Percutaneous fine-needle aspiration (FNA) and core- needle biopsy techniques are frequently used methods for obtaining a histopathologic diagnosis of musculoskeletal tumors. Percutaneous biopsy of previously undiagnosed lesions has multiple advantages compared with open biopsy. It is quick, inexpensive,13,15 less morbid,11 less likely to allow seeding of tumor cells,2 minimally invasive, rarely requires general anesthesia, and often obviates the need for open biopsy.
Recent studies have investigated percutaneous techniques in patients with suspected musculoskeletal tumors and found their diagnostic accuracy between 80% and 97% compared with open biopsy.4,6-9,11,12,16 However, based on our experience we have questioned the accuracy of percutaneous biopsies and our ability to use that information to make definitive treatment decisions regarding radiation, chemotherapy and/or surgery. Other studies have tried to express differences between a successful biopsy and a clinically accurate result.4,5 Fraser-Hill and Renfrew5 found percutaneous needle biopsy of primary tumors must be performed with the knowledge that, even though technically accurate, such biopsies may be of limited clinical value. Duda et al4 reported successful retrieval of an adequate tissue sample in 89%, but a clinically useful histologic result in only 80% (80/100) of computed tomography-guided percutaneous bone biopsies examined in their study.
We hypothesized the clinical utility of this procedure in musculoskeletal oncology patients is lower than that suggested by diagnostic accuracy rates reported previously. Additionally, we hypothesized certain characteristics (FNA without core needle biopsy and soft-tissue, myxoid, and primary tumors) reduce the clinical utility and increase the probability of proceeding to subsequent open biopsy.
MATERIALS AND METHODS
We retrospectively analyzed 120 consecutive patients referred from our musculoskeletal tumor surgery practice with suspected musculoskeletal tumors and who had percutaneous biopsies performed without prior needle biopsy over 1 year by interventional radiologists. Biopsy specimens were analyzed by pathologists in our institution. The seven percutaneous biopsies (two soft tissue, five bone) and 111 open biopsies performed by musculoskeletal tumor surgeons at this hospital during this same time period were not included in this study. These were performed for convenience of the surgeon or patient. The type of biopsy performed was left to the discretion of the orthopaedic surgeon (CMO, EJF, RDL). Although there was no set protocol, open biopsies were more likely to be performed at the time of definitive treatment for metastatic bone disease and benign aggressive bone tumors (eg, giant cell tumor). Patients were referred to interventional radiology (IR) for percutaneous biopsies for a variety of reasons including lesions difficult to access (eg, vertebral body lesions), substantial cardiac risk factors precluding general anesthesia, physician preference, and patient convenience. The type of percutaneous biopsy was left to the discretion of the interventional radiologists and may have been influenced by location and tissue type. We did not have information on tumor depth or bone lesion type (sclerotic, lytic, soft tissue mass present), which may have affected the ease of biopsy and thus the specimen quality and quantity. Complications related to percutaneous biopsy and the number of bone lesions that were associated with a soft-tissue mass were not recorded.
Medical records for each patient were reviewed. Histopathology results from percutaneous biopsies, subsequent open biopsies, and surgical procedures were recorded. The results of percutaneous biopsies were compared with the final diagnoses determined by subsequent surgical pathologic specimens or by clinical evaluation, radiographic assessment, and clinical course. We attempted to define the percutaneous biopsy results in terms of decision making for the orthopaedic surgeon. The percutaneous biopsy result was classified as clinically useful, incorrect, or nondiagnostic. Results were classified as clinically useful if correct treatment could be started based on the biopsy. Treatment in this clinically useful group was nonsurgical (observation, chemotherapy, radiation, antibiotics) and/or surgical. We included patients in this group only if a subsequent biopsy confirmed the initial biopsy. Results were incorrect if surgical specimens did not confirm percutaneous specimen. Nondiagnostic results were those not definite enough to start treatment.
