Fax: (319) 273-5958
Racial differences in treatment and survival from early-stage breast carcinoma†
Version of Record online: 3 OCT 2002
Copyright © 2002 American Cancer Society
Volume 95, Issue 8, pages 1759–1766, 15 October 2002
How to Cite
Joslyn, S. A. (2002), Racial differences in treatment and survival from early-stage breast carcinoma. Cancer, 95: 1759–1766. doi: 10.1002/cncr.10827
Data were provided by the Surveillance, Epidemiology, and End Results (SEER) program public-use CD-ROM (1973–1997), National Cancer Institute, DCCPS, Cancer Surveillance Research Program, Cancer Statistics Branch, released April 2001, based on the August 2000 submission.
- Issue online: 3 OCT 2002
- Version of Record online: 3 OCT 2002
- Manuscript Accepted: 20 MAY 2002
- Manuscript Revised: 7 MAY 2002
- Manuscript Received: 4 FEB 2002
- breast carcinoma;
- breast-conserving surgery;
- radiation therapy;
- African American
African-American women have a significantly worse prognosis from breast carcinoma compared with white women, even when the stage at diagnosis is equivalent. The purpose of this study was to analyze racial differences in the treatment (use of breast-conserving surgery and radiation therapy) of women with early-stage breast carcinoma and the resulting effects on survival rates.
Subjects included 10,073 African-American and 123,127 white women diagnosed with Stage I, IIA, or IIB breast carcinoma in the National Cancer Institute's Surveillance, Epidemiology, and End Results program between 1988 and 1998. Comparisons were made by race with treatment, age, hormone receptor status, and stage at the time of diagnosis. Survival analyses were conducted to compare risk of death for African-American and white women while controlling for age, stage, and hormone receptor status.
Among women diagnosed with early-stage breast carcinoma who receive breast-conserving surgery, African-American women were significantly less likely to receive follow-up radiation therapy in every 10-year age group except in the older than 85 age group. Whether treatment was equivalent or suboptimal, survival for African-American women with early-stage breast carcinoma was significantly worse. However, when treatment was equivalent, the effects of racial differences on survival were significantly less compared with survival associated with suboptimal treatment.
Significant racial differences exist in the treatment of women with early-stage breast carcinoma. Public health efforts to eliminate suboptimal treatment would reduce, but not eliminate, racial disparity in survival. Cancer 2002;95:1759–66. © 2002 American Cancer Society.
Breast carcinoma is the most commonly diagnosed cancer among women of all racial groups in the United States, with an estimated 203,500 new cases that will be diagnosed in 2002.1 Although incidence rates have increased over the past 30 years, the proportion of women with breast carcinomas who are diagnosed in the early stages of the disease has increased as well, resulting in a significant overall downward trend in breast carcinoma mortality since 1990.1
Despite these recent decreases in breast carcinoma mortality rates, significant survival disparities exist between the races and the gap is widening between breast carcinoma mortality rates for African-American and white women. In 1970, chances of dying of breast carcinoma were approximately equivalent for African-American and white women. Since that time, mortality rates have declined 15% for white women and have increased 22% for African-American women.2 Currently, the age-adjusted mortality rate for African-American women is 31.0 per 100,000 compared with 24.3 per 100,000 for white women.1 This widening gap is especially significant because of the lower breast carcinoma incidence rates among African-American women compared with white women (101.5 per 100,000 vs. 115.5 per 100,000, respectively),1 which indicates a poorer prognosis for African-American women diagnosed with breast carcinoma.
Several factors have been examined in relation to this poorer prognosis. One important variable that has been implicated is that African-American women are significantly more likely to be diagnosed with advanced stages of breast carcinoma compared with white women (29.1% vs. 42.7%; diagnosed at Stage I and 7.3% vs. 4.2% diagnosed at Stage IV, respectively).3 However, studies have found that for women diagnosed at the same stage, risk of death is significantly higher for African-American women,3, 4 with race remaining an independent predictor of mortality. From these results, it may be hypothesized that racial differences affect the quality of treatment received by African-American women.
