+ All Categories
Home > Documents > Bone and Soft-Tissue Lesions: What Factors Affect Diagnostic Yield of Image-guided Core-Needle...

Bone and Soft-Tissue Lesions: What Factors Affect Diagnostic Yield of Image-guided Core-Needle...

Date post: 20-Jan-2017
Category:
Upload: mary-g
View: 212 times
Download: 0 times
Share this document with a friend
9
Bone and Soft-Tissue Lesions: What Factors Affect Diagnostic Yield of Image-guided Core-Needle Biopsy? 1 Jim S. Wu, MD Jeffrey D. Goldsmith, MD Perry J. Horwich, MD Sanjay K. Shetty, MD Mary G. Hochman, MD Purpose: To assess lesion-related and technical factors that affect diagnostic yield in image-guided core-needle biopsy (CNB) of bone and soft-tissue lesions. Materials and Methods: Institutional review board approval and verbal informed consent were obtained for a HIPAA-compliant prospective study of 151 consecutive CNBs of bone (n 88) and soft-tissue (n 63) lesions. Each CNB specimen was re- ported separately in the final pathology report. Diagnostic yield (total number of biopsies that yield a diagnosis di- vided by total number of biopsies) was calculated for all lesions and subgroups on the basis of lesion composition (lytic, sclerotic, soft tissue), lesion size (2, 2 to 5, or 5 cm), biopsy needle gauge, image guidance modality, num- ber of specimens obtained, and specimen length (5, 5–10, or 10 mm). The minimum number of specimens required to obtain a diagnosis was determined on the basis of the specimen number at which the diagnostic yield reached a plateau. 2 And Wilcoxon rank-sum tests were performed in bivariate analyses to evaluate associations between each factor and diagnostic yield. Significant fac- tors were evaluated with multivariate logistic regression. Results: Diagnostic yield was 77% for all lesions. Yield was 87% for lytic bone lesions and 57% for sclerotic bone lesions (P .002). Diagnostic yield increased with larger lesions (54% for lesions 2 cm, 75% for lesions 2 to 5 cm, and 86% for lesions 5 cm [P .006]). There was no difference in diagnostic yield for bone versus soft-tissue lesions or ac- cording to needle gauge or image guidance modality. Diag- nostic yield was 77% for bone lesions and 76% for soft- tissue lesions (P .88). Yield was 83%, 72%, 77%, and 83% for biopsies performed with 14-, 15-, 16-, and 18- gauge needles, respectively (P .57). Yield was 77% with computed tomographic guidance and 78% with ultrasono- graphic guidance (P .99). Diagnostic yield increased with number of specimens obtained and with longer spec- imen length; it reached a plateau at three specimens for bone lesions and four specimens for soft-tissue lesions. Conclusion: Diagnostic yield is higher in lytic than in sclerotic bone lesions, in larger lesions, and for longer specimens. Ob- taining a minimum of three specimens in bone lesions and four specimens in soft-tissue lesions optimizes diagnostic yield. RSNA, 2008 1 From the Department of Radiology, Section of Musculo- skeletal Radiology (J.S.W., P.J.H., S.K.S., M.G.H.), and Department of Pathology (J.D.G.), Beth Israel Deaconess Medical Center, 330 Brookline Ave, Boston, MA 02215. From the 2007 RSNA Annual Meeting. Received October 1, 2007; revision requested December 21; revision re- ceived February 4, 2008; accepted March 6; final version accepted March 28. Address correspondence to J.S.W. (e-mail: [email protected] ). RSNA, 2008 ORIGINAL RESEARCH MUSCULOSKELETAL IMAGING 962 Radiology: Volume 248: Number 3—September 2008 Note: This copy is for your personal non-commercial use only. To order presentation-ready copies for distribution to your colleagues or clients, contact us at www.rsna.org/rsnarights.
Transcript
Page 1: Bone and Soft-Tissue Lesions: What Factors Affect Diagnostic Yield of Image-guided Core-Needle Biopsy?               1

Bone and Soft-Tissue Lesions:What Factors Affect Diagnostic Yield ofImage-guided Core-Needle Biopsy?1

Jim S. Wu, MDJeffrey D. Goldsmith, MDPerry J. Horwich, MDSanjay K. Shetty, MDMary G. Hochman, MD

Purpose: To assess lesion-related and technical factors that affectdiagnostic yield in image-guided core-needle biopsy (CNB)of bone and soft-tissue lesions.

Materials andMethods:

Institutional review board approval and verbal informedconsent were obtained for a HIPAA-compliant prospectivestudy of 151 consecutive CNBs of bone (n � 88) andsoft-tissue (n � 63) lesions. Each CNB specimen was re-ported separately in the final pathology report. Diagnosticyield (total number of biopsies that yield a diagnosis di-vided by total number of biopsies) was calculated for alllesions and subgroups on the basis of lesion composition(lytic, sclerotic, soft tissue), lesion size (�2, �2 to 5, or �5cm), biopsy needle gauge, image guidance modality, num-ber of specimens obtained, and specimen length (�5,5–10, or �10 mm). The minimum number of specimensrequired to obtain a diagnosis was determined on the basisof the specimen number at which the diagnostic yieldreached a plateau. �2 And Wilcoxon rank-sum tests wereperformed in bivariate analyses to evaluate associationsbetween each factor and diagnostic yield. Significant fac-tors were evaluated with multivariate logistic regression.

Results: Diagnostic yield was 77% for all lesions. Yield was 87% forlytic bone lesions and 57% for sclerotic bone lesions (P �.002). Diagnostic yield increased with larger lesions (54%for lesions � 2 cm, 75% for lesions � 2 to 5 cm, and 86%for lesions � 5 cm [P � .006]). There was no difference indiagnostic yield for bone versus soft-tissue lesions or ac-cording to needle gauge or image guidance modality. Diag-nostic yield was 77% for bone lesions and 76% for soft-tissue lesions (P � .88). Yield was 83%, 72%, 77%, and83% for biopsies performed with 14-, 15-, 16-, and 18-gauge needles, respectively (P � .57). Yield was 77% withcomputed tomographic guidance and 78% with ultrasono-graphic guidance (P � .99). Diagnostic yield increasedwith number of specimens obtained and with longer spec-imen length; it reached a plateau at three specimens forbone lesions and four specimens for soft-tissue lesions.

