Fax: (212) 360-6965
Factors influencing mortality among young women with second primary breast carcinoma
Article first published online: 31 OCT 2002
Copyright © 2002 American Cancer Society
Volume 95, Issue 10, pages 2051–2058, 15 November 2002
How to Cite
Bernstein, J. L., Lapinski, R., Lynch, C., Holford, T. and Thompson, W. D. (2002), Factors influencing mortality among young women with second primary breast carcinoma. Cancer, 95: 2051–2058. doi: 10.1002/cncr.10950
- Issue published online: 31 OCT 2002
- Article first published online: 31 OCT 2002
- Manuscript Accepted: 17 JUN 2002
- Manuscript Revised: 12 JUN 2002
- Manuscript Received: 15 APR 2002
- National Cancer Institute of the National Institutes of Health. Grant Numbers: CA64245, CA69361
- contralateral breast carcinoma;
- prognostic factors;
- first primary;
- second primary
Tumor characteristics are strong predictors of survival among women with breast carcinoma, yet the variability in prognosis among women presenting with similar stages suggests other factors may also play an important role. We examine the prognostic significance of etiologic risk factors for breast carcinoma to determine whether factors that influence the development of breast carcinoma also affect the course of the disease among a prospective cohort of young women with bilateral breast carcinoma.
The 369 U.S. women included in this study were from the Cancer and Steroid Hormone Study who were diagnosed with an invasive first primary breast carcinoma between 1980 and 1982 and a second primary breast carcinoma before 1999. Cox proportional hazards models were used to evaluate factors known and suspected to be associated with breast carcinoma and with survival, based on reporting at the time of the first primary.
One hundred sixty women died during the 16–18-year follow-up period. The adjusted 1, 5, 10, and 15-year survival rates following diagnosis of second primary breast carcinoma were 94%, 70%, 55%, and 49%, respectively. Survival rates werepoorest among the youngest women, those diagnosed with a second primary within 5 years of their first, poor African American women, women with either primary diagnosed at a later stage, those with less than 12 years of school, single women, and those with major weight gain between age 18 and adulthood.
This study provided little evidence that important etiologic factors for breast carcinoma predict mortality following diagnosis of a second primary breast carcinoma. Cancer 2002;95:2051–8. © 2002 American Cancer Society.
Breast carcinoma is the most common form of cancer among women and the second most common cause of cancer death.1 In 2002, approximately 203,500 new cancers and 39,600 deaths are estimated to occur. Breast carcinoma accounts for about 30% of all new cancers and of all deaths. Although the incidence rate for first primary breast carcinoma has begun to level off, the incidence and mortality rates of second primary breast carcinoma (i.e., a newly arising nonmetastatic carcinoma in the contralateral breast) continue to rise. Approximately 50% of all second primary carcinomas (excluding nonmelanotic skin carcinomas) following a first diagnosis of breast carcinoma will be a second primary breast carcinoma in the contralateral breast. This means that approximately 5–10% of women diagnosed with breast carcinoma will develop a second primary during their lifetime.2 In sharp contrast to the large body of research describing the incidence and survival experience of patients with first primary breast carcinoma, little attention has been paid to survival following diagnosis of contralateral breast carcinoma. In part due to methodologic difficulties associated with studying rare diseases, prognostic factors for second primary breast carcinomas have not been identified consistently and the studies have been limited.
Stage and other characteristics of the tumor are strong predictors of survival among women with first primary breast carcinoma. The variability in prognosis among women presenting with similar stages suggests that other factors may also play an important role.3 Some studies have shown that important etiologic risk factors for breast carcinoma, including age, endogenous and exogenous hormonal factors, family history of cancer, weight, alcohol, and reproductive factors, also have prognostic significance. The purpose of this population-based prospective study was to characterize the mortality experience among women diagnosed with contralateral breast carcinoma and to evaluate the relative prognostic importance of etiologic risk factors, tumor characteristics, and radiation therapy on mortality following diagnosis of second primary breast carcinoma. Our study population is unique because we have followed these women for almost 20 years after the diagnosis of their first primary breast carcinoma. We have detailed risk factor information collected at the time of diagnosis of the first primary breast carcinoma. We describe factors influencing mortality following diagnosis of the second primary breast carcinoma.
