Presented at the 43rd Annual Meeting of the American Society of Clinical Oncology (ASCO), Chicago, Illinois, June 2-6, 2007.
Radiotherapy (RT) has been a risk factor for development of soft tissue sarcomas (STS). The objective of the current study was to quantify the risk of STS after RT and surgery for breast cancer (BCa), assess time trends, and compare long-term survival of patients with RT-associated and non–RT-associated angiosarcoma (AS) using the Surveillance, Epidemiology, and End Results (SEER) database.
Women with BCa reported to SEER in 1973 to 2003 were included. Kaplan-Meier curves and proportional hazards models, reported as hazards ratios (HR) with 95% confidence intervals (95% CI), were used. Survival of patients who developed RT-associated AS was compared with that of patients with primary AS of the thorax and upper extremities.
The cohort of 563,155 BCa patients was divided into 2 groups: those who received RT (37%) and those who received no RT. RT use increased with time (P <. 0001). A total of 948 patients developed STS 1 to 29 years after BCa diagnosis. RT patients had a higher incidence of all STS (31 vs 22 per 100,000 person-years; HR, 1.5 [95%CI, 1.3-1.8]), AS (HR, 7.6; 95% CI, 4.9-11.9), and malignant fibrous histiocytomas (MFH) (HR, 2.5; 95% CI, 1.6-3.9). RT remained a significant predictor after adjustment for covariates (HR, 1.4; 95% CI, 1.2-1.7). Partial mastectomies (HR, 7.1; 95% CI, 3.2-16), RT (HR, 2.2; 95% CI, 1.1-4.3), and lymph node dissections (HR, 2.6; 95% CI, 1.3-5) were found to be independent risk factors for AS. The hazard of STS after RT peaked at 10 years, reaching the non-RT hazard at approximately 23 years. The 5-year survival for STS was 38%. There was no difference in survival noted between RT-induced and non–RT-induced AS.
Several studies have demonstrated that radiotherapy (RT) is a significant risk factor for the development of soft tissue sarcomas (STS), and in particular angiosarcomas (AS), after treatment for breast cancer.1-3 Due to the rarity of STS, most single-institution, retrospective studies include only a small number of patients, thereby making it difficult to accurately quantify this risk. To our knowledge, 2 studies to date have used earlier versions of the Surveillance, Epidemiology, and End Results (SEER) database of the National Cancer Institute to define the risk for development of STS in women previously treated for breast cancer.4, 5 However, the majority of patients in these studies were treated for breast cancer before 1996, when breast conservation therapy (BCT) using partial mastectomy and adjuvant RT was not as prevalent as it is now. Specifically, previous studies were unable to quantify the time-dependent hazard for developing secondary STS beyond 15 years after RT due to limited follow-up. Furthermore, to our knowledge, the studies did not determine the effect of the type of surgery on the subsequent development of STS.
AS are the most common secondary STS encountered after RT.4 In general, AS are aggressive tumors that portend poor survival.6 However, it is still controversial whether the clinical course and survival of RT-associated breast AS (secondary AS) are comparable, better, or worse than that for non–RT-associated breast AS (primary AS). To the best of our knowledge, no large study to date has been able to address this issue.
In the current study, we queried the large multi-institutional SEER database to 1) quantify the risk of developing secondary STS after RT and surgery for breast cancer, 2) assess time trends for the development of STS after a diagnosis of breast cancer, and 3) compare the long-term survival of women with secondary breast AS with that of women with primary breast AS.
