Skeletal Radiol (2002) 31:349–353

DOI 10.1007/s00256-002-0474-3 A RT I C L E

Mohammad Anwar Hau Accuracy of CT-guided biopsies

Jeung Il Kim
Susan Kattapuram in 359 patients with musculoskeletal lesions
Francis J. Hornicek
Andrew E. Rosenberg
Mark C. Gebhardt
Henry J. Mankin

Received: 9 April 2001
Revised: 9 November 2001
Accepted: 10 December 2001
Published online: 7 March 2002
© ISS 2002
M.A. Hau · J.I. Kim · F.J. Hornicek
M.C. Gebhardt · H.J. Mankin (✉)
Oncology Service,
Department of Orthopaedic Surgery,
Massachusetts General Hospital,
Harvard Medical School, Boston,
MA 02114, USA
e-mail: hmankin@partners.org
Tel.: +1-617-7243700
Fax: +1-617-7266823
S. Kattapuram
Musculoskeletal Radiology,
Department of Radiology,
Massachusetts General Hospital,
Harvard Medical School, Boston,
MA 02114, USA
A.E. Rosenberg
Pathology Department,
Massachusetts General Hospital,
Harvard Medical School, Boston,
MA 02114, USA
Abstract The study was undertaken We conclude that these procedures
to assess the diagnostic accuracy and remain the logical and safe choice
clinical usefulness of computed to- for diagnostic studies of patients
mography (CT)-guided biopsies and with lesions of the musculoskeletal
fine needle aspirates of musculoskel- system.
etal lesions. The analysis compared
the accuracy according to anatomical Keywords Biopsy · Needle · Bone ·
location, size, type of lesion, and his- Soft tissue · Tumor · Accuracy
tology. On the basis of the informa-
tion obtained by reviewing the report
of the CT biopsy and comparing it
with the final diagnosis for 359
cases, the overall accuracy was de-
termined to be 71%. The accuracy
for 101 fine needle aspirations was
63% and for 258 CT-guided core bi-
opsies was 74%. It is of note that the
biopsies of 81 pelvic lesions had
higher rates of diagnostic accuracy
(81%) than those of 278 non-pelvic
sites (68%), and especially 94 le-
sions of the spine (61%). The lowest
success rates occurred in 26 patients
with infectious diseases (50%).

Introduction ative impact on both the outcome of treatment and sur-

Percutaneous needle biopsy is a major and valuable
tool in the diagnosis of musculoskeletal lesions. Jaffe in
1958 [1] stated that a biopsy should be regarded as the
final diagnostic procedure, after thorough study of the
patient. A well-planned and executed biopsy provides
an accurate diagnosis and facilitates treatment. A poor-
ly performed biopsy can be disastrous, not only in
terms of failing to provide the correct diagnosis but
also in delaying treatment. If done improperly an open
biopsy may materially affect and limit treatment op-
tions, jeopardize limb preservation and can have a neg-
vival [2, 3, 4, 5].
The essentials of the successful and safe biopsy of
musculoskeletal lesions are an understanding of the be-
havior of tumors and, more specifically, knowledge of
orthopedic oncology, diagnostic musculoskeletal radiolo-
gy, connective tissue pathology and cytopathology [5, 6,
7, 8]. Open biopsy has been the conventional “gold stan-
dard” procedure for obtaining adequate and representa-
tive samples of tissue for the diagnosis of musculoskele-
tal lesions and the reported accuracy rate is as high as
98% [9]. However, this procedure has the potential for
significant complications, particularly when done outside
350

the treatment center [3, 4]. More recently, percutaneous
needle biopsy and fine needle aspiration biopsy, most of-
ten performed under the guidance of computed tomogra-
phy (CT), have become the procedure of choice. These
procedures are relatively painless, safe and low in cost
[10, 11, 12, 13]. The reported accuracy rate for these
procedures is as high as 96% and the complication rate
has been reported to range from 0 to 7.4% [8, 10, 14, 15,
16, 17, 18, 19, 20]. Therefore most tumor treatment cen-
ters advocate core biopsy performed under CT guidance
[6, 7, 8, 9, 10, 14, 15, 17, 21, 22, 23, 24]. However, if the
diagnosis after a percutaneous needle biopsy is in doubt,
or if the results vary from what is suspected clinically,
then a repeat needle biopsy or an open biopsy is indicat-
ed. The reported incidence of an open biopsy required
after needle biopsy in one series is 18% [13].
Terms used in relation to the biopsy report
Non-diagnostic: In these cases, the biopsy reports were either
not consistent with the final diagnosis and therefore rated as
false (“F”) in our result coding system; or declared to be unsat-
isfactory or insufficient in amount of tissue for diagnosis but
not erroneous, and designated “N”.
Diagnostic: For these cases the biopsy report matched the final
surgical pathology report, or the result was considered ade-
quate and proved to be clinically useful in the management of
the patient. They were designated “T” in our coding system.
All these patients remain with the same diagnosis on follow-up
at the present time, supporting the accuracy of the results.
Results
Demographics