MATERIALS AND METHODS
The 369 women included in the analysis were diagnosed with an invasive first primary breast carcinoma between December 1, 1980 and December 31, 1982 and diagnosed with a second primary in the contralateral breast thereafter, but before January 1, 1999. All women were interviewed in person at their homes within 6months of diagnosis of the first breast primary breast carcinoma and were between the ages of 20 and 54 at that time. They were all cases in the eight-center, population-based case–control Cancer and Steroid Hormone (CASH) Study, whose design and conduct have been described.4 The interview instrument consisted of detailed questions related to reproductive and medical history, family history of cancer, smoking, use of oral contraceptives and estrogen replacement therapy, previous breast surgery, and demographic information.
The Surveillance, Epidemiology, and End Results (SEER) Program's SEER Cancer Incidence Public-Use Database 1973–1998, which is updated on an annual basis, was used to determine vital status, follow-up information, tumor characteristics (e.g., stage and histology), and treatment information (i.e., radiation use as the first course of treatment administered or planned within 4 months of the initial course of therapy). The methods used for linkage with interview data have been described elsewhere.5 For consistency with previous studies, a diagnosis of breast carcinoma was considered to represent a second primary if it was an independent ductal carcinoma in situ or an invasive lesion diagnosed in the contralateral breast after diagnosis of a first primary breast carcinoma. We relied on the SEER definition of follow-up reporting of all events and second primary breast carcinoma in the contralateral breast. The definition of second primary breast carcinoma in the contralateral breast includes the date of diagnosis at least 2 months after the first primary or differences in histologic type involving the first three digits of the ICD-O morphology code.6 We had no independent pathologic review and no information on recurrence of the first primary.
Follow-Up for Mortality
Vital status, as of December 31, 1998, was determined for each patient through the SEER program files. For each woman who had died during the follow-up period, data on month and year of diagnosis of the second primary and on month and year of death were used to calculate length of survival (in months). For those known to be alive at last follow-up, the length of follow-up was calculated as the number of months between diagnosis of the second primary and the end of 1998. For women lost to follow-up (n = 28), the length of follow-up was calculated as the number of months between diagnosis of the second primary and the last date when the subject was known to be alive.
To obtain point and interval estimates of the association between risk factors and mortality, rate ratios (RRs) and their corresponding 95% confidence intervals (CIs) were calculated using Cox proportional hazards modeling techniques.7 The outcome of interest was death from any cause following diagnosis of the second primary breast carcinoma. Initial models were evaluated for the effect of each variable while adjusting for age at diagnosis of the second primary and length of the interval between diagnosis of the first and second primaries. Multivariable-adjusted RRs and 95% CIs for each variable were also calculated after adjusting for factors found to be independent predictors of mortality in the initial analysis. The number of women included in each model varied slightly because some covariates had missing values. Possible interaction effects between risk factors were assessed by adding crossproduct terms as covariates to the simple models containing only interval and age at diagnosis of the second primary and the relevant main effects.
To preserve as many patients as possible in the final multivariable models, any patient who had missing information concerning the presence/absence of a certain characteristic was classified as not having that characteristic. Very few cases had missing information. If an association existed, the effect of this type of misclassification would be to attenuate the magnitude of the estimates toward the null. Specifically, the factors for which this assumption was used included family history of cancer, history of benign breast disease, and radiation therapy.
Among the 369 women in this cohort, 160 deaths occurred within the 16–18-year follow-up period. The average interval length between diagnosis of the first and second primaries was 74 months, with 10% of the women presenting with a second primary within the first 6 months and greater than 25% after 10 years. The adjusted 1, 5, 10, and 15-year survival rates following diagnosis of a second primary breast carcinoma were 94%, 70%, 55%, and 49%, respectively. There was a statistically significant trend for the concordance between stage of the first primary and stage of the second (Mantel–Haenszel: x2 = 10.89, P < 0.001) with 163 (41%) presenting at the same stage for both primaries. A similar statistically significant concordant relation was found between those with medullary and lobular histologies for first and second primaries. The histology of the second primary was statistically associated with the interval between primaries (Fisher exact test, P = 0.005). For example, a second primary with lobular histology was diagnosed more frequently between 12 and 59 months after the first primary. Among the women who died, 90% had cancer listed as the primary cause of death and, of these, 87% were from breast carcinoma. The other sites were the pancreas (n = 1), upper airway (n = 3), ovary (n = 1), and unspecified (n = 1).