MATERIALS AND METHODS
This study is based on the SEER database of the National Cancer Institute. The SEER database is a national, multi-institutional, community-based cancer registry containing clinical and pathologic information collected from 18 population-based cancer registries in the United States (Alaska Native, Arizona Indians, Los Angeles, San Francisco-Oakland, San Jose-Monterey, Greater California, Connecticut, Detroit, Atlanta, Rural Georgia, Hawaii, Iowa, Kentucky, Louisiana, New Jersey, New Mexico, Seattle-Puget Sound, and Utah). The database covers approximately 26% of the population in the United States.7
The base cohort for this study comprised all women with microscopically confirmed invasive or in situ breast adenocarcinoma (International Classification of Diseases for Oncology [ICD-O]8 codes 805-854 at sites C500-506 or C508-509) included in the SEER database from 1973 to 2003 (n = 675,705 women).9 Cases with a diagnosis based on autopsy reports or death certificates (n = 201), those with prior malignancies (n = 97,302), and those with no information regarding the use of RT for the treatment of their breast cancer (n = 14,911) were excluded from the study.
The development of subsequent STS in the cohort was identified by querying the SEER database for all microscopically confirmed tumors with ICD-O8 codes 880 through 958 that occurred in women in the cohort during the study period (1973-2003). Women with a diagnosis of STS within 1 year of diagnosis of their breast cancer (n = 136) were excluded from further analyses due to the possibility that their malignancies were synchronous rather than metachronous. Sarcomas were classified according to ICD-O codes into several groups: fibrosarcomas (881.0-881.5), malignant fibrous histiocytomas (MFH) (883.0), dermatofibrosarcomas (883.2-883.3), liposarcomas (885.0-885.8), leiomyosarcomas (889.0, 889.1, and 889.6), AS (912.0 and 917.0), chondrosarcomas (922.0-924.3), endometrial stromal sarcomas (893.0 and 893.1), müllerian mixed tumors (895.0 and 895.1), carcinosarcomas (898.0 and 898.1), phyllodes tumors (902.0), osteosarcomas (918.0-919.4), not specified (880), and others (884.0, 889.4, 890.0-1, 893.3-6, 894.0, 896.4, 901.5, 904.0-4, 914.0, 915.0, 925.0, 926.0, 954.0, 956.0, and 958.0).
The cohort was divided in 2 groups (RT and no RT) based on whether the women received RT for treatment of their initial breast malignancy. The risk for the development of all types of secondary STS in each group was analyzed with Kaplan-Meier curves, log-rank tests, and proportional hazards models, including relevant covariates. Results were expressed in terms of incidences and hazards ratios (HR) with 95% confidence intervals (95% CI). The risk over time of developing STS after breast carcinoma was assessed by plotting the hazards function for each of the groups. Descriptive statistics were performed with the use of analysis of variance (ANOVA) or chi-square tests, as appropriate. A P value ≤ .05 was considered statistically significant.
Results were stratified by location of STS into in-field and out-of-field. In-field STS were defined as those occurring at sites that would be included in the RT field, namely the thorax and ipsilateral upper extremity. The ipsilateral upper extremity was included in the in-field group to include shoulder and upper arm lesions that would be in close proximity to the RT field. Out-of-field STS were those occurring at all other sites, including the contralateral upper extremity.
To evaluate the specific survival difference between secondary and primary AS, a new database was created including all women with a microscopically confirmed diagnosis of AS of the thorax, upper extremities, and all other sites in the SEER database during the study period. Histories of breast cancer and RT for breast cancer were included by querying the main study cohort. Cases of AS diagnosed within 1 year of diagnosis of breast cancer were considered unrelated to the breast cancer or RT. Survival rates were compared between patients with AS of the thorax and ipsilateral upper extremity and a history of breast cancer and RT (secondary AS), and those with AS of the thorax and upper extremities and either no history of breast cancer or a history of breast cancer but no RT (primary AS). Patients with a history of breast cancer and RT but with an AS in the contralateral upper extremity were included in the primary AS group. Kaplan-Meier curves and log-rank tests were used to compare survival rates between patients with primary and secondary AS.
Description of the Cohort
The study included 563,155 women with a mean age of 60 ± 14 years with a diagnosis of invasive or in situ breast cancer. Of these, 211,027 (37%) patients underwent RT for treatment of their breast malignancy. The characteristics of the cohort are shown in Table 1. Overall, the RT group had younger patients and underwent more limited surgical resections than the non-RT group.