Conclusion: Diagnostic yield is higher in lytic than in sclerotic bonelesions, in larger lesions, and for longer specimens. Ob-taining a minimum of three specimens in bone lesions andfour specimens in soft-tissue lesions optimizes diagnosticyield.

� RSNA, 2008

1 From the Department of Radiology, Section of Musculo-skeletal Radiology (J.S.W., P.J.H., S.K.S., M.G.H.), andDepartment of Pathology (J.D.G.), Beth Israel DeaconessMedical Center, 330 Brookline Ave, Boston, MA 02215.From the 2007 RSNA Annual Meeting. Received October1, 2007; revision requested December 21; revision re-ceived February 4, 2008; accepted March 6; final versionaccepted March 28. Address correspondence to J.S.W.(e-mail: [email protected]).

� RSNA, 2008

ORIG

INAL

RESE

ARCH

�M

USCU

LOSK

ELET

ALIM

AGIN

G

962 Radiology: Volume 248: Number 3—September 2008

Note: This copy is for your personal non-commercial use only. To order presentation-ready copies for distribution to your colleagues or clients, contact us at www.rsna.org/rsnarights.

Page 2: Bone and Soft-Tissue Lesions: What Factors Affect Diagnostic Yield of Image-guided Core-Needle Biopsy?               1

Acquiring tissue for histopathologicdiagnosis is an important step inthe management of bone and soft-

tissue lesions. Tissue sampling can beaccomplished with open surgical biopsy,fine-needle aspiration (FNA), or core-needle biopsy (CNB). The choice oftechnique balances the invasiveness,risk, and cost of the procedure againstthe total amount of tissue obtained. Al-though open surgical biopsy is consid-ered the reference standard, it is moreexpensive, may require hospitalization,and has a higher complication rate thanpercutaneous needle biopsy (1–3). Whenperforming FNA, one runs the risk of in-sufficient tissue sampling. FNA is likelyadequate for metastatic lesions; however,its utility in the diagnosis of primary sar-comas of soft tissue and bone is limitedand controversial (1,4–6). Because fine-needle aspirates show only the cytologicfeatures of the lesion and do not showtissue architecture, accurate pathologicdiagnosis is often not possible (1,6).

CNB of bone and soft-tissue lesionsis minimally invasive, can be performedin an outpatient setting, and is less ex-pensive than surgical biopsy (5,7,8).Moreover, CNB creates a smaller bi-opsy track than open surgery, facilitat-ing subsequent resection of the trackat definitive surgery (9,10). Complica-tion rates are low and are estimatedat 0.2%, with a range of 0%–10%(5,7,9,11–19). The diagnostic yield anddiagnostic accuracy of CNB are high andhave improved as the procedure has be-come the standard of care (3–5,9,13–15,19–22).

Bone and soft-tissue lesions vary intheir composition and size, causingtechnical and diagnostic challenges. Di-agnostic success rates can be lower inpredominantly cystic or sclerotic bone

lesions than in lesions with mostly soft-tissue components (7,9,11,14,23–25).In a study by Ayala and Zornosa (23),CNB failed to provide diagnostic tissuein 49% (17 of 35) of sclerotic lesions. Ina report of 110 primary musculoskeletaltumors, Jelinek et al (7) found that cys-tic lesions, or lesions with predomi-nantly necrotic and fluid-filled centers,accounted for 46% (six of 13) of thetumors that were nondiagnostic or mis-diagnosed at CNB.

Currently, to our knowledge, nopublished guidelines exist on the size ornumber of specimens that should be ob-tained during CNB of musculoskeletallesions to optimize the diagnostic yield.Intuitively, obtaining more or largerspecimens should increase the chanceof acquiring a diagnosis; however, thisapproach may also increase the risk ofcomplications (bleeding, fracture, in-fection, pain). Practice patterns varywidely. Some authors advocate the ac-quisition of three or fewer specimens(11,14,15,23). Gil-Sanchez et al (14)obtained only one or two specimens intheir series of 65 bone lesions, achiev-ing a diagnostic yield of 86%, and Ayalaand Zornosa (23) obtained one to threespecimens in their series of 222 percu-taneous bone biopsies, achieving a diag-nostic yield of 79%. Other authors ob-tain a minimum of three specimens(4,5,24), while still others routinely ac-quire five to 10 specimens with fewcomplications (7,16). Jelinek et al (7)acquired three to 10 specimens, with anaverage of 5.5 specimens, in their seriesof 110 primary bone lesions, achieving adiagnostic yield of 88%. To our knowl-edge, no prior study has systematicallyevaluated the influence of specimen sizeor number on the diagnosis of bone andsoft-tissue lesions. In this prospectivestudy, we sought to assess lesion-related and technical factors that affect

diagnostic yield in image-guided CNB ofbone and soft-tissue lesions.

Materials and Methods

PatientsThis Health Insurance Portability andAccountability Act–compliant study wasapproved by our institutional reviewboard. Written informed consent forthe biopsy and verbal consent for inclu-sion of the biopsy data in the study wereobtained. We performed a prospectivestudy of 151 consecutive image-guidedCNBs of bone (n � 88) and soft-tissue(n � 63) lesions in 145 adult patients at asingle institution from January 2006 toAugust 2007. Our institution is a majortertiary referral center for patients withmusculoskeletal neoplasms and is staffedby orthopedic oncologists and an interdis-ciplinary sarcoma team.

