Racial/ethnic variation in clinical presentation, treatment, and survival among breast cancer patients under age 35

Authors

  • Vickie L. Shavers Ph.D.,

    Corresponding author
    1. Applied Research Program, Health Service and Economics Branch, Division of Cancer Control and Population Science, National Cancer Institute, Bethesda, Maryland
    • Applied Research Program, Health Service and Economics Branch, Division of Cancer Control and Population Science, National Cancer Institute, 6130 Executive Boulevard, MSC-7344, EPN Room 4005, Bethesda, MD 20892-7344
    Search for more papers by this author
    • Fax (301) 434-3710

  • Linda C. Harlan Ph.D.,

    1. Applied Research Program, Health Service and Economics Branch, Division of Cancer Control and Population Science, National Cancer Institute, Bethesda, Maryland
    Search for more papers by this author
  • Jennifer L. Stevens B.S.

    1. Information Management Systems, Silver Spring, Maryland
    Search for more papers by this author

  • This article is a US Government work and, as such, is in the public domain in the United States of America.

Abstract

BACKGROUND

The age specific breast cancer incidence rate for African-American women under age 35 is more than twice the rate for white women of similar age, and the mortality rate is more than three times higher. To determine factors that may explain racial/ethnic variation in outcomes among young women diagnosed with breast cancer, the authors examined the clinical presentation, treatment, and survival of African-American, Hispanic, and white women under age 35 years.

METHODS

Surveillance, Epidemiology, and End Results (SEER) Program data for 1990–1998 and SEER Patterns of Care data for 1990, 1991, and 1995 were used for this analysis. Multivariate logistic regression analyses were performed to examine factors associated with the receipt of selected breast cancer treatments. Kaplan–Meier survival analyses and Cox proportional hazards regression analyses were used to examine 5-year overall survival and disease-specific survival.

RESULTS

The authors found racial/ethnic variation in clinical presentation, treatment, and survival. Both African-American and Hispanic women presented with higher disease stage and a higher prevalence of adverse prognostic indicators compared to white women. African-American and Hispanic women received cancer-directed surgery and radiation less frequently after undergoing breast-conserving surgery. Racial/ethnic differences in clinical presentation and treatment were associated with poorer overall survival in unadjusted analyses. African-American and Hispanic women also had poorer overall survival after controlling for clinical and demographic characteristics and type of treatment.

CONCLUSIONS

Future research studies should further examine the factors that influence racial/ethnic differences in incidence, clinical presentation, and treatment differentials among young women diagnosed with breast cancer. A better understanding of these factors will facilitate the development of strategies to help eliminate this health disparity. Cancer 2003;97:134–47. Published 2003 by the American Cancer Society.

DOI 10.1002/cncr.11051

The American Cancer Society estimates that approximately 203,500 women in the United States will be diagnosed with breast cancer and nearly 40,000 will die from breast cancer in 2002.1 The majority of breast cancers occur among women 50 and older. Nonetheless, in 2001, it is estimated that 1600 new in situ cancers and 8000 new invasive breast cancers were diagnosed and that 1200 deaths from breast cancer occurred among women age < 40 years.2

Young women with breast cancer tend to a have a higher prevalence of adverse prognostic indicators, suggesting a more virulent form of disease compared with disease that is seen among older women.3 Young women have larger tumors,4–6 poorer grade tumors,4, 5, 7 more lymph node involvement,4, 5, 7 higher S-phase fractions,4 more aneuploid tumors,4 fewer hormone receptor positive tumors,4, 5, 8 earlier6 and more frequent local recurrences,5, 6 and poorer overall survival9, 10 and disease-specific survival.8 Other studies have reported either better survival or no difference in survival among young women compared with older women.7, 11

In general, African-American women are diagnosed with breast cancer at younger ages than white women.12, 13 In a survey of patients who were treated in hospitals throughout the United States, 6.3% and 6.2% of breast cancers among African-American women and Hispanic women, respectively, occurred in women under age 35 years compared to only 2.8% of breast cancers in white women.14 Furthermore, Surveillance Epidemiology, and End Results (SEER) data also show racial differentials in both breast cancer incidence rates and mortality rates among women under age 35 years. The age-specific breast cancer incidence among African-American women under age 35 years for 1994–1998 (46.7 per 100,000 population) was more than twice the incidence of similar age white women (17.3 per 100,000 population), and the age-specific mortality rate was more than three times higher for African-Americans (11.3 per 100,000 population) than white women (2.6 per 100,000 population).13

The reasons for racial/ethnic variation in breast cancer incidence and mortality rates among women under age 35 years have not been elucidated fully. Previous studies that have examined breast cancer among young women have focused on African-American and white women or did not provide race/ethnic specific results.8, 15, 16 Therefore, very little is known about the clinical presentation, treatment, and survival experiences of young Hispanic women with breast cancer.

To determine factors that may explain racial/ethnic variations in outcomes among young women who are diagnosed with breast cancer, we examined population-based data on the clinical presentation, including stage at diagnosis, hormone receptor status, lymph node status, histology, grade, DNA ploidy, other prognostic indicators, treatment, overall survival, and disease-specific survival among African-American Hispanic, and non-Hispanic white (NH white) women.

MATERIALS AND METHODS

Eligible women were under age 35 years and were newly diagnosed with primary breast cancer from 1990 through 1998 reported to the SEER registries. Women who were not African American, Hispanic, or NH white (n = 509 women), women who were diagnosed by death certificate or at autopsy (n = 8 women), or women with a previous diagnosis of cancer other than nonmelanoma skin malignancy (n = 300 women) were excluded.