In the treatment of early-stage breast carcinoma, breast-conserving surgery (BCS), consisting of a margin-negative lumpectomy (with or without axillary lymph node dissection) and follow-up breast irradiation, is now well established as a safe and frequently preferred treatment, although mastectomy is still a commonly used procedure. Although mastectomy and BCS (with follow-up radiation therapy) have equivalent outcomes (breast carcinoma recurrence rates, survival rates) for treating early-stage breast carcinoma, BCS without follow-up radiation therapy has an increased risk of breast carcinoma recurrence5–11 and mortality.12 Results of studies on racial differences in the use of BCS are inconsistent.13–15 In addition, a significant reduction in the use of irradiation following lumpectomy has been shown in older age groups,12, 16, 17 but there are relatively limited reports of data regarding racial variations in radiation treatment. Therefore, the purpose of this study was to analyze racial differences in treatment (use of BCS and radiation therapy) for early-stage breast carcinoma and the resulting effects on survival.
MATERIALS AND METHODS
Procedures for this study included analysis of data for women diagnosed with breast carcinoma in the National Cancer Institute's Surveillance, Epidemiology, and End Results (SEER) program. Since 1973, this national cancer surveillance program has collected information on all cases of invasive cancer diagnosed in residents of nine geographic areas of the United States: metropolitan Atlanta (since 1975), metropolitan Detroit, the San Francisco/Oakland Standard Metropolitan Statistical Area (SMSA), the Seattle/Puget Sound SMSA (since 1974), and the states of Connecticut, New Mexico, Hawaii, Utah, and Iowa. The geographic areas of the SEER program comprise approximately 12% of the U.S. population and are representative of national demographic characteristics. In addition to collecting cancer-related and demographic information at the time of cancer diagnosis, the SEER program tracks patients annually for survival status. Although breast carcinoma data have been collected since 1973, the focus of this study will be on data collected after 1988. During the mid-1980s, results of randomized clinical trials were available that revealed the equivalence of outcomes between BCS and mastectomy. During this period too the use of BCS increased significantly. Before 1988, information on stage at the time of diagnosis was categorized broadly as in situ, localized, regional, distant, or unstaged. Since 1988, staging has been categorized by the more descriptive American Joint Committee on Cancer (AJCC) TNM method.
Subjects for this study were women diagnosed with primary breast carcinoma in the SEER program areas from 1988 through 1998. These cases included those with simultaneous cancer diagnoses in which primary breast carcinoma was present. In addition to treatment, variables analyzed individually by race included hormone receptor status, stage at time of diagnosis, and age.
The treatment variable describes the first course of treatment received by subjects, including only type of surgery and use of follow-up radiation therapy. Type of surgery was categorized as mastectomy or BCS, with the latter group including lumpectomy, segmental mastectomy, quadrantectomy, tylectomy, wedge resection, nipple resection, excisional biopsy, and partial mastectomy not otherwise specified. BCS may or may not have included axillary lymph node dissection and/or follow-up radiation therapy. Type of surgery (BCS vs. mastectomy) was compared by race. Also, the use of follow-up radiation therapy for subjects receiving BCS was compared by race.
Data on hormone (estrogen and progesterone) receptor status has been recorded only since 1990 in the national SEER program. Hormone receptor status was analyzed separately (estrogen receptor [ER]+, ER-, progesterone receptor [PR]+, and PR-) and in combination (ER+PR+, ER+PR-, ER-PR+, and ER-PR-). The cutoff value for positive hormone receptor status varied between laboratories where assays were performed, with cutoff values ranging from greater than 3 to greater than or equal to 20 femtomoles (fmol) cytosol protein per milligram for both receptors. The most commonly used value was greater than or equal to 10 fmol. Immunohistochemistry methods of determining ER and PR status were in common use by 1998 and it is assumed that many laboratories were utilizing this technology, but information on methods of determining hormone receptor status in laboratories throughout the SEER program was not available. Because receptor-borderline tumors behave clinically like receptor-negative tumors,18, 19 this category was combined with the receptor-negative groups.
Race was categorized as African American or white. Small numbers of cases in the “other races” category precluded meaningful statistical analyses for this group and were not analyzed in this study.
Stage at time of diagnosis was categorized using the AJCC TNM coding scheme. The AJCC system takes into account information on tumor size and extension (T), regional lymph node involvement (N), and the presence of distant metastasis (M). Stage categories analyzed in this study included in situ, Stage I, Stage IIA, and Stage IIB, in which choice of treatment generally (although not always) includes BCS as an alternative to mastectomy. In addition to the AJCC stage variable, lymph node involvement was analyzed separately because Stages IIA and IIB include both lymph node-positive and negative patients.
