Younger women who develop breast cancer are hypothesized to have more biologically aggressive cancer and hence poorer survival rates.1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11 This traditionally held view is of great concern, especially in populations with relatively higher incidences of breast cancer in younger females. In Singapore, for example, the peak age-specific incidence is in the 45- to 49-year age group,12, 13 whereas in the Western Caucasian populations, the peak is in the postmenopausal age groups.14, 15
Breast cancer among younger premenopausal women has higher rates of proliferation, lymphovascular invasion and lower levels of estrogen/progesterone receptors than older patients with cancer of the same stage. Among Caucasian females with lymph node-negative breast cancer, those who are negative for estrogen (ER−) and progesterone receptors (PR−) appear to have the worst prognosis, earliest age of onset and worst grade.2, 3, 5, 6, 7, 8, 9, 16 In the large U.S. study on ER and PR status based on the National Cancer Institute's Surveillance, Epidemiology and End Results (SEER) cancer registries, ER−/PR− cancers have a single peak at a younger age as well as higher proportion of advanced-stage cancer.17 These differences in biologic characteristics could account for the poorer prognosis in the younger patients. It has also been suggested that young age serves as a surrogate measure of adverse prognostic profiles.9
However, the issue of poorer prognosis among younger patients is still controversial. Few studies showed no differences in the survival rates between younger and older patients18, 19, 20 and some even reported the best survival is around the 40–49 age group.21, 22 Various authors have argued the controversy may be due to methodologic issues such as inadequate correction for competing causes of mortality and failure to take into account changing trends in treatment modalities.20, 21
In view of this controversy and the prominent younger peak age-specific incidence in Singapore, we conducted a population-based survival analysis of female breast cancer patients to compare the survival patterns by age groups over a 25-year period, taking into account competing causes of mortality.
MATERIAL AND METHODS
Population-based cancer registration began in Singapore in 1968. Details of the registration procedure have already been well documented.23 The national registry is voluntary notification of all cancer cases diagnosed in Singapore by medical practitioners. All pathology laboratories (public and private) also notify the registry whenever they come across specimens denoting cancer. Trained staff review hospital discharge summaries for cancer cases. The stage of the breast cancer is classified as localized cancer, regional spread and distant metastases based on the notification forms. No attempt was made to access degree of localized invasion or number of regional lymph nodes involved.
There were a total of 6,895 female residents diagnosed with breast cancer from 1 January 1968 to 31 December 1992. Of these, there were 98 (1.4%) who were first notified through death certificates only (DCOs) and a further 14 (0.2%) were diagnosed at autopsy; these 112 were excluded from analysis. Follow-up was done passively till 31 December 1997 by matching with the national death register. Those who remained living were confirmed by matching with the electoral register of 1997; of these, 386 (5.6%) were lost to follow-up and excluded from survival analyses. Of the 386 lost to follow-up, 189 (48.9%) were below 50 years of age. In the final data set of 6,397 patients, there were 3,835 deaths.
The life table method was used to calculate survival rates. The observed survival rates were based on deaths from all causes. Competing causes of death were accounted for by computing the relative survival rates, which is the ratio of observed to expected survival. The relative survival probabilities were calculated using Hakulinen's method24 with the expected survival derived from the general Singapore female population similar to the breast cancer patients in terms of age, ethnicity and calendar period of observation. These were based on life tables derived from all-cause mortality in Singapore.25 A regression model for excess hazards based on relative survival was employed to compare the excess hazard ratios by age group and stage with those < 35 as the reference group.26
The 2-, 5- and 10-year observed survival, relative survival and their 95% confidence intervals by age group and stage are presented in Table I. Of the 6,397 cases who were followed up, the 2-, 5- and 10-year relative survival rates (RSRs) were significantly worse among patients aged > 75 (65%, 48% and 39%, respectively) and the best rates were seen among those aged 40–44 (84%, 67% and 56%, respectively; Fig. 1). Patients younger than 35 years old were only slightly worse off than the highest RSR group; the 2-, 5- and 10-year RSRs were 79%, 60% and 50%, respectively.
Table I. Observed and Relative Survival by Stage and Age
Stage information was available for 71.4% of the cases in this study. Those without stage information did not differ in the age distribution from those with known stage information. The RSRs showed a similar pattern among the various age groups and stages, especially for patients with localized cancer (Fig. 2). However, in the group with distant metastases, the RSRs were unstable because of small numbers and the pattern of higher RSRs in the younger patients was less clear.
The highest risk of excess deaths occurred in the ≥ 75 age group (adjusted excess hazard ratio, or HR, = 3.07 for localized cancer, 2.56 for cancer with regional metastases and 1.74 for cancer with distant metastases). For patients below 45 years of age, the risk of excess deaths from breast cancer was fairly similar, especially in patients with localized cancer and cancer with regional metastases (Fig. 3). In patients with localized cancer and those with regional metastases, those in the 35–39 age group had the lowest risk (adjusted excess HR = 0.82 and 0.90, respectively). In patients with distant metastases, those less than 35 years of age had the lowest excess risk of death.
Cancer rates in Singapore have been well documented with the incidence of female breast cancer rising dramatically from 1968–1972 to 1993–1997 at an average of 3.7% increase in age-standardized rate per annum.12, 13 The remarkable rise of breast cancer in Singapore has made it the most common cancer among women; moreover, a unique pattern emerging in the age-specific incidence is the sharp increase in the premenopausal age group, especially those aged 45–49 years. This increase has been attributed to a strong cohort effect27 supporting the role of lifestyle risk factors such as late child-bearing28 and dietary changes.29, 30 Although these are well-documented risk factors for breast cancer incidence, they have not been shown conclusively to determine survival. In this study, we have analyzed clinical outcome of women with the diagnosis of breast cancer in the defined population of Singapore and report that, unlike previous observations, young women with the disease fare as well or better than postmenopausal women. The risk of excess mortality from breast cancer is not higher in women less than 45 years of age.