Biopsy Technique and AnalysisOne of three musculoskeletal radiolo-gists (J.S.W., P.J.H., S.K.S.) per-formed the biopsies by using standardcoaxial techniques with computed to-mographic (CT) or ultrasonographic(US) guidance. Needle approach wasdiscussed with the referring surgeonprior to biopsy to ensure subsequentresection of the biopsy track. Four nee-dle sizes were used for bone lesions,and three were used for soft-tissue le-sions. Bonopty 15-gauge 1.7-mm-boreneedles (RADI Medical Systems, Upp-

Published online10.1148/radiol.2483071742

Radiology 2008; 248:962–970

Abbreviations:CI � confidence intervalCNB � core-needle biopsy

Author contributions:Guarantor of integrity of entire study, J.S.W.; study con-cepts/study design or data acquisition or data analysis/interpretation, all authors; manuscript drafting or manu-script revision for important intellectual content, all au-thors; manuscript final version approval, all authors;literature research, J.S.W., P.J.H.; clinical studies, J.S.W.,J.D.G., P.J.H., S.K.S.; statistical analysis, J.S.W.; andmanuscript editing, all authors

Authors stated no financial relationship to disclose.

Advances in Knowledge

� Diagnostic yield from core-needlebiopsy (CNB) is higher in lyticversus sclerotic bone lesions, inlarger lesions, and with increasedcore specimen length.

� Nondiagnostic CNBs are morelikely to occur in benign than inmalignant lesions.

Implication for Patient Care

� During CNB, obtaining a minimumof three specimens in bone lesionsand four specimens in soft-tissuelesions is recommended to optimizediagnostic yield.

MUSCULOSKELETAL IMAGING: Core-Needle Biopsy of Bone and Soft-Tissue Lesions Wu et al

Radiology: Volume 248: Number 3—September 2008 963

Page 3: Bone and Soft-Tissue Lesions: What Factors Affect Diagnostic Yield of Image-guided Core-Needle Biopsy?               1

sala, Sweden) were used for bone lesionswith an intact cortex. For soft-tissue lesionsand for bone lesions without an intact cor-tex, a 14-, 16-, or 18-gauge coaxial auto-mated biopsy gun system (Achieve; Cardi-nal Health, Dublin, Ohio) was used. Themajority of patients were in a state ofconscious sedation during the proceduresinduced by using midazolam (Versed;Baxter Healthcare, Glendale, Calif) andfentanyl (Sublimaze; Baxter Healthcare).Complications were documented.

We placed each specimen in an indi-vidual formalin-filled container; eachwas processed and interpreted his-topathologically as if it were from a sep-arate biopsy. The order number of ac-quisition for each specimen was docu-mented. The acquisition of three to fivespecimens is the standard of practice atour institution. We obtained fewerspecimens for small lesions, lesions inhazardous locations (ie, rib, neurovas-cular bundle), and osseous lesions atrisk for fracture. If primary lymphomaof bone was in the differential diagnosis,an additional sample was sent for flowcytometry. Cytology specimens wereobtained in 109 biopsies; however,a cytopathologist was seldom presentduring the biopsies. One experiencedmusculoskeletal pathologist (J.D.G.)evaluated all biopsy specimens for anindividual patient in one session, in theorder of specimen acquisition, and wasnot blinded to the clinical history or im-aging study results. Care was taken bythe pathologist to interpret each speci-men in isolation, despite his having re-viewed additional specimens from thesame biopsy procedure.

Lesion-related FactorsThe lesionsweredivided intoboneand soft-tissue categories. All soft-tissue lesionswere separate from the adjacent bone,without bone involvement or remodeling.Bone lesions were further subcategorizedby composition into lytic or sclerotic lesionson the basis of their appearance on radio-graphs and at CT. We defined a bone lesionas lytic if more than 50% of its volume wasmore lucent than the surrounding normalbone. A lesion was defined as sclerotic ifmore than 50% of its volume was denserthan the surrounding normal bone. In pa-

tients with lesions that appeared normal onradiographs, designation was based on CTdensity. Bone lesion composition wasagreed on by consensus of four musculo-skeletal radiologists (J.S.W., P.J.H.,S.K.S., M.G.H.). All lesions were catego-rized into one of three groups on the basisof size: 2 cm or smaller, greater than 2 to 5cm, and larger than 5 cm. Lesions weremeasured by using the single largest dimen-sion on prebiopsy cross-sectional imagingstudies (CT, magnetic resonance imaging,or US).

Technical FactorsThe biopsy needle gauge (14, 15, 16, or18), imaging guidance modality (CT vsUS), and number of specimens obtainedwere documented for all 151 biopsies.In a subset of 117 biopsies, the length ofeach specimen was documented by theradiologist on the basis of a specimenrating system. Type 1 specimens mea-sured less than 5 mm in length of boneand/or soft tissue. Type 2 specimensmeasured between 5 and 10 mm inlength, and type 3 specimens werelarger than 10 mm. Blood clot in thespecimen was excluded in the specimenlength calculation (Figure).

Determination of Diagnostic YieldDiagnostic yield is the number of bi-opsies that result in a diagnosis di-

vided by the total number of biopsiesperformed. A specimen was consid-ered to be diagnostic when a distinctpathologic diagnosis could be ren-dered from the biopsy tissue that ex-plained the lesion clinically and at im-aging. If this criterion was not met, thespecimen was considered nondiagnos-tic. Moreover, only one specimenneeded to be considered diagnostic forthe biopsy to be considered diagnostic.We then calculated diagnostic yield onthe basis of lesion-related (compositionand size) and technical factors (biopsyneedle gauge, imaging guidance modality,and length of specimen) by dividing thenumber of diagnostic cases by the totalnumber of cases for each subgroup.

Minimum Number of Specimens forDiagnosisTo determine the minimum number ofspecimens required to optimize diag-nostic yield, the cumulative diagnosticyield with each successive specimenwas calculated by dividing the numberof diagnostic biopsies after each speci-men by the total number of biopsies.The specimen number at which the cu-mulative diagnostic yield first reached aplateau indicated the minimal numberof specimens needed for diagnosis. Af-ter this specimen number, the acquisi-tion of additional core specimens did

Three CNB specimens from soft-tissue biopsy performed with 16-gauge biopsy gun and 1.8-cm tray demon-strate the three specimen length categories: type 1 (�5 mm), type 2 (5–10 mm), and type 3 (�10 mm).

