Radiation-induced sarcomas of the breast: A review of a 20-year single-centre experience

Radiation-induced sarcomas (RIS) are histologically proven sarcomas within or around a previously irradiated site, per Cahan’s criteria. RIS incidence is higher in breast cancer compared to other solid cancers and prognosis remains poor given limited treatment options. This study aimed to review 20-year experience with RIS at a large tertiary care centre. Using our institutional cancer registry database, we included patients meeting Cahan’s criteria diagnosed between years 2000 to 2020. Patient demographics, oncologic treatment, and oncologic outcomes data were collected. Descriptive statistics were used to describe demographic data. Oncologic outcomes were assessed using the Kaplan Meier method.


Introduction
Despite the widespread use of radiotherapy in the management of different malignancies, the mechanism of radiation-induced carcinogenesis remains poorly understood, with hypotheses centered around the direct ionizing effect of radiation causing DNA damage, release of reactive oxygen species, or bystander effects leading to release of in ammatory mediators [1]. Radiation-induced sarcomas (RIS) are rare malignancies that comprise only 2.5-5.5% of all sarcomas [2]. RIS were rst de ned by Cahan et al. in 1948, and diagnosis relies on temporal and spatial relation to radiotherapy and a prolonged latency period [3].
Prognosis is generally poor, with 5-year overall survival estimates in the literature varying widely from 17-58% [2,[4][5][6][7][8][9], compared with 54-76% in sporadic sarcomas [2]. Poor outcomes are likely partly due to limited management options in the context of previous treatment as well as local and distant disease aggressiveness [2]. Complete surgical resection is associated with improved disease speci c survival but is only achieved in a minority of patients [10].
Patients who receive radiotherapy for primary breast cancer are known to have a higher incidence of RIS compared to other primary solid cancers [11]. A recent review of Surveillance, Epidemiology, and End Results (SEER) data found an incidence of 0.02% in a population of breast cancer patients diagnosed from 1973 to 2013 [11]. This is lower than older studies which found the incidence ranging from 0.03-0.2% [12][13][14][15]. With improvements in breast cancer survival, it is important to note that RIS typically occur 10 years following breast irradiation, but this latency period can be as long as 20 years [16] or as short as 6 months [17]. Radiation-induced cutaneous angiosarcomas of the breast have a shorter latency period of 4 years [16].
Given the large number of breast cancer patients diagnosed and treated every year, this study aimed to review incidence, risk factors, management, and subsequent oncologic outcomes of RIS of the breast using the 20-year experience at the McGill University Health Centre (MUHC), a large tertiary care centre.

Data Collection
With research ethics board approval, we retrospectively reviewed our institutional cancer registry database to identify all patients with a histologically con rmed breast cancer and sarcoma diagnosed between the years 2000 to 2020. To diagnose RIS, we used Cahan's criteria, which include: 1) history of radiation exposure prior to the development of sarcoma, 2) sarcoma within or around the eld of radiation, 3) prolonged latency period, and 4) histological con rmation of diagnosis that differs from the primary cancer that was treated [3]. Patients that did not meet these criteria were excluded. Using our electronic medical records, we collected patient demographic data, oncologic treatment data for both the primary breast cancer and the secondary sarcoma, as well as oncologic outcomes.

Statistical Analysis
We used descriptive statistics to summarize patient demographic data including mean, median, and standard deviation for continuous variables and frequencies and proportions for categorical variables. Oncologic outcomes including overall survival probability and progression free survival probability, with endpoints such as recurrence or death, and corresponding 95% con dence intervals were generated using the Kaplan-Meier method. [18] Results From 2000 to 2020, approximately 8700 patients were treated with radiotherapy for breast cancer in our institution. Including 3 patients referred from outside centers, we identi ed a total of 19 patients with breast RIS: 11 angiosarcomas (57.9%), 3 osteosarcomas (15.8%), 2 carcinosarcomas (10.5%), 2 undifferentiated pleomorphic sarcomas (10.5%) and 1 high-grade leiomyosarcoma (5.3%). The median age at diagnosis of primary breast cancer was 59 (range 29-78, mean 56). 18 patients underwent partial mastectomy (94.7%), and 1 patient underwent total mastectomy as their primary treatment. All patients received radiotherapy. Seven patients received adjuvant chemotherapy (36.8%) and 8 patients did not (42.1%), while for 4 patients chemotherapy status was unknown (21.1%). 13 patients had hormone receptor positive breast cancer (68.4%) and received endocrine therapy. Data on radiotherapy volumes, dose and fractionation were missing in 3 patients who were treated at outside institutions (15.8%).
Radiotherapy treatments were given as 3D conformal radiotherapy and included breast or chest wall only, or locoregional radiotherapy including axillary, supraclavicular plus or minus internal mammary lymph node irradiation. Four patients (21.1%) had a primary breast boost to their surgical cavity. Only 1 patient had a known genetic cause for their breast cancer (TP53 mutation). Patient and primary disease characteristics are summarized in Table 1. from diagnosis of RIS, 4 patients (21.1%) were deceased and 2 patients (10.5%) were lost to follow-up. RIS diagnosis and treatment data are summarized in Table 2.