The SEER Program was established by the National Cancer Act of 1971 and began case ascertainment in January, 1973. The registries that contribute data for this analysis include the Connecticut Tumor Registry, the University of Iowa Cancer Registry, the New Mexico Tumor Registry, the Utah Cancer Registry, the Hawaii Tumor Registry, Wayne State Cancer Registry, Emory University, Northern California Cancer Center, University of Southern California, and the Fred Hutchison Cancer Research Center. These registries cover the states of Connecticut, Iowa, New Mexico, Utah, and Hawaii and the metropolitan areas of Detroit, Atlanta, San Francisco-Oakland, San Jose-Monterey, Los Angeles County, and Seattle, respectively. Data from the San Jose-Monterey and Los Angeles County registries are included for 1992 through 1998 only.

The SEER registry data from breast cancer cases diagnosed during 1990–1998 were the main source of data for this analysis. SEER Patterns of Care (POC) data for women ages 21–34 who were diagnosed in 1990, 1991, and 1995 were also used to supplement other SEER data. SEER POC data are based on a sample of patients in the SEER registries and are used to describe insurance status, clinical characteristics, and treatments that are not well captured in routinely abstracted SEER data, such as chemotherapy and tamoxifen use. SEER POC data were reabstracted from patient medical records by SEER abstractors, and the treating physicians were contacted to verify information on demographics, tumor characteristics, and treatment.

Ecologic socioeconomic data were obtained from the 1990 United States Census of the population. These included educational data measured as the percent of residents age 25 or older with less than a high school education and median household income of the census tract in which each patient resided. The source of specific data is noted in the footnotes to each table, as appropriate.

Breast-conserving surgery (BCS) includes nipple resection, lumpectomy or excisional biopsy, wedge resection, quadrantectomy, segmental mastectomy, and tylectomy. Mastectomy includes total, modified radical, radical or extended mastectomies. Data on stage at the time of diagnosis were coded using the SEER-modified American Joint Committee on Cancer staging system. Comorbid conditions were abstracted from hospital medical records by SEER abstractors. A single Registered Health Information Administrator centrally coded comorbid conditions using the International Classification of Diseases (9th edition) coding scheme. Comorbidity scores were assigned using these data and the methodology described by Charlson et al.17 The Charlson comorbidity scores were used to examine the influence of comorbidity on treatment and survival.

SEER*STAT software18 and the Statistical Analysis System (version 8.2; SAS, Inc., Cary, NC19) were used to analyze routinely collected SEER data. SEER POC data were weighted to adjust for selection probability and were analyzed with SUDAAN.20 Because of the smaller sample size and the need to weight SEER POC data, only routinely abstracted SEER data were included in the multivariate analyses. Therefore, S-phase fraction, DNA ploidy, comorbidity, and insurance status were not examined in the models presented.

Racial/ethnic differences in the distribution of demographic, clinical, and treatment characteristics were compared using chi-square tests for homogeneity of proportions. We also constructed 2 × 2 tables to calculate race/ethnic specific relative risks with white women as the reference group. Student t tests and analysis of variance (ANOVA) were used to assess the significance of differences in the mean and median values of continuous variables. Logistic regression analyses were performed to examine the association between the clinical and sociodemographic variables and the receipt of radiation after BCS. First, we performed univariate logistic regression analyses to determine which of the independent variables had a significant association with the receipt of radiation therapy after BCS. Then, we performed three separate multivariate logistic regression analyses. Model 1 examined the association of race, Model 2 examined the influence of race and the significant clinical variables from Table 2, and Model 3 added education and income to the variables in Model 2. Our objective was to obtain the most parsimonious model of factors associated with the receipt of radiation therapy after BCS. Kaplan–Meier survival statistics were used to examine overall and disease-specific survival, and Cox proportional hazards regression analyses were used to examine the role of clinical, sociodemographic, and treatment characteristics on survival.

RESULTS

Demographic Characteristics

During 1990–1998, there were 3978 diagnoses of breast cancer among women ages 15–34 years who met our inclusion criteria reported to the SEER registries. Of these, 275 women also were captured in the SEER POC data. Overall, about 18.2% of women were African American, 15.5% were Hispanic, and 66.3% were NH white women. Hispanic and African-American women had a slightly younger mean age at diagnosis, were less frequently married, and more frequently lived in census tracts with low median annual incomes and higher percentages of persons who did not complete high school (Table 1).

Table 1. Demographic Characteristics of Women Under Age 35 Years Diagnosed with Breast Carcinoma: SEER Program, 1990–1998
CharacteristicAfrican-American women (n = 724 patients)Hispanic women (n = 616 patients)White non-Hispanic women (n = 2638 patients)P value
  • SEER: Surveillance, Epidemiology, and End Results; HMO: health maintenance organization.

  • a

    SEER Program, 1990–1998.

  • b

    1990 U.S. Census data.

  • c

    Patterns of Care: 1990-, 1991-, and 1995-weighted proportions.

  • d

    Unweighted number: may differ between analyses because of missing data.

Age (yrs)a    
 Mean30.730.431.3< 0.001
 Median31.031.032.0< 0.001
 Range19–3415–3419–34
Educationb    
 Total no.6835962534
 Mean % of persons in census tract who did not complete high school29.034.716.4< 0.001
 Median % of persons in census tract who did not complete high school27.531.814.6< 0.001
Incomeb    
 Total no.6835962534
 Median annual income in census tract$33,114$35,363$42,268<0.001
Marital status (%)a    
 Single51.231.525.3
 Married37.256.364.6
 Divorced6.26.06.2< 0.001
 Separated1.22.80.8
 Widowed0.60.90.6
 Unknown3.62.62.5
Insurance status (%)c    
 Total no.67d53d155d
 No insurance2.314.83.3
 Medicaid/Medicaid pending20.316.38.0
 Private insurance (non-HMO)9.39.124.9< 0.001
 HMO or other managed care plan, including military37.023.924.1
 Other insurance14.626.534.7
 Unknown16.59.44.9

Insurance status for women who were included in the POC data also differed by race/ethnic group (Table 1). About 15% of Hispanic women did not have health insurance coverage compared with 3.3% of NH white women and 2.3% of African-American women. African-American women (relative risk [RR], 0.4; 95% confidence interval [95%CI], 0.2–0.6) and Hispanic women (RR, 0.4; 95%CI, 0.2–0.7) were significantly less likely to have private nonhealth maintenance organization (HMO) insurance compared with NH white women. African-American women also were enrolled more frequently in an HMO.

