Update on triple-negative breast cancer disparities for the United States: A population-based study from the United States Cancer Statistics database, 2010 through 2014
Abstract
Background
Triple-negative breast cancer (TNBC) has been associated with a more aggressive histology, poorer prognosis, and nonresponsiveness to hormone therapy. It is imperative that cancer research identify factors that drive disparities and focus on prevention.
Methods
Using the United States Cancer Statistics database, the authors examined differences between TNBCs compared with all other breast cancers with regard to age, race/ethnicity, and stage at diagnosis.
Results
A total of 1,151,724 cases of breast cancer were identified from 2010 through 2014, with the triple-negative phenotype accounting for approximately 8.4% of all cases. In unadjusted analyses, non-Hispanic black women (odds ratio [OR], 2.27; 95% CI, 2.23-2.31) and Hispanic women (OR, 1.22; 95% CI, 1.19-1.25) had higher odds of diagnosis when compared with non-Hispanic white women. Women aged <40 years had the highest odds of diagnosis compared with women aged 50 to 64 years (OR, 1.95; 95% CI, 1.90-2.01). Diagnosis at American Joint Committee on Cancer stage III and beyond conferred higher odds of the diagnosis of TNBC (OR for stage III, 1.69 [95% CI, 1.68-1.72]; and OR for stage IV, 1.47 [95% CI, 1.43-1.51]). Results varied slightly in adjusted analyses.
Conclusions
The results of the current study demonstrated that there is a significant burden of disease in TNBC diagnosed among women of color, specifically non-Hispanic black women, and younger women. Additional studies are needed to determine drivers of disparities between race, age, and stage of disease at diagnosis.
Introduction
Triple-negative breast cancer (TNBC) has been found to account for approximately 15% of all breast cancer cases, and is associated with aggressive histology, poorer prognosis, shorter survival, and unresponsiveness to usual hormone therapy.1 A study conducted on a sample of 51,074 women using the California Cancer Registry found that the triple-negative phenotype was statistically significantly associated with younger age, African American race/ethnicity, later stage of disease at the time of diagnosis, lower socioeconomic status, and shortened survival.2 To the best of our knowledge, this study is one of the few to include data regarding personal socioeconomic status. Non-Hispanic black (NHB) women also had a significantly earlier age at diagnosis, high-grade tumors, and a higher percentage of TNBCs.3, 4 The focus on TNBC is due to its aggressive nature and poor prognosis, and therefore it is imperative to continue identifying risk factors, whether environmental or genetic, that exacerbate disparities in breast cancer diagnosis to develop and implement more efficacious population-based prevention strategies.
Although research has established multiple individual-level risk factors that are associated with general breast cancer diagnoses, racial disparities persist even after accounting for these risk factors, particularly between non-Hispanic white (NHW) and NHB populations. Although age-adjusted incidence rates for all types of breast cancer are higher in NHW women, mortality rates are higher in NHB women. According to Surveillance, Epidemiology, and End Results (SEER) data, age-adjusted incidence rates from 2011 through 2015 were 128.6 per 100,000 population for NHW women and 126.9 per 100,000 population for NHB women. The age-adjusted mortality rates from breast cancer in 2011 through 2015 was 28.7 deaths per 100,000 population for NHB women and 20.3 deaths per 100,000 population for NHW women.5 Research indicates that survival in NHB women may be worse due to a higher frequency of adverse histologic features.1 The latest data published by the American Cancer Society have demonstrated that age-adjusted rates of TNBC are 2 times greater among NHB women compared with NHW women.6 The SEER data from 2008 to 2014 indicate that even when disease stage was taken into account, NHB women had lower 5-year relative survival rates compared with NHW women (95.4% vs 99.1% for localized disease, 76.6% vs 86.4% for regional disease, and 19.7% vs 28.1% for distant-stage disease).5
Although older age has been linked to breast cancer overall, this relationship is reversed for TNBC.7 It has been demonstrated that TNBC occurs more frequently in younger women and in NHB women.1, 8-12 However, to our knowledge, the factors that drive these disparities remain unknown.
