Abstract and Introduction

Abstract
We reviewed the clinicopathologic features of 145 consecutive fine-needle aspiration biopsy (FNAB) specimens from 140 patients without a previous diagnosis of sarcoma. Among 138 adequate specimens, 42 bone sarcomas and 80 soft tissue sarcomas were recognized as sarcomas; histologic subtyping was easier in bone than in soft tissue sarcomas and in pediatric than in adult cases. There was no correlation in accuracy of subtyping in low- vs high-grade sarcomas. FNAB was most accurate for subtyping of skeletal osteosarcoma, pediatric small round cell bone/soft tissue sarcomas, synovial sarcoma, skeletal chondrosarcoma, and adult myxoid soft tissue sarcomas. Although almost always recognized as sarcoma, subtyping of adult pleomorphic soft tissue sarcomas generally was not possible but did not influence therapy; all were considered high-grade sarcomas for treatment purposes. There were 4 misinterpretations of subtype in soft tissue sarcomas; none resulted in a change in therapy. Cytogenetic analysis on aspirated material confirmed t(11;22) in 2 Ewing and t(X;18) in 3 synovial sarcomas. No procedure-related complications occurred. Among bone and soft tissue sarcomas, FNAB was sufficient for initiation of definitive therapy in 87% and 83% of patients, respectively. Most FNAB specimens from bone and soft tissue sarcomas are recognized easily as sarcoma, but subtyping seems more accurate in bone sarcomas. Although histologic subtyping of adult soft tissue sarcomas is often impossible, no influence on initial therapy is usually observed. In contrast, subtyping of pediatric sarcomas by FNAB seems highly accurate and is necessary for appropriate therapy.

Introduction
Although it is well-acknowledged that fine-needle aspiration biopsy (FNAB) is a valuable diagnostic tool for locally recurrent and metastatic sarcomas, the use of FNAB for the primary diagnosis of sarcoma, omitting core needle or traditional open incisional biopsies, remains poorly understood and controversial.[1,2] Recently, the Association of Directors of Anatomic and Surgical Pathology (ADASP) published "guidelines" regarding the procedure.[3] However, many of the conclusions drawn by the panel seem vague and unsupported by references. For example, the authors report that regarding FNAB, "Precise typing and grading is often not possible..." and the "morphologic heterogeneity of soft tissue tumors renders them susceptible to misdiagnosis by this technique."[3] Consequently, such statements generate more questions than they answer. How often is subtyping of a sarcoma not possible? How frequently does "morphologic heterogeneity" lead to a misdiagnosis? Does the inability to determine a specific histologic subtype affect therapy in all cases? Other equally important questions about FNAB were not addressed by the ADASP. Should the patient's interests (as well as risks) also be considered when deciding on the method of biopsy? Finally, in this age of cost containment, is the procedure cost-effective, even when taking into account the risk for sampling error?

The answers to these questions about FNAB and the primary diagnosis of sarcoma have not been addressed completely in the literature.[1] In part, this is related to the fact that sarcomas are relatively rare lesions, and any attempt to study them requires large numbers of cases, ensuring that the full spectrum of histologic subtypes is represented.

We report herein our experience with 140 patients with sarcoma (procured from a total of 1,246 cases of benign and malignant bone and soft tissue aspirates), initially analyzed by FNAB, with emphasis on the importance of the multidisciplinary approach, accuracy of histologic subtyping, and the relationship of FNAB interpretation to therapeutic significance. We should also point out that the purpose of our study is not to disprove the contentions of the ADASP, but rather attempt to answer, if only through our experience, many of the questions raised about the role and limitations of FNAB in the management of musculoskeletal sarcoma. This study represents an expansion of our previously reported experience limited to 67 patients with soft tissue sarcoma.[4]

Materials and Methods
We report our ongoing experience involving 145 consecutive FNAB bone (49 aspirates) and soft tissue (96 aspirates) sarcoma specimens (135 primary tumors, 1 local recurrence, 9 metastases) from 140 patients (91 patients with soft tissue sarcoma; 49 patients with skeletal sarcoma), none of whom had a previously established sarcoma diagnosis. All cytologic samples were obtained without general anesthesia using 23- and 25-gauge needles by the standard manual method (124 aspirates) and computed tomography-guided or ultrasound-guided needle placement (21 aspirates) at Wake Forest University Baptist (WFU) Medical Center, Winston-Salem, NC (109 aspirates), and the University of North Carolina (UNC) Hospitals, Chapel Hill (35 aspirates). One case was obtained at an outside hospital. Five patients with primary soft tissue sarcoma underwent concomitant FNABs of the primary tumor and a presumed metastatic site. A total of 7 experienced pathologists participated in the interpretation of these specimens; however, all material was reviewed by one of us (S.E.K.), who also served as the primary pathologist or the consultant on most (>90%) cases. No individual pathologist was responsible for a disproportionate number of inaccurate diagnoses. The cytologic diagnosis used for statistical purposes represented the diagnosis (not the retrospective review) given by the pathologist(s) who signed out the report. Histologic material in the form of cell blocks, incisional biopsy specimens (only for nondiagnostic FNAB specimens), and/or excisional biopsy specimens/resection specimens were available for review in all cases.

