Race, socioeconomic status, and breast carcinoma in the U.S.

What have we learned from clinical studies?

Authors

  • Chaundré K. Cross M.D.,

    Corresponding author
    1. Joint Center for Radiation Therapy, Beth Israel Deaconess Medical Center, Brigham and Women's Hospital, Dana-Farber Cancer Institute, and Harvard Medical School, Boston, Massachusetts
    • Joint Center for Radiation Therapy, Beth Israel Deaconess Medical Center, Brigham and Women's Hospital, Dana-Farber Cancer Institute, and Harvard Medical School 75 Francis Street, ASB-1, L2 level, Boston, MA 02115
    Search for more papers by this author
    • Fax: (617) 732-7347

  • Jay Harris M.D.,

    1. Department of Radiation Oncology, Dana-Farber Cancer Institute, Brigham and Women's Hospital, and Harvard Medical School, Boston, Massachusetts
    Search for more papers by this author
  • Abram Recht M.D.

    1. Department of Radiation Oncology, Beth Israel Deaconess Medical Center and Harvard Medical School, Boston, Massachusetts
    Search for more papers by this author

Abstract

BACKGROUND

Whether African-American women have biologically more aggressive breast carcinoma compared with white women and whether race acts as a significant independent prognostic factor for survival have not been determined. Alternatively, race merely may be a surrogate for socioeconomic status (SES).

METHODS

A literature review was performed of clinical trials and retrospective studies in the U.S. that compared survival between white women and black women with breast carcinoma after adjustment for known prognostic factors (patient age, disease stage, lymph node status, and estrogen receptor status) to assess the impact of race and SES.

RESULTS

Single institutional and clinical studies suggest that, when black patients are treated appropriately and other prognostic variables are controlled, their survival is similar to the survival of white patients. Twelve retrospective studies and 1 analysis of a clinical trial included SES and race as variables for survival. Only three of those studies revealed race as a significant prognostic factor for survival after adjusting for SES.

CONCLUSIONS

SES replaces race as a predictor of worse outcome after women are diagnosed with breast carcinoma in many studies. However, black women present with more advanced disease that appear more aggressive biologically, and they present at a younger age compared with white women. Further research should be conducted concerning the precise elements of SES that account for the incidence of breast carcinoma, age at diagnosis, hormone receptor status, and survival to devise better strategies to improve outcome. Cancer 2002;95:1988–99. © 2002 American Cancer Society.

DOI 10.1002/cncr.10830

Breast carcinoma is a major issue for all women, regardless of race, socioeconomic status (SES), or country of origin. However, breast carcinoma incidence and outcomes are not the same in all patient populations.1 The incidence of breast carcinoma is lower in African-American (black) women compared with white women. Conversely, black women have a higher mortality rate from breast carcinoma compared with white women (31.4 per 100,000 population vs. 25.3 per 1000,000 population, respectively) and are less likely to survive 5 years after diagnosis compared with white women (71.0% vs. 86.0%, respectively).1 Even when known prognostic factors, such as tumor size and lymph node status, are accounted for, differences persist in terms of mortality. Possible reasons for this discrepancy include younger age at diagnosis, later stage at detection, different incidences of negative tumor receptors for progesterone and estrogen, and tumors that are more aggressive and less responsive to treatment in black women.1, 2 The impact of SES also may help us understand these findings.

This review focused on whether black women have biologically more aggressive breast disease compared with white women and whether race acts as a significant independent prognostic factor for survival. Alternatively, we wanted to determine whether race is merely a surrogate for SES. We reviewed outcome studies comparing survival between white women and black women with breast carcinoma that adjusted for known prognostic factors (age, stage, lymph node status, and estrogen receptor [ER] status) to assess the impact of race and SES. We also briefly reviewed the issue of differences in diagnosis and treatment in relation to race and SES.

Epidemiology of Breast Cancer Risk Factors in Black Women

An estimated 7% of all breast malignancies are associated with inherited mutations in BRCA1 and BRCA2.3 There are currently no known recurring breast carcinoma-related gene mutations that have an increased frequency in black women.3–10 However, almost nothing is known regarding the interaction of various genes and environmental factors, both exogenous and endogenous, and the association with SES that may be different between black women and white women.8

One important but complex factor is the role of endogenous hormones in the etiology of breast carcinoma.11, 12 It has been theorized that obesity leads to increased endogenous estrogen levels, due to conversion of androstenedione in adipose tissue, and, thus, to a greater incidence of breast carcinoma.13 Several studies have explored the association of obesity and breast carcinoma in black women.14–17 Obesity has been associated with an increased risk of postmenopausal breast carcinoma for both black women and white women. How obesity in black women relates to tumor biology and prognosis after they are diagnosed with breast carcinoma is not understood fully. It is unlikely that obesity may explain the increased incidence of breast carcinoma in premenopausal black women.