MUSCULOSKELETAL IMAGING: Core-Needle Biopsy of Bone and Soft-Tissue Lesions Wu et al

964 Radiology: Volume 248: Number 3—September 2008

Page 4: Bone and Soft-Tissue Lesions: What Factors Affect Diagnostic Yield of Image-guided Core-Needle Biopsy?               1

not improve the diagnostic yield. Thisminimum specimen number was calcu-lated for the various lesion-related andtechnical factor subgroups.

Surgical Correlation and DiagnosticAccuracyThe number of patients continuing tosurgery was documented, and histologicresults from the surgical resection orsurgical biopsy were compared with theCNB results. Using the histologic resultsin the patients who underwent surgeryas the reference standard, we calcu-lated diagnostic accuracy by dividing thenumber of CNBs that matched the sur-gical biopsy results by the number ofpatients with surgical correlation.

Statistical AnalysisAnalyses were performed by using soft-ware (SAS, version 8.02; SAS Institute,

Cary, NC). We used the �2 or Fisherexact test for categoric variables andthe Wilcoxon rank-sum test for ordinalor continuous variables. We estimatedodds ratios for our main outcome mea-sure, diagnostic yield. In multivariateanalyses, we built iterative, forced-entrymodels. Candidate variables with P � .20in the bivariate analyses were entered inmultivariate logistic regression.

Results

PatientsA total of 151 consecutive CNBs wereperformed in 145 patients (74 womenand 71 men). Three patients underwentbiopsy for the same lesion twice, andthree patients underwent biopsy for twodifferent lesions. The mean patient agewas 53.4 years (range, 19–93 years).

Lesion-related FactorsThere were 88 bone lesions and 63 soft-tissue lesions. Of the bone lesions, 60were lytic and 28 were sclerotic. In re-gard to lesion size, 16% (24 of 151) oflesions were 2 cm or smaller, 38% (57of 151) of lesions measured between 2and 5 cm, and 46% (70 of 151) of le-sions were larger than 5 cm. The mostcommon bone lesion sites were the iliacbone (n � 22), the femur (n � 20), andthe tibia (n � 10), and the most com-mon soft-tissue lesion sites were thepelvis or thigh (n � 21), the knee (n �10), and the ankle (n � 7).

Technical FactorsThe biopsy needle gauge ranged from14 to 18, with the 15-gauge needle, usedin 45% (68 of 151) of biopsies, being themost common. The majority of biopsies(88% [133 of 151]) were performedwith CT rather than US (12% [18 of151]) guidance. A total of 599 speci-mens were obtained during the 151biopsies, with a mean of 4.0 speci-mens per biopsy. One specimen wasobtained in 10 (7%) of the 151 biop-sies; two specimens, in four (3%) biop-sies; three specimens, in 19 (13%)biopsies; four specimens, in 72 (48%) biop-sies; five specimens, in 40 (26%) biop-sies; and six specimens, in six (4%) bi-opsies. In four biopsies, because of thepatient’s request, only a single specimenwas obtained.

In a subset of 117 biopsies thatyielded 463 specimens, the specimenlength was documented on the basis of aspecimen rating system. Of the 463 spec-imens, 13% (60 of 463) were type 1specimens, 29% (132 of 463) were type2 specimens, and 59% (271 of 463)were type 3 specimens. We did not in-clude specimen length in the initialstudy design; thus, these data were notrecorded in the first 34 biopsies.

Diagnostic YieldThe overall diagnostic yield was 77%(116 of 151). There was no significantdifference in diagnostic yield betweenbone (77% [68 of 88]) and soft-tissue(76% [48 of 63]) lesions (P � .88). Lyticbone lesions had a significantly higherdiagnostic yield (87% [52 of 60]) than

Table 1

Diagnostic Yield: Lesion-related Factors

Lesion-related Factor No. of Diagnostic Cases Total No. of Cases Diagnostic Yield (%) P Value

All lesions 116 151 77General lesion subtype .88

Soft tissue 48 63 76Bone 68 88 77

Bone lesion subtype �.001Sclerotic 16 28 57Lytic 52 60 87

Size (cm) �.001�2 13 24 54�2 To 5 43 57 75�5 60 70 86

Table 2

Diagnostic Yield: Technical Factors

Technical Factor No. of Diagnostic Cases Total No. of Cases Diagnostic Yield (%) P Value

Biopsy needle gauge .5714 5 6 8315 49 68 7216 23 30 7718 39 47 83

Imaging guidance modality .99CT 102 133 77US 14 18 78

Specimen size (mm) �.001�5 (Type 1) 25 60 425–10 (Type 2) 81 132 61�10 (Type 3) 222 271 82

MUSCULOSKELETAL IMAGING: Core-Needle Biopsy of Bone and Soft-Tissue Lesions Wu et al

Radiology: Volume 248: Number 3—September 2008 965

Page 5: Bone and Soft-Tissue Lesions: What Factors Affect Diagnostic Yield of Image-guided Core-Needle Biopsy?               1

sclerotic bone lesions (57% [16 of 28])(P � .002). Diagnostic yield increasedwith larger lesions; it was 54% (13 of24) for lesions 2 cm or smaller, 75% (43of 57) for lesions between 2 and 5 cm,and 86% (60 of 70) for lesions largerthan 5 cm (P � .006) (Table 1).

