Increased Risk for Second Primary Malignancies in Women with Breast Cancer Diagnosed at Young Age: A Population-Based Study in Taiwan

Studies conducted in Western countries have reported excess risks for second primary malignancies after breast cancer. However, there is little documentation of ethnic differences in these excess risks. Asian women are characterized by younger age at diagnosis of breast cancer, but very few reports are available on the incidences and risks for second primary cancers in this region. Using population-based data from the Taiwan National Cancer Registry (TNCR) for the period 1979 to 2003, we quantified standardized incidence ratios and cumulative incidence of second cancers among 53,783 women with initial diagnoses of breast cancer. Age-specific incidences showed peaks among women in their 40s, and 1,085 cases (2.02%) developed nonbreast second primary cancers. The risk for second cancers differs significantly according to age at diagnosis of breast cancer. In comparison with women diagnosed when ≥50 years (standardized incidence ratio, 0.96; 95% confidence interval; 0.89-1.04), there were significantly greater risks for bone, corpus uterine, ovarian, thyroid, esophageal, kidney and lung cancers, nonmelanoma skin cancer, and leukemia or lymphoma in women diagnosed when <50 years (standardized incidence ratio, 1.43; 95% confidence interval, 1.29-1.58). The survival probabilities differed between breast cancer patients with and without second cancers (P < 0.001). After diagnosis of the second cancer, the median survival time was only 2.87 years. In conclusion, we confirmed that young age at diagnosis of breast cancer predicted a subsequently increased risk for second malignancies, and the second cancers indeed worsen survivorship in patients who survived breast cancer. (Cancer Epidemiol Biomarkers Prev 2008;17(10):2647–55)

The incidence of breast cancer has increased among women in most Asian countries (1-3). In Taiwan, breast cancer has become the leading cancer in females, with age-adjusted incidence increasing from 11.72 per 100,000 women in 1980 to 45.89 per 100,000 in 2003 (4). Concurrently, the survival time of breast cancer patients has increased; the five year survival rate among women in Taiwan diagnosed during the period 1991 to 1997 was 81% (5, 6). Second primary malignancies following breast cancer are becoming an issue of concern to clinicians and to patients and their families.

Intriguingly, during the past two decades, the median age at first diagnosis of female breast cancer has remained at 45 to 49 years in Asian countries such as Taiwan (4, 6), China (6), Korea (7), and Japan (8, 9), which is at least 10 years younger than their counterparts in Western countries and Australia (>60 years; refs. 10-12). There is no satisfactory explanation of this disparity in the age of onset. The age-specific breast cancer incidence rate also differed between African-American and White women of similar ages in the United States (13, 14), indicating that, in addition to lifestyle and hormone use, other biological factors could contribute to racial or ethnic differences in clinical presentation among breast cancer patients.

A 20% to 30% excess risk for a second nonbreast primary malignancy in female breast cancer patients has been reported, mostly in Europe (10, 15-18) and the United States (19-21). The relevance of ethnic diversity to such second primary cancers has been much less well described, particularly in Asian countries, so we launched a large-scale epidemiologic study in Taiwan, an area characterized by early age of breast cancer onset (22, 23). We aimed to define the incidences and patterns of second primary cancers in patients with breast cancer according to age at onset and time interval since breast cancer diagnosis and to explore the effect of developing second cancers on the survival of patients who survived breast cancer.

We quantified second cancer incidences among 59,001 patients with initial diagnoses of breast cancer (International Classification of Diseases, 9th Revision, 174) who were reported to the TNCR⁷

from January 1, 1979, to 31 December 2003. The TNCR was founded in 1979 and financially supported by the National Department of Health with the aim of estimating cancer incidence in Taiwan. It is a population-based cancer registry that covered 22 million people in 2003. Hospitals with >50 beds are obligated to submit information on newly diagnosed cancer patients to the TNCR, which reimburses the hospitals on the basis of numbers of cases reported to reduce the likelihood of underreporting. All cancer registry databases in the TNCR have been systemically converted to International Classification of Diseases, 9th Revision, codes (24), and linked with death certificates from the National Death Database. Persons not identified by this process were therefore considered to be alive for the purpose of the current study (passive follow-up). Coding of multiple primaries followed the principles of the International Agency of Cancer Registries and the IARC (25). The lymphoma and leukemia incidence rates were combined because they are both classified in the TNCR as cancers of the hematopoietic and reticuloendothelial systems. Informed consent was not required because all registry records are anonymous and open to the public.