With the exception of the single outside case, the remaining specimens were processed as follows: Aspirated material was expelled onto glass slides and smeared using a second glass slide. Half of the smears were air dried and stained with Diff-Quik (Fisher Scientific Biomedical Sciences, Swedesboro, NJ). An equal number of slides were immersed immediately in 95% ethanol for staining by the Papanicolaou method. Cell blocks were prepared from 103 FNAB specimens. The procedures for producing a cell block varied between the 2 institutions.

At WFU, aspirated material was dispensed onto a glass slide, allowing the specimen to clot and placing the clot into formalin for processing as a routine surgical specimen. At UNC, aspirated material was dispensed directly into Plasma-Lyte with 5% dextrose (Baxter Healthcare, Deerfield, IL), spun down to form a pellet, resuspended, and mixed with coagulation factors, including thrombin. The resulting clot then was placed in a cassette in formalin for processing as a routine surgical specimen. We routinely order between 6 and 10 unstained recut slides for immunocytochemistry if needed.

Ancillary studies (eg, immunocytochemistry, flow cytometry, cytogenetics) were performed in a total of 41 cases. When available and necessary (31 cases), sections of cell block material were examined immunocytochemically using the avidin-biotin-peroxidase complex and using commercially available antibodies to the following antigens: S-100 protein (polyclonal, 1:2; Dakopatts, Carpinteria, CA), cytokeratin (monoclonal, AE1/AE3, 1:100; Boehringer Mannheim, Indianapolis, IN), vimentin (monoclonal, 1:25; BioGenex, San Ramon, CA), desmin (monoclonal, D33, 1:300; Dakopatts), muscle-specific actin (monoclonal, HHF-35, 1:25,000; Enzo Diagnostics, Farmingdale, NY), and 013 (monoclonal, CD99, 1:40; Signet, Dedham, MA). Appropriate positive and negative controls were used throughout these procedures. In 2 tumors, aspirated material was placed in 6% glutaraldehyde and processed for electron microscopy by standard protocol.

At UNC and WFU Medical Center, we routinely perform FNAB on accessible bone and soft tissue lesions in adults and children, in outpatient and inpatient settings. Our cytopathology services maintain mobile FNAB carts equipped with Diff-Quik stain for immediate assessment, 95% ethanol for Papanicolaou staining, glass slides, syringes and syringe holder, and a double-headed microscope. On notification of an eligible patient for FNAB, the cart (at UNC a handheld basket also is used) is taken to the patient's room by a cytotechnologist and/or pathology resident accompanied by an attending pathologist. Pertinent radiographs, including plain films, computed tomography scans, and magnetic resonance images, if available, are reviewed and clinical data are retrieved. Immediately following the FNAB, a portion of the sample is screened using Diff-Quik stain, and a preliminary interpretation is given. If necessary, a second or third FNAB may be performed for additional material or studies (eg, flow cytometry, electron microscopy, cytogenetics, ploidy analysis). Each study requires an additional complete FNAB pass. For immunohistochemical evaluation, we prefer a formalin-fixed block to cytologic smears, as adequate positive and negative histologic controls are readily available. As with ancillary studies, cell blocks were obtained only if they were thought, based on preliminary interpretation, to be clinically necessary.

Specimens for cytogenetics were collected in RPMI 1640 tissue culture medium and processed according to the methods of Kilpatrick et al.[2] For immunophenotyping and ploidy analysis by flow cytometry, aspirated material was collected in RPMI media and processed by standard protocol in 2 cases. The flow cytometry markers included CD5, CD7, CD10, CD13, CD19, CD20, CD45, kappa, and lambda.

With the exception of the myxoid sarcomas, we generally did not transcribe a histologic grade in the cytologic report of soft tissue sarcomas. A low, intermediate, or high grade was rendered in cases of myxofibrosarcoma, based on the tumor hypercellularity and the amount of background myxoid granular film, as these parameters usually are inversely related.[5] For most sarcomas, if the subtype (eg, synovial sarcoma) or cytomorphologic group (eg, adult pleomorphic sarcoma) could be ascertained by cytologic material, a specific grade was not necessary for treatment. However, as some sarcomas may manifest as low- or high-grade tumors, if clinically indicated, we attempted to favor either low or high grade based on the degree of nuclear pleomorphism and the presence (not the percentage) or absence of tumor necrosis. Table 1 illustrates this approach, the results of which have been reported.[4] Among bone sarcomas, 2 aspirates of conventional skeletal chondrosarcoma were designated as low grade based on the abundance of matrix material, the low cellular yield, and the relative uniformity and minimal nuclear pleomorphism of the tumor cells. Osteosarcomas were graded on a 4-tiered scale based on the Broders scheme,[6] emphasizing cytomorphologic features, including nuclear atypia and pleomorphism.