Clinical Characteristics

Stage

There were racial/ethnic differences in stage distribution (Table 2). Hispanic women were less likely than white women to be diagnosed with in situ breast cancer (RR, 0.5; 95%CI, 0.3–0.7), as were African-American women (RR, 0.8; 95% CI, 0.6–1.1). African-American women (RR, 0.96; 95%CI, 0.91–1.0) and Hispanic women (RR, 0.97; 95%CI, 0.92–1.0) also were somewhat less likely to be diagnosed with Stage I–II disease than NH white women, a statistically marginal finding.

Table 2. Initial Clinical Characteristics of Women Under Age 35 Years Diagnosed with Breast Cancer: SEER Program, 1990–1998
Clinical variableAfrican-American women (n = 724 patients)aHispanic women (n = 616 patients)aWhite non-Hispanic women (n = 2638 patients)aP value
  • SEER; Surveillance, Epidemiology, and End Results; AJCC: American Joint Committee on Cancer; ER: estrogen receptor; NA: not available; PR: progesterone receptor.

  • a

    Applies to SEER data; for SEER Patterns of Care, data sample sizes (Total no.) are shown as unweighted numbers (numbers may differ between analyses because of missing data).

  • b

    SEER Program, 1990–1998

  • c

    Patterns of Care: 1990-, 1991-, and 1995-weighted proportions.

  • d

    There were too few observations in this group for statistical testing.

  • e Note that estrogen receptor and progesterone receptor status were determined for women with primary tumor that measured ≥ 1 cm in greatest dimension.

AJCC stage (%)b    
 In situ7.04.18.8 
 Stage I16.415.926.7 
 Stage II52.454.145.1< 0.001
 Stage III10.211.28.6 
 Stage IV6.56.73.7 
 Unknown stage7.58.17.1 
Size of primary tumor (%) (mutually exclusive categories)b    
 Microscopic focus/foci2.50.72.7 
 Diagnosed with mammography/xerography only0.10.50.2 
 Paget disease of the nipple0.10.00.1 
 Diffuse, widespread1.83.91.6< 0.001
 < 1 cm2.94.16.4 
 ≥ 1 cm78.979.676.0 
 Unknown/unstated13.711.412.9 
 Mean size (cm)3.43.63.0< 0.001
 Median size (cm)2.83.02.5< 0.001
Laterality at diagnosis (%)b    
 Unilateral99.599.799.4 
 Bilateral0.10.00.10.892
 Unknown0.40.30.5 
No. of lymph nodes examined (%)b    
 None21.323.518.6 
 1–32.42.12.0 
 4–1011.210.113.70.004
 ≥ 1162.760.463.6 
 No. unknown1.42.91.2 
 Unknown1.11.01.0 
No. of positive lymph nodes (%)b    
 None36.532.644.6 
 1–322.423.421.1 
 4–10 +12.411.59.3 
 ≥ 115.45.75.0 
 Exact no. unknown (at least 1)1.12.30.5< 0.001
 None examined21.123.518.6 
 Unknown1.11.01.0 
Histology (%)b    
 Ductal cancer in situ5.43.47.2 
 Lobular cancer in situ1.20.71.1 
 Mixed ductal/lobular cancer in situ0.40.00.20.007
 Other in situ0.00.00.2 
 Invasive cancer93.095.991.2 
Tumor grade (%)b    
 Grade 1 (well differentiated)2.13.43.4 
 Grade 2 (moderately differentiated)16.319.622.3 
 Grade 3 (poorly differentiated)54.851.543.3< 0.001
 Undifferentiated2.65.25.9 
 Unknown grade24.220.325.1 
S-phase fraction (%)c    
 Total no.a564891 
 Low (< 10%)11.415.019.8 
 High (> 10%)11.431.76.2< 0.001
 Not done/unknown77.253.374.1 
DNA ploidy (%)c    
 Total no.a564891 
 Aneuploid23.726.217.2 
 Diploid4.328.420.8< 0.001
 Not done/unknown72.045.462.0 
ER status (%)b    
 All ages    
  ER testing not done8.68.49.0 
  ER positive30.529.638.9 
  ER negative34.432.033.1 
  Borderline (negative or positive undetermined)1.40.81.6< 0.001
  Ordered, results unknown5.83.43.2 
  Unknown19.325.814.3 
Proportion of ER positive tumors by age group for women with known ER status (%)    
 Total no.4803841940 
 Age group (yrs)    
  15–190.00.0100NAd
  20–2463.236.447.70.230
  25–2944.342.753.30.074
  30–3448.752.255.70.051
PR status (%)b    
 PR testing not done9.99.49.4 
 PR positive27.427.835.6 
 PR negative35.632.535.4< 0.001
 Borderline (negative or positive undetermined)1.20.81.4 
 Ordered, results unknown5.93.43.3 
 Unknown19.926.114.9 
Proportion of PR positive tumors by age group for women with known PR status (%) by age group    
 Total no.4653761911 
 Age group (yrs)    
  15–190.00.0100NAd
  20–2455.636.445.50.478
  25–2944.641.746.90.639
  30–3444.050.052.20.024
Charlson Comorbidity Score (%)c    
 Total no.6753155 
 Proportion with a score of 1 or more6.06.41.90.007

Tumor size

Median tumor sizes differed by race/ethnic group. African-American women were significantly less likely than NH white women to be diagnosed with a primary tumor measuring < 1 cm (RR, 0.6; 95%CI, 0.4–0.8) as were Hispanic women (RR, 0.6; 95%CI, 0.4–0.8). The median sizes of primary tumors were 3.0 cm, 2.8 cm and 2.5 cm for Hispanic, African-American, and NH white women, respectively (Table 2).

