Long-term outcomes in patients with radiation-associated angiosarcomas of the breast following surgery and radiotherapy for breast cancer

Abstract

Background

Radiation-associated angiosarcoma (RAAS) is a devastating disease occasionally observed in breast cancer patients treated with radiation. Due to its rarity, our knowledge—of disease risk factors, epidemiology, treatment, and outcome—is extremely limited. Therefore, we sought to identify clinicopathologic factors associated with local and distant recurrence, and disease-specific survival (DSS).

Methods

Radiation-associated angiosarcoma was defined as pathologically confirmed breast or chest wall angiosarcoma arising within a previously irradiated field. A comprehensive search of our institutional tumor registry (1/1/93 through 2/28/11) was used to identify patients (n=95 females); patient, original tumor, RAAS treatment, and outcome variables were retrospectively retrieved and assembled into a database.

Results

The median follow-up for all RAAS patients was 10.3 years (range, 2.4 – 31.8 years). The latency period following radiation exposure ranged from 1.4 to 26 years (median = 7 years). One- and five-year DSS rates were 93.5% and 62.6%, respectively. Reduced risk of local recurrence was observed in patients who received chemotherapy (P = 0.0003). In multivariable analysis, size was found to be an independent predictor of adverse outcome (P = 0.015).

Discussion

Our study demonstrates that RAAS exhibits high recurrence rates. It also highlights the need for well-designed multicenter clinical trials to inform the true utility of chemotherapy in this disease.

INTRODUCTION

Angiosarcoma of the breast occurs as primary tumors or secondary to previous radiotherapy. Primary breast angiosarcomas, typically observed in women 30-40 years old, present as ill-defined parenchymal masses. Secondary breast angiosarcomas usually occur in older women several years after breast cancer treatment. These lesions involve the dermis and are often confused with benign conditions; e.g., infection or hemorrhage.

The term radiation-associated angiosarcoma (RAAS) is used instead of radiation-induced angiosarcoma (RIAS), since RIAS implies that radiation is the sole etiologic factor, which may disregard other unknown or poorly characterized contributory effects. It is worth noting that the association between radiation and sarcoma development was first proposed by Cahan, et al., who established criteria for the diagnosis of postradiation sarcoma.[1] These criteria were later modified by Arlen, et al., to include the tissues adjacent to the path of the radiation beam also at risk for the development of sarcoma, and a latency period of at least 3 to 4 years.[2]

Angiosarcomas are a rare known complication of radiotherapy for breast carcinoma (<1% of breast tumors) with an estimated incidence of 0.04-0.15%.[3-7] Yap, et al., reported a Surveillance, Epidemiology and End Results (SEER) study of breast cancer patient second malignancies (n=270,000), demonstrating that angiosarcoma is the most prevalent postradiation secondary sarcoma at this site.[8] Given their long latency period (range: 3-20 years;[7, 9-11]), RAAS under-reporting is likely, and its true incidence therefore higher. Moreover, as the number of patients treated with adjuvant radiotherapy who become long-term survivors increases, RAAS incidence may likewise increase.

Due to the rarity of RAAS, most reports are either anecdotal or small institutional series. Accordingly, we reviewed our experience with breast RAAS, the largest single institutional series reported to date, seeking to identify clinicopathologic factors prognostic for disease-free, local, and distant recurrence-free survival.

METHODS

With the approval of the Institutional Review Board of The University of Texas MD Anderson Cancer Center (UTMDACC), we identified all patients diagnosed with RAAS of the breast or chest wall from 1/1/93 to 2/28/11 who met these criteria: 1) history of radiation to the breast/chest prior to development of angiosarcoma; 2) angiosarcoma occurrence within radiotherapy fields; and 3) angiosarcoma histologically verified by a UTMDACC pathologist.

The date of RAAS histologic confirmation was defined as the date of diagnosis. For tumor size, the maximum dimension determined by radiologic or pathologic assessment was utilized. A microscopically positive surgical margin was defined as tumor within <1mm of the inked margin. Patients with more than one lesion were considered to have multifocal disease. Local recurrence was considered as any recurrence at the primary site without metastasis.

Clinicopathologic variables included: 1) patient factors: age at diagnosis (≤50 or >50 years); 2) tumor factors; e.g., microscopic margins (negative or positive), grade (low/intermediate, high) and size (≤10cm or >10cm); 3) surgical procedure; 4) neoadjuvant/adjuvant treatment. The tumor grade was based on that stated in the pathology report given at the time the slides were reviewed originally. Locally advanced disease was defined as tumor involving the chest wall or axilla, and/or measuring greater than 10cm in size. Patients with atypical vascular lesions (AVLs) were excluded, given that AVLs and cutaneous angiosarcomas may represent two histologically distinct entities.