Results

Histologic Subtyping
There were 7 (4 bone and 3 soft tissue) unsatisfactory samples (defined as absent or too few cells for diagnosis). Table 2 summarizes the accuracy of FNAB for various histologic subtypes. The first column represents the final diagnosis based on histologic material. An accurate diagnosis was defined as an exact "line" diagnosis or, less commonly, the terminology could include the phrase "consistent with" the specific entity. A cytopathology report that included the designation "favor" for a specific sarcoma was not considered an accurate diagnosis. In all cases, an "accurate diagnosis" was followed by definitive therapy based solely on the cytologic diagnosis. Among adequate specimens, 42 (93%) bone sarcomas and 80 (86%) soft tissue sarcomas were diagnosed as at least sarcoma; however, histologic subtyping was more easily accomplished in bone than in soft tissue (82% vs 54%) sarcomas.

Table 3 summarizes the accuracy of FNAB for histologic subtyping for pediatric (19 years of age or younger) vs adult cases of sarcoma. Histologic subtyping of pediatric sarcomas was more accurate than that observed in adult (36/39 [92%] vs 51/99 [52%]) cases. More specifically, FNAB was most accurate for histologic subtyping of skeletal osteosarcoma (25/28 [89%]), pediatric small round cell bone or soft tissue sarcomas (18/20 [90%]), synovial sarcoma (8/9 [89%]), skeletal chondrosarcoma (9/9 [100%]), and adult myxoid soft tissue sarcomas (12/16 [75%]). Although almost always recognized as sarcoma (19/20 [95%]), histologic subtyping of adult pleomorphic soft tissue sarcomas was generally not possible but had no influence on therapy, as used at our institutions. Paraffin-embedded cell blocks were generally useful only from the standpoint of providing material for ancillary studies (specifically immunohistochemistry).

Among bone sarcomas, there were 3 cases of sampling error, possibly secondary to morphologic heterogeneity. A 58-year-old man with a clinically and radiologically typical giant cell tumor initially was given a diagnosis by FNAB as simply "giant cell tumor." Subsequent curettage revealed an osteosarcoma arising in a giant cell tumor (or so-called malignant or dedifferentiated giant cell tumor). Two cases of dedifferentiated chondrosarcoma were clearly recognized as high-grade sarcoma, but the low-grade cartilaginous component was not present in the cytologic preparations. In 1 patient, the FNAB was performed concomitantly with a core needle biopsy, also revealing a nonspecific high-grade sarcoma without the concomitant low-grade chondrosarcoma. Open incisional biopsy and resection confirmed the diagnosis of dedifferentiated chondrosarcoma in both cases. There were no cases of sampling errors due to the heterogeneity of the tumor (eg, extensive necrosis) among soft tissue sarcomas. When x-ray films and scans were available, we preferentially sampled areas thought to contain viable tumor, performing multiple passes if deemed necessary.

Ultrastructural examination helped confirm the diagnosis of 1 case of Ewing sarcoma. Immunophenotyping by flow cytometry effectively excluded a hematopoietic process in 2 cases of rhabdomyosarcoma. Cytogenetic analysis on aspirated material confirmed the t(11;22) in 2 cases of Ewing sarcoma and the t(X;18) in 3 cases of synovial sarcoma. Interestingly, all of the latter cases were diagnosed easily, and treatment was initiated based on the cytomorphologic features, clinical features, and the results of immunohistochemical testing. Cytogenetic analysis was attempted but was unsuccessful for the evaluation of 1 case each of embryonal rhabdomyosarcoma, neuroblastoma, Ewing sarcoma, extraskeletal myxoid chondrosarcoma, and well-differentiated liposarcoma (atypical lipoma). However, excluding the single case of atypical lipoma, the inability to perform cytogenetic analysis did not preclude the rendering of an accurate diagnosis.

Histologic Grading
Table 4 summarizes the accuracy rate for histologic subtyping by FNAB for low- vs high-grade sarcomas. There was no correlation between the accuracy of histologic subtyping in low-grade (22/34 [65%]) vs high-grade (65/104 [62.5%]) sarcomas. A histologic grade (eg, low, intermediate, or high) was applied and documented in the cytopathology report for skeletal chondrosarcoma (2 cases), myxofibrosarcoma (3 cases), and leiomyosarcoma (1 case). Resection specimens confirmed the histologic grade in all of these cases. However, 1 case of myxofibrosarcoma contained a focal area (