We compared our study to the literature regarding musculo- skeletal percutaneous biopsy accuracy rates performed predominantly by interventional radiologists. We examined the definitions of accuracy in the literature and compared it to our definition of accurate. We focused our comparisons of accuracy rates to studies on musculoskeletal biopsies that included tumor in the differential diagnosis.
Patients underwent core biopsy (3/120), fine-needle-aspirate (FNA) biopsy (50/120), or both (67/120). When core and FNA biopsies both were performed, they were performed at the same time. Of the 120 patients, 59 were men and 61 were women. There were 34 tumors located in the upper extremity, 50 in the lower extremity, eight in the spine, 22 in the pelvis, six located elsewhere (axilla, clavicle, groin, sternum). Biopsies sampled bone in 71 patients and soft tissue in 49. Nine percutaneous biopsies were described as having myxoid features, whereas 111 did not. The final diagnoses were primary benign tumor in 38 patients, primary malignant tumor in 20, metastatic tumor in 28, infection in four, and nonneoplastic processes in 30 patients (Table 1). The specific diagnoses included two fibrous dysplasias, four ganglion cysts, 22 metastases, one malignant glomus tumor, three chondrosarcomas, one myxoid liposarcoma, two atypical lipomas, one lipoma, four plasmacytomas, one desmo- plastic fibroblastoma, four osteosarcomas, seven enchondromas, two pigment villonodular synovitis, one Ewing's sarcoma, one osteoblastoma, three lymphoma, one osteochondroma, three giant cell tumors, two hemangiomas, one hemangiopericytoma, two hemangioendothelioma, one high-grade sarcoma, one soft tissue sarcoma, three myxomas, two fibromyxoid sarcomas, one synovial sarcoma, one neurofibroma, one desmoid, one angioleiomyoma, one malignant fibrous histiocytoma, five infections, one noncaseating granuloma, one synovitis, one rheumatoid pannus, four schwannomas, one Baker's cyst, one tumoral calcinosis, one brown tumor, one case of wear debris, and one neural tumor. Twenty-three patients had normal tissue.
The type of percutaneous biopsy (core, FNA, or both), tissue type (bone versus soft-tissue), myxoid type (myxoid versus nonmyxoid), and tumor type (primary benign, primary malignant, metastatic, infection, or nontumor) were recorded. Patients with abscesses or osteomyelitis were grouped in the infection tumor type, whereas those without infection or tumor (patients with bone bruises, stress fractures, etc) were grouped into the nontumor category.
The percentages of clinically useful, nondiagnostic, and incorrect biopsies and the percentage of biopsies leading to open biopsy were calculated for each subtype. The effects of biopsy type, tissue type, myxoid type, and tumor type on clinical utility and probability of proceeding to open biopsy were tested using the Fisher's exact test and logistic regression. Because only three patients underwent core biopsy alone, these patients were excluded from logistic regression to avoid bias. Only p values of < 0.05 were considered significant.
We calculated sensitivity, specificity, and positive and negative predicted value by categorizing our percutaneous needle biopsy results into diagnostic categories (ie, metastatic lesion) based on histologic features, clinical history, and radiologic findings. Using metastatic lesions as the example, a biopsy that showed metastatic disease and was consistent with a final diagnosis of metastatic disease was considered a true-positive (TP) result. A biopsy that showed metastatic disease and was not consistent with a final diagnosis of metastatic disease was considered a false-positive (FP) result. A biopsy that did not show metastatic disease and did not have a final diagnosis of meta- static disease was considered a true-negative (TN) result. A biopsy that did not show metastatic disease but had a final diagnosis of metastatic disease was considered a false-negative (FN) result.
Ninety of the 120 percutaneous biopsy results (75%) were clinically useful. Twenty-seven (22.5%) were nondiagnostic, and three (2.5%) were incorrect. Thirty (25%) were followed by open biopsy (Table 1). Clinical considerations weighed heavily in the interpretation of needle biopsies, therefore open biopsy was occasionally required to mitigate between clinical and percutaneous diagnoses. This clinical suspicion of doubt in the percutaneous biopsy results was validated in three patients, despite a definitive diagnosis from the percutaneous biopsies.
