Postradiation sarcoma: Morphological findings on fine-needle aspiration with clinical correlation




The current study was conducted to describe the clinical features and presentation, cytomorphological characteristics with histological correlation, and prognosis of patients who undergo fine-needle aspiration (FNA) for postradiation sarcoma (PRS).


A retrospective review was performed of 13 individual patients who were pooled from the FNA services of 3 academic institutions between 2001 and 2012. Cases were reviewed for the primary tumor, radiation history, latency period, and other distinguishing clinical features. The frequency of the various cytological preparations as well as the use of immunohistochemistry (IHC) on this material were reviewed. The cytopathology diagnosis was compared with the resection diagnosis, and the survival time was reviewed.


The median age of the patients was 61 years (range, 35 years-94 years) and no significant gender predilection was noted. The median latency period was 11 years (range, 5 years to > 50 years). Patients generally presented with large tumors (median, 8 cm [range, 3 cm-12 cm]), and the median survival was 14 months (range, 6 months-46 months). Nine of 13 patients died of their disease and 1 was lost to follow-up. The tumors were morphologically heterogeneous. IHC played an important role in excluding other diagnoses in those cases in which sufficient material was available.


PRS is a morphologically heterogeneous entity that can be diagnosed by FNA. It is a diagnosis of exclusion that requires a history of therapeutic radiation and often requires IHC to rule out locally recurrent malignancy. Cancer (Cancer Cytopathol) 2012. © 2012 American Cancer Society.


Postradiation sarcoma (PRS) is a nosologic entity comprised of any sarcoma that arises in an irradiated body site after a latency period.1-5 The term is interchangeable with “postirradiation sarcoma” and “radiation-associated sarcoma.” Because the evolution of this phenomenon is multifactorial and poorly understood, most investigators do not favor the overly simplistic term “radiation-induced sarcoma.”6, 7 The most common primary tumors for which radiotherapy is associated with PRS include breast carcinoma, Hodgkin lymphoma, cervical carcinoma, and bone and soft tissue sarcomas.8 Histologically, PRS can be any type of sarcoma, including malignant fibrous histiocytoma, angiosarcoma, leiomyosarcoma, fibrosarcoma, malignant peripheral nerve sheath tumor, myxofibrosarcoma, chondrosarcoma, and osteosarcoma.9-12 Distinguishing PRS from spontaneous sarcoma is important because PRS tends to present at a more advanced stage and behaves more aggressively than a morphologically equivalent spontaneous sarcoma.8, 13 To our knowledge, this is the first cytomorphologic study on a series of PRS.


Case Selection and Information

All available electronic records from the pathology data systems at The Johns Hopkins Hospital (7 cases), Emory Medical Laboratories (2 cases), and Ohio State University Medical Center (4 cases) were queried for cytological specimens that were signed out as PRS or were derived from tumors that were signed out as PRS on the resection specimen. The cases identified originated between the years 2001 and 2012. If a patient had undergone repeated biopsies, only the first diagnosis of PRS by fine-needle aspiration (FNA) was reviewed and included for the purposes of the features noted at presentation.

The history of radiotherapy and chemotherapy, latency period, primary tumor type, and the stage of the PRS at the time of presentation were derived from the electronic medical record. In 1 case, the radiotherapy history was so remote that the age of the patient was not known definitively, so the latency period was approximated as at least 50 years. The tumor size was derived from the dimensions of the tumor noted at the time of resection (10 cases), by radiology (2 cases), or by clinical examination (1 case). Of the 9 patients who died, all deaths were due to disease. The survival time was calculated as the time elapsed from the first diagnosis of PRS to the date of death. For cases from The Johns Hopkins Hospital, the date of death was either documented by hospital clinicians or obtained from the Social Security Death Index.


The FNA procedures were performed by interventional radiologists (9 cases) and pathologists (4 cases) using ultrasound guidance with on-site evaluation of adequacy. Smears evaluated on-site for adequacy were prepared in duplicate and included at least 1 air-dried slide for Diff-Quik staining and on-site evaluation and an alcohol-fixed slide for Papanicolaou staining for each pass as material allowed. Additional aspirated material was diverted for studies such as cell block, immunohistochemistry (IHC), and cytospin as demonstrated in Table 1. IHC was performed on the core biopsy (2 cases), cell block (5 cases), or destained direct smear (1 case) according to standard procedures. After the procedure, all the slides were previewed by a cytotechnologist and signed out by a board-certified cytopathologist. If a core biopsy was taken (4 of 11 cases), the specimen was interpreted separately by an expert surgical pathologist with expertise in soft tissue tumors in communication with the attending cytopathologist.