Age was analyzed in three ways. First, mean age as a continuous variable was compared by race and analyzed further by treatment categories. Second, distribution of treatment by 10-year age categories for African-American and white women was compared. Finally, age was used as a surrogate measure of menopausal status, with ages 50 years and younger categorized as premenopausal and ages older than 50 years as postmenopausal. Although these categories are subject to misclassification, this age category definition is a reasonable surrogate for menopausal status when actual information is unavailable.20
Data were analyzed using SAS version 8 statistical software. The numbers and percentages of subjects in race categories were calculated for stage at time of diagnosis, treatment categories, age categories, and hormone receptor status. Indicator variables were created for age categories and tumor stage variables so that each category could be compared with a single reference category. The chi-square statistic was used to determine significant differences between race and other variables. Age (continuous) was analyzed by race and type of treatment using multiple-factor analysis of variance (ANOVA). Significant differences in age between groups were determined following significant ANOVA results using the Scheffe multiple comparison post hoc test.
When comparing survival by race, log rank statistics were calculated.21 To determine whether race was a significant predictor of breast carcinoma mortality, independent of age, stage, and hormone receptor status, Cox proportional hazards multiple regression analyses were done.22 In addition, two separate Cox proportional hazards analyses were conducted for radiation and no radiation categories for women receiving BCS. Survival analyses included death from any cancer as the outcome of interest (“failures”), with all other causes of death and survivors categorized as “censored.” In survival analyses with the Cox regression technique, it is preferable to use all competing causes of death (in this case, all cancer deaths) as the failure outcome for the most conservative explanation of results because death from any cancer may not be an independent event in a patient with existing breast carcinoma. A survival time variable was calculated by determining the number of months between the date of diagnosis and the most recent follow-up. Subjects with unknown survival status or who had died but for whom cancer status was undetermined were categorized with missing survival data.
Estimates of risk ratios were estimated for each independent variable in the full multivariate model as the exponentiation (exp) of the regression coefficient (β). Confidence intervals (95%) were calculated for each estimated risk ratio as exp (β ± 1.96 * standard error [β])
Subjects included 10,073 African-American women and 123,127 white women. Patient and tumor characteristics for the population of women diagnosed with early-stage breast carcinoma are described in Table 1. Compared with white women, African-American women were significantly younger at the time of diagnosis. More than one-third of African-American women were premenopausal compared with fewer than 1.4 of white women. African-American women were significantly more likely to be diagnosed at a more advanced stage and with lymph node-positive disease. Racial differences in type of surgery were statistically significant. African-American women were slightly more likely to receive BCS, but these differences were not clinically significant. In addition, African-American women were significantly more likely to be diagnosed with ER-PR- tumors compared with white women.
|Variable||White (%)||African American (%)|
|Age category, yrs; (P < 0.0001)|
|< 35||2228 (1.8)||462 (4.6)|
|35–44||12,569 (10.2)||1696 (16.8)|
|45–54||23,168 (18.8)||2393 (23.8)|
|55–64||25,022 (20.3)||2071 (20.6)|
|65–74||31,105 (25.3)||2065 (20.5)|
|75–84||22,562 (18.3)||1133 (11.2)|
|85+||6473 (5.3)||253 (2.5)|
|Stage at diagnosis (P < 0.0001)|
|I||71,151 (57.8)||4504 (44.7)|
|IIA||35,618 (28.9)||3466 (34.4)|
|IIB||16,358 (13.3)||2103 (20.9)|
|Surgery (P = 0.0387; n = 132, 428)|
|Breast-conserving surgery||54,556 (44.6)||4545 (45.6)|
|Mastectomy||67,909 (55.4)||5418 (54.4)|
|ER (P < 0.0001; n = 91,481)|
|Positive||67,210 (79.1)||4021 (61.6)|
|Negative||17,744 (20.9)||2506 (38.4)|
|PR (P < 0.0001; n = 88,942)|
|Positive||57,425 (69.6)||3527 (54.9)|
|Negative||25,096 (30.4)||2894 (45.4)|
|Combined hormone receptors (P < 0.0001; n = 88,816)|
|ER+PR+||54,462 (66.1)||3171 (49.5)|
|ER+PR−||10,601 (12.9)||767 (11.9)|
|ER−PR+||2882 (3.5)||350 (5.5)|
|ER−PR−||14,460 (17.5)||2123 (33.1)|
|Lymph node involvement (P < 0.0001; n = 131,753)|
|No||84,231 (73.8)||6101 (66.4)|
|Yes||29,928 (26.2)||3091 (33.6)|
|Menopausal status (P < 0.0001)|
|Premenopausal||28,710 (23.3)||3662 (36.4)|
|Postmenopausal||94,417 (76.7)||6411 (63.6)|
Of the 59,101 women in this population who received BCS (44.6% of the total sample of women diagnosed with early-stage breast carcinoma), 29.9% (n = 1357) of African-American women did not receive follow-up radiation, compared with 24.2% (n = 13,228) of white women (P < 0.0001). Table 2 describes the proportion of women in each age category who received BCS and the proportion who received follow-up radiation. African-American women were significantly less likely to receive radiation therapy in each age category except in the age category of 85 years and older.