Local recurrence-free survival was determined as the time from the initial RAAS treatment to recurrence at any previously irradiated site. Deaths due to disease were treated as a disease-specific survival (DSS) endpoint; other deaths were treated as censored observations. Overall survival (OS) was defined as time from initial RAAS diagnosis until death from any cause. Surviving patients and those lost to follow-up were censored on the date of last follow-up. DSS and OS were calculated by Kaplan and Meier methods.[12] Significance between survival curves was evaluated using log-rank tests and multivariable Cox proportional hazards regression analysis.[13] P<0.05 was considered significant.

RESULTS

Breast cancer history

Ninety-five patients previously treated with breast conserving surgery or mastectomy and radiation therapy were evaluated at UTMDACC for RAAS of the breast or chest from 1/1/93 to 2/28/11; (Table 1). Most patients presented with early stage disease. The median age at diagnosis was 62 years. Ninety-three percent of the patients were Caucasian, 3% were African-American, 3% Hispanic, and 1% were other races. The majority of the patients (93%) received primary breast cancer treatment at a non-UTMDACC facility. Most patients underwent breast-conserving therapy with or without axillary lymph node dissection. Ten patients underwent modified radical mastectomy (MRM); the type of surgical procedure was not documented in four patients. Only 31 patients had information regarding the radiation dose. All of which received postoperative external beam radiation (median dose: 60 Gray (Gy); range, 45-65 Gy).

Table 1.

Patient and primary breast cancer characteristics for all cases

Characteristics	Number of patients (n = 95) (%)
Age at the time of Breast Cancer* diagnosis (years)
Mean	60
Median (range)	62 (31-88)
Breast Cancer Tumor size (cm)
Mean	2.1
Median (range)	2 (0.5-11)
Breast Cancer tumor stage
0	5 (5.3)
IA	35 (36.8)
IB	0 (0.0)
IIA	18 (18.9)
IIB	11 (11.6)
Unknown	26 (27.4)
T1NX	1 (1.1)
TXN0	3 (3.2)
TXN1	4 (4.2)
Unknown	18 (18.9)
Surgical Treatment for the Breast Cancer
Lumpectomy	27 (28.4)
Lumpectomy & SLND+ or ALND0	50 (52.6)
Modified radical mastectomy	10 (10.5)
Total mastectomy	2 (2.1)
Bilateral mastectomy	2 (2.1)
Unknown	4 (4.2)
Radiation Received (Gy)*
Mean	56
Median (range)	60 (46-65)

⁺SLND, Sentinel lymph node dissection

⁰ALND, Axillary lymph node dissection

^*Data regarding radiation therapy was available for 31 patients.

Radiation-associated angiosarcoma presentation

All patients (median age: 71 years; range, 34-92) were free of breast carcinoma at the time of RAAS diagnosis (Table 2). The median latency period was 7 (range,1.4-26) years. All patients presented with skin changes including skin nodules (n=60), rash/erythema (n=19), macular lesion (n=7), or ulceration (n=3). The median time from symptom to diagnosis was 3 months (range, 0-11 months). Patients commonly presented with multifocal disease (70%), with primary lesions located in the breast (62%) and in the chest wall/axilla/arm (38%). At presentation, 96% of patients had localized disease; only 4% had metastases. Ten patients had clinically apparent lymphedema in the ipsilateral arm.

Table 2.