Table 1. Preparations and Techniques Used for the 13 Cases of PRS in the Current Study
 Diff-QuikPapanicolaouCell BlockCytospinCore BiopsyIHC
  • Abbreviations: IHC, immunohistochemistry; PRS, postradiation sarcoma.

  • a

    The median values for the number of cytospin and core needle biopsy preparations are both 0 because > 50% of the cases did not undergo these procedures.

No. of cases131110347
Median no. of slides/case (range)4 (3-8)4 (0-10)1 (0-3)0 (0-2)a0 (0-1)a1 (0-13)


Thirteen FNA cases from PRS were identified at 3 institutions. Of these, 3 cases were aspirations performed on a known PRS, and 10 were aspirations performed for the purposes of establishing a first diagnosis. For 1 institution (Johns Hopkins Hospital), the percentage of total known PRS cases (23 cases) that underwent FNA (7 cases) is approximately 30%. PRS cases that did not undergo FNA most commonly included tumor sites in the bone (4 cases), bladder (2 cases), and brain (2 cases). By contrast, only 1 tumor in this series was in the bone, and the other 12 tumors were present in soft tissue at sites close to the radiation fields of the primary tumor. Hence, this study clearly represents a subset of PRS that included tumor sites that underwent FNA at the study institutions. The median age of the patients at the time of the diagnosis of PRS was 61 years (range, 35 years-94 years) with no gender predilection. The median latency period was 11 years (range, 5 years to > 50 years). Of these cases, 46% (6 of 13 cases) had a history of chemotherapy accompanying the radiation. The primary tumors, cytopathologic diagnoses, and resection diagnoses are given in Table 2.

Table 2. Primary Tumor, Cytopathologic Diagnosis, and Resection Diagnosis of PRS
Primary TumorCytopathology DiagnosisResection Diagnosis
  • Abbreviations: BrCA, breast adenocarcinoma; MFH, malignant fibrous histiocytoma; MFS, myxofibrosarcoma; PrCA, prostate adenocarcinoma; PRS, postradiation sarcoma; SCC, squamous cell carcinoma; SCN, spindle cell neoplasm.

  • a

    Surgery was not performed.

  • b

    Cytopathologic diagnosis was made on an aspiration sample of a known PRS.

  • c

    Grade 2 by the NCI grading system.

BrCAAtypical cellsAngiosarcoma
BrCAEpithelioid angiosarcomaa
BrCAPRSbGrade 2 MFHc
Hodgkin diseaseMFHPRS osteosarcoma
Juvenile angiofibromaPoorly differentiated SCNbMFH
Kaposi sarcomaMFHPRS osteosarcoma
Larynx SCCHigh-grade MFSHigh-grade MFS
Lymphoepithelial cavernous sinus tumorSCNMFH
PrCASpindle and epithelioid high-grade sarcomaSpindle and epithelioid high-grade sarcoma
RhabdomyosarcomaHigh-grade PRSHigh-grade PRS
Synovial sarcomaHigh-grade sarcomabChondroblastic osteosarcoma
Synovial sarcomaHigh-grade sarcomaSarcoma with rhabdomyosarcomatous differentiation

An imaging study of 1 case of PRS is shown in Figure 1. The median tumor size at the time of presentation was 8 cm (range, 3 cm-12 cm), and 23% of patients (3 of 13 patients) presented with metastatic disease. Not surprisingly, the overall survival in this group was poor. At the time of last follow-up, 9 of the 13 patients (69%) had died of disease with a median survival of 14 months (range, 6 months-46 months). One patient was lost to follow-up and the 3 remaining patients were < 36 months from the time of aspiration.

Figure 1.

(A) A typical plain radiograph of postradiation sarcoma in the arm demonstrates the bony destruction of the adjacent soft tissue tumor. (B) The T2-weighted magnetic resonance image demonstrates an infiltrating soft tissue mass that is hyperintense to muscle.

Overall, a various amount of cellularity was noted, with the majority of cases displaying abundant material. The common denominator was the presence of a highly pleomorphic spindle cell population. In the case of epithelioid angiosarcoma, there were scattered cells with round to oval shapes and eccentric nuclei (Fig. 2). Long and tapering cytoplasmic processes were also evident. Myxofibrosarcoma predominantly contained spindle cells embedded in a loose myxoid stroma. Focal marked anisonucleosis was apparent (Fig. 3), as well as occasional, curvilinear, fine capillary vessels. Osteosarcoma displayed pleomorphic naked nuclei or plasmacytoid cells juxtaposed or embedded in immature osteoid matrix (Fig. 4). Because the observation of immature osteoid matrix is difficult to make with certainty on routine direct smears, the cases of osteosarcoma were diagnosed as such after resection. Malignant fibrous histiocytoma demonstrated bizarre multinucleation, brisk abnormal mitoses, and highly pleomorphic spindle cells focally associated with loose matrix material (Fig. 5). A histologic section of MFH showed sheets of large multinucleated cells with pleomorphic nuclei and macronucleoli, abundant bizarre mitoses, and focal necrosis (Fig. 6).