Women's reproductive experiences may influence the age at which they are diagnosed with breast carcinoma. Pathak et al. discussed the effect of early age of first full-term pregnancy.18 It appears that, at a younger age, parous women are at a greater risk of developing breast carcinoma compared with nulliparous women; this pattern reverses at older ages. However, higher parity and early age at first pregnancy confer a lifetime decrease in the risk of breast carcinoma compared with nulliparity. This most likely is secondary to endogenous estrogen during pregnancy.

No studies have shown an increased risk of breast carcinoma in black women who use oral contraceptives compared with white women.18, 19 The full extent of interaction between reproductive experiences and hormonal risk factors, including age at first birth, obesity, estrogen levels, and age at menarche and menopause, deserves further investigation to understand the incidence and aggressiveness of breast carcinoma in black women. The relation between SES and reproductive experiences and their influence on age at the time of diagnosis with breast carcinoma also should be explored further.

A number of studies have shown that black women have more poorly differentiated breast tumors and more ER negative breast tumors compared with white women.10, 20–30 The Black/White Cancer Survival Study (BWCSS)10 comparison of 506 black patients with 457 white patients who had the same disease stage found that the former group was significantly more likely to have higher nuclear grade, higher mitotic activity, necrosis, and ER negative tumors. In the study from the University of Texas,20 1016 black patients were compared with 4885 white patients. Again, black patients were significantly more likely to have lymph node involvement, more involved lymph nodes, higher S-phase fraction, and ER negative and progesterone receptor (PR) negative tumor cells. There were no significant differences in DNA ploidy, histologic type, HER-2/neu, or p53 expression. Another study from Seattle found a greater incidence of ER-negative tumors in both young black women and young white women compared with older women; however, there was an excess of ER negative tumors in young black women.24 Investigators at the National Cancer Institute reviewed the distribution of ER and PR status in 112,588 women with breast carcinoma between 1992 and 1997 for 8 ethnic/racial groups. ER negative and PR negative breast tumors were more common in black women compared with white women (23% vs. 14%, respectively). Among black women, the average age of patients with ER negative breast carcinoma was 47 years, compared with an average age of 50 years for white patients. Also, ER negative tumors in black women were more likely to be Stage II rather than Stage I and Grade 3 rather than Grade 2 compared with to the tumors in white women.23

SES factors, such as low income and low educational level, have been associated with the diagnosis of ER negative breast tumors.26, 27 The association of race, age, SES, parity, and ER negative breast carcinoma deserves further investigation. In theory, one would expect an increase in the proportion of ER positive breast tumors after the prolonged estrogen exposure of single or multiple pregnancies. At this time, it is unclear why black women present with more ER negative breast tumors at a younger age.

Association of Race with Age, Stage, and Treatment

There is a difference in presentation with breast carcinoma between black women and white women. Although the yearly incidence rate of female breast carcinoma is lower among black women (100.2 per 100,000 population) compared with white women (114.0 per 100,000 population),1 breast carcinoma typically develops at an earlier age among black women. The ratio of incidence rates in black women compared with white women is 3.1 at ages 20–24 years and 1.4 at ages 30–34 years.1 A study by Joslyn and West found that more than one-third of black patients with breast carcinoma were age < 50 years at the time of diagnosis, compared with one-fourth of white patients.2

During 1990–1997, mortality rates decreased less for the black population (− 0.1% per year) compared with the white population (− 0.7% per year).1 Among those age ≥ 50 years, black women had a higher mortality rate during that period compared with white women. The higher mortality rate among black women compared with white women has been attributed in part to the advanced stage of disease at presentation and treatment received. Surveillance, Epidemiology, and End Results (SEER) data from 1988 to 1995 reveal that 55% of black women with breast carcinoma are diagnosed above Stage I compared with 45% of white women.2 This discrepancy in presentation may be explained by SES factors or more aggressive tumor biology.

Previous studies have shown that black women receive less aggressive adjuvant treatment compared with white women.31–33 This may explain part of the discrepancy in outcome between races in population-based studies, because single institutional studies and clinical trials in which all patients receive uniform treatment have shown equal outcomes for all races.33–38 Population-based studies are limited secondary to incomplete data on actual treatments given to patients. However, Breen et al. revealed that black women received less adequate care compared with white women in the general population.39 One reason for less aggressive treatment for African Americans or people of lower SES may be comorbid illnesses.33 The effect of stage specific treatment differences on outcome among African-Americans and whites in the general population remains to be determined.40 The impact of SES on treatment differences will be discussed below.

