Introduction
Breast angiosarcoma is a rare malignancy of endovascular origin and a subtype of soft-tissue sarcoma, accounting for less than 1% of all breast malignancies and less than 5% of all sarcomas [
1,
2]. Clinically, breast angiosarcoma is divided into two categories: primary breast angiosarcoma (PBAS) and secondary breast angiosarcoma (SBAS). PBAS frequently arises from the breast gland, breast soft tissue, subcutis, or dermis; it is difficult to distinguish from benign disease and usually occurs in young women. It often presents clinically as painless or palpable injury and is sometimes accompanied by a change in the color of the skin, such as skin depigmentation [
3]. However, SBAS is always related to two aetiological factors: radiotherapy after breast-conserving mastectomy, in which the latency period for onset is usually less than 10 years; and ipsilateral limb lymphedema called Stewart-Treves syndrome (STS). SBAS usually occurs in older women; the tumor develops in the skin tissue and may invade the parenchyma of the mammary gland, and it is sometimes associated with erythaematous plaques or nodules [
4,
5].
Given the rarity of this disease, many institutions have inadequate samples to research. The majority of relevant studies have examined the clinicopathological features of PBSA, but few studies have explored SBSA and the treatment of breast angiosarcoma. Additionally, many of these articles are small case reviews and case reports, and we cannot derive definitive clinical features that affect prognosis. A study from The University of Texas M. D. Anderson Cancer Center examined of 55 patients with angiosarcoma to determine its clinical features and prognosis, and only SBAB was found to occur predominantly in young patients receiving radiotherapy, but there was no difference in OS between PBSA and SBAB [
6].
Clinically, there is occult pathogenesis of angiosarcoma, and there is nonspecific imaging on ultrasound and mammography, posing a great challenge to its diagnosis. Treatment in the sarcoma service is currently recommended for the management of breast angiosarcoma [
7]. In past research, some authors have thought that multidisciplinary combination therapy including surgery, radiation, and systemic chemotherapy can improve survival in patients [
8]; however, due to the small sample size, there are inherent limitations in the experimental results. A review has shown that there is no clear evidence of specific treatments for specific subtypes of angiosarcoma and mostly symptomatic treatment, and the inability to treat specifically for a particular type may also be one of the reasons for the worse OS of breast angiosarcoma [
9]. Similarly, in multivariate analysis, it was found that radiotherapy and systematic chemotherapy do not significantly prolong the overall survival of patients in other studies.
Based on the current research status, we chose to use the Surveillance, Epidemiology, and End Results Program (SEER) database to extract and analyze data on PBSA and SBAB. The main objective of this study was to explore the clinical characteristics and treatment decisions of PBSA and SBAB to solve the remaining treatment problems, and to improve the survival of patients.
Methods
This study was performed with approval from the Surveillance, Epidemiology, and End Results Program (SEER) database. The SEER database is a publicly available, comprehensive database of demographic information for selected US states and counties (approximately 35% of the US population), which compiles information on prevalence, morbidity, mortality, and survival rates (
http://seer.cancer.gov/about/overview.html). We obtained permission from SEER to access the study data (reference number 14492-Nov2020). Given that the study is based on a strict registration process, the need for informed consent was waived. In addition, the study was exempted from Institutional Review Board approval, as the use of SEER data does not enable the identification of patient information.
We retrospectively reviewed the information we collected and finally determined that 394 patients had angiosarcoma who received treatment between 2004 and 2015 due to the limitation of the years of tumor size and tumor extension inclusion in the database. Studies have divided breast angiosarcoma into PBSA and SBAB, and the determination of PBSA and SBAB was based on the criteria of Saira et al. [
10] and Taimur et al. [
11]. The inclusion criteria were as follows: (1) the patient had only one primary angiosarcoma or multiple malignancies, but angiosarcoma was the first diagnosis of PBSA, and (2) the patient had a non-first-stage malignancy that had previous breast cancer, lung cancer, or soft tissue cancer in the chest wall as SBAB. The exclusion criteria were as follows: (1) patients with previous cancer that was not related to the breast or chest wall but a malignancy in other parts, such as brain cancer, skin cancer, and bone cancer, were excluded, and (2) patients with incomplete clinical information on tumor size, stage, and grade were excluded. Ultimately, we identified 247 patients for inclusion in this study for further analysis.
