Introduction

Biopsy is often necessary to diagnose a mass that is indeterminate based on history, physical, laboratory, and imaging studies alone. The goal of biopsy is to obtain diagnostic tissue while minimizing morbidity, limiting potential tumor spread, and avoiding interference with future treatments. Techniques that have evolved to accomplish these goals include open surgical biopsy, core biopsy, and fine-needle aspiration (FNA). Open (incisional) biopsy has long been the gold standard for soft tissue mass diagnosis, with a diagnostic accuracy of 94% to 99% [1, 48]; however, it is expensive ($4321.25 to $7234.00) and carries a complication rate of up to 16%, including hematoma, tumor spread, and wound problems that may interfere with adjuvant treatments [1, 48, 53]. Therefore, less invasive methods have emerged.

Defined as the sampling of tissue through a 20-gauge or smaller needle, FNA has the advantages of speed, convenience, decreased cost (average $1060 per case), minimal morbidity, and a theoretically lower risk of local contamination [1, 10, 20, 52]. Downsides include the limited sample, inaccessibility of some masses, and variable accuracy, especially in the diagnosis of sarcoma. In regard to FNA of general soft tissue masses, the literature reports a wide range of sensitivities (86%–100%), specificities (36%–100%), and diagnostic accuracies (21.9%–98%) [3, 5, 8, 10, 16–18, 20, 32–35, 37, 39, 44, 45, 49, 54]. In these studies, however, nondiagnostic samples have generally been excluded, artificially improving results.

Because of the limited tissue retrieved with FNA, core biopsy has evolved as an alternative, using a 10- to 14-gauge coring needle to obtain cylindrical tissue blocks. A block of tissue allows the pathologist to examine tumor architecture and cellular interrelation, improving the diagnosis of histologic subtype and grade compared to FNA [13, 19, 56]. Other advantages of core biopsy, as with FNA, include speed, convenience, decreased cost (average $1106 per case), minimal morbidity, minimal contamination, and a 0.1% to 1.1% complication rate; disadvantages are also similar to FNA and include limited sampling and inaccessibility of some masses (secondary to size, depth, density, or location) [8, 24, 25, 40, 50, 53, 56]. Improved from FNA, core biopsy’s soft tissue mass sensitivity ranges from 81.8% to 100%, specificity from 91% to 100%, and diagnostic accuracy from 72.7% to 100% [6–9, 12, 19, 23–26, 31, 36, 40, 42, 43, 50, 53, 57–59]. However, as with FNA, these studies often excluded nondiagnostic samples, improving apparent accuracies. In the only previously published Level I study evaluating soft tissue mass biopsy techniques, Yang and Damron [57] elegantly compared FNA and core biopsy to each other in the diagnosis of the same soft tissue mass and found core biopsy to be more accurate than FNA on all accounts, with FNA 64% accurate and core 83% accurate in establishing the specific diagnosis.

No previous study has prospectively evaluated FNA, core biopsy, and open surgical biopsy in the diagnosis of the same soft tissue mass. Therefore, to determine and compare the diagnostic accuracies of these biopsy techniques, we asked the following questions: How do FNA, core biopsy, and open biopsy compare to the final clinical diagnosis (and to each other) in regard to (1) identifying malignancy, (2) establishing the exact diagnosis (grade and subtype), and (3) guiding appropriate treatment?

Patients and Methods

From January 2007 to January 2009, we invited all 106 patients evaluated by the Orthopedic Oncology Service with a palpable primary soft tissue mass not previously diagnosed to participate in this prospective study. Indications for biopsy were inability to confidently characterize the nature of the soft tissue mass with history, physical, laboratory, and imaging studies alone and patient desire for diagnosis. Before initiation of the study, a power analysis was performed. The choice of sample size was made on the basis of the primary outcome of ability to determine malignancy in the tissue sample. We excluded 32 patients who were unwilling to participate in the study and 17 patients with poorly or relatively inaccessible masses adjacent to vital structures. Fifty-seven of the 106 patients met these criteria. Assuming a beta error of 0.05 and a power of 0.80, it was anticipated 45 specimens would be required in each group of FNA, core biopsy, and open biopsy to demonstrate a 10% difference in the accuracy between the groups. Complications were defined as those that caused patient morbidity, required clinical intervention, or interfered with future treatment. Patients were followed for at least 6 months, and no patients were lost. Two masses were in the neck, seven in the back, 18 in the upper extremity, and 30 in the lower extremity (Table 1). The Institutional Review Board and Cancer Research Committee approved this study protocol and all patients were also individually registered with our institution’s Cancer Center Research Participant Registry.