Association of SES with Age, Stage, and Treatment

A higher percentage of the black population, compared with the white population, comes from the lower end of the SES strata.29, 41 Lower SES has been correlated with poorer dietary habits, smoking, and less physical activity, which lead to increased rates of comorbid illnesses and, thus, to higher mortality rates in women with breast carcinoma.42 Lower SES also is associated with a lower incidence of breast carcinoma. This association can be explained in part by known reproductive risk factors. Women in the lower socioeconomic strata are more likely to have more children, as discussed above, and to have them at a younger age compared with women in higher socioeconomic strata.40 Some authors have hypothesized that a short-term increase in breast carcinoma risk immediately after pregnancy is consistent with the higher incidences of breast carcinoma among young black women compared with young white women. The overall lower lifetime incidence of breast carcinoma among black women is consistent with a long-term benefit of early and repeated pregnancy.18, 40

Axtell and Myers43 were the first to hypothesize that the observed trends in the SEER data base for African-American females with breast carcinoma to present at a younger age, and to present with later stage disease, and to have a lower survival rate compared with white women may be due to socioeconomic factors.44–50 Several studies have shown that SES predicts for breast carcinoma incidence and outcome more strongly than race.33, 42, 47–51

Breast carcinoma outcome is affected by stage of disease and actual treatment given. The disproportionate number of black women who present with more advanced breast disease compared with white women has been associated with SES in several studies.50, 52–57 Specifically, socioeconomic factors, such as lower educational level and income,52 cultural beliefs,54–57 lower insurance status,50, 51 diet,58 and screening practices,59, 60 appear to affect the stage of disease at the time of presentation and, thus, presumably affect outcome.

Whether treatment differences secondary to recommendations,61 actual treatment,33, 39 or compliance62 affect mortality from breast carcinoma remains to be demonstrated. One study from Michigan that linked SEER registry data to Medicaid enrollment files demonstrated that African-American women and others of lower SES were less likely to undergo surgery or radiation therapy after breast-conserving surgery.33 Given this finding, it is useful to examine outcomes in clinical trials that control for treatment and biology.

Race and Outcome in Studies with Adjustments for Prognostic Factors

Numerous studies have been published regarding breast carcinoma survival in African-American women. However, many factors other than race need to be considered to gauge the true meaning of differences in outcome between black patients and patients in other ethnic groups. These include stage, age, lymph node status, ER status, SES, and the actual treatments given. A literature search was conducted on Medline for articles comparing survival outcomes between white patients and black patients with early-stage breast carcinoma that adjusted for these known prognostic factors to assess the impact of race and SES.

Race and Outcome in Retrospective Studies with Adjustments for Stage and Age

Four studies have included race, stage, and age as variables (Table 1).2, 63–65 Two of those studies found that race was not a significant factor in outcome. The first, from Howard et al.,63 reviewed patients in three health maintenance organizations (HMOs) in which it was presumed that access and use of health care and the SES of the patients were similar. There was not a significant difference in stage of disease or ER status between the white patients and the black patients. A study by Burri et al.65 retrospectively analyzed patients with early-stage breast carcinoma who were seen at the academic hospital affiliated with Emory University and an inner city hospital. In their study, there also were no differences in ER status, the use of adjuvant therapy, or surgical margin status between the races.

Table 1. Race and Outcome in Studies with Adjustments for Stage and Age
Study (yr)Population (study yrs)Racial compositionM/URace as a prognostic factor for survivalP value
  • M/U: multivariate or univariate analysis; HMO: health maintenance organization; B: black: W: white: BCT: breast-conserving therapy;

  • a

    Adjustment for health maintenance organization, stage of disease, and missed appointments.

  • b

    Relative risk.

  • c

    No significant differences in local control rates (95.5% vs. 94.8%) and disease free survival rates (90.3% vs. 91.7%) was seen for black patients versus white patients, respectively.

  • d

    Adjustment for age only.

Howard et al.63 (1998)Three HMOs; mastectomy for Stage II or higher (1986–1990)89 B, 157 WMNot significant0.067a
Wojick et al.64 (1998)U.S. military health care facilities (1975–1994)698 B, 5879 WMSignificant1.41b
Burri et al.65 (2000)Emory University and inner city hospitals; patients received BCT (1993–1996)102 B, 162 WNot significantc
Joslyn and West2 (2000)National SEER study (1988–1995)6537 B, 75,078 WMSignificantd0.02

In contrast, Wojick et al.64 analyzed the Department of Defense Tumor Registry. Those authors found that, after adjusting for age and stage, race was still a significant factor for survival. However, there was a significant difference in lymph node status between the races, and this variable was not included in their analysis. A recent study by Joslyn and West2 using SEER data also revealed that race was a significant factor in outcome. In that study, African-Americans more often presented with tumors that were ER negative and had advanced stage disease. Like other large, population-based studies, differences in treatment given, length of follow-up by racial group, lymph node status, and SES were not known. However, those two studies were larger than the other studies and, thus, had greater power to detect differences.