Lymph node status

African-American and Hispanic women received axillary lymph node examinations less frequently compared with NH white women. About 77% of Hispanic and 78.7% of African-American women received axillary lymph node examinations compared with 81.4% of NH white women (Table 2). Among women who underwent axillary lymph node dissection, African-American (RR, 1.2; 95%CI, 1.1–1.3) and Hispanic women (RR, 1.3; 95%CI, 1.1–1.4) were more likely to have positive lymph nodes compared with NH white women. More than 57% of Hispanic women had at least one positive lymph node compared with 53.1% of African-American women and 44.5% of NH white women. African-American women (RR, 1.3; 95%CI, 1.1–1.5) and Hispanic women (RR, 1.3; 95%CI, 1.1–1.6) also were more likely to be diagnosed with four or more positive lymph nodes compared with NH white women.

Grade and other proliferative indices

African-American and Hispanic women had a higher prevalence of factors associated with highly aggressive tumors (Table 2). Compared with NH white women, African-American women were somewhat more likely to have Grade 3 or undifferentiated tumors (RR, 1.2; 95%CI, 1.1–1.3), tumors with S-phase fractions > 10% (RR, 1.5; 95%CI, 1.03–2.3), and aneuploid tumors (RR, 3.6; 95%CI, 1.6–8.0) among patients for whom these data were known. Hispanic women also were more likely than NH white women to have Grade 3 or undifferentiated tumors (RR, 1.2; 95%CI, 1.1–1.3), tumors with S-phase fractions > 10% (RR, 2.4; 95%CI, 1.5–3.8), and aneuploid tumors (RR, 1.1; 95%CI, 0.8–1.5).

ER/PR status

We found racial/ethnic differences in hormone receptor status that persisted when examined by age groups. Both African-American women and Hispanic women more frequently had hormone receptor positive tumors less frequently compared with NH white women and had tumors with unknown receptor status more frequently (Table 2). Compared with NH white women, African-American women were more likely to have estrogen receptor (ER) negative tumors (RR, 1.1; 95%CI, 1.03–1.3) or progesterone receptor (PR) negative tumors (RR, 1.1; 95%CI, 1.02–1.3). Hispanic women also were more likely to have ER negative or PR negative tumors compared with NH white women (Hispanic women: RR, 1.1; 95%CI, 1.0–1.3; NH white women: RR, 1.1; 95%CI, 1.0–1.2). To evaluate the role of race/ethnicity and age on hormone receptor status, we examined ER and PR status by race/ethnicity for four age categories (i.e., ages 15–19 years, 20–24 years, 25–29 years, and 30–34 years). The frequency of ER positive tumors increased with increasing age only among Hispanic women and NH white women (Table 2). For African-American women, the proportion of ER positive tumors was highest among women ages 20–24 years. When stratified by 5-year age groups, racial/ethnic differences in the prevalence of ER positive tumors were significant only for women ages 30–34 years. A similar pattern was observed for PR status (Table 2).

Comorbidity

There was a higher prevalence of women with Charlson comorbidity scores ≥ 1 more among African-American and Hispanic women compared with than NH white women. Six percent of African-American women and 6.4% of Hispanic women had Charlson comorbidity scores ≥ 1 compared with 2% of NH white women (Table 2).

Treatment for in situ breast cancer

There were no statistically significant racial/ethnic differences in treatment for patients with in situ breast cancer (data not presented). Cancer-directed surgery was included in the treatment regimen for in situ breast cancer for all African-American and Hispanic women and for 99% of NH white women. Mastectomy without radiation was the most frequently received treatment regimen for women with in situ breast cancer among all three racial/ethnic groups. There were no statistically significant racial/ethnic differences in the receipt of radiation therapy after BCS for in situ cancer when either African-American women or Hispanic women were compared with NH white women. Chemotherapy seldom was received, and women from the three racial/ethnic groups did not receive immunotherapy as treatment for in situ breast cancer.

Treatment for Invasive Breast Cancer

There were racial/ethnic differences in the receipt of cancer-directed surgery, radiation therapy after BCS, and use of tamoxifen among women with ER positive disease, Table 3. Compared with NH white women, African-American women (RR, 1.8; 95%CI, 1.3–2.5) and Hispanic women (RR, 1.7; 95%CI, 1.3–2.5) were nearly twice as likely to receive no cancer-directed surgery for invasive breast cancer. The most common reason that women did not receive cancer-directed surgery was because it was not recommended. Among women who did not receive cancer-directed surgery, African-American women (70.2%), more frequently than Hispanic women (40.5%) and NH white women (41.5%), did not undergo surgery because it was not recommended. Less than 3% of women from each of the three racial/ethnic groups who did not undergo surgery had contraindications to surgery; and 7% of Hispanic and 8.5% of NH white women refused surgery, but no African American women refused surgery. Mastectomy was the most frequent procedure for women from all three racial/ethnic groups who underwent surgery. Rates of BCS were similar among the study groups (Table 3).

Table 3. Patterns of Care by Race/Ethnic Group for Women Under Age 35 Years Diagnosed with Invasive Breast Cancer: SEER Program, 1990–1998
Pattern of careAfrican-American women (n = 724 patients)aHispanic women (n = 616 patients)aWhite non-Hispanic women (n = 2638 patients)aP value
  • SEER: Surveillance, Epidemiology, and End Results; BMT: bone marrow transplantation; NOS: not otherwise specified.