Table 1. Characteristics of the Breast Cancer Cohort
RT (n = 211,027) No.
No RT (n = 352,128) No.
RT indicates radiotherapy; SD, standard deviation; AJCC, American Joint Committee on Cancer 6th edition grading system; LN, lymph node.
Numbers in parentheses indicate the population without missing values on which the analyses were performed.
Partial or subcutaneous mastectomies.
Year of diagnosis was grouped in 5-year intervals except for the last 2 groups (1998-2000 and 2001-2003), which encompass 3-year intervals each.
Among the 480,504 patients who underwent curative surgery (partial, total, or radical mastectomy) for invasive breast carcinoma, the use of partial mastectomies as the initial treatment increased from 30% between 1973 and 1977 to 57% between 2001 and 2003 (P <. 0001) (Fig. 1). During the same period, the use of RT increased from 23% to 50% (P <. 0001).
Risk of Secondary Sarcoma
During the study period, 948 patients from the cohort developed STS (at any site) within a median latent period of 7 years after the diagnosis of breast cancer (range, 1 year-29 years). The use of RT for treatment of breast cancer was associated with an increased risk for the development of STS (31 cases vs 22 cases per 100,000 person-years) (HR, 1.54; 95% CI, 1.3-1.8) (Table 2) (Fig. 2). This increased risk was particularly evident for AS (HR, 7.63; 95% CI, 4.9-11.9) and MFH (HR, 2.46; 95% CI, 1.6-3.9). No correlation was found between the use of RT and the development of fibrosarcomas or leiomyosarcomas.
Table 2. Incidence and Hazards Ratios for the Development of Sarcoma After Breast Cancer
The specific location of STS was known in 884 (93%) patients who developed STS. Of these, 235 (26%) patients had STS involving the thorax or the ipsilateral upper extremity (in-field) in relation to their primary breast cancer. RT increased the risk of developing a secondary in-field STS for all histologies (HR, 4.1; 95% CI, 3.2-5.4), AS (HR, 8.97; 95% CI, 5.5-14.6), MFH (HR, 4.99; 95% CI, 2.2-11.4), and other STS (HR, 2.09; 95% CI, 1.4-3.2) (Table 2) (Fig. 2). RT did not increase the risk for development of STS in areas outside of the RT field.
On univariate analysis, the use of partial mastectomies for the treatment of breast cancer was associated with an increased risk for the development of secondary STS at all sites (HR, 1.3; 95% CI, 1.1-1.5) and at in-field sites (HR, 2.94; 95% CI, 2.2-4.0) when compared with total or radical mastectomies. No effect was found at out-of-field sites. The use of lymph node dissections was not associated with the development of STS. On multivariate analysis, age, use of RT, and type of surgery (for in-field tumors) were found to be independent risk factors for the development of STS after breast cancer (Table 3).
Table 3. Multivariate Analysis for the Development of Sarcoma After Breast Cancer
The use of partial mastectomies significantly increased the risk of developing AS at all sites (HR, 10.27; 95% CI, 5.5-19.3) and at in-field sites (HR, 10.86; 95% CI, 5.6-21.0). The use of lymph node dissections was not found to be associated with the development of AS on univariate analyses. However, on multivariate analyses, RT (HR, 2.23; 95% CI, 1.1-4.3), partial mastectomies (HR, 7.12; 95% CI, 3.2-16.0), and lymph node dissections (HR, 2.57; 95% CI, 1.3-5.0) were all found to be independent risk factors for the development of AS at all sites. Similar results were found for in-field AS. None of these risk factors were associated with the development of out-of-field AS.
The hazard for the development of secondary STS (at all sites) after RT increased in the third year after diagnosis of breast cancer, peaked at 8 to 12 years, and then decreased again, approximating the hazard of the non-RT group at approximately 23 years (Fig. 3).