Race and Outcome in Retrospective Studies with Adjustments for Stage, Age, and Lymph Node Status

Table 2 illustrates the importance of including lymph node status as a predictor for survival. Two large studies from the American College of Surgeons22, 66 found that race, stage, age, and ER status all were significant predictors of survival. Differences in treatment between the races were not investigated, however. In the later study,22 lymph node status was not a significant factor in the regression analysis. However, if stage was removed from the analysis, then lymph node status became a significant predictor for survival. In the earlier study,66 black patients with lymph node positive tumors measuring < 2 cm in greatest dimension and with lymph node negative tumors measuring < 3 cm in greatest dimension did as well as white patients.

Table 2. Race and Outcome in Studies with Adjustments for Stage, Age, and Lymph Node Status
Study (year)Population (study yrs)Racial compositionM/URace as a prognostic factor for survivalP value
  • M/U: multivariate or univariate analysis; B: black; W: white; BCT: breast-conserving therapy.

  • a

    Cure rates were similar in patients with negative lymph nodes and tumors measuring < 3 cm in greatest dimension or with positive lymph nodes and tumors measuring < 2 cm in greatest dimension.

  • b

    The 5-year and 10-year survival rates for black patients were 84% and 76.6%, respectively, for those with lymph node negative disease. Black patients with lymph node positive disease had survival rates of 58% and 35.2% at 5 years and 10 years, respectively. These rates were not significantly different from the survival rates reported in the literature for nonblack patients.

  • c

    A significant difference was found for a subgroup of patients with one to three positive lymph nodes (P = 0.008).

  • d

    After adjusting for race with age, tumor size, lymph node status, hormone receptor status, and histology on multivariate analysis, the effect of race as a prognostic factor for survival remained significant. The risk ratios for local, regional, and distant disease for black patients compared with white patients were 1.9, 1.52, and 2.21, respectively.

Nemoto et al.66 (1980)Survey of 498 hospitals, (1978)1023 B, 11,958 WNot significanta
Natarajan et al.22 (1985)Survey of 565 hospitals (1981)1223 B, 14,706 WMSignificant0.012
Valanis et al.69 (1987)University Cincinnati (1969–1979)67 B, 144 WMNot significant0.42
Briele et al.70 (1990)Cook County Hospital: black patients compared with white patients in the literature, (1973–1987)526 BNot significantb
Pierce et al.67 (1991)University of Pennsylvania/Fox Chase Cancer Center: BCT for patients with Stage I–II disease (1977–1985)75 B, 615 WMSignificant0.04
Elias et al.37 (1994)University of Maryland: Mastectomy for patients with Stage I–II disease (1978–1986)60 B, 112 WMNot significantc0.83
Heimann et al.38 (1997)University of Chicago: Mastectomy for patients with Stage I–III disease (1946–1987)481 B, 1037 WMNot significant0.90
Newman et al.68 (2001)Detroit SEER data base: Patients age < 40 years (1990–1999)492 B, 1322 WMSignificantd

Pierce et al.67 performed the first study reviewing the patterns of failure difference between blacks and whites undergoing breast-conserving therapy (BCT). Those authors found that race was a significant factor for survival, but black patients were significantly more likely to present with more advanced disease and were given chemotherapy less often compared with white patients. It is interesting to note that the patterns of regional lymph node failure were different between the races; black patients were more likely to fail in the supraclavicular fossa, and white patients were more likely to fail in the axilla.

A recent abstract using the SEER database68 regarding outcome for patients with breast carcinoma age < 40 years from the metropolitan Detroit region who were treated between 1990 and 1999 found that race was a significant predictor for worse survival, adjusting for age, tumor size, lymph node status, and hormone receptor status. African-American women had significantly larger tumors, more ER negative tumors, and lower rates of localized disease. That study did not provide information on possible variability in treatment and did not report disease specific mortality rates.

Race was not a significant factor for survival in several studies from single institutions in which all patients had similar tumor characteristics, uniform treatment, and were presumed by the respective authors to have comparable SES.37, 38, 69, 70 However, in the study by Elias et al.,37 black patients with pathologic involvement of one to three lymph nodes fared worse than comparable white patients. In a study from Cook County Hospital in Chicago by Briele et al.,70 the survival of 526 black patients with breast carcinoma was compared with survival rates reported in the literature for nonblack patients. The proportion of these patients with positive ER status was similar to the proportion reported for nonblack patients. Outcomes also were similar between the two groups. Heimann et al.38 performed a detailed analysis of the experience of the University of Chicago between 1946–1987 for 481 black patients and 1037 white patients with breast carcinoma. Patients were followed for a median of 8.8 years. The number of axillary lymph nodes involved did not vary significantly between ethnic groups when analyzed in relation to tumor size. On multivariate analysis, race was not a statistically significant prognostic factor for survival.