The use of FNA and core biopsies together was more clinically useful (81%) than the use of FNA alone (68%) (p = 0.037). We found no difference between the use of both biopsies and core alone because of the low number of patients (3) in the latter group. The use of both biopsies produced clinically useful results 81% of the time, whereas FNA alone and core alone were useful 68% and 67% of the time, respectively (Fig 1). Although patients undergoing both biopsies tended to undergo open biopsy less often (18%) than those having either alone (34% and 33%) (Table 1), there was no difference in the open biopsy rates among the three groups (Table 2).
Percutaneous biopsies of bone lesions were more useful (p = 0.044) than those of soft-tissue lesions. Bone biopsies were clinically useful 83% of the time, whereas soft- tissue biopsies were useful 63% of the time. There was no difference in the open biopsy rate between the bone and soft tissue groups. Percutaneous biopsies of nonmyxoid lesions were more useful (p = 0.001) than biopsies of lesions displaying myxoid features. Only one of the nine (11%) myxoid lesion biopsies was found to be clinically useful, compared to 80% of nonmyxoid lesions. The open biopsy rates for nonmyxoid (23%) and myxoid lesions (44%) were not statistically different.
Tumor type had no effect on clinical utility. Clinical utility rates ranged from 50% for lesions caused by infection to 86% for metastatic lesions (Table 1). The percentage of patients undergoing subsequent open biopsy was 25% and ranged from 13% for benign primary tumors to 75% for lesions related to infection. However, there was no effect of tumor type on the probability of proceeding to open biopsy.
Sensitivity and negative predicted value (NPV) were varied, but specificity and positive predicted value (PPV) were relatively consistent (Table 3). Sensitivity ranged from 85% for metastatic lesions to 50% for infection.
In the cases where the biopsy was incorrect, one patient had a FNA consistent with pigmented villonodular synovitis, but the final pathology was not consistent with pigmented villonodular synovitis. The second patient had an FNA that was consistent with a low-grade myxoid lesion, but the final pathology showed hemangioma. The third incorrect biopsy was a core that showed fragments of skeletal muscle but the final pathology showed desmoplastic fibroblastoma.
Percutaneous biopsies are frequently used for musculo- skeletal lesions and have advantages to open biopsy. They are often correct and convincing enough to direct clinical treatment, obviating the need for open biopsy. Recent literature suggests the accuracy rate for percutaneous biopsies in musculoskeletal lesions is between 80% and 97%.4,6-9,11,12,16 However, we believe the clinical utility of percutaneous biopsies is less than suggested by most of the previous literature. We also thought the percutaneous biopsy utility rate would be affected by biopsy type, tissue type, and tumor type.
Important limitations of the study include its retrospective nature, selection bias of the surgeon's decision regarding biopsy type, and the lack of followup open biopsies on many patients. The retrospective nature of the study is most limiting in the lack of strict criteria for needle or open biopsy. Surgeon bias may have resulted in the more difficult to access lesions being sent for needle biopsy, potentially lowering our reported rate of clinical utility. Subsequent open biopsy or surgical procedure in which pathologic tissue could be obtained was not thought necessary in 58 patients (48%), primarily in nonoperative management of benign and metastatic lesions. No other studies avoided this limitation. This limitation exposes the study to error in the case of incorrect pathologic or radiologic interpretation.
Published accuracy rates for percutaneous biopsies of musculoskeletal lesions vary and differ from our study largely because of the definition of accuracy. Studies that reported high accuracy rates (95-97%)9,12 excluded non- diagnostic biopsy results in their calculation of accuracy, whereas those that included nondiagnostic biopsies in their calculation have lower reported accuracy rates (74- 79%).1,17 One study that considered clinical utility found percutaneous biopsies may be less clinically useful (71%) than the accuracy rates suggest.5 Because we included nondiagnostic biopsies and considered whether the biopsy result allowed treatment to proceed, we believe our clinical utility rate is more indicative of the usefulness of percutaneous biopsies than many accuracy rates published previously. The effect of setting on biopsy is undetermined, and the majority of the literature concerns biopsies performed by interventional radiologists, as in our study.5,7-9 Some of the literature reports accuracy rates of biopsies performed by orthopaedic oncologists.16
Similar to Schweitzer et al,14 we found the combination of FNA and core biopsies were better than FNA alone, indicating the complementary role of FNA and core biopsies in patients with suspected musculoskeletal lesions. Another study, with 43 patients with sclerotic bone lesions, showed 13% of patients had one correct biopsy and one incorrect biopsy.10 These authors agreed FNA and core biopsies are complementary, and a substantial number of their patients would have been misdiagnosed if they had not undergone both FNA and core biopsies.