The median overall survival (OS) after the diagnosis of secondary STS (at all sites) was 31 months, with a 5-year OS rate of 38%.
During the study period, 476 women in the SEER database developed AS of the thorax and upper extremities (regardless of history of breast cancer). Of these, 77 (16%) patients developed secondary AS of the thorax or ipsilateral upper extremity after undergoing RT for the treatment of breast cancer. The OS after diagnosis of secondary AS (median, 35 months; 5-year OS, 38%) was not found to be significantly different from that after the diagnosis of primary AS (median, 48 months; 5-year OS, 46%) of the thorax and arms (P = .81) (Fig. 4). However, the OS of the 476 patients with AS of the thorax and arms in the SEER database (primary and secondary) (median, 46 months; 5-year OS, 45%) was significantly better than the survival of the 458 patients with AS in other locations (median, 10 months; 5-year OS, 22%) (P <. 0001).
The results of the current study demonstrate that RT increases the risk for the development of all types of STS by 1.5-fold and specifically of AS by 9-fold in women treated for breast cancer. The incidence of treatment-associated STS has increased over time, likely due to the increasing use of RT for the treatment of breast cancer. The hazard of developing secondary STS after RT peaks at approximately 10 years and remains elevated up to 20 years after RT. Survival of patients with secondary AS is not significantly different from that of patients with primary AS of the thorax and upper extremities.
The results of this study are in agreement with prior reports based on earlier versions of the SEER database and other cancer populations. In an analysis of 194,798 women with breast cancer in the SEER database (1973-1995), Huang and Mackillop4 found that women treated for breast cancer had a higher incidence of STS and AS than the overall population, regardless of whether they received RT. Patients who received RT had an even higher incidence of all sarcomas and AS. The risk peaked between 5 and 10 years after treatment of breast cancer.
In a similar study, Yap et al5 analyzed 274,572 patients with invasive breast cancer in the SEER database (1973-1997) and found that the cumulative incidence of secondary STS at 15 years was 3.2 per 1000 patients after RT, which was significantly higher than that in patients with no RT (2.3 per 1000 patients). The most common histology for patients who underwent RT was AS; >50% of secondary STS in this group were AS, compared with only 6% among patients who did not receive RT. These results are also supported by several smaller studies, including case series and national registries.1, 3, 10
Despite the significantly higher risk of STS after RT in patients with breast cancer, the absolute incidence of these neoplasms is low. In the current study, the incidence of all secondary STS and AS in patients who received RT was only 31 and 7, respectively, per 100,000 person-years. This low risk does not outweigh the known benefits of RT in the management of breast cancer. In a meta-analysis of RT for breast cancer conducted by the Early Breast Cancer Trialists' Collaborative Group,11 RT led to significant reductions in local recurrence and 15-year mortality among patients with breast cancer, mainly those with BCT and with positive lymph nodes after mastectomy. Women who received RT had a higher incidence of contralateral breast cancer (HR, 1.2), lung cancer (HR, 1.6), esophageal cancer (HR, 2.1), leukemia (HR, 1.7), and STS (HR, 2.3). However, even after taking into account all these increased risks, the use of RT translated into a definite reduction in 15-year overall mortality among women with breast cancer.