Race and Outcome in Retrospective Studies with Adjustments for Stage, Age, Lymph Node Status, and SES

Race and SES are correlated strongly in many studies, as shown in Table 3. Hence, differences in outcome that appear to be related to race may be due to SES. The roles of both race and SES in the survival of patients with breast carcinoma have been examined by a number of investigators.26, 29, 44–50, 71–74 In 1982, Dayal et al.44 reported outcomes on 206 black patients and 117 white patients with breast carcinoma. After adjusting for age and stage, there was a persistent difference between the two races (P = 0.053). After creating a socioeconomic index that included median school years, median income, percentage of high school graduates, median rent, median house value, and percentage living below poverty level, racial differences for survival became insignificant (P = 0.780). More recent studies have confirmed the findings of Dayal et al.33, 47, 48, 50

Table 3. Race and Outcome in Studies with Adjustments for Age, Stage, Lymph Node Status, and Socioeconomic Status
Study (yr)Population (study yrs)Racial compositionM/URace as a prognostic factor for survivalP value
  • M/U: multivariate or univariate analysis; B: black; W: white; HMO: health maintenance organization.

  • a

    Relative risk.

  • b

    Equivalent survivals, except in black patients and white patients with Stage III disease (5-year survival rates, 33% vs. 45%, respectively).

  • c

    Relative risk for patients age < 50 years.

  • d

    Relative risk for patients age ≥ 50 years.

Dayal et al.44 (1982)Medical College of Virginia (1968–1977)388 B, 515 WMNot significant0.780
Vernon et al.45 (1985)M. D. Anderson Cancer Center (1949–1968)912 B, 4618 WMSignificant< 0.01
Bain et al.46 (1986)SEER data, metropolitan Atlanta, Emory University (1978–1982)536 B, 2322 WMSignificant0.01
Basset and Krieger71 (1986)Western Washington Cancer Surveillance (1973–1983)251 B, 1255 WMNot significant1.10a
Polednak73 (1988)New York at Stony Brook (1976–1981)890 B, 24372 WNot significantb
Eley at al.29 (1994)Black/White Cancer Survival Study (1985–1986)612 B, 518 WMNot significant1.30a
Perkins et al.72 (1996)M. D. Anderson Cancer Center (1958–1987)801 B, 2581 WMNot significant0.048
Simon and Severson74 (1996)Michigan SEER Registry (1988–1992)1880 B, 8662 WMSignificant1.68,c 1.33d
Franzini et al.50 (1997)M. D. Anderson Cancer Center (1987–1991)163 B, 964 WMNot significant0.15
Yood et al.48 (1999)HMO in Michigan (1986–1996)255 B, 548 WMNot significant1.00a
El-Tamer et al.47 (1999)State University of NY-Brooklyn/Kings County Hospital (1982–1995)1297 B, 448 WMNot significant0.80
Bradley et al.33 (2002)Michigan SEER Registry and Medicaid files (1996–1997)1110 B, 4609 WMNot significant0.149

Several studies have been performed at the M. D. Anderson Cancer Center on the impact of race and SES on the survival of patients with breast carcinoma.45, 50, 72 In 1985, Vernon et al.45 reviewed patients who were treated from 1949 to 1968, placing patients into three SES categories based on their insurance status and ability to pay hospital bills. Multivariate analysis revealed that ethnicity was an independent predictor for survival after adjusting for stage, age, SES, and delay of starting treatment from the first sign or symptom of disease. However, only a crude staging system was used, tumor characteristics were reported incompletely, and data on treatment and follow-up were not described. Later studies from the same institution by Perkins et al.72 and Franzini et al.50 did not find that ethnicity was a significant prognostic factor for survival. The more recent studies revealed no treatment or age differences between the different ethnic populations.