To our knowledge, no studies to date have examined racial disparities in cases with TNBC across the United States using the United States Cancer Statistics (USCS) database because few have searched beyond the scope of one state. In the current study, after unavoidable sample exclusions as described below, the USCS database yielded 39 states with complete data for this study. Previous findings have been limited due to their small sample sizes from either local cancer registries or SEER data, with the smallest being 474 cases from a clinical trial to what to our knowledge was the largest at 452,215 cases over a 10-year period, and thus these findings are neither spatially representative nor generalizable.11-13
Using data from a more comprehensive population data set, the objective of the current study was to validate previous findings in the literature, confirming the percentage of breast cancer cases that are triple negative, and the effects of age, race, and stage of disease at the time of diagnosis on the likelihood that the breast cancer is of TNBC type. There were several research questions of interest. Are the underlying distributions of age, race, and stage of disease at the time of diagnosis different for women with TNBC compared with women with all other types of breast cancer? Do the odds of a TNBC diagnosis among women with breast cancer differ by race, age, or stage of disease at the time of diagnosis at the individual level?
We used the following research hypotheses: 1) NHB women would have higher odds of a TNBC diagnosis than their NHW counterparts in nationally aggregated data analysis; 2) younger women would have higher odds of receiving a TNBC diagnosis; and 3) women diagnosed at late and distant stages of disease would have higher odds of receiving a diagnosis of TNBC.
Materials and Methods
We examined all breast cancer cases diagnosed from 2010 through 2014 from the restricted-use USCS database, which is a population-based surveillance system of cancer registries with data representing approximately 99% of the US population and combines data from both the National Cancer Institute's SEER program and the Centers for Disease Control and Prevention's National Program of Cancer Registries.14 All states participate in the USCS registry data system, but 5 states (Illinois, Kansas, Michigan, Minnesota, and Missouri) did not provide county-level breast cancer data and 4 states (Connecticut, Iowa, New Mexico, and Utah) did not code for triple-negative data. Alaska and Hawaii were excluded from analysis due to missing contextual data. Because this study population is a part of a larger geographically focused study, these states were excluded to ensure comparability. The remaining 39 states were used for analysis. The study was limited to the years between 2010 and 2014 because the USCS did not collect data regarding human epidermal growth factor receptor 2 (HER2) status until 2010.
Statistical Analysis
The data set was analyzed using SAS statistical software (version 9.4; SAS Institute Inc, Cary, North Carolina). Cases of TNBC were identified using site-specific factors 1, 2, and 15. Site-specific factors 1, 2, and 15 identify estrogen receptor, progesterone receptor, and HER2, respectively. The classification variables are categorical for “positive,” “negative,” “borderline,” and “other.” Those coded as negative (value ‘020’) for each site-specific factor were considered triple negative. Late-stage disease was defined as diagnosis at stage ≥III according to the seventh edition of the American Joint Committee on Cancer cancer staging manual,15 whereas distant-stage disease was defined as diagnosis at stage IV. Age groups were defined as <40 years, 40 to 49 years, 50 to 64 years, 65 to 74 years, and ≥75 years, with age 50 to 64 years serving as the referent group. There were 6 race/ethnicity categories in the study: 1) NHW; 2) Hispanic; 3) NHB; 4) American Indian/Alaska Native; 5) Asian; and 6) other; NHW served as the referent category. Descriptive statistics were calculated for age, race, and stage variables in the data set. Chi-square tests and Student t tests were used to compare differences with regard to the distribution of age, race, and stage of disease in cases of TNBC versus all other breast cancer cases. Logistic regression then was used to determine the odds of a diagnosis of TNBC given breast cancer and its variations by race, age, and stage. Unadjusted models included race, age, and stage as individual independent predictors of TNBC given breast cancer. Adjusted models that included age, race, and stage variables simultaneously were run, and late stage and distant stage of disease were included in separate models to avoid perfect collinearity because “distant” is a subset of “late stage.” This approach is similar to that of Bauer et al, who conducted the California Cancer Registry study.1
Results
A total of 1,151,724 breast cancer cases from 2010 through 2014 in the 39 states were identified in the data set, with a mean case age at diagnosis of 61.8 years. In this comprehensive collection of breast cancer cases, approximately 75% of the cases were NHW women; among all races/ethnicities, approximately 27.7% were diagnosed at late stage and approximately 5% were diagnosed at distant stage. In the examined time period, TNBC cases accounted for 8.4% of all breast cancer cases (Table 1). NHB women accounted for a larger percentage of TNBC cases, 21.4%, compared to overall breast cancer cases, 10.9% (Table 2). The triple-negative group had a lower mean age at diagnosis of 59.3 years compared with the non-TNBC breast cancer cases at 62.1 years (P < .0001) (Table 2). As expected, TNBC cases had higher percentages of late-stage and distant-stage diagnoses (Table 2).