The clinical characteristics of the follow-up patients were reviewed: ICD-O-3 histology codes as haemangiosarcoma (9120), race, tumor onset location, tumor grade, tumor size, tumor stage, depth, number of primary tumors, chemotherapy, radiation, and type of initial surgery. Classify tumors at the nipple in the central group of the breast. The return of the SEER database divides the pathology grading into Grade I, Grade II, Grade III, and Grade x. Based on the maximum dimension of the tumor, we divided it as ≤ 5, 5–10, or > 10 cm. The historic tumor stage used was local, regional, and distant. In the tumor extension group, tumors that were confined to breast tissue and fat were defined as confined, tumors that were infiltration of lymph nodes around the tumor and skin involvement were defined as local infiltration, and tumors that were attachment or fixation to pectoral muscle or underlying tissue were defined as deep. To describe specific surgeries in the database, we divide the surgeries into partial mastectomy, mastectomy, and unknown. Unknown refers to patients who have not undergone surgery or who have refused surgical treatment. Patients who have received radiation therapy in the past are unknown about the dose of radiation therapy, and the specific regimen of chemotherapy and the cycle of chemotherapy are also unknown.
For the 2 patient groups, the clinical characteristics of patients are mainly realized as frequencies and percentages as categorical variables. Overall survival (OS) was defined from the initial angiosarcoma diagnosis to the time of death. OS was assessed by the Kaplan‒Meier method, and the differences between the different groups were compared using the log-rank test. The median of the results is included in the table as an estimate. Chi-square testing was applied to compare distributions between groups. Univariate analysis was performed for PBSA and SBAB prognostic variables with the log-rank test and Cox regression. Variables that were significant in univariate analysis were incorporated into multivariate Cox proportional hazards regression analysis. The extractor of the data was SEER 18, and the statistical analysis was performed with SPSS 22. P values < 0.05 or less were considered significant.
Discussion
Breast angiosarcoma is a rare disease in the clinic, and we divided it into PBSA and SBAB types based on similar studies in previous literature [
10,
11]. PBSA accounts for less than 1% of breast cancer cases, with an incidence of 4.5 cases per million, but the incidence of SBAB was reported to be less than 0.5% [
12]. Due to the rarity of the disease, it causes great challenges in the diagnosis, prognosis, and treatment of the disease. This study sought to determine the correlation between PBSA and SBAB by comparing 100 primary patients from the SEER database with 147 secondary patients, as well as to explore the diagnosis and treatment decisions of the disease. Our results suggested that the size of the tumor, the stage of the tumor, and the depth of the tumor in breast angiosarcoma often mark a progressive tumor phenotype. The size of the tumor in PBSA led to a worse prognosis, and the grading of the tumor in the SBAB context represents a worse prognosis with the depth of the tumor. However, the 1-, 3-, and 5-year OS of the two groups of patients were compared, and no significant difference was found. Under different monotherapy and combination therapy models, we found that simple surgical treatment, partial mastectomy combined with radiation and partial mastectomy combined with chemotherapy in PBSA can improve OS outcomes, but different treatments in SBAB were not statistically significant.
In a review of past research, the incidence of angiosarcoma in recent years has gradually increased from 1.52 to 2.04 per year, and breast cancer patients account for a relatively high proportion [
13]. At present, the pathogenesis of PBSA is unclear, but SBAB is usually associated with postoperative edema, lymphatic blockade, perioperative infection, and radiation therapy. The onset of the disease is usually 4 to 7 years after breast-conserving therapy. PBSA usually presents clinically as a palpable mass, while SBAB usually presents with cutaneous changes such as abrasions, violaceous skin rash, and red papules or pain and discomfort in deep tissues, and the presentation is polycentric [
14]. These clinical manifestations are similar to the manifestations of many benign tumors, and it is difficult for us to diagnose them through these manifestations, which also leads to our inability to diagnose and treat them through early clinical manifestations, resulting in delays in the condition and poor prognosis. On the other hand, due to the polycentric nature of SBAB, we were also unable to accurately measure the size of the tumor and the edge of the tumor in the clinic. Therefore, it was difficult for us to achieve R0 resection clinically, which also led to the risk of local recurrence and metastasis in patients in disguise, and some patients may have the possibility of reoperation.