Figure 2.

Highly pleomorphic spindle cells of an epithelioid angiosarcoma are shown (H & E, original magnification × 40).

Figure 3.

The pleomorphic spindle cells and myxoid matrix of a myxofibrosarcoma are shown (Diff-Quik stain, original magnification × 40).

Figure 4.

Pleomorphic naked nuclei, plasmacytoid cells, and osteoid are seen in an osteosarcoma specimen (Papanicolaou stain, original magnification × 40).

Figure 5.

Pleomorphic spindle cells and multinucleated cells in a malignant fibrous histiocytoma are shown (Papanicolaou stain, × 40).

Figure 6.

A histologic section of malignant fibrous histiocytoma from the resection specimen is shown and demonstrates multinucleated cells, nuclear and cellular anisocytosis, and atypical mitotic figures (H & E, original magnification × 40).

As indicated in Table 1, soft tissue tumors often require multiple cytological modalities for proper workup. The higher number of Diff-Quik slides reflects the use of the Diff-Quik inspection of direct smears for the on-site evaluation of adequacy. A cell block was made in 10 cases (77%), and IHC was performed in 4 of these cases. Two cases were found to have sufficient material on core needle biopsy for IHC staining, and 1 case was destained for IHC staining. The most common IHC stains are listed in Table 3 and demonstrate the common practice of staining for pancytokeratins to rule out carcinoma in these tumors. The use of the vascular markers cluster of differentiation 31 (CD31) and CD34 reflects the clinical suspicion for angiosarcoma in this clinical context. IHC was also used for establishing smooth muscle markers in cases of rhabdomyosarcoma and for ruling out local recurrences of the primary tumors with tumor-specific markers.

Table 3. Most Commonly Ordered IHC Stains and Their Frequenciesa
IHC AntigenNo. of Cases
  • Abbreviations: CD31, cluster of differentiation 31; CD34, cluster of differentiation 34; IHC, immunohistochemistry;

  • a

    A total of 7 cases underwent IHC. Stains that were ordered only once are not listed and include cytokeratin (CK) 7, CK20, desmin, myogenin, B-cell lymphoma 2 (Bcl-2), mouse anti-CD99 (O13), prostate-specific antigen (PSA), prostatic-specific acid phosphatase (PSAP), prostein (P501S), p63, thyroid transcription factor-1 (TTF-1), leukocyte common antigen (LCA), and renal cell carcinoma (RwCC).

CD31 or CD343
S-100 protein3
Cam 5.22


PRS is a rare disease that comprises < 5% of all sarcomas; the overall risk of PRS after radiotherapy for any reason is estimated to be between 0.03% to 0.8% of all patients treated with radiotherapy.10, 14 These statistics are complicated by the variability in radiotherapy regimens and latency periods; the postradiation latency interval varies from only a few months to > 50 years.15, 16 Older radiotherapy regimens comprised of low doses of orthovoltage radiation for superficial diseases appear to be associated with longer latency intervals compared with currently favored radiotherapy protocols, which use higher doses of deeply penetrating megavoltage radiation. The latency periods with current protocols is shorter, but there does not appear to be a difference in other parameters such as histological grade or overall survival.17, 18 The incidence of PRS appears to be related most closely to the dose; there does not appear to be a difference between orthovoltage and megavoltage regimens if the dose is equal.10

The identification of a soft tissue tumor as PRS is relevant to an FNA service for 2 main reasons. First, PRS presents at a higher stage of disease and may have a poorer prognosis than de novo sarcoma,8, 13 and therefore identifying it can have treatment implications. Second, the consideration of PRS in a patient and the subsequent clarification of the clinical history can lead to a faster and more accurate diagnosis with less need for ancillary testing. Thus, a history of radiotherapy and malignancy are important components of the patient record in the workup of soft tissue neoplasms by FNA. The current study presents the key clinical and pathological features of patients who have presented for the first diagnosis of PRS by FNA; to the best of our knowledge, the clinical and pathological features of PRS in patients who present for FNA have not been described in a comprehensive publication to date. This could be due to several factors. First, these are rare tumors; the largest studies are comprised of a few hundred cases,12, 19 and the majority of others have smaller numbers. Second, the morphological spectrum of PRS is very broad, and therefore there is not a constellation of findings that can be used to definitively identify a tumor as PRS instead of a spontaneous sarcoma. Third, the malignancy and radiotherapy history can be so remote that they are often not available at the time of FNA or core needle biopsy, and therefore PRS may be underreported.