We found percutaneous biopsies of bone lesions to be more useful than those of soft-tissue, but other studies with fewer patients showed mixed results. One study of 69 musculoskeletal percutaneous biopsies found similar accuracy rates in bone lesions (86%) and in soft-tissue lesions (77%),6 although they had a small number (13) of soft-tissue lesions. Fraser-Hill and Renfrew5 examined accuracy and effective accuracy, which they considered to be a measure of clinical utility and to provide a better reflection of the performance in the clinical setting. They found a higher accuracy in soft-tissue lesions (93% or 14/15) when compared with bony lesions (83% or 72/87). However, when measuring the effective accuracy, these authors found no difference between soft-tissue (73% or 11/15) and bone (70% or 61/87) biopsies. Twenty-seven of the 71 bone lesions were metastatic malignancies compared with one metastatic malignancy of 49 soft tissue tumors. Although one might assume this would bias our results in favor of showing percutaneous biopsies of bone to be more clinically useful, our analysis failed to show percutaneous biopsies of metastatic lesions were statistically more clinically useful.
Biopsies containing myxoid features were less useful than nonmyxoid lesions, and no published studies have examined the effect of myxoid features on the diagnostic accuracy or utility of percutaneous biopsies. Myxoid tissue rarely helps to guide definitive treatment because hypo- cellular myxoid biopsy tissue can be representative of benign (intramuscular myxoma) or malignant tumors (myxoid sarcoma).
We found tumor type had no effect on clinical utility. Other studies showed biopsies of metastatic lesions tended to be more accurate than those of primary lesions but without statistical significance. One study found primary tumors were correctly diagnosed by FNA 75% of the time, and metastatic bone tumors were correct 84% of the time.9 Fraser-Hill and Renfrew,5 using their measure of clinical utility, found higher rates in metastatic lesions (77%) compared with primary lesions (59%). The clinical usefulness of metastatic lesion biopsies in our study (86%) was not statistically different from the biopsies of primary lesions (71%). Some studies show higher accuracy in malignant lesions.1,3,4 Our data showed no differences, but showed the opposite trend with a higher utility in primary benign lesions (74%) than primary malignant lesions (65%). The poor clinical utility (50%) of infection was influenced by the small number of cases (4) and the difficulty interpreting the pathology without positive culture results. Open biopsies in these cases increased clinical suspicion of infection, provided an opportunity to collect additional cultures, and allowed surgical treatment of the infection.
Our open biopsy rate (25%) was slightly higher than the 18% reported by one other study,17 and was an indicator of the treating physician's confidence in the interpretation of the biopsy and clinical data. We were unable to find any single factor that affected the open biopsy rate. The decision to obtain an open biopsy was made by the treating surgeon based on the information from the history, examination, imaging, and pathology. Clinical considerations weighed heavily in the interpretation of needle biopsies, as evidenced by 6% of our correct percutaneous biopsies being followed with an open biopsy. These occurred because of doubt in the percutaneous biopsy result despite a definitive pathology report. This phenomenon has been observed in another study17 and illustrates that, even with good percutaneous biopsy accuracy rates, clinical judgment still plays a large role in the clinical management of patients after biopsy. Clinical vigilance can reduce adverse consequences for a patient that may come from a nondi- agnostic biopsy or incorrect percutaneous biopsy. Consequences from an incorrect biopsy include delayed treatment, unnecessary treatment, incorrect treatment, and progression of disease. The consequences of a nondiagnostic biopsy are the same as those above if incorrect assumptions are made about the diagnosis, and otherwise they are the effects of a subsequent open biopsy (ie, surgical and wound complications).