| Variable | No. | Percentage | No. With TNBC | Percentage With TNBCa |
|---|---|---|---|---|
| Race | ||||
| NHW | 862,205 | 74.86 | 63,579 | 7.37 |
| Hispanic | 95,507 | 8.29 | 8438 | 8.83 |
| NHB | 135,389 | 11.76 | 20,726 | 15.31 |
| AI/AN | 5504 | 0.48 | 502 | 9.12 |
| Asian | 44,424 | 3.86 | 3042 | 6.85 |
| Other | 8695 | 0.75 | 462 | 5.31 |
| Age, y | ||||
| Mean age groupsc | 1,151,724 | 61.84 | 96,749 | 59.28 |
| <40 | 47,329 | 4.11 | 7474 | 15.79 |
| 40-49 | 179,715 | 15.6 | 17,065 | 9.50 |
| 50-64 | 433,798 | 37.67 | 37,993 | 8.76 |
| 65-74 | 275,982 | 23.96 | 19,806 | 7.18 |
| ≥75 | 214,900 | 18.66 | 14,411 | 6.71 |
| Late-stage disease | ||||
| Yes | 310,588 | 27.68 | 36,659 | 11.80 |
| No | 811,633 | 72.32 | 59,292 | 7.31 |
| Distant-stage disease | ||||
| Yes | 54,073 | 4.82 | 6412 | 11.86 |
| No | 1,068,148 | 95.18 | 89,539 | 8.38 |
| Missing datab | 29,503 | 2.6 | 798 | 2.70 |
| TNBC | ||||
| Yes | 96,749 | 8.4 | ||
| No | 1,054,975 | 91.6 |
- Abbreviations: AI/AN, American Indian/Alaska Native; NHB, non-Hispanic black; NHW, non-Hispanic white; TNBC, triple-negative breast cancer.
- a This is the row percentage of TNBC cases.
- b This represents the number of cases for whom data regarding stage of disease were missing.
- c Percentage is the mean age for each group.
| Variable | Other Breast Cancer N = 1,054,975 | TNBC N = 96,749 | Test for Difference | |||
|---|---|---|---|---|---|---|
| No. | Percentage | No. | Percentage | Statistic | P | |
| Race | Chi-square, 9851.15 | <.0001 | ||||
| NHW | 798,626 | 75.70 | 63,579 | 65.72 | ||
| Hispanic | 87,069 | 8.25 | 8438 | 8.72 | ||
| NHB | 114,663 | 10.87 | 20,726 | 21.42 | ||
| AI/AN | 5002 | 0.47 | 502 | 0.52 | ||
| Asian | 41,382 | 3.92 | 3042 | 3.14 | ||
| Other | 8233 | 0.78 | 462 | 0.48 | ||
| Age groups, y | Chi-square, 5051.62 | <.0001 | ||||
| <40 | 39,855 | 3.78 | 7474 | 7.73 | ||
| 40-49 | 162,650 | 15.42 | 17,065 | 17.64 | ||
| 50-64 | 395,805 | 37.52 | 37,993 | 39.27 | ||
| 65-74 | 256,176 | 24.28 | 19,806 | 20.47 | ||
| ≥75 | 200,489 | 19.00 | 14,411 | 14.90 | ||
| Mean age, ya | 62.07 | 59.28 | Student t test, 59.86 | <.0001 | ||
| Late-stage disease | Chi-square, 7182.29 | <.0001 | ||||
| Yes | 273,929 | 25.97 | 36,659 | 37.89 | ||
| No | 752,341 | 71.31 | 59,292 | 61.28 | ||
| Distant-stage disease | Chi-square, 2311.43 | <.0001 | ||||
| Yes | 47,661 | 4.52 | 6412 | 6.63 | ||
| No | 978,609 | 92.76 | 89,539 | 92.55 | ||
| Missing datab | 28,705 | 2.72 | 798 | 0.82 | ||
- Abbreviations: AI/AN, American Indian/Alaska Native; NHB, non-Hispanic black; NHW, non-Hispanic white; TNBC, triple-negative breast cancer.
- a Welch-Satterthwaite Student t test was used to compare mean age differences (F = 1.08; P < .0001 test for equal variance). Percentage is the mean age for each group.
- b This represents the number of cases for whom data regarding stage of disease were missing.