To assess the age of onset accurately, estimate person-year follow-up and minimize potentially unconfirmed cancer diagnosis in this study cohort, 5,218 patients were excluded from analysis because they met one or more of the following criteria: (a) missing birth dates (519 cases), (b) missing follow-up date or death status (3,955 cases), or (c) second cancer diagnosis or death occurring <1 month after breast cancer diagnosis (798 cases). Subsequent breast cancers (75 cases) were also excluded because they may have represented a progression of the initial tumor. Second breast cancers could be underreported partly because IARC or International Agency of Cancer Registries rules do not permit coding of more than one tumor in paired organs if the histology is identical. Thus, a total of 53,783 patients were included in the analysis. For subsequent analyses, these cases were stratified by the age at diagnosis of breast cancer (<50 or ≥50). The cutoff age of 50 years was chosen because 48.2% of female breast cancer cases diagnosed in 2005 were in patients <50 years old (Annual Report of Taiwan Cancer Data Base, published by Department of Health in 2007).

To quantify the excess of second malignancies after diagnosis of breast cancer, we calculated the standardized incidence ratios (26) and the corresponding 95% confidence intervals (95% CI) for all types of second primary cancers except breast. Standardized incidence ratios were taken as the ratio of the observed number of second cancers to the expected number, which was obtained by assuming that these persons experienced the same cancer incidence as the corresponding general female population. The number of person-years at risk was defined as the number of years from the date of breast cancer diagnosis to the date of death, date of diagnosis of second primary cancer, or the end of the study period (December 31, 2003), whichever came first. The person-years of observation for each 5-year age group, 5-year period (1979-1983, 1984-1988, 1989-1993, 1994-1998, 1999-2003), and time since entry to the cohort (≤5, 6-10, or >10 years) were multiplied by the incidence rates of cancers for the Taiwanese female population. The corresponding products were summed over all ages and calendar years to yield the expected numbers of second cancer at each site. Confidence intervals of standardized incidence ratios were based on the assumption of a Poisson distribution of second cancer cases.

Cumulative incidence rates for occurrence of second cancers were calculated in the survivors' cohort, with death treated as a competing risk according to the method described by Kalbfleisch and Prentice (2002; ref. 27). Briefly, this method allows for the fact that patients who die are no longer at risk for second cancers, so it differs from the cumulative incidence estimated by the Kaplan-Meier method, which treats competing events as censored at the time they occurred. Gray's test (28) was used to assess the statistical significance of differences in cumulative incidence between the two onset-age strata. The survival curves of breast cancer patients diagnosed at the age of <50 and ≥50 years were calculated by the Kaplan-Meier method and the differences between these two age groups were examined by a log-rank test. A Cox model with a time-dependent covariate (29), allocating follow-up time for each patient to the nonsecond cancer group until second cancer occurrence, was used to compare the survival between patients with and without a second cancer. All statistical tests were two-sided, and P < 0.05 was considered statistically significant.

Of the total 53,783 women with initial diagnoses of breast cancer (28,528 cases diagnosed at the age of <50 years and 25,255 at the age of ≥50 years) and complete data available for study, 1,085 cases (2.02%) developed second primary cancers during 290,966 person-years of follow-up (Table 1). Overall, breast cancer was diagnosed at mean the age of 50.35 years (median, 49 years). The mean age at diagnosis of second primaries was 60.56 years. The average follow-up time was 5.41 years, including 46,709 cases (87%) followed up for at least 1 year, 13,140 (24%) for 5 to 10 years, and 9,024 (17%) for >10 years. The average interval between first and second cancers was 5.99 years with an SD of 4.81 years.