In terms of the grouping definition of tumor size, our classification divides it into < 5, 5–10, or > 10 cm groups based on previous literature (
p = 0.15). Because tumors smaller than 5 cm in primary patients are the most reliable prognostic criterion for prognosis, the size of the tumor is closely related to the prognosis of the primary patients [
15,
16]. This is consistent with the results of this study in univariate analysis and multivariate analysis. However, Keila [
17] found that tumor size with 10 cm as the defining point was also prognostically correlated in secondary patients in univariate analysis and multivariate analysis. Through the analysis of the data, it was found that the tumor size in this study was 9.5 cm as the cutoff point, which was statistically significant (
p < 0.001). However, from the perspective of clinical and collected data, tumors larger than 9.5 cm have exceeded the scope of the breast, the number of patients with large tumors in the clinic is few, and the prognosis is relatively worse; thus, the results obtained herein indicate only a “nominally” poor prognosis, and we did not use these data.
Our study revealed that the OS of angiosarcoma is poor and is not sensitive to various treatments, especially among patients with SBAB. The tumor itself grew quickly and was prone to distant metastasis, and early diagnosis and intervention are particularly important. However, traditional imaging had a certain degree of invisibility, such as mass under the mammograms, asymmetrical density, parenchymal disorders, skin thickening and microcalcification, and under ultrasound, and they were mostly manifested as masses and thickening of the skin, which are similar to many benign diseases. In the literature on Sona A, it is shown that MRI is diagnostically specific, manifested by low T2-weighted imaging in the rapidly enhancing part of the dermis and parenchyma lesions and T1-weighted imaging characterized by a fast initial and delayed washout, which can determine the extent of the tumor [
18]. Therefore, it is recommended that patients with suspected angiosarcoma can use MRI for localization to determine the extent of the tumor, and at the same time, if the patient has a history of breast cancer radiation therapy, skin changes should be observed carefully, and if necessary, puncture biopsy should be performed to confirm the diagnosis.
There are also differences in prognosis due to the difference in PBSA and SBAB performance. In primary patients, in a review article of 42 cases of PBSA, the most significant prognostic factor was the size and grade of the tumor, while no difference in prognosis for radiotherapy-induced angiosarcoma was found [
3]. This conclusion is similar to a previous article. In a review of 57 cases, the overall survival rate at 10 years was 62%, and pathological grade and tumor size were found to be associated with prognosis in univariate analysis, while multivariate analysis found that only pathological grade was associated with prognosis [
19]. In another study of 16 cases of PBSA, the authors found that all patients underwent surgery with curative intent, and the overall survival rate was 13.6 months. The univariate analysis suggested that high-grade tumors predicted a worse prognosis, but tumor metastases had no effect on prognosis [
20]. However, in our study, consistent with previous studies, the 1-, 3-, and 5-year OS rates for primary angiosarcoma were 80%, 39%, and 25%, respectively. Univariate analysis found that tumor size, tumor spread, and surgery were associated with prognosis, and multivariate analysis found that tumor size was independently associated with prognosis. Our research has demonstrated that in patients with PBSA, early diagnosis (when the tumor < 5 cm) and surgery can effectively improve the prognosis and survival of patients. In secondary patients, in a review article of 112 cases of SBAB, high-grade tumor patients (OS: 36 months) had a worse prognosis than lower-grade tumor patients (OS: 48 months) and a 5-year overall survival rate of 50.5% [
6]. In addition, a review of the Jeffrey [
21] article in 176 secondary patients found that OS at 3 years was 74%, and it was proven that positive surgical margins, tumor depth, and high grade were associated with prognosis. However, in another article review of the George [
22], we found that age, radical resection, and margins were associated with prognosis in univariate and multivariate analyses, while tumor grading was not statistically significant. In our study, the 1-, 3-, and 5-year OS rates for secondary angiosarcoma were 80%, 42%, and 34%, respectively. Tumor grade, age, tumor spread, and tumor depth were found to be associated with prognosis in univariate analyses, and tumor grade, tumor stage, and tumor depth were associated with prognosis in multivariate analysis, consistent with previous studies. In this study, it was found that there were obvious differences in prognostic factors between PBSA and SBAB, but no significant differences were found in OS outcomes; thus, it was not recommended to manage the clinical and prognostic management of primary and secondary patients as usual. Among primary and secondary patients, SBAB is not related to tumor size; perhaps, it was not accurate in the measurement range due to its multicentric characteristics affecting the assessment of prognosis. Age is associated with prognosis, but we should consider comorbidities in older patients. It is possible that other comorbidities cause the OS of the patient to decline. Secondary patients are associated with the depth of the tumor in prognosis, possibly because the patient received radiation therapy in the past, resulting in an increased risk of developing disease in all areas of radiation therapy.