The unadjusted logistic regression models confirmed the results presented in Table 3. NHB, Hispanic, and American Indian/Alaska Native women had higher odds of receiving a diagnosis of TNBC, whereas women of Asian and other races were found to have lower odds when using NHW women as the referent (Table 3). Compared with NHW women, NHB women had the highest odds of a TNBC diagnosis (odds ratio [OR], 2.27; 95% CI, 2.23-2.31), whereas women of a race other than NHB had the lowest odds of being diagnosed with TNBC (OR, 0.71; 95% CI, 0.64-0.77). Comparing age groups using women aged 50 to 64 years as the referent group, women aged <40 years were found to have the highest odds of a TNBC diagnosis (OR, 1.95; 95% CI, 1.90-2.01), whereas those aged ≥75 years had the lowest odds (OR, 0.75; 95% CI, 0.73-0.76). Women diagnosed at a late stage were 69% more likely to be diagnosed with TNBC (OR, 1.69; 95% CI, 1.68-1.72), and women diagnosed with distant-stage disease were approximately 47% more likely to be diagnosed with TNBC (OR, 1.47; 95% CI, 1.43-1.51).
| Variable | Unadjusted | Adjusted: Late Stage | Adjusted: Distant Stage | |||
|---|---|---|---|---|---|---|
| OR | 95% CI | OR | 95% CI | OR | 95% CI | |
| Race | ||||||
| NHW | Referent | Referent | Referent | |||
| Hispanic | 1.22 | 1.189-1.247 | 1.11 | 1.08-1.134 | 1.13 | 1.103-1.158 |
| NHB | 2.27 | 2.233-2.309 | 2.11 | 2.075-2.148 | 2.16 | 2.122-2.196 |
| AI/AN | 1.26 | 1.15-1.382 | 1.19 | 1.083-1.305 | 1.21 | 1.105-1.332 |
| Asian | 0.92 | 0.889-0.959 | 0.85 | 0.819-0.885 | 0.85 | 0.816-0.88 |
| Other | 0.71 | 0.642-0.774 | 0.71 | 0.646-0.788 | 0.7 | 0.634-0.772 |
| Age, y | ||||||
| <40 | 1.95 | 1.902-2.007 | 1.76 | 1.711-1.808 | 1.89 | 1.835-1.939 |
| 40-49 | 1.09 | 1.072-1.114 | 1.07 | 1.046-1.087 | 1.09 | 1.065-1.106 |
| 50-64 | Referent | Referent | Referent | |||
| 65-74 | 0.81 | 0.791-0.82 | 0.85 | 0.833-0.864 | 0.83 | 0.817-0.847 |
| ≥75 | 0.75 | 0.734-0.764 | 0.82 | 0.8-0.833 | 0.81 | 0.792-0.824 |
| Late-stage disease | ||||||
| Yes | 1.70 | 1.675-1.722 | 1.58 | 1.558-1.602 | — | |
| No | Referent | Referent | ||||
| Distant stage disease | ||||||
| Yes | 1.47 | 1.431-1.511 | — | 1.39 | 1.351-1.426 | |
| No | Referent | Referent | ||||
- Abbreviations: AI/AN, American Indian/Alaska Native; NHB, non-Hispanic black; OR, odds ratio; NHW, non-Hispanic white; TNBC, triple-negative breast cancer.
After controlling for late-stage diagnosis and case age, NHB women had approximately twice the odds of diagnosis with TNBC compared with NHW women (Table 3), whereas Hispanic women had odds of a TNBC diagnosis that were similar to those of NHW women. The youngest age group had the highest odds of a TNBC diagnosis, whereas the oldest age group had the lowest odds (Table 3). Women aged 40 to 49 years did not have different odds of diagnosis compared with women aged 50 to 64 years. For those women diagnosed at late stage, the odds of a triple-negative diagnosis were 1.58 times the odds for those diagnosed earlier than stage III disease. In the model adjusted for distant-stage disease, the results were similar (Table 3). Women aged 40 to 49 years had a slightly higher odds of diagnosis with TNBC at 1.09, whereas the results remained the same for Hispanic women. Those diagnosed at distant stage had 1.39 times the odds of a diagnosis of TNBC.
Discussion
To the best of our knowledge, the current study is both the most geographically broad in terms of the scope of the states included and the most recent in terms of the data examined. Using more current data ensures the consistency of TNBC coding and provides an improved estimate of the overall burden of this highly problematic type of breast cancer. Having a broader sample brings greater heterogeneity in terms of local context into the realm of analysis, which can be fruitful when using variations across predictors to explain variations across geography in future studies.16
In the 39-state subset of the entire US breast cancer population, approximately 8.5% of cases were classified as triple negative using site-specific factors. Triple-negative cases were found to account for fewer breast cancer cases (8.5%) compared with previous studies, which used smaller samples and/or focused on a single state or city, with estimates ranging from 10% to 15%.1, 17-23 The localized studies used data from California1, 20; Atlanta, Georgia19, 22; and Detroit, Michigan.21 The distribution by race was different for the triple-negative cases compared with all other breast cancer cases (P < .0001). NHB women accounted for approximately 10.9% of the other breast cancer cases, but 21.4% of the TNBC cases. By contrast, NHW women accounted for approximately 75.7% of the other breast cancer cases, but only 65.7% of TNBC cases. In the study by Bauer et al,1 NHB women accounted for approximately 4.4% of other breast cancer cases and 10% of TNBC cases.