Table 1 Demographic and pathologic information for the 57 patients, with the FNA, core biopsy, open biopsy, and final diagnoses
Full size table

After consent was obtained, each patient was taken to the operating room and placed under general anesthesia. According to a standardized protocol, FNA was performed on the tumor mass in line with the planned surgical incision by a cytopathologist or surgeon trained in FNA technique. After the skin was prepared with an alcohol pad, FNA was performed using a 1.5-inch 23-gauge needle attached to a 20-mL syringe in a standard syringe holder according to standardized guidelines [22]. Three to five passes of the lesion were performed to provide sampling from circumferential areas of the tumor without breaching the far wall. The aspirates were divided into two sets: one air-dried and the other fixed in 95% ethanol. Both sets were stained with hematoxylin and eosin (H&E). Tissue fragments were retrieved from needle rinse and embedded in paraffin for cell blocks and H&E-stained sections. The cell blocks were also used for molecular biology and tumor marker analysis (keratin, vimentin, smooth muscle actin, myoglobin, Factor VIII, CD34, S100, HMB45, CD117, desmin). All FNA biopsies were reviewed by a senior cytopathologist trained in oncology who was given a complete clinical history. The FNA smears were assigned to one of the following categories: malignant, benign, or nondiagnostic. Smears were deemed nondiagnostic when the cells obtained were insufficient for any type of diagnosis. Sarcomas were graded as low grade, high grade, or not gradable. Grading criteria included the presence of mitoses, cellularity, differentiation, nuclear pleomorphism, and necrosis [2, 15, 41]. When possible, a specific histologic diagnosis was reported.

Directly after completion of FNA, core biopsy was performed on the tumor mass in line with the planned incision by the orthopaedic oncology team. The core biopsies were performed using a Tru-Cut® soft tissue biopsy needle (Cardinal Health, Dublin, OH), through the FNA insertion site, taking multiple samples (three to five passes) throughout the tumor circumferentially with care to obtain adequate tissue for evaluation, but not to breach the far wall of the tumor. The biopsies were not sent for frozen section analysis, were fixed immediately in 10% buffered formalin, and were stained routinely with H&E. Histochemical stains (alkaline phosphatase and Prussian blue) were applied. Special stains, such as van Gieson, McManus, and reticulin (Gordon-Sweet), were applied when appropriate. For those specimens suggestive of sarcoma, immunohistochemical stains were selected from a panel of antibodies: cytokeratins (AE1/AE3, CAM1.2, MNF116, CK5, CK7, CK20), epithelial membrane antigen, S100, HMB45, melanin A, desmin, actin (muscle-specific actin), α-smooth muscle actin, h-caldesmon, vimentin, CD30, CD15, CD45, CD45-RO, CD20, CD3, CD10, CD5, CD23, bcl-2, MIB-1, CD34, CD31, Factor VIII, kappa- and lambda-light chains, and osteonectin. Two senior musculoskeletal pathologists specializing in orthopaedic oncology independently, and blindly with regard to other specimens taken from the same patient, reviewed the core biopsy specimens. The biopsy specimens were classified as diagnostic or nondiagnostic based on the adequacy of the tissue obtained for histologic analysis to yield any basic diagnosis. The specimens were evaluated for the nature of the lesion (benign or malignant), specific histologic diagnosis, and grade. Sarcomas were graded according to the American Joint Committee on Cancer grading criteria as either low grade (Grades 1 and 2) or high grade (Grades 3 and 4) [22].

After FNA and core biopsy procedures were performed, open biopsies were performed by a trained orthopaedic oncologist in accordance with sarcoma principles. An incision was made through the FNA and core puncture site and in line with the planned resection, and specimens were sampled from the tumor periphery until frozen section revealed adequate sampling. The nonfrozen tissue was immediately placed in 10% buffered formalin. The tissue handling, fixation, and staining and pathologic analysis were identical to those of the core biopsy specimens. The results of the FNA, core, and open biopsy were compared to both the complete resection final pathology reports and the final clinical diagnosis given to the patient.