|Age category (yrs)||White||African American|
|BCS (%)||Radiation (%)||BCS (%)||Radiation (%)|
Table 3 shows the results of two-factor ANOVA comparing age by treatment group. African-American women are significantly younger than white women in each treatment group (P < 0.0001). Women who received mastectomy were significantly older than those who received BCS (P < 0.0001) and women who received radiation therapy were significantly younger than those who did not (P < 0.0001).
|Overall||62.8 ± 14.3||57.4 ± 14.5|
|Mastectomy||63.3 ± 14.2||57.5 ± 14.4|
|BCS||62.1 ± 14.3||57.0 ± 14.5|
|Radiation||60.1 ± 13.1||55.7 ± 13.3|
|No radiation||68.1 ± 16.3||60.0 ± 16.8|
Results of log rank statistical comparisons of mortality are presented in Table 4. Overall, African-American women had a significantly higher risk of cancer mortality compared with white women. This disparity was also found for women who received BCS. When stratifying the BCS group by follow-up radiation category, racial differences in survival were significant, although the magnitude of differences in survival by race for women who received radiation therapy was approximately one half that of those who did not receive radiation therapy. Similar differences in magnitude exist when the effects of radiation on survival were stratified by age category.
|Age category||Percent Survived||P value|
|BCS (radiation categories combined)||87.9||93.2||< 0.0001|
|BCS, no radiation (yrs)||81.8||89.5||< 0.0001|
|BCS, radiation (yrs)||90.4||94.4||< 0.0001|
Multivariate modeling of factors associated with survival indicated that race remains a statistically significant predictor of survival, even when controlling for other variables (type of surgery, radiation therapy, hormone receptor status, age, menopausal status, lymph node involvement, and stage at diagnosis) in the model. African-American women diagnosed with early-stage breast carcinoma experienced a 43% increase in cancer mortality compared with white women (Table 5).
|Variable||Risk of death||95% confidence interval|
|Age category (yrs)|
|Stage at diagnosis|
|Lymph node involvement|
In the full regression model, BCS was associated with a slightly increased risk of death, whereas follow-up radiation therapy was not significantly associated with a reduced risk of cancer mortality. Women with hormone receptor-positive tumors had significantly better survival, as did younger women, premenopausal women, women with no lymph node involvement, and women with an earlier stage of disese diagnosis. When Cox regression analyses included stratification by radiation versus no radiation for women who received BCS, results were similar to the overall model presented in Table 5. That is, the magnitude of the effect of race decreased whereas the magnitude of the effects of hormone receptor status and stage increased.
In this large, population-based study of racial differences in treatment of early-stage breast carcinoma, African-American women were significantly more likely to receive BCS than white women (statistically significant although clinically insignificant). However, for women diagnosed with early-stage breast carcinoma who received BCS, African-American women were significantly less likely to receive follow-up radiation. Whether treatment was equivalent or suboptimal among African-American women, survival from early-stage breast carcinoma was significantly worse for African-American women.
Several studies found equivalent survival from breast carcinoma between African-American and white women with equivalent treatment.23–29 However, the results of the current study support results of a study by Wojcik et al.30 who reported that survival remains significantly worse for African-American women even with equivalent treatment. In the Wojcik et al. study, survival was significantly better for patients who received BCS with radiation therapy than for those who received BCS alone. Similarly, in the current study, the use of radiation therapy after BCS reduced the magnitude of racial differences in survival by nearly one-half. This indicates that increasing the use of follow-up radiation therapy in African-American women would decrease the survival disparity, but would not eliminate it. Other factors such as tumor biology, systemic therapy (chemotherapy and hormonal treatment), and differences in breast carcinoma risk profile by race are important contributors to the disparity. In addition, previous research indicates that patient factors related to socioeconomic status (SES), such as child/family care concerns, literacy/education levels, and lack of transportation, may contribute to noncompliance among patients in the treatment of breast carcinoma and other diseases.31–35 Noncompliance may contribute to the significantly poorer survival among African-American women who receive equivalent treatment.
The significantly lower use of radiation for African-American women may be due to physician noncompliance with treatment prescription standards. Previous studies of use and prescription of appropriate therapy by clinicians revealed racial differences in the treatment of breast carcinoma36 and other chronic diseases37–44 for African-American patients.