  • a

    Applies to SEER data; for SEER Patterns of Care data, sample sizes are indicated by the rows headed Total no.

  • b

    SEER Program, 1990–1998.

  • c

    Patterns of Care: 1990, 1991, and 1995 weighted percent.

  • d

    Weighted proportions.

  • e

    Unweighted numbers: may differ between analyses because of missing data.

Site specific surgery (%)b    
 No cancer-directed surgery7.77.34.2
 Mastectomy    
  With radiation10.914.212.2
  Without radiation42.840.141.0< 0.001
 Breast-conserving surgery (%)    
  With radiation23.821.531.3
  Without radiation14.916.911.4
Adjuvant therapy (%)cd    
 Total no.e5852125
 None20.617.020.4
 Chemotherapy, multiple agents46.552.467.0
 Tamoxifen only6.49.50.6
 Chemotherapy and tamoxifen13.417.810.6< 0.001
 Refused chemotherapy6.81.70.4
 Chemotherapy recommended, unknown if received3.41.50.0
 Chemotherapy unknown2.90.01.0
Radiation (%)bd    
 None58.458.253.0
 Received radiation37.237.644.4
 Refused radiation0.60.50.5
 Recommended, unknown if received2.83.42.0< 0.001
 Unknown1.00.30.2
Immunotherapy (%)b    
 None98.495.895.6
 Biologic response modifier0.52.42.2
 BMT, autologous0.00.00.2
 BMT, NOS0.20.00.0
 Stem cell transplantation0.31.21.30.059
 Combined immunotherapy0.00.20.2
 Refused immunotherapy0.00.00.1
 Recommended, unknown if received0.60.30.3
 Unknown0.20.20.2
Treated on a clinical trial protocol (%)cd    
 Total no.e5852125
 Yes1.810.711.0
 No70.674.869.3
 Refused0.00.01.10.006
 Recommended, unknown if treated on protocol0.00.00.6
 Unknown27.714.618.0

Although there were no statistically significant racial differences in the refusal of radiation therapy, African-American (RR, 0.8; 95%CI, 0.8–0.9) and Hispanic women (RR, 0.8; 95%CI, 0.7–0.9) were significantly less likely than NH white women to receive adjuvant radiation after undergoing BCS for invasive breast cancer. To gain insight into the possible role of comorbid conditions, using POC data only, we examined the association of the receipt of adjuvant radiation therapy after BCS and the Charlson comorbidity score (i.e. < 1, 1 +) in a univariate analysis. A comorbidity score ≥ 1 was not significantly associated with the receipt of radiation therapy after BCS. Similarly, using POC data, we also examined the role of Medicaid and the receipt of radiation therapy after BCS. Women with Medicaid health insurance were significantly less likely to receive adjuvant radiation therapy after undergoing BCS (RR, 0.7; 95%CI, 0.5–0.98) compared with women who had other types of insurance.

African-American women and white women received systemic therapy at similar rates, whereas the percentage of Hispanic women who received systemic therapy was slightly lower. Multiple-agent chemotherapy was the most frequently received systemic treatment for women from all three racial/ethnic groups and was received by 46.5% of African-American women, 52.4% of Hispanic women, and 67.0% of NH white women with invasive breast cancer. About 7% of African-American, 1.7% of Hispanic, and 0.4% of NH white women refused chemotherapy (Table 3). Overall, 19.8% of African-American women, 27.3% of Hispanic women. and 11.2% of NH white women received tamoxifen alone or in combination with chemotherapy for their invasive breast cancer. Among women with ER positive disease, 40.1% of African-American, 55.6% of Hispanic, and 15.7% of NH white women received tamoxifen alone or in combination with chemotherapy (P < 0.001). African-American women with ER positive disease (RR, 2.6; 95%CI, 1.6–4.1) and Hispanic women with ER positive disease (RR, 3.5; 95%CI, 2.4–5.3) were significantly more likely to receive tamoxifen compared with NH white women. Women who were treated on a clinical trial protocol were also more likely to receive tamoxifen (RR, 2.0; 95%CI, 1.3–3.0), BCS (RR, 2.0; 95%CI, 1.8–2.3), and immunotherapy (RR, 2.9; 95%CI, 1.1–7.6) compared with women who were not treated on a clinical trial protocol. Overall, about 9% of patients were treated on a clinical trial protocol. African-American women were the least likely of all treated women to have received treatment on a clinical trial protocol (Table 3).

To examine the role of clinical and demographic characteristics on the receipt of radiation therapy after BCS, we performed logistic regression analyses (Tables 4, 5). In unadjusted analyses, race, stage, tumor size, and lymph node status were significantly associated with the receipt of radiation therapy after BCS; but grade, education, income, marital were not. African-American and Hispanic women were significantly less likely than white women to receive radiation therapy after undergoing BCS (Model 1; African-American and Hispanic women: odds ratio [OR], 0.6; 95%CI, 0.5–0.8; NH white women: OR, 0.5; 95%CI, 0.4–0.7) in unadjusted analyses and after controlling for stage and lymph node status in multivariate analyses (Model 2; African-American and Hispanic women: OR, 0.7; 95%CI, 0.5–0.9; white women: OR, 0.5; 95%CI, 0.4–0.7). With the addition of education to the clinical characteristics in Model 2, the association of African-American race became marginally significant, whereas the association of Hispanic ethnicity and the receipt of radiation therapy after BCS did not change (Model 3).