Outcome variables of determining malignancy, determining exact diagnosis, and guiding eventual treatment for FNA, core biopsy, and open biopsy were measured against the final clinical diagnosis determined by analysis of the completely resected specimen in combination with the final clinical impression. Malignancy and exact diagnosis (subtype and grade) were determined by a cytopathologist (in the case of FNA) and by a musculoskeletal pathologist (in the case of core and open biopsy) as previously described. Concordance with indicated treatment was determined by comparing the indicated treatment determined by a trained orthopaedic oncologist with the given diagnosis resulting from the FNA, core, and open biopsy final pathologic results.

All information collected in this study was recorded and analyzed using SPSS® software (SPSS Inc, Chicago, IL). Sensitivities, specificities, positive predictive values (PPVs), negative predictive values (NPVs), and concordances were determined. These values were then compared with the t test for proportions, set to a 95% confidence interval.

Results

Adequate tissue sample to determine any kind of diagnosis was obtained in 50 of the 57 (87.7%) FNAs, in 49 of 57 (86.0%) core biopsies, and in 57 of 57 (100%) open biopsy specimens. These diagnostic sample proportions were similar in both benign (29 of 33 [87.9%] for FNA and 28 of 33 [84.9%] for core) and malignant (21 of 24 [87.5%] for each) cases. Open surgical biopsy determined malignancy (or a benign diagnosis) correctly 100% of the time when compared to the complete resection and final clinical diagnosis results, with FNA only 75.4% (p < 0.0002) accurate and core biopsy only 80.7% (p < 0.0015) accurate in this regard. Open biopsy had better results than both percutaneous techniques in terms of sensitivity, specificity, PPV, NPV, and concordance with the final diagnosis (Table 2).

Table 2 Accuracy of biopsy techniques in regard to determining malignancy when compared to the final diagnosis
Full size table

As with determining malignancy, open biopsy was able to determine the correct grade and subtype in 100% of cases, with no discrepancies after full resection and final clinical diagnosis. FNA and core biopsy were concordant with the final exact diagnosis in 33.3% and 45.6% of cases, respectively, which were both less (p < 0.0001) concordant than open surgical biopsy (Table 3).

Table 3 Summary of FNA, core biopsy, and open biopsy concordances with the final diagnosis
Full size table

Correct treatment would have been initiated in 38.6% and 49.1% of cases on the basis of FNA specimens and core specimens, respectively, while all patients who underwent open biopsy would have had correct treatment initiated based on the biopsy results. Compared to open biopsy, FNA and core biopsy were both less (p < 0.0001) accurate in this regard (Table 3).

Five of the 23 FNA samples that reported malignancy turned out to be pigmented villonodular synovitis (two), schwannoma (two), and myxoma (one) on final pathology (Table 1). Of the five pigmented villodular synovitis and schwannoma cases, four were reported as malignant and one as nondiagnostic on FNA, while on core biopsy, one was reported as malignant and two as nondiagnostic. None of the open surgical biopsy results were changed after complete resection or final clinical impression.

There was one complication of the 57 cases (1.8%) in which a patient developed a wound dehiscence 10 days after the procedure on the posterior neck, which was successfully treated nonoperatively with dressing changes and oral antibiotics, with no effects on the patient’s ultimate treatment and outcome. There were no postbiopsy hematoma complications causing morbidity, requiring intervention, or compromising treatment or outcome.

Discussion

Open biopsy has long been considered the gold standard for diagnosis of an extremity soft tissue mass [1, 11, 14, 38, 51]; however, proponents of percutaneous techniques suggest FNA or core biopsy is just as effective and should replace open biopsy as the method of choice [28–30, 40, 56]. No study has prospectively compared the accuracy of these three biopsy techniques in a standardized fashion. Therefore, we asked how accurate FNA, core biopsy, and open surgical biopsy are and how they compare to each other with regard to determining malignancy, establishing the exact diagnosis, and guiding the appropriate treatment.

We note some limitations to our study. First, we studied only palpable and safe soft tissue masses, which excludes deeper masses and those in close proximity to neurovascular structures, which may be more challenging to sample. Second, the need for patient consent could potentially make the cohort less representative, since a patient with an aggressive tumor may be less likely to engage in “experimental” surgery. Third, the diagnostic standard by which all three diagnostic techniques were judged was based on the complete surgical resection and the final clinical impression of the orthopaedic oncologist; although this measure is the best we have, this diagnosis could still be wrong and skew our comparative results. Lipomas are diagnostic based on MRI alone and are notoriously difficult to diagnose through percutaneous techniques or on frozen sections; our study included eight lipomas, and their inclusion may skew results and detract from data regarding masses that cannot be diagnosed using available noninvasive techniques. Finally, the accuracy of percutaneous biopsies depends on the operator technique of biopsy and pathologic analysis; much attention was placed toward proper biopsy technique in all cases and having the pathology read by a well-informed and specialized cytopathologist (in the case of FNA) and musculoskeletal pathologist (in the case of core and open biopsies).

