEER Program uses the self-identified race/ethnicity information in the US Census (the 2000 Census for the study period we analyzed) and on death certificates. 14 In this study, the race/ ethnicity classifications were non-Hispanic White, Hispanic White, Black, Asian/Pacific Islander, American Indian/Alaskan Native, and unknown. Persons of unknown race/ethnicity were excluded from the primary statistical analysis.

Statistical Analysis
To calculate age-standardized incidence rates and incidence rate ratios (IRRs) with 95% CIs, we used SEER*Stat, version 8.3.6 software (National Cancer Institute). We adopted 2-sided tests to match

JAMA Network Open | Oncology
Variation in Breast Cancer Subtype Incidence and Distribution by Race/Ethnicity in the United States the race/ethnicity of patients with subtypes of breast cancer. To calculate case to case odds ratios (ORs) for racial/ethnic groups, we used polytomous logistic regression compared with the non-Hispanic White group (the reference group), adjusting for age, geographical region, and year of diagnosis. As the comparison outcome-for example, if the HR-positive and ERBB2-negative subtype was identified-the case-to-case ORs were equal to a modified ratio for the subtype of breast cancer and the race/ethnicity of a subpopulation.
A 2-sided P < .05 or an IRR whose 95% CI excluded 1.0 was considered statistically significant.
Data were analyzed from January 1, 2010, to December 31, 2015.  (Table). More diagnoses were recorded in later years than in earlier years of the study period except for the last year of study for American Indian/Alaskan Native patients.

Incidence of Breast Cancer by Race/Ethnicity
The annual incidence rate of all breast cancers in non-Hispanic White women was 31.3 (95% CI, 31.2-31.5) per 100 000 people, which was higher compared with the incidence among Black women (IRR,

Tumor Molecular Subtype and Histological Grade
The incidence of HR-positive and ERBB2-negative, HR-negative and ERBB2-positive, HR-positive and ERBB2-positive, and TNBC molecular subtypes varied markedly according to race/ethnicity ( Figure 1). In Black patients, the incidences of HR-positive and ERBB2-positive (IRR, 1.12; 95% CI,

Tumor Size and TNM Stage
The incidence of different sizes of breast tumors varied considerably by race/ethnicity (Figure 3)
With respect to the AJCC TNM stages (Figure 4)

Distribution of Patients by Race/Ethnicity Tumor Molecular Subtype and Histological Grade
The proportion of Black patients with the HR-positive and ERBB2-negative subtype was smaller than in non-Hispanic White patients (16 372

Discussion
Breast cancer is the most commonly diagnosed cancer in women worldwide. 2 It is also the top cause of cancer death in women. 15 High incidence rates have been found in North America, Australia and New Zealand, Western and Northern Europe, and Asia and Sub-Saharan Africa. 16 Regional disparities in breast cancer incidence are likely associated with societal changes brought on by industrialization. 16 Breast cancer is among a diverse group of malignant neoplasms that have characteristics associated with race/ethnicity and region and whose subtype results show substantial variations. 17 This disease exhibits various morphological characteristics, immunohistochemical profiles, and

JAMA Network Open | Oncology
Variation in Breast Cancer Subtype Incidence and Distribution by Race/Ethnicity in the United States molecular subtypes with particular clinical course and outcome. 17 Considering the diversity of the US population and the availability of data from the SEER database, we analyzed the incidences of breast cancer subtypes after diagnosis and distributions of patients across clinicopathological variables.
To our knowledge, this study is the first to carry out a comprehensive yet detailed comparative analysis of breast cancer incidence by race/ethnicity from the perspectives of molecular subtype, histological grade, pathological pattern, tumor size, TNM stage, and tumor location. As an extension of similar previous studies, [18][19][20] this study also observed the variations in these clinicopathological characters. We believe that this study was an important step toward ameliorating, not just describing, racial/ethnic differences in breast cancer. The results suggest that combining epidemiologic data with genomic and molecular profiling data may be a critical future research direction.

Strengths and Limitations
This study has some strengths. The most notable strength is the use of the population-based SEER database, which classifies more than 95% of patients with cancer in their geographically dispersed catchment areas, representing approximately 28% of the US population. 21,22 Hence, the findings of this study reflect current breast cancer incidence in the US. Although Warner et al 11 previously published a similar study, the non-population-based NCCN database that the authors used only examined patients who had access to and were receiving treatments at key cancer centers. In the Warner et al 11 report, the median age of patients with breast cancer was nearly 10 years younger than the national median age.
This study has some limitations. The percentage of patients with breast cancer was relatively small because of the relatively small American Indian/Alaskan Native group. Nonetheless, we found statistically significant differences in the proportions and incidence rates in the subtypes of breast cancer. In comparison, misclassifications of race/ethnicity in cancer studies could affect the outcomes; for instance, if the American Indian/Alaskan Native group were misclassified, the group may be mistaken as belonging to the Hispanic White or unknown race group. However, case-to-case ORs with recovery analyses use only medical information and should be less affected by race/ ethnicity mismanagement. Also, incidence estimates are sensitive to ascertainment bias and the biases linked to screening and the overdetection of low-grade disease because of differential access to health care or the variation in diagnostic procedure. Thus, incidence rates and IRRs should be interpreted with caution for the American Indian/Alaskan Native group and other racial/ethnic groups.

Conclusions
In this cohort study, disparities were found in the incidences and proportions of different molecular subtypes, histological grades, pathological patterns, tumor sizes, TNM stages, and tumor sites of breast cancers across different racial/ethnic populations in the United States. This study could serve as a step toward ameliorating the racial/ethnic differences in breast cancer. Future research that combines epidemiologic data with genomic and molecular profiling data may serve as a future research direction and help in closing the race/ethnicity-based gap in the incidence and distribution of breast cancer subtypes.