Table 4. Logistic Regression Analysis of Women who Received of Radiation Therapy after Undergoing Breast-Conserving Surgery, SEER Program 1990–1998
VariableRelative risk (95% confidence interval)
Model 1 (unadjusted)Model 2 (adjusted for disease stage and lymph node status)Model 3 (adjusted for disease stage, lymph node status, and education)a
  • a

    The ecologic income variable was not associated with the receipt of radiation therapy after breast-conserving surgery in univariate analyses and, thus, was not included in the multivariate model.

  • b There were 1028 white patients, 260 African-American patients, and 227 Hispanics patients.

Race/ethnic group   
 White1.01.01.0
 African American0.6 (0.5–0.8)0.7 (0.5–0.9)0.7 (0.5–1.0)
 Hispanic (black or white)0.5 (0.4–0.7)0.5 (0.3–0.6)0.5 (0.4–0.7)
Disease stage   
 In situ0.3 (0.2–0.4)0.3 (0.2–0.4)
 Stage I1.01.0
 Stage II0.9 (0.7–1.3)0.9 (0.7–1.3)
 Stage III0.9 (0.5–1.8)1.0 (0.5–1.8)
 Stage IV0.3 (0.1–0.6)0.3 (0.1–0.6)
No. of positive lymph nodes   
 None1.01.0
 1–30.9 (0.7–1.3)0.9 (0.7–1.3)
 4–100.7 (0.4–1.1)0.7 (0.4–1.1)
 ≥ 110.3 (0.1–0.6)0.3 (0.1–0.6)
Education (% of persons who did not complete high school)   
 ≥ 25.11.0
 ≥ 16.2 and < 25.10.8 (0.6–1.1)
 ≥ 9.5 and < 16.20.8 (0.6–1.1)
 < 9.50.7 (0.5–1.0)
Table 5. Cox Proportional Hazards Regression Analysis of the Hazard of Death from All Causes among Women Under Age 35 Years with Breast Cancer
VariableHazard ratio (95% confidence interval)
Model 1 (unadjusted)Model 2 (adjusted for stage, grade, lymph node status, and treatment)Model 3 (adjusted for stage, grade, lymph node status, treatment, age, income, and education)
  1. BCS: breast-conserving surgery.

Race/ethnic group   
 White1.01.01.0
 African American1.4 (1.2–1.6)1.3 (1.1–1.5)1.2 (1.1–1.4)
 Hispanic (black or white)1.5 (1.3–1.7)1.2 (1.0–1.4)1.1 (1.0–1.3)
Stage   
 In situ0.7 (0.5–0.9)0.7 (0.5–0.9)
 Stage I1.01.0
 Stage II1.1 (0.9–1.4)1.1 (0.9–1.4)
 Stage III1.7 (1.3–2.1)1.6 (1.3–2.1)
 Stage IV3.6 (2.7–4.7)3.6 (2.7–4.8)
 Unknown stage1.3 (1.0–1.7)1.3 (1.0–1.7)
Grade   
 Well differentiated1.01.0
 Moderately differentiated0.9 (0.6–1.4)0.9 (0.6–1.3)
 Poorly differentiated1.1 (0.7–1.5)1.1 (0.7–1.5)
 Undifferentiated1.0 (0.7–1.6)1.0 (0.7–1.6)
 Unknown grade0.8 (0.6–1.2)0.8 (0.6–1.2)
No. of positive lymph nodes   
 None1.01.0
 1–31.4 (1.1–1.7)1.4 (1.1–1.7)
 4–101.8 (1.5–2.3)1.8 (1.5–2.3)
 ≥ 112.4 (1.9–3.0)2.4 (1.9–3.0)
 Exact no. unknown (at least 1)2.4 (1.5–3.9)2.4 (1.5–3.9)
 Unknown status2.1 (1.8–2.6)2.1 (1.8–2.6)
Treatment   
 No surgery1.5 (1.2–1.9)1.5 (1.1–1.9)
 Mastectomy without radiation1.01.0
 Mastectomy with radiation0.9 (0.8–1.1)1.0 (0.8–1.2)
 BCS with radiation0.8 (0.7–0.9)0.8 (0.7–0.9)
 BCS surgery without radiation1.3 (1.1–1.5)1.3 (1.1–1.5)
Age (yrs)1.0 (0.98–1.01)
Median income of persons in census tract   
 $4210–$31,6251.0
 $31,626–$39,6280.9 (0.8–1.1)
 $39,269–$48,5611.1 (0.9–1.3)
 $48,562–84,8831.0 (0.8–1.2)
Education (% of persons in census tract who did not complete high school)   
 < 10.551.0
 10.56–17.731.0 (0.8–1.1)
 17.74–28.211.0 (0.8–1.2)
 28.21–83.581.1 (0.9–1.3)

Mortality and Survival

African-American and Hispanic women experienced poorer outcomes compared with white women. Just over 23% of African-American and 21.9% of Hispanic women, compared with 15.6% of NH white women, had breast cancer listed as the underlying cause of death. In addition, 4.2% of African-American, 2.9% of Hispanic, and 2.0% of NH white women died from other causes. The 5-year overall survival rates were 42.1% for African-American women, 43.1% for Hispanic women, and 54.1% for NH white women (P ≤ 0.001). Similarly, the 5-year disease-specific survival rates were 66.6% for African-American women, 64.7% for Hispanic women, and 76.5% for NH white women (P < 0.001).

In an unadjusted Cox regression analysis, African-American women and Hispanic women had a significantly increased overall and disease specific hazard of death compared with white women (Tables 5–6. The overall hazard ratio (HR) for African-American women was 1.4 (95%CI, 1.2–1.6), and the disease specific HR was 1.6 (95%CI, 1.3–1.9). Hispanic women also had an increased overall and diseases-specific hazard of death compared with white women (Hispanic women: HR, 1.5; 95%CI, 1.3–1.7; NH white women: HR, 1.6; 95%CI, 1.3–1.9).