Standardized incidence ratios and corresponding 95% CIs for second primary cancers at all sites except breast were calculated and stratified by the age at diagnosis of breast cancer (<50 and ≥50 years; Table 2). Irrespective of site, the overall standardized incidence ratio for developing a subsequent nonbreast second cancer was 1.09 (95% CI, 1.03-1.16). The standardized incidence ratio did not increase for patients with the onset age of ≥50 years (standardized incidence ratio, 0.96; 95% CI, 0.89-1.04) but significantly increased for patients with the onset age of <50 years (standardized incidence ratio, 1.43; 95% CI, 1.29-1.58).

For all patients, risks for the following second cancers showed statistically significant increases over the population: bone (standardized incidence ratio, 4.55; 95% CI, 2.70-8.63), corpus uteri (standardized incidence ratio, 2.75; 95% CI, 2.19-3.40), ovary (standardized incidence ratio, 2.09; 95% CI, 1.62-2.65), skin (nonmelanoma; standardized incidence ratio, 1.52; 95% CI, 1.15-1.96), and thyroid (standardized incidence ratio, 1.42; 95% CI, 1.04-1.90). For patients <50 years old at the time of diagnosis of breast cancer, the following 5 second cancers rates remained significantly higher than in the general population: bone (standardized incidence ratio, 7.69; 95% CI, 2.81-16.74), corpus uteri (standardized incidence ratio, 1.91; 95% CI, 1.27-2.76), ovary (standardized incidence ratio, 2.61; 95% CI, 1.85-3.56), skin (nonmelanoma; standardized incidence ratio, 3.56; 95% CI, 2.30-5.25), and thyroid (standardized incidence ratio, 1.87; 95% CI, 1.28-2.64). There were significant excess risks for 4 additional second cancers: esophagus (standardized incidence ratio, 4.82; 95% CI, 1.3-12.34), kidney (standardized incidence ratio, 2.11; 95% CI, 1.12-3.61), lung (standardized incidence ratio, 1.93; 95% CI, 1.40-2.59), and leukemia and lymphoma (standardized incidence ratio, 2.17; 95% CI, 1.18-3.64). For patients diagnosed at the age of ≥50 years, the risk increased only for corpus uteri (standardized incidence ratio, 3.52; 95% CI, 2.66-4.57) and ovary (standardized incidence ratio, 1.63; 95% CI, 1.08-2.36). In general, the standardized incidence ratio was significantly decreased for cervical cancer (standardized incidence ratio, 0.71; 95% CI, 0.59-0.86), suggesting that it occurred less frequently than expected. Although the absolute numbers of second primary cancers in stomach (n = 75), colon and rectum (n = 151), and liver (n = 109) were large, the risks for these did not exceed those in the general population.

The standardized incidence ratio estimates were stratified by interval since first diagnosis of breast cancer (Table 3). Nine selected sites with excess risk were evaluated in patients with breast cancer diagnosis at the age of <50 years and two sites in those diagnosed at the age of ≥50 years. The entire follow-up period was divided into consecutive 5-year intervals. A higher risk for developing a second cancer was seen in the first 5 years after breast cancer diagnosis, but the risk decreased as time elapsed. For those with initial diagnosis at the age of <50 years, risks for bone, corpus uteri, thyroid cancers, and leukemia or lymphoma increased only in the first 5 years. Risks for lung, nonmelanoma skin, and ovarian cancers were significantly elevated during the first 10 years. Risks for esophagus and kidney cancers were not significant at any follow-up period, perhaps because only limited numbers were observed. For those with initial diagnosis at the age of ≥50 years, risks for corpus uteri and ovarian cancers were significant during the first 10 years.

To study the trend of two most related second cancers (corpus uteri and ovary) with age, we stratified their standardized incidence ratios according to 4 age groups (<40, 40-49, 50-59, and ≥60 years) at breast cancer diagnosis (Table 4). Interestingly, these two cancers clearly exhibited a trend with age. The risk for corpus uteri cancer was significantly elevated with increasing age at diagnosis (standardized incidence ratio, 0.93, 2.19, 2.41, and 5.34), whereas the risk for ovarian cancer was greater with younger age at diagnosis (standardized incidence ratio, 2.76, 2.55, 1.95, and 1.30).