At present, there is still great controversy over the treatment of angiosarcoma. In a review article of 9 patients among secondary patients, surgery combined with radiotherapy improved survival, but chemotherapy did not have a survival benefit. Surgery to R0 in primary and secondary patients improves long-term survival and can reduce the risk of recurrence [
19]. However, there were also articles reviewing the treatment of 28 patients, 15 of whom underwent axillary dissection, and suggested that chemotherapy was more inclined to use high-grade or lesions greater than 9.5 cm. A total of 22 patients were included in the review of Joshua [
10], and neoadjuvant chemotherapy was found to have a survival benefit and demonstrated that multidisciplinary combination therapy could prolong long-term survival in patients. A review of the treatment of 16 patients in the Qun-Chao Hu [
20] found that partial resection and total resection had no effect on survival in primary patients, and it was proven that patients who did not reach the R0 edge after surgery were only associated with postoperative recurrence and had no effect on survival. In secondary patients, the authors reviewed 176 patients in whom chemotherapy was used less frequently (41 vs. 4%) than other types of sarcomas and radical resection improved survival, but radiation therapy was not beneficial [
21]. The results of this study are not consistent with those of previous studies. We only found in univariate analysis that surgical treatment was significant for primary patients, and the other variables were not statistically significant. Therefore, our analysis of the mode of combined therapy found that in primary patients, partial resection, partial resection combined with chemotherapy and partial resection combined with radiotherapy are all statistically significant in the treatment of primary patients. Although it was not statistically significant in the secondary group, we cannot deny the therapeutic significance of total resection combined with related adjuvant therapy clinically. Based on the results of this study, we believe that in primary patients, in surgery combined with chemotherapy and surgery combined radiotherapy, the HR of partial resection combined chemotherapy is 0.016, while the total resection combined chemotherapy HR is 0.440, and the partial resection combined with radiotherapy HR is 0.105, while the total resection combined with radiotherapy HR is 0.412. The reason for the occurrence of this treatment may be related to the size of the tumor. In patients with tumors smaller than 5 cm, most patients will choose partial resection, and the prognosis of patients with tumors < 5 cm is better than that of patients with tumors ≥ 5 cm, which may also indirectly affect the patient’s treatment outcome.
Although this study is the largest collection of risk factors to date, there are still data limitations. First, it is well established that radiotherapy is associated with SBAB, and information on the dose, extent, duration, and period of radiotherapy plays a crucial role in the evaluation of secondary angiosarcoma; however, this information was not available in the database. It has been shown that chronic lymphedema is an important risk factor for the development of angiosarcoma of the breast (STS), and fibrosis of the surrounding skin tissue caused by radiotherapy may be a factor in its development [
23]. It is evident from previous cases that STS has a worse prognosis than other skin malignancies, and early diagnosis and treatment are imperative; however, STS is insidious, and many diagnoses rely on subjective assessment [
24,
25]. In recent years, studies have been performed using TDC and indocyanine green (ICG) lymphography to objectively assess the status of lymph node circulation, providing new ideas for early diagnosis of the disease [
26,
27]. Unfortunately, information on lymph node edema is not available in the SEER database.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.