Patient and radiation-associated angiosarcoma characteristics for all cases

Characteristics	Number of patients (n = 95) (%)
Age at RAAS* diagnosis (year)
Mean	68
Median (range)	71 (34-92)
Latency (years)
Mean	8
Median (range)	7 (1.4-26)
Location
Breast	59 (62.1)
Chest wall/axilla/arm	36 (37.9)
Clinical presentation
Skin nodules	60 (63.2)
Rash erythema	19 (20.0)
Macular lesion (bruise/skin thickening)	7 (7.4)
Ulceration	3 (3.2)
Not documented	6 (6.2)
Foci of disease
Solitary nodule	29 (30.5)
Multifocal	66 (69.5)
RAAS status at presentation
Localized	91 (95.7)
Metastasis	4 (4.3)
Treatment for RAAS
Surgery alone	40 (42.1)
Neoadjuvant chemotherapy/surgery	12 (12.6)
Surgery/adjuvant chemotherapy	31 (32.6)
Neoadjuvant chemotherapy/surgery/adjuvant chemotherapy	5 (5.3)
Limb perfusion and surgery	1 (1.0)
Chemotherapy alone	5 (5.3)
Palliative radiation	1 (1.0)
Surgical Treatment for RAAS (n=89)
Total mastectomy	49 (55.1)
Wide local excision	27 (30.3)
Bilateral mastectomy	7 (7.9)
Modified radical mastectomy	4 (4.5)
Forequarter amputation	2 (2.2)
RAAS Tumor size ( cm)
Mean	6.9
Median (range)	5 (0.2-24)
≤10cm	60 (63.2)
>10cm	20 (21.1)
Grade
Low/Intermediate	18 (18.9)
High	54 (56.9)
Not documented	23 (24.2)

^*RAAS, Radiation-associated angiosarcoma

Treatment characteristics

All patients presenting with localized RAAS underwent surgical evaluation; 89 patients (94%) underwent resection. Sixty-two (70%) underwent surgery elsewhere and 27 (30%) underwent definitive surgery at UTMDACC. Total mastectomy without axillary node dissection was most commonly performed (55%), followed by wide local excision for those patients with prior mastectomy (30%; Table 2). Four patients underwent MRM, seven had bilateral mastectomies and two were treated with forequarter amputation. Reconstructive surgery was performed on 24 patients, including 14 who underwent myocutaneous flaps (i.e., advancement flap, latissimus dorsi or transverse rectus abdominus myocutaneous flap); ten patients received skin grafts. Two patients with locally advanced disease did not undergo surgical treatment due to disease progression and medical comorbidities. There were no postoperative mortalities.

Forty-nine patients (52%) received chemotherapy in addition to surgical resection (Table 2). Twelve of the patients with localized disease (13%) received neoadjuvant therapy, all of whom presented with extensive cutaneous lesions and skin involvement. Neoadjuvant chemotherapy was given if the possibility of R0/R1 resection was dubious. Thirty-one patients received adjuvant therapy in addition to surgical intervention. Adjuvant chemotherapy was offered to patients with large and/or high grade lesions, or margins <1cm. Five patients received both neoadjuvant and adjuvant chemotherapy. Five patients with unresectable, locally advanced/metastatic disease were treated with chemotherapy alone.

The chemotherapy regimens varied: 13 patients received gemcitabine/docetaxel, 12 patients received paclitaxel alone, eight received doxorubicin/ifosfomide +/− paclitaxel, eight received doxorubicin +/− dacarbazine, five received interferon alone, two were treated with gemcitabine alone, and six received other multidrug regimens. One patient received palliative radiation for bleeding from ulcerated tumors involving the medial arm; due to comorbidities, systemic treatment or surgery was not recommended.

Pathologic findings

The diagnosis of RAAS was established histologically in all cases (57% high grade; 19% low/intermediate grade; in 24%, grade was not reported; Table 2). All tumors involved skin; no patients had tumors limited to the breast parenchyma. Tumors ranged in size from 0.2-24cm (median=5cm). Eight patients had positive margins (four each R1 or R2 resections). Two of the R1 patients were re-excised with resultant R0 margins, one of whom received subsequent adjuvant therapy; the other two R1 resection patients were treated with adjuvant therapy and close follow-up. Of the four R2 margin patients, one underwent re-excision with subsequent negative margins and three were treated with adjuvant therapy alone, due to unresectability or surgically prohibitive medical comorbidities. For the four patients who underwent MRM, none of the lymph nodes examined were positive for angiosarcoma.

Local and distant treatment failure

After median follow-up of 10.8 years (range, 2.4-31.8), local recurrence following R0/R1 resection was observed in 48% of patients (n=46). Eleven patients developed second local recurrences. Distant metastasis occurred in 26 patients (27%). Median time to local and distant recurrence was 1.8 years (range, 0.1-15.8) and 3.0 years (0.2-15.8), respectively. Thirteen patients (14%) who developed local recurrence had synchronous distant metastasis. The most common site for distant metastasis was the lung (n=18; three of which patients also had liver metastasis), followed by the liver (n=5).