In the current study, age group distributions were different between cases of other breast cancers and TNBC (P < .0001). The youngest age group, those aged <40 years, accounted for approximately 3.8% of other breast cancer cases and 7.7% of triple-negative cases. In what to our knowledge is the most comparable registry study, which covered California Cancer Registry data from 1999 through 2003,1 this age group accounted for more breast cancer (5.7%) and TNBC (12.2%) cases. The percentage of those diagnosed at late stage and distant stage was higher in the TNBC group compared with the other breast cancer cases. Late-stage diagnosis occurred in approximately 37.9% of triple-negative cases, and distant-stage diagnosis occurred in 6.6%. This finding is contrary to the results of the study by Bauer et al,1 which found that late-stage cases of TNBC accounted for approximately 15% of the cases and distant stage accounted for 4%. We found evidence of different distributions by age, race, and stage at diagnosis compared with previous studies that used older data from more limited geographic areas. These stark differences indicate the importance of national population-based studies to demonstrate generalizable findings with which to inform comprehensive cancer control efforts.
More generally, the results of the current descriptive analysis confirmed disparities previously found in the literature and demonstrated that, with regard to the diagnosis of TNBC, there are significant burdens among women of color, specifically NHB women; younger women; and women diagnosed at a later stage. The current study found that those burdens were higher among these groups than had been previously estimated because the disparities among groups were greater than previously found, even though the overall impact on breast cancer cases was lower, potentially due to the use of a more comprehensive population and more current data. These differences were confirmed in the logistic regression analyses. In both adjusted and unadjusted models, NHB women were found to have significantly higher odds of a diagnosis of TNBC compared with NHW women. Women in the youngest age group also had significantly higher odds of a triple-negative diagnosis, as did women diagnosed at late and distant stages of disease.
The current study was limited in scope because it focused solely on one disease subtype: TNBC. Future studies can expand the scope to examine additional subtypes of breast cancer and explore differences between these subtypes. In addition, the current study focused on disease of stage ≥III, in which morbidity and mortality generally are higher. The inclusion of additional stage information could provide insight into the distribution of disease in this particular population. Given the large sample size and geospatial coverage of the data, these results are somewhat different from and also more generalizable compared with data from previous studies. The lower incidence of TNBC potentially could be due to changes in data collection from registries because 2010 was the first year that SEER began collecting data regarding HER2 status for breast cancer cases. The current analysis focused solely on individual characteristics to update the literature with more comprehensive and recent data regarding TNBC. It has been found that breast cancer subtype was a significant factor in prognosis with regard to survival, and that the triple-negative subtype demonstrated the worst prognosis irrespective of race, age, or stage of disease.23 With regard to these results, it is important to consider what additional external factors may influence individual-level variations in diagnosis. Examining external factors such as physical and social environmental characteristics may elucidate drivers of disparities further because the literature consistently has examined individual biological factors. Due to the aggressive nature of TNBC and the lack of therapeutic options, it is important to know which groups experience a higher risk to better provide interventions.
Funding Support
Funded by the National Institute on Minority Health and Health Disparities of the National Institutes of Health under award 1F31MD012752 (to Lia C. Scott). The Centers for Disease Control and Prevention's National Program of Cancer Registries contributed funds to cover the standard research data center fees for researchers conducting analyses under approved research projects. The content is solely the responsibility of the authors and does not necessarily represent the official views of Georgia State University, the University of North Carolina, the National Center for Health Statistics, the National Institute on Minority Health and Health Disparities, or the National Institutes of Health.
Conflict of Interest Disclosures
Lia C. Scott received a Dissertation Training Grant (F31-Diversity) from the National Institutes of Health. The other authors made no disclosures.
Author Contributions
Lia C. Scott: Conceptualization, formal analysis, funding acquisition, methodology, visualization, writing–original draft, and writing–review and editing. Lee R. Mobley: Funding acquisition, supervision, and writing–review and editing. Tzy-Mey Kuo: Data curation, validation, visualization, and writing–review and editing. Dora Il’yasova: Funding acquisition, methodology, validation, and writing–review and editing.