Table 1 shows the multivariable adjusted RRs and 95% CIs for risk of mortality following diagnosis of the second primary breast carcinoma according to tumor characteristics and treatment. Survival was poorest among women who had either primary diagnosed at a later stage and had the second primary breast carcinoma diagnosed within 5 years of the first primary. Receiving radiation as treatment for either primary had no substantial impact on mortality, nor did histology (ductal, medullary, or lobular) of either cancer or history of benign breast disease before the diagnosis of the first primary.
|Specific factor||No. at risk||RRa||95% CI||Missing|
|Time interval between first and second primaries (mo)|
|Stage of first primaryc|
|Histology of first primary|
|Stage of second primaryd|
|Histology of second primary|
|History of benign breast disease|
|Radiation following first primary||7|
|Younger than 45 yrs old at first diagnosis|
|Time since radiation (per 5 yrs)||1.13||(0.66–1.92)|
|45+ yrs old at first diagnosis|
|Time since radiation (per 5 yrs)||0.94||(0.50–1.75)|
|Radiation following second primary||6|
|Younger than 45 yrs old at first diagnosis|
|Time since radiation (per 5 year)||1.08||(0.52–2.25)|
|45+ yrs old at first diagnosis|
|Time since radiation (per 5 year)||0.47||(0.12–1.80)|
Table 2 presents the mortality risk following diagnosis of second primary breast carcinoma according to etiologic risk factors for breast carcinoma, including demographic and reproductive factors and family history of cancer. Risk of mortality was greatest among women younger than 35 who were diagnosed with a first primary, women with less then a high school education, and women who were single at the time they were diagnosed with their first primary. Although the main effects for race and household income level were not statistically significant, there was a statistically significant interaction effect (African American RR = 0.71; Income < $20,000 RR=1.00; and African American × Income < $20,000 RR=3.00; 95% CI 1.08–8.28). Women who had a 20–29% increase in Quetelet's index from age 18 to adulthood (time of their diagnosis of their first primary carcinoma) were at a substantially increased risk of death compared with women with less of a change. Neither smoking nor alcohol consumption affected risk. Risk of death according to family history of cancer and reproductive and menstrual history factors are also presented and there was not a substantial association for any of these factors.
|Specific factor||No. at risk||RRa||95% CI||Missing|
|Age at first primary diagnosis|
|Age at second primary diagnosis|
|Annual Income ($1000)|
|Income by race interaction|
|Income < $20,000||97||1.11||(0.73–1.69)|
|Interaction: Income < $20,000 by African American||24||2.98||(1.08–8.26)|
|Marital status at first primary|
|History of smoking and alcohol use at first primary|
|Age began smoking|
|Pack years smoked|
|History of body size and Quetelet's index at first primary|
|Adult weight (pounds)|
|Quetelet's index at age 18|
|Quetelet's index as adult|
|Percent change in Quetelet's index from age 18 to adult|
|Family history of cancer at first primary|
|Second degree only||8||0.68||(0.21–2.14)|
|Second degree only||14||0.94||(0.41–2.15)|
|Any breast carcinoma||144||0.75||(0.54–1.06)||0|
|Any ovarian carcinoma||17||1.44||(0.75–2.79)|
|Any endometrial carcinoma||29||0.87||(0.48–1.60)|
|Reproductive factors at first primary|
|Age at menarche|
|Age at first pregnancy (minimum 6-mo duration)|
|Age at menopause|
|Hormone therapy-all women|
|Estrogen replacement therapy||127||0.80||(0.55–1.15)||0|
|oral contraceptive use|
Our results indicate that a young age at second primary breast carcinoma is independently associated with a worse prognosis after adjusting for stage at diagnosis of both first and second breast primaries and length of the interval between primaries. This finding is consistent with other studies that evaluated survival among young women with unilateral and bilateral breast carcinoma. Those studies found that breast carcinoma among young women was more biologically aggressive than among older women.8–10 Our finding that a shorter interval between primaries is associated with a greater chance of mortality lends some support to this suggestion of aggressive tumor behavior among young women. We found no interaction between age at diagnosis and etiologic factors. This suggeststhat the development of specific types of cancer that may have different prognostic potential was not a function of age-related hormonal factors, such as menarche, parity, or menopause. We also failed to find an association with family history of cancer, suggesting that cancer patients with an inherited genetic susceptibility may not experience differential survival following second primary breast carcinoma.