In the BWCSS,29 in which SES was determined for each patient individually, black patients were more likely than white patients to be single and to have lower SES. In multivariate analysis, disease stage was the strongest factor related to survival. After controlling for both stage and SES, there was no difference in survival rates between black patients and white patients. In contrast, in a recent study of patients at an HMO48 in which SES was determined by the patient's address in relation to census data, SES (but not stage) was a strong contributor to the observed survival difference between black patients and white patients. The different results of these two studies may have been due to disease stage. In the BWCSS, there was nearly a 20% difference in the proportion of early-stage disease between white patients and black patients, whereas, in the HMO study, there was only an 11% difference. Therefore, the effect of disease stage on survival rates may have obscured the contribution of SES in the BWCSS. Also, black patients in the BWCSS were matched by age with white patients, which makes it impossible to establish the contribution of this important variable to the chance of survival.2

A study by Simon and Severson combined Michigan SEER registry data from the period 1988–1992 and census tract data to stratify patients into low, medium, and high SES levels based on the proportion of high school graduates and median household income.74 There were no significant differences in survival by race after adjustment for SES and stage (P = 0.67) for patients who survived longer than 48 months. There was a significant correlation between age at diagnosis and race (P = 0.005). After adjustments, the relative risks for death in black women compared with white women were 1.68 among women age < 50 years and 1.33 for women age ≥ 50 years. The lack of information on treatment and tumor characteristics and the use of surrogate data, rather than individualized data, on SES weaken these conclusions.

Contrary findings were reported by the authors of another study that linked data from the 1996–1997 Michigan SEER registry to Medicaid files to measure SES and comorbid illnesses.33 Black women did not have a statistically significant increase in the odds of having late-stage disease at diagnosis or death compared with white women after adjusting for SES, age, race, and stage (P = 0.149). In addition, regardless of race, lower SES contributed to a greater incidence of comorbid illnesses, late-stage disease at diagnosis, and worse survival.

Work by Bain et al.46 found that race was a significant predictor of survival despite adjustment for SES, stage, age, tumor size, and lymph node status. However, this was a population-based study in which patients received heterogeneous treatments, median follow-up was short (24 months), and only a crude measure of SES was used (county of residence). Also, there was no available information on the number of histologically involved lymph nodes for 13% of white women and for 26% of black women.

Race and Outcome in Prospective Clinical Trials

Clinical trials are useful tools for comparing differences in outcome between black patients and white patients (Table 4). The patients in such trials are homogeneous with respect to disease stage, receive uniform treatment, and are stratified with regard to clinical factors, such as ER status, lymph node status, and tumor size. However, there are some disadvantages to using clinical trials to probe differences in outcome related to race. These include a lack of detailed social and economic data and the small number of nonwhite patients who usually participate in these studies.75

Table 4. Race and Outcomes in Clinical Trials
Study (yr)Population (study yrs)Racial compositionRace as a prognostic factor for survivalFive-year survival rate (%)
  • B: black; W: white; NSABP: National Surgical Adjuvant Breast Project; PF:PFT: the ratio of patients who received chemotherapy with melphalan and 5-fluorouracil (PF) and patients who received chemotherapy with PF plus tamoxifen (PFT); CALBG: Cancer and Leukemia Group B.

  • a

    Black patients with positive lymph nodes fared significantly worse compared with white patients (survival rates were not given).

  • b

    Includes only patients with negative lymph node status

  • c

    Not significant after adjustments (survival rates were not given).

  • d

    P value for race after adjusting for socioeconomic status, lymph node status, tumor size, hormone status, and age on multivariate analysis.

  • e

    Relative risk.

Dignam75 (2000)NSABP Protocol B-04 (1971–1975)191 B, 442 WNot significanta85,b 85
Fisher et al.76 (2001)NSABP Protocol B-06 (1976–1984)86 B, 899 WSignificant74,b 89
Constantino et al.35 (1987)NSABP Protocol B-09 (1977–1980)146 B, 896 WNot significantcPF:PFT: 52:57 B, 68:65 W
Gordon et al.26 (1992)Case Western Reserve (1974–1985)253 B, 1132 WNot significant0.43d
Dignam75 (1997)NSABP Protocol B-13 (1981–1988)108 B, 916 WNot significant83, 85
Dignam75 (1997)NSABP Protocol B-14 (1982–1988)203 B, 3709 WNot significant93, 92
Roach et al.36 (1997)CALGB 8541 (1985–1988)185 B, 1325 WNot significant1.14e

The National Surgical Adjuvant Breast Project (NSABP) has performed a number of retrospective reviews of outcome differences between black patients and white patients using data from their prospective trials.25, 26, 34–36, 75, 76 Protocol B-04 compared radical mastectomy, total mastectomy, and total mastectomy plus chest wall and lymph node irradiation for patients with clinically lymph node negative disease. A review of outcomes of 191 black women and 442 white women found that, even though black women were more likely to be younger, have larger tumors, and have positive lymph nodes, after adjustment for prognostic factors, the survival rate for black patients was within 10% of the survival rate for white patients. The survival among women with pathologically lymph node negative disease who underwent radical mastectomy was similar for the two groups, with about 85% of women surviving for 5 years.75 Protocol B-06 compared total mastectomy with lumpectomy and breast irradiation; all patients underwent axillary dissection. Results for 86 black patients and 899 white patients showed a worse prognosis for black patients with lymph node negative disease (74%) compared with white patients with lymph node negative disease (89%) at 5 years. Black patients on this protocol were more likely to have ER negative tumors and tumors of a poorer nuclear grade.76