There was no difference in diagnos-tic yield on the basis of needle gauge orimaging guidance modality. Diagnosticyield was 83% (five of six) for biopsiesperformed with a 14-gauge needle, 72%(49 of 68) for biopsies performed with a15-gauge needle, 77% (23 of 30) forbiopsies performed with a 16-gaugeneedle, and 83% (39 of 47) for biopsiesperformed with an 18-gauge needle(P � .57). Diagnostic yield was 77%(102 of 133) with CT guidance and 78%(14 of 18) with US guidance (P � .99).For the subset of 117 biopsies in whichthe specimen size was documented, theoverall diagnostic yield was 77% (90 of117), and the yield increased signifi-cantly with larger specimen length (P �.001) (Table 2). The odds of obtaining

a diagnostic specimen was 2.2 timesgreater for a type 2 specimen comparedwith a type 1 specimen (95% confidenceinterval [CI]: 1.2, 4.1) and 6.3 timesgreater for a type 3 specimen comparedwith a type 1 specimen (95% CI: 3.5,11.5). When we evaluated the effect ofspecimen length on the diagnostic yieldfor the various lesion-related and tech-nical factors, we found that diagnosticyield increased with longer specimensfor all lesion subgroups except lyticbone lesions (Table 3).

Minimum Number of Specimens forDiagnosisThe cumulative diagnostic yield reached aplateau at either three or four specimensfor all lesions and subtypes (Table 4). Inthe 88 bone biopsies, the cumulative di-agnostic yield reached a plateau by thethird specimen. For the 63 soft-tissuelesions, four specimens were necessarybefore the cumulative diagnostic yieldreached a plateau, which occurred intwo patients. One patient had a high-

grade malignant peripheral nervesheath tumor in the calf with largecentral areas of necrosis. In this pa-tient, diagnostic tissue was seen onlywith the fourth specimen (five totalspecimens were taken). The secondpatient had a calcified juxtaarticularchondroma of the knee in which diag-nostic tissue was seen only with thefourth and fifth specimens.

Histopathologic DiagnosesThere were 116 diagnostic biopsies, with37% (43 of 116) of the biopsies yieldingbenign histologic findings and 63% (73 of116) of biopsies yielding malignant histo-logic findings (Tables 5, 6). The mostcommon benign lesions were intramus-cular myxoma (n � 7), schwannoma(n � 6), and fibromatosis (n � 4). Themost common malignant lesions weremetastases, accounting for 47% (34 of73) of malignant lesions. Metastasesfrom the lung (n � 14) and breast (n �7) were the most common metastases.Of the primary malignancies of bone,

Table 3

Diagnostic Yield according to Specimen Length for Various Lesion-related and Technical Factors

Lesion-related or Technical FactorNo. ofBiopsies

No. ofSpecimens

Diagnostic Yield (%)Type 1Specimens � 5 mm

Type 2Specimens 5–10 mm

Type 3Specimens � 10 mm P Value

All lesions 117 463 42 61 82 �.001General lesion subtype

Soft tissue 51 202 29 56 83 �.001Bone 66 261 49 65 81 .001

Bone lesion subtypeSclerotic 18 72 13 50 70 �.001Lytic 48 189 71 70 85 .055

Lesion size (cm)�2 20 65 25 35 68 .011�2 To 5 45 172 46 55 88 �.001�5 52 226 56 79 80 .057

Biopsy needle gauge14 5 24 0 75 84 .1315 47 174 36 53 75 �.00116 29 123 50 67 78 .1218 36 142 54 67 91 �.001

Imaging guidance modalityCT 105 420 43 62 83 �.001US 12 43 0 0 75 .024

Histologic diagnosisBenign 32 132 53 79 90 .001Malignant primary cancer 31 136 75 97 90 .15Malignant metastasis 27 101 67 100 100 �.001

MUSCULOSKELETAL IMAGING: Core-Needle Biopsy of Bone and Soft-Tissue Lesions Wu et al

966 Radiology: Volume 248: Number 3—September 2008

Page 6: Bone and Soft-Tissue Lesions: What Factors Affect Diagnostic Yield of Image-guided Core-Needle Biopsy?               1

lymphoma of bone (n � 17) and plas-macytoma (n � 6) were the most com-mon.

Cytology specimens were obtainedin 72% (109 of 151) of the biopsies. Inall 109 patients, the cytology speci-mens did not provide additional diag-nostic information. Specifically, therewere no CNBs considered to be nondi-agnostic that were diagnostic at cyto-logic analysis. Conversely, in 31 cases,the cytology specimen was nondiag-nostic and the CNB was diagnostic.

Surgical Correlation and DiagnosticAccuracySurgical biopsy or resection was per-formed in 37% (56 of 151) of patients.Of the 116 patients with diagnostic CNBresults, 34% (39 of 116) continued on tosurgery. In 36 patients, the final patho-logic finding at surgery was identical tothe histologic finding from the image-guided biopsy, including tumor grade.Thus, the diagnostic accuracy of image-guided CNB in this series was 92% (36of 39). There were three patients inwhom the CNB histologic finding was

different from the final histologic findingat surgery. In one patient, a benign car-tilaginous lesion at CNB was found to bea low-grade chondrosarcoma at resec-tion. In a second patient, a lesion diag-nosed as fibromatosis at CNB and fro-zen section was found to be a synovialsarcoma at resection. In the third pa-tient, a malignant low-grade spindle cellneoplasm diagnosed at CNB was foundto be a benign perineurioma at surgicalresection.

Among the 35 patients with nondi-agnostic CNBs, 17 (49%) continued onto surgery (Table 7). Fifteen (88%) ofthese 17 patients had lesions with be-nign histologic findings at surgery. Themost common lesion was intramuscularhemangioma. Maximum time of fol-low-up was 21 months.

ComplicationsOne minor complication occurred in ourstudy. One patient with a small myxoidliposarcoma of the thigh developed asmall biopsy-related hematoma that re-quired overnight observation, mostlyfor pain control.

Multivariate Statistical AnalysisBecause it is possible that some factorsexplained other factors, leading to indi-rect bias, we estimated odds ratios forour main outcome measure, diagnosticyield, by using multivariate logistic re-gression. The significant candidate vari-ables were (a) bone lesion subtype,(b) lesion size, and (c) specimen length.Lesion subtype, biopsy needle gauge,and imaging guidance modality were notsignificant in the bivariate analyses andhence were not included in the multivar-iate logistic regression.