The most consistent finding in the literature on mortality following first primary breast carcinoma is the association between the late stage of breast carcinoma and poor prognosis.11–16 Subsequent to diagnosis of second primary breast carcinoma, we found that late stage, of either tumor, was indicative of poor prognosis. Specifically in the multivariable-adjusted models that adjusted for stage at diagnosis of both primaries, late stage at diagnosis of both the first primary and the second primary was independently associated with a dramatic increase in mortality compared with women diagnosed with localized second primary breast carcinoma. We also found that early-stage diagnosis of the second primary breast carcinoma was associated with reduced mortality. Ernster et al.17 reported that the lower risk is most likely due to the effectiveness of treatment and the nature of in situ disease.
After multivariable adjustment, which included stage at diagnosis, the mortality rate increased significantly among the women in our study who had gained a substantial amount of weight between the age of 18 and the time of diagnosis of their first primary breast carcinoma. The magnitude of effect was similar among premenopausal and postmenopausal women and it was independent of height and weight. The literature on mortality and body weight suggests that among postmenopausal women, the relation of a high body mass idex (BMI) to mortality may be a result of a higher production of extraovarian estrogens in fat cells18–20 that accelerate tumor growth. However, the mechanisms whereby a high BMI mediates breast carcinoma among younger women are less straightforward. Daling et al.21 found a significant increase in mortality among heavy young women and the tumors were larger and more likely to have markers of high cellular proliferation. In that study, the information on weight concerned the period before the diagnosis of the first primary. For this proliferation to affect prognosis following second primary breast carcinoma, the action would need to be sustained. Nevertheless, the mechanism is unclear. Previous studies have demonstrated an etiologic association between second primary breast carcinoma and obesity,2, 5 which might affect prognosis, even after the first primary. Future studies should examine obesity in the interval between primaries to ascertain if this is biologically plausible. In the postmenopausal period, the major source of estrogen is the extraglandular (largely adipose tissue) conversion of the adrenal androgen, androstenedione, to estrone.22, 23
We observed that after adjusting for age and stage of both primaries, poor African American women, women with little education, and single women had an increased risk of mortality. Although the risk of a second primary breast carcinoma among African American women has been understudied, the mortality experience of African American women compared with white women has been documented. After adjusting for age at diagnosis, tumor stage, estrogen/progesterone receptor status, menopausal status, and histology, African American women experience a significant increase in mortality risk compared with white women.24 Furthermore, lower socioeconomic status (SES) has been associated with less cancer screening, resulting in delayed diagnosis and later stage of disease at presentation.24, 25 In some analyses, African American women have a two to five times increased risk of mortality after adjusting for tumor site and stage. Many investigators have studied the association between a patient's social support system and her resulting mortality experience.26–28 Kvikstad et al.26 showed that mortality following breast carcinoma diagnosis was significantly greater among divorced women compared with married women. Neale et al.27 showed that white widows also had worse survival. Whatever the mechanisms associated with these factors, those that mediate breast carcinoma prognosis are likely to be the same regardless of whether it is a first or second primary.