The estimated risk of developing a second cancer in the breast cancer survivors' cohort was calculated for women who had been diagnosed with breast cancer at the age of <50 or ≥50 years, with death treated as a competing risk. The cumulative incidences and the confidence intervals at 5, 10, 15, 20, and 25 years are shown (Fig. 1). There was no risk plateau and the cumulative incidence rates over time in patients diagnosed <50 years of age were lower than in patients diagnosed later (P < 0.001).

The median overall survival was 19.92 ± 0.22 years for the cohort (Fig. 2). Patients diagnosed primary breast cancer at the age of <50 years had more favorable survival than patients diagnosed later (21.45 ± 0.25 versus 15.10 ± 0.32 years; P < 0.001). According to the Cox model with a time-dependent covariate, the patients with occurrence of second cancers seemed to be associated with a significant increased risk for death compared with those without second cancers (Table 5). The hazard ratio was 2.67 for all breast cancer patients (P < 0.001), 2.66 for patients with the age of <50 years at diagnosis (P < 0.001), and 2.44 for patients with the age of ≥50 years at diagnosis (P < 0.001). The result suggests that second cancer occurrence was associated with worse survival.

The overall survival time for 1,085 patients after diagnosis of second cancers was calculated by the Kaplan-Meier method (Fig. 3). The median survival was 2.87 years (95% CI, 2.40-3.35). The 1-, 5-, 10-, and 15-year survival rates were 71%, 40%, 30%, and 23%, respectively.

Breast cancer is currently regarded as a heterogeneous disease. Numerous studies have shown racial or ethnic disparities in incidence rate, age of onset, survival rate, and tumor characteristics, including hormone receptor status, histology, and stage (30, 31). However, possible ethnic differences in trends toward the development of second primary cancers are less well documented. It is noted that breast cancer in Chinese women is characterized by tumor onset at a younger age. In our cohort over the past 25 years in Taiwan, 53% of breast cancers were diagnosed in women <50 years. In the United States, only 23% of breast cancers were diagnosed at the age of <50 years (all races combined), according to data from the Surveillance, Epidemiology, and End Results Program (32). Despite a 4-fold increase in breast cancer incidence from 1979 to 2003 in Taiwan, 48.2% of female breast cancer cases diagnosed in 2005 remained under the age of 50 years, indicating that the preponderance of breast cancer in young females has not changed as lifestyle changes. To explain this early age of tumor onset, genetic polymorphisms (23, 33-36) and the association between lipid profiles and breast cancer (37) have been rigorously studied; however, the etiologic factors remain obscure. Exploring the pathogenesis of second primary cancers might provide some insight into breast cancer etiologies in this region.

We found that the second cancer risk is substantially greater among female patients with initial diagnosis of breast cancer at the age of <50 years. The risk for cancer at all sites was 43% higher than in the general population (standardized incidence ratio, 1.43; 95% CI, 1.29-1.58), of which bone, corpus uteri, ovary, skin (nonmelanoma), thyroid, esophagus, kidney and lung cancers, and leukemia or lymphoma all significantly increased. Patients diagnosed with breast cancer when they were >50 years old showed excess risks only at two sites, corpus uteri and ovary. The mean follow-up of our cohort was relatively short (5.41 years); the magnitude of risk (as measured by standardized incidence ratio and cumulative incidence) might increase with longer follow-up. During the past 25 years, the clinical practice and breast cancer treatment, including surgery, chemotherapy, tamoxifen, and radiation therapy were similar to Western countries. Although the trend toward that the risk for second malignancies was higher among younger patients was consistent with Western patients (17, 38, 39), various cancer sites at excess risks have been less consistent except for ovary and corpus uteri. For example, bone, skin (nonmelanoma), thyroid, esophagus, kidney, and lung cancers, and leukemia or lymphoma were the concordant sites as compared with Western studies (38, 40); however, the excess risks for soft tissue sarcoma, salivary gland, stomach, colon, and rectum cancer from previous reports (10, 16, 29, 41) were not observed in our cohort. Therefore, second cancer screening for Western patients cannot be uniformly applied to Asian population, probably affected by different genetic or lifestyle factor.