For bone lesion subtype, a lytic le-sion was 3.4 times more likely than asclerotic lesion to have a diagnosis (95%CI: 1.8, 6.2), even after adjustment forlesion size and specimen length. For le-sion size, with each increase in lesionsize subgroup (�2 cm, �2 to 5 cm, �5cm), there was a 1.8 times higher likeli-hood of achieving a diagnosis (95% CI:1.2, 2.9), even after adjustment forbone lesion subtype and specimen

Table 4

Cumulative Diagnostic Yield for Bone and Soft-Tissue Lesions with Increasing Numberof Specimens

Lesion-related or Technical Factor

Cumulative Diagnostic Yield according toNo. of Specimens Obtained (%)

No. ofSpecimens toReach DiagnosticYield Plateau1 2 3 4 5 6

All lesions (n � 151) 65 71 75 77 77 77 4General lesion subtype

Soft tissue (n � 63) 67 71 73 76 76 76 4Bone (n � 88) 64 70 77 77 77 77 3

Bone lesion subtypeSclerotic (n � 28) 39 46 57 57 57 57 3Lytic (n � 60) 75 82 87 87 87 87 3

Lesion size (cm)�2 (n � 24) 42 50 54 54 54 54 3�2 To 5 (n � 57) 67 72 74 75 75 75 4�5 (n � 70) 71 77 84 86 86 86 4

Biopsy needle gauge14 (n � 6) 67 67 67 83 83 . . . 415 (n � 68) 54 63 72 72 72 72 316 (n � 30) 67 73 73 77 77 77 418 (n � 47) 79 81 83 83 83 83 3

Imaging guidance modalityCT (n � 133) 65 71 76 77 77 77 4US (n � 18) 67 72 72 78 78 78 4

Table 5

Histologic Results for 43 Benign Boneand Soft-Tissue Lesions withDiagnostic Biopsies

Diagnosis*No. ofLesions

Intramuscular myxoma 7Schwannoma 6Fibromatosis, desmoid type 4Giant cell tumor of tendon sheath 3Fibrous dysplasia 3Ossifying fibroma 2Lymph node 2Leiomyomata cutis 2Benign cartilage lesion 2Hematoma 2Nodular fasciitis 1Acute osteomyelitis 1Epithelioid hemangioma 1Giant-cell tumor of bone 1Perineurioma 1Rheumatoid nodule 1Bone infarct 1Juxtaarticular chondroma 1Pseudogout (CPPD) 1Tenosynovitis 1

* CPPD � calcium pyrophosphate dihydrate depositiondisease.

MUSCULOSKELETAL IMAGING: Core-Needle Biopsy of Bone and Soft-Tissue Lesions Wu et al

Radiology: Volume 248: Number 3—September 2008 967

Page 7: Bone and Soft-Tissue Lesions: What Factors Affect Diagnostic Yield of Image-guided Core-Needle Biopsy?               1

length. Finally, with each increase inspecimen length (�5 mm, 5 to 10 mm,�10 mm), there was a 1.7 times higherlikelihood of achieving a diagnosis (95%CI: 1.2, 2.6), even after adjustment forbone lesion subtype and lesion size.

Discussion

When examining lesion-related factorsthat affect diagnostic yield, we, likeother researchers (3,4,18), found nosignificant difference in diagnostic yieldfor biopsies performed in bone versussoft-tissue lesions. The significantlylower diagnostic yield for sclerotic ver-sus lytic bone lesions may be due to a“masking” of the underlying lesion byreactive sclerosis (9). In these patients,targeting of the tumor is difficult, par-ticularly if there is no identifiable soft-tissue component. Moreover, scleroticlesions often require a cutting needle ordrill to breach the cortex and/or reac-tive bone, and these samples can bemacerated and degraded by crush ar-tifacts, hampering histologic evalua-tion and thereby lowering diagnosticyield and diagnostic accuracy (9,11,14,23,25).

The significant increase in diagnos-tic yield with increasing lesion size maybe related to the difficulty in targetingsmaller lesions, introducing a greaterchance of sampling error. Many auto-

mated handheld gun systems have a1.8-cm sample tray; thus, for lesionssmaller than 1.8 cm, the samples willlikely contain some normal tissue, pos-sibly complicating diagnosis.

As for technical factors, we ex-pected to see an increase in diagnosticyield with decreasing biopsy needlegauge; however, this was not observed.The lowest diagnostic yield was with the15-gauge Bonopty bone biopsy needle,which can be technically more challeng-ing to use than the automated biopsyguns, possibly accounting for the lowerdiagnostic yield. We did see a significantincrease in diagnostic yield with longerspecimens: Only 42% of specimens lessthan 5 mm in length were diagnostic, asopposed to 82% of specimens longerthan 10 mm. On the basis of these re-sults, an awareness by the operator ofthe length of the core specimens mayhelp improve diagnostic yield, and,when appropriate, attempts should bemade to obtain longer specimens.

The cumulative diagnostic yieldreached a plateau at either three or fourspecimens for all lesions and lesion sub-groups. In 88 bone biopsies, no addi-tional diagnostic yield was obtained af-ter the third specimen. For the 63 soft-tissue lesions, four specimens werenecessary before the diagnostic yieldreached a plateau. On the basis of theseresults, we suggest obtaining a mini-mum of three specimens in bone lesionsand four specimens in soft-tissue lesions

during CNB. While, to the best of ourknowledge, there are no similar studiesfor comparison in the musculoskeletalliterature, analogous studies have beenperformed on breast biopsy procedures(26–28). Fishman et al (27) performedUS-guided biopsy of 73 breast masseswith a 14-gauge automated biopsy de-vice and processed each sample sepa-rately. They attempted to obtain fivesamples for the majority of biopsies andfound that four samples were necessaryto obtain a diagnosis in 100% of biop-sies.