Information on patient comorbidity was not available in this database. Therefore, one potential limitation of this study is that differences in the prevalence of comorbid conditions between African-American and white women may have had an effect on prescribing patterns. However, the use of survival statistics reduces the effects of this confounder by considering comorbid conditions that lead to death as censored data in the log rank and Cox analyses, thereby statistically controlling for these most serious comorbid conditions that may have had the greatest impact on prescribing patterns.
Results of this study support and extend several major theoretical concepts. First, results of randomized clinical trials (RCTs) in the 1980s concluded that BCS and mastectomy have equivalent overall and disease-free survival, leading to clinical guidelines on the use of BCS with follow-up radiation therapy in the early 1990s. Generalizability of RCTs to real life is often not realistic or representative, given stringent controls compulsory to randomized trials that are quite unlike the multiple known and unknown confounding factors present within diverse communities in real life. However, results of this large population-based analysis, although not controlling for all confounding factors, are at least an accurate description of the equivalence of survival following these two procedures in the real world.
Second, this study emphasizes the importance of follow-up radiation therapy in patients who receive BCS and the significant positive impact that this treatment has on survival. Although the use of radiation therapy improved survival among African-American women, significant racial differences remain, leading to the third major concept — race has a major significant independent impact on breast carcinoma survival. It is imperative to determine the unknown factors associated with race (e.g., SES) that have a negative effect on patient and tumor characteristics (e.g., age at diagnosis, stage at diagnosis, and tumor hormone receptor status) that affect survival.
In determining the factors associated with race that have a negative effect on survival, it is equally important to recognize the risk by those interpreting results of these studies of creating an arbitrary and capricious designation of a “black breast cancer.”45 This designation is based on significant differences in distribution of variables such as stage, age, and hormone receptors by race. The obvious dangers associated with this designation is the tendency to refrain from further research and claim that poorer survival in African-American women diagnosed with breast carcinoma is a “genetic inherited” defect and therefore not amenable to innovations in medical and public health interventions.
However, most researchers will agree that race is a surrogate measure of factors such as SES, access to health care, and cultural systems. Smith et al.46 found that breast carcinoma survival among African-American women of lower SES is more similar to that among white women of lower SES than to women of both races in higher SES classifications. Other studies of racial and socioeconomic differences in all-cause mortality found similar results.47–49 Two hypotheses arise from these studies: 1) SESis the true cause of factor in racial differences in survival or 2) the health experiences of African-American women are different from that of white women. This second hypothesis has been proposed in studies of the “weathering conceptual framework” that purports premature health deterioration among African-American women due to a multitude of societal circumstances.50
Both of these hypotheses have significant social and public health implications. In either case, social/societal factors are implicated in breast carcinoma survival differences rather than any arbitrary construct such as race classification of black and white. Macroepidemiologic studies that examine societal structure and influence are needed to examine patterns and outcomes associated with racial differences in individual biomedical and genetic risk factors and breast carcinoma etiology.
In conclusion, racial differences exist in the treatment of early-stage breast carcinoma. Public health efforts to eliminate these treatment differences would reduce, but not eliminate, racial disparity in survival. Further research is warranted to determine ways to eliminate racial differences in breast carcinoma survival.
- 1Cancer statistics, 2002. CA. 2002; 52: 23–47., , , .
- 9Conservation therapy of breast cancer [letter]. Lancet. 1989; 8655: 168–169., , .
- 22Regression models and life tables. J Stat Soc B. 1972; 34: 187–220..
- 27Prognosis among African American and white women with lymph node negative breast carcinoma: findings from two randomized clinical trials of the National Surgical Adjuvant Breast and Bowel Project (NSABP). Cancer. 1997; 80: 80–90., , , , .
- 36Impact of race on cardiac care and outcomes in veterans with acute myocardial infarction. Med Care. 2002; 40 (suppl 1): 86–96., , , .
- 37Lower extremity nontraumatic amputation among veterans with peripheral arterial disease: is race an independent factor? Med Care. 2002; 40 (suppl 1): 106–116., , , , .
- 38Race, treatment, and survival of veterans with cancer of the distal esophagus and gastric cardia. Med Care. 2002; 40 (suppl 1): 14–26., , , .
- 40Research on the provider contribution to race/ethnicity disparities in medical care. Med Care. 2002; 40 (suppl 1): 140–151..
- 45Why are so many black women dying of breast cancer? MAMM. 2001; 4: 34–41..