The strength of this present study lies in the comprehensive nationwide cancer registry with accurate follow-up information through effective electronic record linkages. Singapore is a small city-state where the more than 3 million resident population has a unique personal national registration identification card number. This enables comprehensive record linkage between the cancer registry, the Death Register and the Parliamentary Electoral Register. The Singapore Cancer Registry receives voluntary notifications from all medical practitioners, histopathologic laboratories and hospital discharges. The DCO rate over the study period was less than 5% and more than 80% of all notified cancer cases were histologically confirmed.12, 13 In this cohort of breast cancer patients, there was only 1.4% (98 cases) DCOs. Follow-up using record linkage with the Death Register and the Parliamentary Electoral Register was nearly complete as the Death Register covers all deaths occurring in Singapore and parliamentary election is compulsory for all Singaporeans. Utilizing these tools, the vital status of all citizens could be easily confirmed. The loss to follow-up was only 5.6% and after accounting for DCOs and autopsy-diagnosed cases, information was available for 92.8% of all breast cancer cases diagnosed from 1968 to 1992. This small proportion of loss to follow-up is unlikely to bias the RSR estimates.
Several studies among Caucasian women have shown that breast cancer in young women, especially those less than 40 years old, has a poorer prognosis.1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11 In the SEER's data, for the period 1992 to 1999, those less than 45 years old had a 5-year RSR of 83.1% and the peak 5-year RSR was 88.9% in those aged 65–74.4 It has been argued that detecting early breast cancer in younger women is difficult because of the density of mammary glands, especially during pregnancy and while lactating.19, 31 It has also been reported that patients below the age of 35 tend to be estrogen and progesterone receptor-negative, positive for p53 protein and have a high proliferative status.2, 6, 9, 16, 32 These may be markers of more aggressive cancer that have been shown to be less responsive to current management options.33
In our study, the proportions of localized tumors in the < 35 age group compared to the 40–44 age group (35.4% in the < 35 years and 40.0% in the 40–44 years) were not remarkably different over the 25-year period. This finding suggests that there was not much difference in the early detection of breast cancer in different age groups. This is probably due to the high literacy rate, easy accessibility of health care and the widespread use of public media for health education in a small city-state like Singapore.
Overall, the risk of excess death is lower in patients who are less than 50 years old even after adjustment for ethnicity and calendar year period. These findings are in contrast to other studies.1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11 Moreover, the EUROCARE II collaborative project also found higher 1- and 5-year RSR in those aged below 45.34
Among premenopausal women, those aged < 35 years had a slightly lower RSR, rising to a peak in the perimenopausal period and declining sharply postmenopausally. This observation is similar to an earlier Swedish study.21 Using similar relative survival approach, they found the best relative survival in the 45–49 age group and significantly lower relative survival in those aged 50–59. The authors suggested that as the observed trend coincides with the age at menopause, events associated with menopausal changes might account for the difference. In our study, the best relative survival was in the 40–44 age group and this also supports the hypothesis that menopausal events may contribute to the differences in survival between pre- and postmenopausal patients.
However, the relationship between tumor biology and survival is doubtlessly very complex and other biologic mechanisms and prognostic factors may account for the survival differences among the age groups.9, 35 The effect of adjuvant therapy was also explored in a Danish study.1 They found that the increased risk of dying in younger women was confined to those who did not receive adjuvant chemotherapy.
Singapore is a small city-state with easily accessible oncology services. The treatment for breast cancer in Singapore has evolved over the last 25 years from surgical mastectomy to breast conservation.36 Adjuvant chemotherapy and radiotherapy are easily available through various secondary and tertiary health facilities. However, the use of adjuvant tamoxifen treatment for breast cancer became widespread only in the early 1990s. In our data analyses, the use of calendar-year period in the regression model will to some extent adjust for temporal changes in treatment modalities. The favorable survival outcome in the younger patients may be due to a larger proportion opting for adjuvant therapy, as they are more educated and motivated. Unfortunately, the Singapore Cancer Registry has not comprehensively captured treatment modalities for all patients.
It has also been suggested that weight or body mass index (BMI) may be a prognostic factor for breast cancer survival. Although these are well-established risk factors for breast cancer incidence, their role in the survival of breast cancer is still controversial. Daling et al.37 in their population-based case-control study reported that in younger women with breast cancer, those in the highest quartile of self-reported precancer BMI had a 2.5 times risk of dying compared with those in the lowest quartile. Even after adjustments for various biologic characteristics such as estrogen receptor status and mitotic cell count, those in the highest quartile have a 1.7 times higher risk of dying. Others have argued that obesity is unlikely to be an independent risk factor and may be a surrogate for delayed diagnosis.38 As weight and BMI are very closely related to the various stages of cancer development, it is difficult to assess if it is an independent prognostic marker, especially in case-control studies.
There are some data supporting the hypothesis that mutations in the BRCA1 and -2 genes have an effect on breast cancer survival.39 Small-scale hospital-based studies in Singapore show BRCA1 mutations of around 8% in breast cancer patients with family history of breast or ovarian cancer. There are no population-based data on BRCA mutations or estrogen receptor positivity and other genetic markers such as p53 mutations and c-erbB-2 overexpression are lacking. Perhaps the use of high-throughput cDNA microarray technology in future molecular epidemiologic studies will provide further information on the complex relationship of age and prognosis in female breast cancer patients.
An International Cancer Transfer fellowship awarded to Wen Bo Du by the International Union Against Cancer, Geneva, Switzerland, allowed analysis of the data to be carried out at the International Agency for Research on Cancer, Lyon, France.