Our results indicate that the histo-logic nature of the lesions can affect di-agnostic yield. Of the 35 nondiagnosticCNBs in our study, 33 appeared benign,either at surgery (n � 15) or on thebasis of clinical and radiographic fol-low-up (n � 18). Although the maxi-mum time of follow-up was only 21months, the 18 patients who did notcontinue on to surgery were consideredto have low suspicion for malignancy.Interestingly, 15 of the 17 lesions con-tinuing to surgery had a benign final his-tologic diagnosis. The two lesions withmalignant histologic findings at surgerywere two separate adamantinomas inthe tibia in one patient. Because theprebiopsy imaging studies were highlysuggestive of adamantinoma, the patientwent on to surgical biopsy after nondi-agnostic CNB. So why are nondiagnos-tic biopsies more likely to occur in be-nign lesions? A partial answer may be

Table 6

Histologic Results for 73 MalignantBone and Soft-Tissue Lesions withDiagnostic Biopsies

Diagnosis*No. ofLesions

Metastasis 34B-cell lymphoma 17Plasmacytoma 6Liposarcoma 4Chondrosarcoma 3Osteosarcoma 2Sarcoma NOS, high grade 2Leiomyosarcoma 2Melanoma 1Malignant peripheral nerve sheath tumor 1Sarcoma NOS, low grade 1

* NOS � not otherwise specified.

Table 7

Follow-up Results after 35 Nondiagnostic CNBs

Follow-up Method and Diagnosis* No. of Biopsies Malignant or Benign

Surgical follow-up 17Intramuscular hemangioma 4 BenignCystic lesion (ganglion and paralabral) 3 BenignOsteomyelitis 2 BenignAdamantinoma 2 MalignantReactive osseous changes from traumaand/or DJD 2 BenignUnicameral bone cyst 1 BenignAneurysmal bone cyst 1 BenignLow-grade chondrosarcoma 1 BenignFibro-osseous lesion 1 Benign

Clinical follow-up (no surgery) 18 Benign

* DJD � degenerative joint disease.

MUSCULOSKELETAL IMAGING: Core-Needle Biopsy of Bone and Soft-Tissue Lesions Wu et al

968 Radiology: Volume 248: Number 3—September 2008

Page 8: Bone and Soft-Tissue Lesions: What Factors Affect Diagnostic Yield of Image-guided Core-Needle Biopsy?               1

that certain benign lesions cannot bedefinitively diagnosed with CNB. Forthe 15 nondiagnostic CNBs with benignhistologic findings at surgery, the twomost common diagnoses were intramus-cular hemangioma (n � 4) and ganglionand paralabral cyst (n � 3). Both ofthese lesions may require that substan-tial amounts of tissue capable of show-ing tissue architecture be obtained sothat a definitive histologic diagnosis canbe made, likely explaining why the sam-ples were nondiagnostic at CNB. Ourresults indicate that nondiagnostic biop-sies are more likely to occur with benignversus malignant lesions. Although notconclusive, these results suggest a po-tential role for nondiagnostic biopsy re-sults in guiding clinical management bypreventing unnecessary follow-up surgi-cal biopsy.

There were limitations to our study.One limitation involved not obtainingthe same number of specimens in everybiopsy. A few biopsies yielded only oneor two specimens. However, three ormore specimens were obtained in 91%of the biopsies. We typically obtainedthree to five specimens during CNB anddid not find it appropriate to alter ourpractice habits to obtain six specimensfor every biopsy. For lesions at risk forfracture or in hazardous locations (ie,rib, neurovascular structures), three orfewer specimens were obtained. In fourbiopsies, because of the patient’s re-quest, the procedures were terminatedafter a single specimen was obtained.

Another limitation was that only onemusculoskeletal pathologist reviewedthe slides; however, all efforts weremade at independent interpretation ofthe individual samples. There is onlyone dedicated musculoskeletal patholo-gist at our institution; hence, we did notperform a retrospective second inter-pretation. Moreover, we realize thatmany sites do not have dedicated mus-culoskeletal pathologists or cytopatholo-gists. The size and number of specimensrequired to optimize diagnostic yieldmay vary according to expertise in theinterpretation of tissue.

It is possible that similar studies indifferent patient populations may yielddifferent results. The patients described

here were drawn from an adult tertiaryreferral orthopedic oncology service,and we did not include pediatric pa-tients. We had only two patients withosteosarcoma, a common primary ma-lignant tumor in the pediatric popula-tion. Also, patients with spinal lesionswere not included in this study, as spi-nal biopsies are performed by our inter-ventional neuroradiologists. Neverthe-less, we feel that our study providesuseful data for clinical care of relevantpatient populations and a reasonablesource of comparative data for futurestudies.

In conclusion, lesion-related andtechnical factors affecting diagnosticyield with CNBs in musculoskeletal le-sions were identified. Diagnostic yield ishigher in lytic than in sclerotic bone le-sions, with larger lesions, and withlonger specimen core length. There wasno difference in diagnostic yield be-tween soft-tissue versus bone lesionsor according to biopsy needle gaugeor imaging modality. Diagnostic yieldreached a plateau after three speci-mens for bone lesions and after fourspecimens for soft-tissue lesions. Thus,during image-guided CNB, we suggestobtaining a minimum of three speci-mens in bone lesions and four speci-mens in soft-tissue lesions when clini-cally feasible, to optimize diagnosticyield.

Acknowledgments: We thank Ann Wu, MD,for statistical analysis and manuscript review andNeil Rofsky, MD, for manuscript review.

References1. Kilpatrick SE, Cappellari JO, Bos GD, Gold

SH, Ward WG. Is fine-needle aspiration bi-opsy a practical alternative to open biopsyfor the primary diagnosis of sarcoma? expe-rience with 140 patients. Am J Clin Pathol2001;115:59–68.

2. Hau A, Kim I, Kattapuram S, et al. Accuracyof CT-guided biopsies in 359 patients withmusculoskeletal lesions. Skeletal Radiol2002;31:349–353.