It is always a concern in population-based studies about misclassification of metastases in some anatomic sites such as bone, liver, and lung as new primary cancers. Among 53,783 of breast cancer patients in our cohort, there were only 10 bone cancers, 109 liver cancers (Taiwan is an endemic area for hepatocellular carcinoma), and 122 lung cancers. Considering the incidences of primary breast cancer metastasizing to these distant sites should be far higher than these numbers, we speculate that possible misclassification of metastases as new second primaries does not seem to have been a particular problem in TNCR. However, subsequent breast cancers were excluded because the data of the registry did not specify whether the second breast cancer occurred ipsilaterally or contralaterally to the first primary. It is difficult to discriminate whether they are truly second primary breast cancers or recurrences if the histology is identical. Our cohort assesses only the risk for secondary nonbreast cancers.

Many cancer sites are likely to be associated with treatment modalities for breast cancer. For example, organs close to the breast such as esophagus (42), lung (43), thyroid (44), and bone (45) are likely to be affected by radiotherapy. However, our data do not support the hypothesis that the excess cancer rates at these sites are therapy induced because their risks did not increase by latency. Moreover, the magnitude of risk at these sites was very low in breast cancer patients diagnosed at the age of >50 years, suggesting that other factors such as shared genetic predisposition or shared environmental risk factors could be involved. The differential risks between patients <50 and ≥50 years also exclude the possibility that the observed excesses of second cancers are due to bias caused by increased medical surveillance during follow-up after breast cancer.

A significantly increased risk for cancers of the uterine corpus and ovary, and a decreased risk for cervical cancer were observed in our cohort, as well as many studies conducted in Western countries (10, 15, 16, 38, 41). Tamoxifen (46) and shared reproductive risks such as early menarche, late menopause, and nulliparity probably contribute to the excess of endometrial cancer, the risk for which was higher among women ≥50 years at breast cancer diagnosis (standardized incidence ratio, 3.52; 95% CI, 2.66-4.57) than those <50 years (standardized incidence ratio, 1.91; 95% CI, 1.27-2.76). Second primary ovarian cancers showed the opposite sequence in which young patients were at higher risk (standardized incidence ratio, 2.61; 95% CI, 1.85-3.56) than the older ones (standardized incidence ratio, 1.63; 95% CI, 1.08-2.36), suggesting that genetic factors such as BRCA-1 and BRCA-2 (47) may play a role. The age-trend analysis showed that the risk for cancer of the uterus corpus increased as the age at breast cancer diagnosis increased, whereas the risk for ovarian cancer had an inverse trend. An analogous trend with age was also described in other studies such as a cohort of 525,527 breast cancer patients from cancer registries in Europe, Australia, Canada, and Singapore (10), and the other cohort of 376,825 breast cancer patients in 4 Scandinavian cancer registries (39). Age trend should be taken into account for gynecologic surveillance in patients with breast cancer. The decreasing trend in the incidence of cervical cancer could be explained by the difference in background incidence because cervical cancer is usually associated with lower socioeconomic status and human papilloma virus.

A major finding of our study is that second primary cancers indeed have a negative impact on survival in breast cancer patients (hazard ratio, 2.67). Notably, the elevated risk of death was not changed using a dichotomous age at diagnosis (hazard ratio, 2.66 for ages <50 years and 2.44 for ages ≥50 years). An increased hazard of death (hazard ratio, 3.98; 95% CI, 3.77-4.20) in breast cancer patients with second cancers was also reported by the Dutch population-based study (29).

In conclusion, we have confirmed that young age at diagnosis predicts for an increased risk for second malignancies after breast cancer. The findings that second cancers indeed worsened survivorship in cancer survivors will justify a more comprehensive research task in the epidemiology, screening, and chemoprevention of second primary cancers (48).

No potential conflicts of interest were disclosed.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

We thank the staffs in the TNCR, Department of Health, Taiwan, R.O.C., and the reviewers for improving this paper.