Although the availability of prospectively obtained questionnaire data and tumor characteristics on all subjects and the long follow-up period were strengths of this study, as with any epidemiologic study, there were limitations. While this is the largest study of its kind involving direct interviews of 369 patients with almost 20 years of follow-up information, the sample size was relatively small. Nevertheless, there was a lack of information on treatment subsequent to the first course and on the use of chemotherapy or hormonal treatment for the initial primary. All treatment information was derived from SEER public-use records and was limited in detail (these records excluded chemotherapy or hormonal treatment information). We evaluated the effect of radiation therapy, both in terms of sequence and source, and found no substantial relationship to survival. This finding is consistent with the results reported by other studies of unilateral breast carcinoma.29–32 Another related limitation is that SEER began to collect data on the extent of the disease (e.g., size of the tumor, grade, and extent of tumor including lymph node involvement, number of positive lymph nodes, and number of lymph nodes evaluated) in 1988. Therefore, data were only available for about 50% of the second primaries. No data were recorded for the first primaries. The absence of information for these important prognostic indicators is a significant drawback, yet the data on stage and histology were informative. Finally, although metastatic spread to the contralateral breast is an uncommon occurrence33 and the SEER program follows strict guidelines for diagnosing an independent primary, it is possible that a second primary may have been a contralateral metastasis.
We assessed the joint effects of epidemiologic risk factors, tumor characteristics, and treatment on risk of mortality following a second primary breast carcinoma. In general, prognosis was favorable after a second primary breast carcinoma. After adjusting for stage at diagnosis and the interval between primaries, young age, weight, measures of SES, race, and marital status were independent predictors of mortality following a second primary breast carcinoma. Factors that have been associated with incidence of first and second primary breast carcinomas in the past34–40 did not predict mortality in this study. These included family history of breast carcinoma, oral contraceptive use, smoking, hormonal factors, and reproductive and menstrual factors. Considering exposure to most of these risk factors will have occurred before the detection of the first primary, any effect would be mediated through the biologic characteristics of the tumors.41 Etiologic factors (e.g., family history of cancer), which may distinguish between subtypes of breast carcinoma, did not lead to significant differences in survival.
The authors thank Philip J. Landrigan, M.D., for his valuable input and critical reading of the text and Dr. Ira Bleiweiss for his help interpreting the pathology information. They also acknowledge the contributors to the Cancer and Steroid Hormone (CASH) Study. CASH study design and coordination were done at the Division of Reproductive Health, Center for Chronic Disease Prevention and Health Promotion, Centers for Disease Control: principal investigator, George L. Rubin; project director, Phyllis A. Wingo; project associates, Nancy C. Lee, Michele G. Mandel, and Herbert B. Peterson. The following data collection centers and principal investigators were involved: Atlanta, Raymond Greenberg; Connecticut, J. Wister Meigs and W. Douglas Thompson; Detroit, G. Marie Swanson; Iowa, Elaine Smith; New Mexico, Charles Key and Dorothy Pathak; San Francisco, Donald Austin; Seattle, David Thomas; Utah, Joseph Lyon and Dee West. Pathology review principal investigators were Fred Gorstein, Robert McDivett, and Stanley J. Robboy. Project consultants were Lonnie Burnett, Robert Hoover, Peter M. Layde, Howard W. Ory, James J. Schlesselman, David Schottenfeld, Bruce Stadel, Linda Webster, and Colin White. Pathology consultants were Walter Bauer, William Christopherson, Deborah Gersell, Robert Kurman, Allen Paris, and Frank Vellios. The CASH Study was supported by interagency agreement 3-Y01-HD-8-1037 between the Centers for Disease Control and the National Institute of Child Health and Human Development, with additional support from the National Cancer Institute.
- 1American Cancer Society. 2000 facts and figures: selected cancers, breast. http://www.cancer.org/statistics/cff2000/selectedbreastcancers.html#breast. 6-15-0200.
- 6ICD-O. International classification of diseases for onology, 2nd ed. Geneva: World Health Organization, 1990.
- 7Regression models and life-tables. J R Stat Soc. 1972; 34: 187–220..
- 10Survival patterns among younger women with breast cancer: the effects of age, race, stage, and treatment. J Natl Cancer Inst Monogr. 1994; 16: 69–77., .
- 22Breast cancer. In: SchottenfeldD, FraumeniJ. Cancer epidemiology and prevention, 2nd ed. New York: Oxford University Press, 2002: 1022–1039., , , .
- 23Increased availability of serum estorgens in breast cancer: a new hypothesis. In: PikeM, SiiteriP, WelshC. Hormones and breast cancer. Cold Spring Harbor: Cold Spring Harbor Laboratories, 1981: 87–106., .