The study data indicate that these cases are comprised of aspirates that are readily diagnosed as malignant. Only 1 of the 13 cases received a cytopathologic diagnosis that was merely atypical rather than definitive for the diagnosis of sarcoma, and the cellularity of this aspirate was poor. No false-negative results were found. Because these neoplasms are high grade, the diagnosis of a malignant process is highly reproducible across the 3 institutions participating in this study. The more difficult distinction in this context is the classification of the malignant process as a sarcoma and more importantly as a PRS. In this regard, the combination of the clinical history and IHC is effective. Given the varied radiological and histological appearances of PRS, the need to pursue a history of radiotherapy may not be consistently apparent. Because the malignancy history is often known at the time of FNA, awareness of this uncommon entity is important for pursuing the treatment history for any neoplasm. As the experience in the current study demonstrates, the role of IHC in the initial diagnosis of PRS can be useful in that it can rule out pseudosarcomatous changes in a poorly differentiated carcinoma with cytokeratin immunostains. However, in the absence of highly specific antigens for sarcoma, the value of IHC lies more in its exclusionary power.

Imaging studies are known to have a high degree of interpatient variability due to the heterogeneous histological subtypes of PRS that are possible.20, 21 The current study was no exception. The radiographic appearance of these neoplasms ranged from that of a soft tissue sarcoma to osteosarcoma with ossification. This detail reflects a key observation in the tumor presentation, specifically that the malignancy and radiotherapy history in combination with the cytopathologic findings is the most appropriate means of arriving at the diagnosis of PRS. The radiographic studies are more important for staging than for determining the original diagnosis.

Although 2 of the cases were typical of the classic connection between angiosarcoma and radiotherapy for breast cancer, the numbers were insufficient to make any association between primary tumor type and the morphology of the PRS. Moreover, the observation that one of the cases of irradiated breast cancer led to the diagnosis of MFH serves as a reminder of the extremely heterogeneous nature of PRS. Thus, radiotherapy history and the exclusion of carcinoma remain the most important features in the workup of these cases. There is currently no standardized method for predicting which primary tumors lead to a particular morphology of PRS.

The current study has selection bias both because the real number of PRS cases most likely exceeds the number of recognized PRS cases and because the number of PRS cases clearly exceeds the number of FNA procedures on PRS. The bias reflects practice limitations of FNA. For example, PRS in the brain has been described,7, 20, 22 but this site is not aspirated and therefore no cases exist in this series. Similarly, osteosarcoma in the bone after irradiation is the oldest example of PRS.4 Although the current series had 3 cases of osteosarcoma, only 1 was aspirated from bone. These examples underlie the importance of characterizing patients with PRS who undergo FNA so that the identification of other potential cases of PRS can be identified more promptly in the future.

PRS cases have a higher likelihood of IHC features with poor prognostic implications such as p53 and a high proliferative index.23 Although not conclusive proof, these findings, in conjunction with the worse clinical behavior of PRS, indicate that these tumors may be genetically distinct from the primary sarcomas they resemble morphologically. This is an important consideration given that the patients who develop PRS have already received maximal doses of chemotherapy and radiotherapy. Therefore, they receive less extensive therapy than patients with de novo sarcoma. Thus, the search for molecular diagnostic modalities and targeted therapies could provide helpful improvements. Recently, investigators have revealed targetable mutations in postradiation angiosarcoma such as fms-like tyrosine kinase receptor (FLT) and kinase insert domain receptor (a type III receptor tyrosine kinase) (KDR),24 which may provide a rationale for better therapeutic guidance in these tumors in the future. It has yet to be shown whether the tumors accessible to FNA harbor these mutations. Given the relative ease with which FNA material can be obtained and used for mutational analysis, aspiration may prove to be a more prominent diagnostic modality in PRS if these findings are maintained in subsequent investigations.


To the best of our knowledge, the current series is the first of its kind to detail the cases of PRS that undergo FNA. Commensurate with the findings of other publications that are comprised of histopathological findings, the PRS cases in the current study were heterogeneous with regard to their clinical, radiographic, and histological presentations. Thus, the findings reinforce the concept that a history of cancer and radiotherapy should be part of the clinical knowledge base. When feasible, IHC was useful in establishing this diagnosis of exclusion.


No specific funding was disclosed.


The authors made no disclosures.