Table 6. Cox Proportional Hazards Regression of the Hazard of Death from Breast Cancer among Women Under Age 35 Years with Breast Cancer
VariableHazard ratio (95% confidence interval)
Model 1 (unadjusted)Model 2 (adjusted for stage, grade, lymph node status, and treatment)Model 3 (adjusted for stage, grade, lymph node status, treatment, age, income, and education)
  1. BCS: breast-conserving surgery.

Race/ethnic group   
 White1.01.01.0
 African American1.6 (1.3–1.9)1.4 (1.1–1.7)1.3 (1.0–1.6)
 Hispanic (black or white)1.6 (1.3–1.9)1.1 (0.9–1.3)1.0 (0.8–1.3)
Stage   
 In situ0.7 (0.02–0.3)0.7 (0.02–0.3)
 Stage I1.01.0
 Stage II2.2 (1.5–3.1)2.1 (1.5–3.1)
 Stage III3.5 (2.3–5.3)3.4 (2.3–5.2)
 Stage IV9.8 (6.2–14.7)9.5 (6.2–14.6)
 Unknown stage1.9 (1.2–2.9)1.9 (1.2–2.9)
Grade   
 Well differentiated1.01.0
 Moderately differentiated2.5 (0.9–6.9)2.5 (0.9–6.9)
 Poorly differentiated3.2 (1.2–8.5)3.2 (1.2–8.6)
 Undifferentiated3.3 (1.2–9.3)3.3 (1.2–9.3)
 Unknown grade2.3 (0.0–6.1)2.2 (0.8–6.1)
No. of positive lymph nodes   
 None1.01.0
 1–32.1 (1.5–2.8)2.1 (1.6–2.8)
 4–103.3 (2.4–4.5)3.3 (2.4–4.5)
 ≥ 114.7 (3.3–6.6)4.8 (3.4–6.7)
 Exact no. unknown (at least 1)5.0 (2.8–8.7)5.1 (2.9–8.9)
 Unknown status4.4 (3.3–6.0)4.5 (3.3–9.0)
Treatment   
 No surgery1.0 (0.7–1.4)1.0 (0.7–1.3)
 Mastectomy without radiation1.01.0
 Mastectomy with radiation1.0 (0.8–1.2)1.0 (0.8–1.2)
 BCS with radiation0.7 (0.6–0.9)0.7 (0.6–0.9)
 BCS surgery without radiation0.8 (0.6–1.1)0.8 (0.6–1.1)
Age (yrs)1.003 (0.98–1.03)
Median income of persons in census tract   
 $4210–$31,6251.0
 $31,626–$39,6280.9 (0.7–1.2)
 $39,269–$48,5611.0 (0.7–1.2)
 $48.562–84.8830.9 (0.7–1.2)
Education (% of persons in census tract who did not complete high school)   
 < 10.551.0
 10.56–17.731.1 (0.9–1.4)
 17.74–28.211.2 (0.9–1.5)
 28.21–83.581.2 (0.9–1.6)

After adjusting for stage, number of positive lymph nodes, and type of treatment in multivariate analyses, African-American and Hispanic race/ethnicity remained significantly associated with an increased overall hazard of death (Table 6). Women who did not undergo surgery for their breast cancer had nearly two times the overall hazard of death compared with women who received mastectomy (HR, 1.5; 95%CI, 1.2–1.9). Women who underwent BCS without radiation also had an increased overall hazard of death compared with women who underwent mastectomy (HR, 1.3; 95%CI, 1.1–1.5).

African-American women and Hispanic women also had an increased hazard of death from breast cancer in an unadjusted analysis. In multivariate analyses, African-American women had an increased hazard of death from breast cancer compared with NH white women after adjustment for stage, grade, lymph node status, and treatment (HR, 1.4; 95%CI, 1.1–1.7). After also adjusting the analysis for age, education, and income, African-American women continued to have a significantly elevated risk of death from breast cancer (HR, 1.3; 95%CI, 1.03–1.6). In contrast, there was no statistically significant increase in the hazard of death from breast cancer in multivariate analyses for Hispanic women compared with NH white women. Furthermore, age, income, and education were not significantly associated with the risk of overall death or death from breast cancer.

DISCUSSION

We found racial/ethnic disparities in clinical presentation, treatment, and survival. Both African-American and Hispanic women presented with higher disease stage and had a higher prevalence of adverse prognostic indicators compared with NH white women. The reason for the higher proportion of Stage III–IV disease we observed among African-American and Hispanic women is not clear. Studies among older women have shown that, in part, racial/ethnic differences in disease stage at diagnosis are a result of the underutilization of screening mammography among racial/ethnic minorities.12 Routine screening mammography, however, is not currently recommended for women under 40 unless there is a family history of breast cancer diagnosed at an early age. Furthermore, mammography is less sensitive in detecting breast cancer among younger women.7, 21–23 Nevertheless, results from a population-based case–control study showed that, among women age under 45, 52% of NH white, 40% of African-American, and 33% of women in other racial/ethnic groups had received at least 1 mammogram; and 15% of NH white women, compared with 9% of African-American women and 6% of women in other racial/ethnic groups, received 3 or more mammograms within the 5-year period prior to 1 year before the date of diagnosis for patients in the case group and prior to the date of the telephone screening call for individuals in the control group.24 Furthermore, racial/ethnic differentials in prior knowledge of a family history of breast cancer, as suggested by research showing a tendency among racial/ethnic minorities to not discuss cancer diagnoses,25 may have resulted in racial/ethnic differences in the receipt of recommendations for earlier screening mammograms for women at increased risk of developing breast cancer. In the study by Weiss et al.,24 only 28% of women who had a family history of breast cancer among first-degree relatives had not received a mammogram, 43% had received 1 or 2 mammograms, and 29% had received 3 or more mammograms. The association of fatalism with screening behavior26–28 and the higher prevalence of fatalism among racial/ethnic minority populations27, 29, 30 suggest that fatalism about cancer also may contribute to the differential use of screening mammography among young women who are at increased risk for breast cancer.