Cox regression multivariable analyses were used to identify factors associated with local and distant recurrence-free survival. We observed a local recurrence risk reduction in patients who received chemotherapy and surgical resection (hazard ratio [HR]=0.35, 95% confidence interval (CI):0.15-0.80, P=0.012). Age, grade, margin status and tumor size were not associated with local recurrence-free survival. The limited number of patients precluded any definite conclusion regarding the role of chemotherapy in the prevention of multiple recurrences. High grade was associated with decreased distant recurrence-free survival (HR=4.23, 95% CI:1.39-12.89, P=0.011). Age, margin status, tumor size and use of chemotherapy were not associated with distant recurrence-free survival.

Survival outcomes

Of the 95 patients, 30 died of disease and 17 patients died of other causes. The one-, two- and five-year OS was 91%, 78%, and 54%, respectively. The one-, two- and five-year DSS was 94%, 84%, and 63%, respectively (Table 3). Patients who developed metastatic disease or local recurrence had significantly worse DSS than those who did not (P=0.0002; Figure 1A); patients who presented initially with locally advanced disease or metastasis had worse DSS than patients who presented with localized primary tumors (P=0.01; Figure 1B). Median survival time for patients with localized tumors was 7.3 years (range; 0.2-15.8) versus 4.7 years (range; 0.7-10.1) for those presenting with locally advanced disease. Univariable analysis demonstrated that tumor size >10cm, development of local and distant recurrence were adverse prognostic factors (Table 4; Kaplan-Meir curves, Figure 1A, 1D). Grade and adjuvant/neoadjuvant treatment were not associated with improved survival. Size was the only independent adverse outcome predictor on multivariable analysis (P=0.015) (Table 4).

Table 3.

Follow up and survival outcomes

	TOTAL (n= 95)	SURGERY ALONE (n= 40)	SURGERY +CHEMO (n= 49)	P value
Time to recurrence (year)				0.7
Mean	3.0	2.8	3.1
Median (range)	1.3 (0.1-15.8)	1.3 (0.1-15.1)	1.5 (0.2-15.8)
Disease-specific survival				0.15*
1 year	94.7%	93.1%	100.0%
5 year	56.3%	52.2%	60.0%
Overall survival				0.59
1 year	%	%	%
5 year	%	%	%
Local recurrence-free survival				0.0003*
1 year	70.9%	56.2%	81.4%
2 year	60.0%	46.3%	70.6%
5 year	50.7%	36.9%	62.8%

^*Log-rank test for equality of survivor functions.

Figure 1.

Radiation-associated angiosarcoma outcome analysis. Kaplan-Meier survival curves for angiosarcoma-specific survival for: A, Patients that did not develop recurrence versus recurrence only versus metastatic disease, P = 0.0002; B, Localized versus locally advanced versus metastatic, P = 0.01; C, Low/intermediate grade versus high grade, P = 0.2; D, Size ≤ 10cm versus > 10 cm, P = 0.005.

Table 4.

Univariable and multivariable Cox proportional hazard models for disease-specific mortality

Variable	n	HR+	95% CI#	P value
Univariable analysis
Age (years)
≤ 55	27
> 55	34	1.75	0.883-3.495	0.108
Grade
Low/Intermediate
High
Margin
Negative	11
Positive	20	0.86	0.300-2.454	0.292
Tumor size
≤ 10 cm	20
> 10 cm	13	1.03	0.3513-3.028	0.448
Treatment
Surgery alone	29
Surgery and chemotherapy	10	1.92	0.800-4.582	0.144

⁺HR, hazard ratio

^#CI, confidence interval

^*This group includes those patients that received neoadjuvant, adjuvant or both. No statistical difference was found when segregating these groups.

DISCUSSION

Given the rarity and variability of clinical presentation, RAAS diagnosis is often delayed and management strategies are inconsistent. Some studies have proposed that one of the differences between RAAS of the breast and other radiation-associated sarcomas is the relatively short interval between radiation therapy and the development of the angiosarcoma.[7, 14-18] The mean time interval between radiation therapy and subsequent sarcoma onset is approximately 10 years.[1, 19, 20] For instance, postradiation bone sarcoma series report a median latency range from 14 - 17 years.[1, 19, 20] We observed eight cases occurring in less than four years after radiation treatment. Although RAAS presents a distinct clinical pattern, the difference between radiation-induced versus sporadic lesions remains unverified. The establishment of genetic differences between sporadic and radiation-induced angiosarcomas will facilitate discrimination between these two entities. Recent data have demonstrated that radiation-induced angiosarcomas are characterized by a consistent amplification of the c-MYC oncogene, which is less often observed in sporadic lesions.[21] Furthermore, a signature of 135 genes has been identified which distinguishes radiation-induced from sporadic sarcomas.[22] In the future, better understanding of these molecular features will assist us in the differentiation of radiation-induced versus sporadic lesions.