The specific pathophysiology of secondary sarcomas after treatment of breast cancer is not entirely clear. Lymphedema has been recognized as a major risk factor for the development of AS since Stewart and Treves reported the development of cutaneous AS after mastectomy.12 The results of the current study and others have demonstrated that prior treatment of breast cancer is associated with a higher incidence of secondary malignancies, in particular STS,13, 14 and that RT is a major risk factor.15 However, to our knowledge, it is still unknown whether the effect of RT is mediated solely by direct tissue damage or by the development of lymphedema as a sequela of obstructed lymphatic channels from RT, from surgery, or from a combination of both.16 In an analysis of the Swedish Cancer Registry, Karlsson et al1 found that the amount of radiation energy given for breast cancer correlated with the risk of developing non-AS secondary STS, whereas upper extremity edema was the only risk factor related to the development of AS. In another study analyzing >6500 women with breast cancer, the risk of developing STS, including MFH, fibrosarcomas, and AS, increased with increasing radiation dose, regardless of edema.2
BCT has been associated with the development of secondary STS, in particular AS.17 This effect is believed to be directly related to RT. To the best of our knowledge, no prior studies have assessed the effect of partial mastectomies independent of RT. In the current study, patients who underwent partial mastectomies had a 7-fold higher risk of developing secondary AS when compared with those who underwent mastectomies, after adjusting for RT and lymph node dissection. It is difficult to understand why AS incidence would be higher after partial mastectomy than total mastectomy, independent of lymph node dissection or RT use. It is possible that disruption of the lymphatic channels in the breast while leaving breast tissue behind may cause subclinical lymphedema that could aggravate the effect of RT (and lymph node dissection) on the development of AS. Another explanation for this result could be that SEER data involve under-reporting of RT utilization during BCT.
The time interval between the diagnosis of breast cancer and the development of secondary STS is variable and has been reported to be between 2 to 30 years.2, 3, 5, 6, 10, 17-19 In the current study, the risk significantly increased 3 years after a diagnosis of breast cancer, peaked at approximately 10 years, and then decreased to approximately the non-RT risk after 20 years from breast cancer diagnosis. Therefore, close follow-up of women who receive RT for the treatment of breast cancer should be continued for >20 years after diagnosis. Whether the interval to peak incidence will extend beyond 10 years with even longer follow-up is unknown. Although patients should be followed closely, the relative rarity of secondary STS makes it hard to justify a regular schedule of sarcoma-specific surveillance imaging.
It has been suggested that the OS rates of women diagnosed with secondary AS may be different from that of primary lesions.20 In the current study, patients with secondary AS had a median and 5-year OS of 35 months and 38%, respectively, which is not significantly different from that of patients with primary AS. In a recent study including 55 patients with breast AS, Vorburger et al6 found that, even though radiation-naive patients presented more frequently with metastases, they had a slightly more favorable OS over the first 3 years after diagnosis, but the OS curves were not significantly different. The median OS for all patients was approximately 3 years. Based on this information and the findings of the current study, it appears that, despite the possible biologic and anatomic differences that may exist between primary and secondary AS, the response to therapy, as defined by OS, is equivalent.
Due to its methodology and structure, the current study has several limitations. The SEER database does not provide specific information regarding RT such as dose of radiation or the size of the radiation fields. This makes causality and the determination of dose-related effects impossible to ascertain. Furthermore, there is no information regarding clinical factors such as the presence or absence of lymphedema and the administration of chemotherapy for the treatment of the initial breast cancer or secondary sarcomas. Although the SEER-Medicare Database could have provided more clinical information than the database used, >65% of the patients analyzed in the current study were aged <65 years at the time of their initial malignancy, therefore limiting the use of the SEER-Medicare Database.
Based on the results of this and other studies, we conclude that RT and partial mastectomies significantly increase the risk for the development of secondary STS for up to 20 years after the treatment of breast cancer. The relatively low overall incidence of secondary STS coupled with the relatively high incidence of breast cancer limits the efficacy of a routine schedule of sarcoma-specific surveillance. Nevertheless, healthcare providers should be aware of the low risk of developing STS in the long term and their aggressive behavior, in addition to metachronous breast cancer, in their routine follow-up of women with a prior history of breast cancer and RT. The incidence of secondary STS may continue to increase as these modalities become more prevalent. Future studies are warranted to further define the specific factors that make partial mastectomies and RT increase the risk of developing STS and whether changes in these factors such as the current use of smaller radiation fields will have an impact in the future incidence of the disease.