Discussion
Our center is a large tertiary care center that treated approximately 8700 breast cancer patients with radiotherapy in the 20-year period included in this study. We are also a referral center for sarcoma patients diagnosed at outside centers. We are specialized in sarcoma management, employing a multidisciplinary approach which includes surgical oncology, medical oncology and radiation oncology. To our knowledge, this is one of the rst studies exploring RIS of the breast in a cohort of patients diagnosed and treated with modern techniques between 2000 to 2020.
In our study, we found 19 patients with RIS of the breast over the 20-year period. If we used the 8700 patients that were diagnosed and treated for breast cancer at our centre over this period as a comparison, this would result in an incidence of 0.2%. Taghian et al. examined 7000 breast cancer patients treated between 1954-1983 in France [14]. They found 11 patients that developed secondary sarcoma, resulting in a cumulative incidence of 0.2% at 10 years. However, noticeably different radiotherapy techniques were used in this historic cohort, as some patients received total body irradiation in addition to breast radiotherapy, resulting in a higher total radiotherapy dose [14]. In their 2002 publication, Yap et al. used SEER data to identify breast cancer patients who underwent RT from 1973-1997 (n = 275 000) [15]. The cumulative incidence of developing RIS in patients who received radiotherapy was 0.32% at 15 years, compared to 0.23% in patients who did not receive radiotherapy [15]. In a more recent analysis of the SEER database, Snow et al. found the incidence of RIS of the breast to be 0.02% at a median follow-up of 9.6 years [11]. A major limitation of SEER data is that it does not report data on radiation treatment or systemic treatment received [11,15]. The SEER data analysis also excluded patients for whom data were missing, therefore likely underestimating the true incidence. With 19 breast RIS patients in our study period, our 0.2% incidence would appear higher than the recent SEER analysis, but is comparable to historical data [14,15]. However, had we not included the small number of patients that were not initially treated for their breast cancer at our center, our center-speci c incidence would be slightly lower. Due to the retrospective nature of our study, it is impossible to establish the true denominator, limiting our calculation of true incidence during this time period.
While less than 10% of secondary malignancies are radiation-induced, the relative risk of developing RIS varies by treatment site and was found to be correlated with higher dose, younger age and increased time since diagnosis in one review [19]. Dose is an important factor, with risk increasing linearly after 40Gy [19]. The risk of sarcoma was 30.6 times higher for doses more than 44Gy compared to doses less than 15Gy in breast cancer patients in one study [20]. All of the patients included in our analysis in which radiotherapy dose was known received doses of at least 40Gy, consistent with standards of care for breast radiotherapy.
Breast cancer patients were found to have the highest incidence of RIS compared to other primary solid cancers in a recent review of SEER data [11]. Hereditary breast cancer syndromes (Li Fraumeni, Retinoblastoma, Nijmegen breakage syndrome) and BRCA are thought to play a role [21]. Kadouri et al. reviewed 473 BRCA and p53 mutation carriers and found the rate of RIS to be 0.43% (n = 7 women), but the overall rate is still low and not signi cantly different compared to all breast cancer patients with RIS reported in other studies [21]. Schlosser et al. retrospectively reviewed 230 women with BRCA mutations who underwent RT for breast cancer and found the incidence of secondary malignancies to be 0.32 per 1000 women-years [22]. However, of the six women who developed secondary malignancies, none were RIS [22]. In our study, we identi ed one patient with a TP53 mutation. However, it is di cult to draw conclusions on hereditary breast cancer and incidence of RIS given the rarity of both.
While this study yields important insights into the incidence and outcomes of RIS of the breast, it has several limitations. Firstly, inherent to its retrospective nature, data was missing, for example pertaining to previous radiation treatment details for patients that were treated at outside institutions. Two patients were lost to follow-up, further limiting our data analysis. Furthermore, we included data from electronic health records only, and therefore did not have access to paper charts which may have been used in the early 2000s. Given that this is single institutional data and incidence of breast RIS is rare, the total number of patients included in our analysis is small. This small number of patients renders it di cult to draw conclusions on risk factors associated with poorer outcomes. Our overall survival was favorable, with a 2-year disease-speci c survival of 88.9%. Our PFS of 56.1% at 2 years illustrates that a large proportion of patients recur locally or distantly despite aggressive multimodality treatment, suggesting that longer follow-up time is needed for survival data maturation.

Conclusion
Overall survival and progression-free survival for RIS of the breast appear favorable when salvage is possible. Patients should ideally be managed in a high patient-volume tertiary centre with multidisciplinary expertise. Further studies should evaluate multi-institutional data, since the rarity of RIS renders it di cult to draw conclusions on risk factors.

Declarations
Funding Figure 1 Kaplan-Meier estimates of PFS probability

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download. SupplementaryData.docx