3. Mitsuyoshi G, Naito N, Kawai A, et al. Accu-rate diagnosis of musculoskeletal lesions bycore needle biopsy. J Surg Oncol 2006;94:21–27.

4. Yao L, Nelson SD, Seeger LL, Eckardt JJ,Eilber FR. Primary musculoskeletal neo-

plasms: effectiveness of core-needle biopsy.Radiology 1999;212:682–686.

5. Welker JA, Henshaw RM, Jelinek J,Shmookler BM, Malawer MM. The percuta-neous needle biopsy is safe and recom-mended in the diagnosis of musculoskeletalmasses. Cancer 2000;89:2677–2686.

6. Domanski HA, Akerman M, Carlen B, et al.Core-needle biopsy performed by thecytopathologist: a technique to complementfine-needle aspiration of soft tissue and bonelesions. Cancer 2005;105:229–239.

7. Jelinek JS, Murphey MD, Welker JA, et al.Diagnosis of primary bone tumors with im-age-guided percutaneous biopsy: experiencewith 110 tumors. Radiology 2002;223:731–737.

8. Fraser-Hill MA, Renfrew DL, Hilsenrath PE.Percutaneous needle biopsy of musculoskel-etal lesions. 2. Cost-effectiveness. AJR Am JRoentgenol 1992;158:813–818.

9. Berning W, Freyschmidt J, Ostertag H. Per-cutaneous bone biopsy, techniques and indi-cations. Eur Radiol 1996;6:875–881.

10. Leffler SG, Chew FS. CT-guided percutane-ous biopsy of sclerotic bone lesions: diagnos-tic yield and accuracy. AJR Am J Roentgenol1999;172:1389–1392.

11. Kattapuram SV, Rosenthal DI. Percutaneousbiopsy of skeletal lesions. AJR Am J Roent-genol 1991;157:935–942.

12. Murphy WA, Destouet JM, Gilula LA. Percu-taneous skeletal biopsy 1981: a procedurefor radiologists—results, review, and recom-mendations. Radiology 1981;139:545–549.

13. Tehranzadeh J, Freiberger RH, Ghelman B.Closed skeletal needle biopsy: review of 120cases. AJR Am J Roentgenol 1983;140:113–115.

14. Gil-Sanchez S, Marco-Domenech SF, Irurzun-Lopez J, Fernandez-Garcia P, de la Iglesia-Cardena P, Ambit-Capdevila S. Ultrasound-guided skeletal biopsies. Skeletal Radiol 2001;30:615–619.

15. Pramesh CS, Deshpande MS, Pardiwala DN,Agarwal MG, Puri A. Core needle biopsy forbone tumours. Eur J Surg Oncol 2001;27:668–671.

16. Torriani M, Etchebehere M, Amstalden E.Sonographically guided core needle biopsy ofbone and soft tissue tumors. J UltrasoundMed 2002;21:275–281.

17. Thanos L, Mylona S, Kalioras V, Pomoni M,Batakis N. Percutaneous CT-guided inter-ventional procedures in musculoskeletal sys-tem (our experience). Eur J Radiol 2004;50:273–277.

18. Altuntas AO, Slavin J, Smith PJ, et al. Accu-

MUSCULOSKELETAL IMAGING: Core-Needle Biopsy of Bone and Soft-Tissue Lesions Wu et al

Radiology: Volume 248: Number 3—September 2008 969

Page 9: Bone and Soft-Tissue Lesions: What Factors Affect Diagnostic Yield of Image-guided Core-Needle Biopsy?               1

racy of computed tomography guided coreneedle biopsy of musculoskeletal tumours.ANZ J Surg 2005;75:187–191.

19. Puri A, Shingade VU, Agarwal MG, et al.CT-guided percutaneous core needle biopsyin deep seated musculoskeletal lesions: aprospective study of 128 cases. Skeletal Ra-diol 2006;35:138–143.

20. Zornoza J, Bernardino ME, Ordonez NG,Thomas JL, Cohen MA. Percutaneous needlebiopsy of soft tissue tumors guided by ultra-sound and computed tomography. SkeletalRadiol 1982;9:33–36.

21. Fraser-Hill MA, Renfrew DL. Percutaneousneedle biopsy of musculoskeletal lesions. 1.Effective accuracy and diagnostic utility. AJRAm J Roentgenol 1992;158:809–812.

22. Ng CS, Salisbury JR, Darby AJ, Gishen P.Radiologically guided bone biopsy: results of502 biopsies. Cardiovasc Intervent Radiol1998;21:122–128.

23. Ayala AG, Zornosa J. Primary bone tumors:percutaneous needle biopsy—radiologic-pathologic study of 222 biopsies. Radiology1983;149:675–679.

24. Logan PM, Connell DG, O’Connell JX, MunkPL, Janzen DL. Image-guided percutaneousbiopsy of musculoskeletal tumors: an algo-rithm for selection of specific biopsy tech-niques. AJR Am J Roentgenol 1996;166:137–141.

25. McCarthy EF. T-guided needle biopsies ofbone and soft tissue tumors: a pathologist’sperspective. Skeletal Radiol 2007;36:181–182.

26. Helbich TH, Rudas M, Haitel A, et al. Eval-uation of needle size for breast biopsy:comparison of 14-, 16-, and 18-gauge bi-opsy needles. AJR Am J Roentgenol 1998;171:59–63.

27. Fishman JE, Milikowski C, Ramsinghani R,Velasquez MV, Aviram G. US-guided core-needle biopsy of the breast: how many spec-imens are necessary? Radiology 2003;226:779–782.

28. Koskela AK, Sudah M, Berg MH, et al.Add-on device for stereotactic core-needlebreast biopsy: how many biopsy specimensare needed for a reliable diagnosis? Radiol-ogy 2005;236:801–809.

MUSCULOSKELETAL IMAGING: Core-Needle Biopsy of Bone and Soft-Tissue Lesions Wu et al

970 Radiology: Volume 248: Number 3—September 2008


Recommended