Our rate of clinical utility is comparable to reported rates of diagnostic utility or effective accuracy5,17 and lower than simple diagnostic accuracy rates. Simple diagnostic accuracy rates, those that do not consider clinical utility, of percutaneous biopsies may overestimate the ability to make a treatment decision. The clinical utility of myxoid features is low. Core biopsies combined with FNA are more clinically useful than FNA alone. Although the clinical utility is less than the simple diagnostic accuracy, percutaneous biopsy has many advantages and its use remains recommended.
The authors thank Neha Sachdev, BA, for her help with this manuscript.
1. Ayala AG, Zornosa J. Primary bone tumors: percutaneous needle biopsy. Radiologic-pathologic study of 222 biopsies. Radiology
2. Cara del Rosal JA, Canadell J. Biopsy technique in the treatment of osteosarcoma. Int Orthop
3. Carrasco CH, Wallace S, Richli WR. Percutaneous skeletal biopsy. Cardiovasc Intervent Radiol
4. Duda SH, Johst U, Krahmer K, Pereira P, Konig C, Schafer J, Huppert P, Schott U, Bohm P, Claussen CD. Technique and results of CT-guided percutaneous bone biopsy. Orthopade
. 2001;30: 545-550.
5. Fraser-Hill MA, Renfrew DL. Percutaneous needle biopsy of musculoskeletal lesions. 1. Effective accuracy and diagnostic utility. AJR Am J Roentgenol
6. Hodge JC. Percutaneous biopsy of the musculoskeletal system: a review of 77 cases. Can Assoc Radiol J
7. Issakov J, Flusser G, Kollender Y, Merimsky O, Lifschitz-Mercer B, Meller I. Computed tomography-guided core needle biopsy for bone and soft tissue tumors. Isr Med Assoc J
8. Jelinek JS, Murphey MD, Welker JA, Henshaw RM, Kransdorf MJ, Shmookler BM, Malawer MM. Diagnosis of primary bone tumors with image-guided percutaneous biopsy: experience with 110 tumors. Radiology
9. Jorda M, Rey L, Hanly A, Ganjei-Azar P. Fine-needle aspiration cytology of bone: accuracy and pitfalls of cytodiagnosis. Cancer
10. Leffler SG, Chew FS. CT-guided percutaneous biopsy of sclerotic bone lesions: diagnostic yield and accuracy. AJR Am J Roentgenol
11. Logan PM, Connell DG, O'Connell JX, Munk PL, Janzen DL. Image-guided percutaneous biopsy of musculoskeletal tumors: an algorithm for selection of specific biopsy techniques. AJR Am J Roentgenol
12. Pramesh CS, Deshpande MS, Pardiwala DN, Agarwal MG, Puri A. Core needle biopsy for bone tumours. Eur J Surg Oncol
. 2001;27: 668-671.
13. Ruhs SA, El-Khoury GY, Chrischilles EA. A cost minimization approach to the diagnosis of skeletal neoplasms. Skeletal Radiol
14. Schweitzer ME, Gannon FH, Deely DM, O'Hara BJ, Juneja V. Percutaneous skeletal aspiration and core biopsy: complementary techniques. AJR Am J Roentgenol
15. Skrzynski MC, Biermann JS, Montag A, Simon MA. Diagnostic accuracy and charge-savings of outpatient core needle biopsy compared with open biopsy of musculoskeletal tumors. J Bone Joint Surg Am
16. Tsukushi S, Katagiri H, Nakashima H, Shido Y, Arai E. Application and utility of computed tomography-guided needle biopsy with musculoskeletal lesions. J Orthop Sci
17. Yao L, Nelson SD, Seeger LL, Eckardt JJ, Eilber FR. Primary musculoskeletal neoplasms: effectiveness of core-needle biopsy. Radiology