Adequate tissue sample to determine any kind of diagnosis was obtained in all of our open biopsy specimens but in only 87.7% of the FNAs and 86.0% of the core biopsies; furthermore, open biopsy was more sensitive, specific, predictive, and accurate in regard to determining malignancy than the percutaneous techniques. Literature supports the disconcerting fact that a percutaneous biopsy result negative for malignancy is not confirmatory [55], and our FNA and core biopsy NPVs also support this claim (Table 2). As sarcomas enlarge, they typically outgrow their blood supply, leading to areas of central necrosis, and inadvertent sampling of these areas may lead to nondiagnostic specimens [11, 14, 38, 51]. In addition, mesenchymal tumors or sarcomas are difficult to diagnose (even basically) on cellular morphology alone without visualizing the stromal structure [19, 21, 56], especially in regard to spindle cell tumors [46]. The improved diagnostic sampling in the open biopsy group is likely due to the sending of frozen sections until histologic evidence of diagnostic sampling was achieved, a procedure not followed with the percutaneous techniques. We compared our findings with those of the most applicable published studies regarding FNA (Table 4) and core biopsy (Table 5) of extremity soft tissue masses, although it should be noted previous studies often excluded nondiagnostic samples in their accuracy calculations.

Table 4 Comparison of our study with published literature regarding FNA of extremity soft tissue masses
Full size table
Table 5 Comparison of our study with published literature regarding core biopsy of extremity soft tissue masses
Full size table

The exact diagnosis (matching grade and subtype) was obtained in 100% of open biopsies and in only 33.3% and 45.6% of FNA and core biopsies, respectively. Decreased accuracy of FNA (Table 4) and core biopsy (Table 5) in regard to determining the grade and subtype was also found in other studies, even with most of the studies excluding nondiagnostic samples. Identifying the correct grade and exact histologic subtype is critical in the case of malignancies [2]. Proponents of percutaneous techniques suggest taking multiple samples from multiple different locations within the tumor to obtain a representative sample [25] or combining both FNA and core biopsy to improve the accuracy in subtype diagnosis [19]. However, multiple passes from varying locations have the potential to increase contamination and thereby theoretically increase the risk for local recurrence or theoretically cause distant spread by introducing tumor cells directly into the vasculature [11, 14, 38, 51].

Previous studies have looked at accuracy of biopsy techniques in terms of determination of malignancy, grade, and subtype, but none have looked at the effect on eventual treatment. Since guiding the management plan is the ultimate benefit of any test, we added this outcome measure. Again, open biopsy guided the appropriate treatment 100% of the time; however, based on FNA and core biopsy specimens alone, correct treatment would have been initiated only 38.6% and 49.1% of the time, respectively.

Proponents of percutaneous techniques cite less morbidity and fewer complications [2]. Prior studies of open biopsy have reported complications ranging between 0% and 17% [11, 14, 38, 51]. Most studies on FNA either do not specifically state complications or have a very low complication rate (0%–1%) [2–5, 10, 15–18, 20, 27–30, 32–35, 37, 39, 44, 45, 49, 54, 55], with most complications related to local tenderness and bleeding. Core biopsy has a reported complication rate ranging between 0% and 7.4%, most commonly hematoma, bleeding, and infection [9, 47, 56]. In our series, there was only one complication of 57 cases (1.8%), which was a wound dehiscence treated nonoperatively without any effect on the patient’s further treatment.

In summary, core biopsy had greater sensitivity, specificity, predictive value, and accuracy than FNA in regard to determining malignancy; however, both were inferior to open surgical biopsy, with FNA only 75.4% accurate and core biopsy only 80.7% accurate in this regard; therefore, a negative FNA or core biopsy result does not ensure absence of malignancy. Core biopsy had greater accuracy than FNA in regard to establishing the exact diagnosis and guiding the appropriate treatment; however, these accuracy values for both FNA and core were very low (< 50%) and both were inferior to open biopsy. Therefore, we recommend open biopsy of indeterminate soft tissue masses as a more reliable, accurate, and confirmatory means of determining malignancy, establishing the exact diagnosis, and guiding their appropriate treatment.