These contradictory findings led these NSABP investigators to perform a more detailed analysis of patients with lymph node negative disease in two other trials, Protocols B-13 and B-14, first published in 1989. Protocol B-13, for patients with ER negative tumors, compared surgical treatment alone with surgery plus 12 courses of sequential methotrexate and 5-fluorouracil. Protocol B-14 compared surgery plus placebo with surgery plus tamoxifen among patients with ER positive tumors. Dignam75 retrospectively reviewed differences in prognosis between black patients and white patients, controlling for hormone receptor status, treatment, and other pathologic and clinical factors. That analysis included 916 white patients and 108 black patients from Protocol B-13 and 3079 white patients and 203 black patients from Protocol B-14. In both trials, black women were more likely than white women to have been treated with total mastectomy than with lumpectomy and radiation. For women with ER negative tumors, the 5-year disease free survival (DFS) rate and the overall survival rate were 71% and 83%, respectively, for black patients and 74% and 85%, respectively, for white patients. For women with ER positive tumors, the 5-year DFS rate and the overall survival rate were 81% and 93%, respectively, for black patients and 80% and 92%, respectively, for white patients.75 These differences were not statistically significant.

Protocol B-09 compared outcome for patients with lymph node positive tumors who were treated either with melphalan and 5-fluorouracil (PF) or with the same chemotherapy plus tamoxifen (PFT). That trial included 896 white women and 146 black women with Stage II breast carcinoma accrued from 1977 to 1980. Tumors were larger and more commonly ER negative in the black women. In the PF arm, Constantino et al.35 reported that the unadjusted 5-year survival rates were 52% for black women and 68% for white women; whereas, in the PFT arm, 57% of black women and 65% of white women were alive at 5 years. However, the 5-year survival rates adjusted for prognostic factors were not different between the two populations.

In another retrospective review of a clinical trial for patients with Stage II disease (Cancer and Leukemia Group B study 8541), Roach et al.36 performed multivariate analyses for overall survival and DFS. When adjusted for dose intensity of chemotherapy, number of lymph nodes involved, ER status, and age, race was not prognostically significant for outcome.

Crowe et al.28 evaluated age and race as predictors of survival in two multi-institutional studies performed from 1974 to 1985. A total of 253 black women and 1132 white women were entered into those prospective clinical trials during 1974–1979 and 1980–1985. Multivariate analysis evaluating age, ER status, number of involved lymph nodes, race, and greatest tumor dimension found that age was the only variable that was not statistically significant. A more detailed analysis of this data base26 reported on recurrent breast carcinoma and overall survival in relation to ER status, number of lymph nodes involved, age, race, SES, and tumor size. A Cox proportional hazards model indicated that a greater number of positive lymph nodes, a large tumor size, lower SES, and negative ER status were significant prognostic factors for DFS. After adjustment for SES, race ceased to be a significant predictor for either DFS or overall survival. Patients of either race who were of a lower SES were more likely to have a recurrence and to die of breast carcinoma.

DISCUSSION

Several studies have shown that black patients with breast carcinoma have worse pathologic and clinical factors compared with white patients. Breast carcinoma in black women is more often of advanced stage and is ER negative, and black women present at a younger age. The explanation for these observations is uncertain. Two recent studies, as noted above, did not show an increased incidence of BRCA1 mutations in African-Americans with breast carcinoma.3, 4 Hormonal factors may play a role in the incidence of breast carcinoma in black women (e.g., age at first birth and age at menarche and menopause), although it is difficult to see how this would explain an increase in the risk of ER negative breast carcinoma.

A critical issue is whether these observations regarding race can be explained by SES. Future research should focus on how SES contributes to the stage, age at diagnosis, and biology of breast carcinoma. Access to and use of health care seems to be related to level of education, insurance status, and income. Higher parity rates seen in groups with lower SES may be associated with educational level and income. It remains unclear how increased exposure to endogenous estrogen at a younger age may cause an increased incidence of ER negative breast tumors. Also, future studies should incorporate assessment of variation in the incidence of comorbid illnesses between races as a possible variable causing disparities in outcome.