Most of our patients (94%) underwent surgical resection; only one received palliative radiation; re-irradiation is usually contraindicated in RAAS due to the cumulative risk of severe toxicity. However, complete RAAS response rates using radiotherapy with hyperthermia have been reported [23, 24]. Additionally, a recent study observed local recurrence in only 1 of 13 patients who received radiation for their RAAS versus 10 of 29 who underwent surgery alone.[25] Notwithstanding, the radiotherapy role in RAAS therapy remains uncertain.

Consistent with other reports, our analysis demonstrated significantly better survival for patients presenting with small and/or localized disease [7]; moreover, our breast RAAS patients had a five-year DSS better than that achieved in other post-radiation sarcoma histologies; e.g., malignant peripheral nerve sheath tumor, osteosarcoma, fibrosarcoma, malignant fibrous histiocytoma (5-year cumulative survival rates: 12%, 31%, 35%, and 45%, respectively).[18, 20, 26] The behavior of RAAS remains uncertain; some reports suggesting prognosis inferior to sporadic counterparts [25, 27, 28], whereas others report converse findings.[15, 16, 29] We observed a relatively indolent, good prognosis disease course, albeit with a 48% local recurrence rate even in R0 patients, which is comparable to other RAAS series [14, 15, 29] but higher than for other STS contexts.[30, 31] Angiosarcoma is a multifocal disease characterized by microsatellite lesions that may comprise occult sarcoma spread beyond apparent R0 margins, as noted in other sarcoma subtypes [32-35]; hence, the significance of negative surgical margins is unclear, highlighting that surgery alone may not eradicate all disease.

In our series, 52% of patients received chemotherapy in addition to surgery. In older series, lack of up-front chemotherapy or usage only for unresectable disease could underlie observed inferior outcomes.[28] However, other contemporary studies report favorable outcomes similar to ours; most patients in those series did not receive chemotherapy [15, 16, 29, 36] and so the role chemotherapy for RAAS remains uncertain. Yet to be established unique features of breast angiosarcoma versus those originating in other anatomic regions may also be contributory, as noted in other reports.[7, 14, 15, 28]

Although, the risk of local recurrence was lower in patients who received chemotherapy, beneficial effects not reflected in disease-specific or overall survival. However, our study is limited in its ability to establish possible chemotherapy benefits in RAAS. Only six pathology reports for patients who received neoadjuvant therapy indicated therapy-induced necrosis responses. Different chemotherapy regimens were utilized; limited numbers of patients in each subgroup suggest that these findings need to be explored in larger patient cohorts receiving standardized chemotherapy regimens, stratified for tumor burden. Phase II trials have demonstrated clinical benefit of paclitaxel treatment for unresectable angiosarcoma [37-39] with a 74% rate of non-progression after two cycles.[37] Other groups have described promising responses to paclitaxel or gemcitabine-taxane (including complete remissions) for resectable primary or recurrent RAAS breast disease.[40-43] These findings highlight the need for well-designed trials to inform the true utility of chemotherapy in this disease.

In summary, although the incidence of RAAS is low, the number of cases reported is increasing as a function of the improved likelihood of surviving early stage breast carcinoma. Furthermore, with increased use of external beam radiation in the management of breast cancer patients, the incidence of post-radiation sarcomas may increase in the future. Long-term follow-up is needed for early disease detection. Complete surgical excision of angiosarcoma is the treatment of choice, without axillary dissection due to the low likelihood of axillary metastases.[7, 36, 44] Given the significant local recurrence rates, close monitoring is recommended.[14, 15, 36] Locally recurrent disease is often treatable; we have shown elsewhere that complete resection in this setting can potentially prolong survival.[45] Chemotherapy and radiotherapy may have roles to play in cases of unsuccessful surgical clearance; their beneficial effects remain to be defined.[24, 25, 37, 39, 41, 43, 46, 47] Accordingly, well-designed multicenter clinical trials are now urgently needed to advance RAAS care.

Synopsis.

We evaluated a large cohort of patients with radiation associated-angiosarcomas of the breast in order to identify clinicopathologic prognostic factors for local and distant recurrence-free survival, and disease-free survival.