Racial/ethnic differences in the utilization of clinical breast examination or breast self-examination also may underlie the differentials in stage we observed. In one report, only 4% of the breast tumors among women under age 35 years were detected with mammograms, 9% were detected with clinical breast examinations, and 80% were detected by the patient or her partner.24 Furthermore, differences in access to a usual source of care may contribute to racial differentials in the use of clinical breast examination (due to the lack of opportunity to have a breast examination) and in the use of breast self-examination, because self-examination may not be recommended or because the proper technique for breast self-examination is not taught. Data from the Medical Expenditure Panel Survey showed that 29.6% of Hispanic women and 20.2% of African-American women, compared with 15.2% of NH white women, have no regular source of care.31

We found that young African-American and Hispanic women had higher proportion of tumors with clinical characteristics associated with a poor outcome, including tumors that were aneuploid, with high S-phase fractions and higher grade tumors compared with NH white women. These findings suggest that breast cancer may be a more aggressive disease in young African-American and Hispanic women. African-American and Hispanic women also more frequently, had ER negative tumors, which likely influenced overall tamoxifen utilization. Other research has shown that the frequency of ER positive tumors increases with age.32 We also observed this general trend; however, the proportion of ER positive tumors among African-American women was greatest for women ages 20–24, and the frequency of ER positive tumors was lower among African-American women compared with NH white women for all other age groups.

We also found racial/ethnic treatment differences among women who were diagnosed with invasive breast cancer. Although they had larger tumors, African-American and Hispanic women underwent axillary lymph node dissection less frequently compared with NH white women. The proportion of African-American and Hispanic women who did not undergo cancer-directed surgery for their invasive breast cancer and the proportion of women from all racial/ethnic groups who underwent BCS without receiving radiation therapy are of particular concern. Observational research studies show that BCS without radiation therapy is associated with a higher frequency of local recurrence compared with BCS with radiation therapy,33, 34 especially among younger women.35, 36 The fact that < 1% of patients from each racial/ethnic group refused radiation therapy suggests that the absence of a recommendation to have radiation therapy after undergoing BCS contributed to the high proportion of women who did not receive this treatment. Other possible explanations include insurance status/type of insurance and the presence of comorbid conditions, which differed significantly by race/ethnic group. Although we examined the type of insurance and the presence of comorbid conditions among the subset of women who also were included in POC data, we could not incorporate these variables into multivariate analyses because of the problems detailed above (see Materials and Methods). However, we found that African-American and Hispanic women in this study had health care coverage through Medicaid more frequently compared with white women and that women with Medicaid received radiation therapy less frequently after undergoing BCS. These findings, coupled with other reports of the less frequent receipt of adjuvant radiation therapy after BCS among Medicaid beneficiaries compared with women with private insurance,37, 38 suggests that insurance type is a likely contributor to the treatment differences we noted. We also noted a higher prevalence of women with Charlson comorbidity scores ≥ 1 among African-American and Hispanic women compared with NH white women. Research studies among older women show that patients with comorbid conditions receive adjuvant radiation therapy less frequently after undergoing BCS.39 Analysis of the POC data from our study showed no statistically significant association between comorbidity score and the receipt of radiation therapy after BCS among our group of young women.

In multivariate analyses of regularly abstracted SEER data, African-American race, Hispanic ethnicity, disease stage, and education were significantly associated with the receipt of BCS without radiation therapy. Another interesting finding was the higher proportion of ER positive African-American and Hispanic women who received tamoxifen compared with ER positive NH white women.

Hispanic and African-American women also had poorer overall survival and disease specific survival compared with NH white women. Given the overall poorer clinical presentation we observed for Hispanic and African-American women compared with NH white women and the treatment differentials noted above, this is not a surprising finding. However, after adjusting for clinical characteristics, type of treatment, income, and education, African-American race was independently associated with an increased hazard of death in multivariate analyses that examined overall survival, whereas the association with Hispanic ethnicity was marginal. In a similar model that examined disease specific survival, Hispanic ethnicity was not associated with survival, whereas the association with African-American race was marginal.

Limitations

Data on DNA ploidy, S-phase fraction, insurance status, clinical trial participation, and comorbidity (from SEER POC data) represent approximately 7% of the regularly abstracted SEER cases for the study period. These data, however, were obtained from a random sample of SEER cases and were adjusted to account for the probability of selection, which should have reduced sampling bias. African-American women and Hispanic women were over sampled to assure an adequate number for analyses. Because these data represent a sample of cases, findings should be interpreted with the level of caution attributed to other sample data.

Directions for Future Research

Future research studies should examine factors that influence racial/ethnic differences in incidence, stage at diagnosis, and the prevalence of poor prognostic indicators among women who are diagnosed with breast cancer at under age 35. An analysis of prior knowledge of family history of breast cancer, in conjunction with racial/ethnic screening practices, also may provide insight into the racial/ethnic differentials in disease stage at diagnosis among these young women. Furthermore, future studies should evaluate more fully reasons for the failure to recommend radiation therapy after BCS for young women with breast cancer, particularly in light of the higher prevalence of local recurrences that occur among women in this age group compared with older women. A better understanding of factors that influence both racial/ethnic disparities in incidence and mortality from breast cancer among young women will facilitate the development of strategies to help eliminate this health disparity.

Acknowledgements

The authors thank Martin L. Brown, Ph.D., Chief, Health Services and Economics Branch, for his comments and suggestions on various drafts of this article. The authors also thank Rachel Ballard-Barbash, Assistant Director, Applied Research Program for her comments on the final draft of this article.

Ancillary