Roach and Alexander created a reliability scoring system for analyses of race as a prognostic factor.77 This system evaluates studies by whether or not they adjust for tumor size, lymph node status, SES, and uniformity of treatment. More weight is given to single institutional studies than multi-institutional studies or studies that use population-based registries. The population-based studies that were included in this review all revealed race as a significant predictor for survival.2, 22, 66, 68, 74 However, reliability scores for these studies were low due to heterogeneous treatment, lack of adjustment for SES, and inadequate staging information. In contrast, studies of race and outcome conducted at single institutions or using clinical trial data bases had high reliability scores. Single institutional studies and clinical trials usually have more uniform treatment and accessible information on compliance and follow-up. However, these studies may contain a biased sample of both black patients and white patients, which would decrease the generalizability of their results. In addition, the smaller numbers of African-American women in these studies lead to lower statistical power to detect meaningful differences.2 Nonetheless, such studies suggest that, when black patients are treated appropriately and other prognostic variables are controlled for, their outcomes are similar to the outcomes of white patients. Therefore, even if African-American women present with more aggressive tumors, current therapies are adequate to control their disease.

Twelve studies and one analysis of a clinical trial included SES and race as variables for survival.26, 29, 33, 44–48, 50, 71–74 Only three studies revealed race as a significant factor for survival after adjusting for SES.45, 46, 74 Dayal et al.44 concluded that, once the disease has occurred, survival depends on elements that determine the likelihood of early diagnosis and affect the chances of living with the disease. These factors include limited education (which may prevent early detection), compliance with treatment recommendations, and low income (which may inhibit access to screening or health professionals). For example, Ayanian et al.51 reported on clinical outcomes for uninsured patients (including patients on Medicaid) and insured patients with breast carcinoma in the New Jersey tumor registry. Uninsured patients presented with more advanced disease and had a lower survival rate. The authors provide two possible explanations: lead-time bias, meaning survival for insured patients seems longer because their tumors are diagnosed sooner, even if earlier treatment is not prolonging their lives; and length-time bias, which results from higher screening rates in insured patients, leading to the detection of tumors that spread more slowly and are less aggressive. Both lead-time bias and length-time bias may explain why SES does not correct totally for outcome differences between ethnic groups. In addition, the increased likelihood of having comorbid illnesses in the lower SES strata may account for patients receiving less aggressive adjuvant therapy and, thus, having a worse outcome. However, taken together, these clinical trials suggest that factors related to SES, such as education, income, and access to health care, rather than race, may account primarily for differences in outcome between black women and white women with breast carcinoma.

One limitation of the studies that were included in this review was the variation in the methods used to measure SES. Measuring income status and educational level by census tracts and blocks and insurance-based data is not as accurate as self-reported data. One aspect of SES that is difficult to measure is how it affects patient behaviors that may influence outcomes (e.g., use of screening mammography). Future studies should incorporate self-reported educational level, income status, insurance data, and individual histories of health care utilization to fully understand how SES influences health outcomes.

In 1997, Harold Freeman, M.D., Chairman of the President's Cancer Panel, convened a meeting on The Meaning of Race in Science-Considerations for Cancer Research. Among the conclusions reached by that panel were that race is a social construct and that biologically distinct races do not exist.78 Rather than focusing on racial and ethnic groups as discrete categories defined largely by visible phenotypic characteristics (e.g., skin color), viewing race and ethnicity as multidimensional psychological and social constructs may help to move our understanding of how culture, ethnic identity, and minority status influence individual risks of developing and dying from breast carcinoma.3

CONCLUSIONS

African-American women have a lower incidence of breast carcinoma but present at a younger age. They also have a higher breast carcinoma mortality rate. However, the correlation between race, SES, and breast carcinoma incidence and survival remains unclear. The clinical studies reviewed here suggest that patients of any race with lower SES tend to present with worse tumor pathologic factors, higher stage of disease, and a lower likelihood of survival. Whether the responsible variables are nutrition, environmental factors, hormonal factors, education, screening practices, treatment recommendations, or income remains to be determined.

Race is a social construct, and the variability of genomic make-up between different ethnic groups is small. It seems doubtful that genetic mutations are responsible for lower survival rates in African-American women with breast carcinoma. Outcome studies from single institutions and clinical trials with uniform treatment have shown that race is not a significant predictor for survival after adjustment for stage, tumor size, lymph node status, hormone receptor status, age, and SES. SES replaced race as a predictor of worse outcome after women were diagnosed with breast carcinoma in many studies. Further research should be conducted concerning the precise elements of SES that account for the incidence of breast carcinoma, age at diagnosis, hormone receptor status, and survival to devise better strategies to improve outcome. In addition, further studies of how to improve access to treatment and compliance with treatment recommendations in relation to SES are warranted.

Acknowledgements

The authors thank Barbara Silver, Clinical Research Coordinator, Dana-Farber Cancer Institute, Harvard Medical School, for editing the article.

Ancillary