The total number of patients included in this retrospec-

tive study was 359. Two hundred and twenty-six were
Materials and methods male and 133 were female. The mean age was
A retrospective study was performed on 359 patients who under-
49±20 years, with a range of 10–100 years. Two hundred
went CT-guided biopsy between January 1999 and August 2000. and fifty-three biopsies were of bone and 106 of soft tis-
These biopsies included fine needle aspirations (FNA), core needle sue lesions. The distribution of biopsy sites according to
biopsy, or a combination of both FNA and core biopsy, performed anatomical location is shown in Table 1. Ninety-two pa-
for bone and soft tissue lesions. The data were gathered from the tients had both FNA and core biopsy, another 258 pa-
Hospital computerized patient management records, our FOXPRO
Orthopaedic Oncology Service computerized system and the re- tients had core biopsy only, and 9 patients had only a
cords of the Massachusetts General Hospital Pathology Department. FNA performed. The final major diagnoses for all 359
Four hundred and two CT images from patients who had had percu- patients are shown in Table 2.
taneous needle biopsies were reviewed by the musculoskeletal radi-
ologist. On the basis of the imaging and other criteria 43 patients
who had had procedures for therapeutic aspiration of abscesses, ste-
roid injection of cystic lesions or epidural steroid injection were Success rate
eliminated, leaving 359 patients in the series who were referred for
CT-guided biopsy because of suspected primary, recurrent or meta- The statistical analysis of accuracy of all the biopsies ac-
static lesions. The patient’s pre-biopsy examination, including his- cording to type of biopsy, anatomical location and dis-
torical data and physical examination, was performed by the attend-
ing orthopedic surgeon, who reviewed the patient’s imaging studies ease is shown in Table 3. The result demonstrates that of
with the musculoskeletal radiologist and often the pathologist. the 359 cases, 103 were designated as either “F” (inaccu-
The majority of the biopsies (322) were performed under local
anesthesia; only a small group of 37 children or anxious adults re-
quired conscious sedation. All the procedures were performed by Table 1 Anatomical sites bio-
or under the supervision of a senior musculoskeletal radiologist. psied Pelvis 81
The tissue from bone was obtained using 12–16 gauge needles Thigh 63
while 13–18 gauge needles were used for soft tissue tumors. Nee- Lumbar spine 37
dle aspiration biopsies were performed using disposable 18–21 Thoracic spine 27
gauge needles. For each biopsy, the tissue samples were sent to the Leg 27
Pathology Department for interpretation by cytopathologists or Sacrum 22
clinical musculoskeletal pathologists. The cytopathologists re- Arm 18
sponded quickly after study of the FNA, as did the pathologists af- Scapula 14
ter a frozen section. If there was limited tissue available from the Rib 11
core biopsy or the tissue was mostly bone or the diagnosis on fro- Abdomen 10
zen section was questionable, the pathologists preferred to report Shoulder 9
on permanent sections, sometimes with special stains. Under these Cervical spine 8
circumstances, the results were not available for two or more days. Knee 8
One hundred and seventy-seven patients in our series subse- Clavicle 7
quently underwent surgery following the biopsy so that the accu- Sternum 4
racy of the biopsy could be directly correlated with the final histo- Hip 3
logical diagnosis. For the remaining 185 patients, no surgical pro- Thorax 3
cedures were performed because the biopsy results were diagnos- Forearm 3
tic, or provided sufficient information to allow non-surgical treat- Foot and ankle 3
ment. For these cases, the clinical course and response to treat- Hand 1
ment were studied to provide confirmation of the accuracy of the Total 359
biopsy results.
 351

Table 2 Statistical analysis of the accuracy of needle biopsies for lignant tumors the diagnosis was correct in 164 (90%)
the major diagnoses encountered (total series=359 cases) cases (P<0.00001).
Metastatic carcinoma (89) 88% The accuracy (“T”) rates for 253 bone and 106 soft
Soft tissue sarcoma (33) 94% tissue biopsies are 68% and 79% respectively, which
Round cell tumors (34) 94% provide a statistically significant difference (P<0.04).
Chondrosarcoma (11) 95% More importantly, the difference between the percentage
Osteosarcoma (15) 87%
Benign bone tumors (23) 81%
of inaccurate and possibly harmful biopsies of bone
Benign soft tissue tumors (22) 77% (19%) and of soft tissue (7%) is also statistically signifi-
Infection (26) 50% cant (P<0.003). Comparing the FNA and core biopsy re-
Unknown (63) 47% sults, the accuracy rate for core biopsy (74%) is better
than for FNA (63%), and the rate of inaccurate and pos-
sibly harmful biopsies for core biopsy (12%) is much
Table 3 Statistical analysis of the accuracy of biopsy (total lower than for FNA (23%). Both the differences are sta-
series=359 cases). T accurate, F not accurate and possibly harm- tistically significant (P<0.04 and P<0.01, respectively).
ful, N not sure but no harm
As shown in Table 3, anatomical site has a significant
Overall: entire series Overall without unknowns P value effect on accuracy of the biopsy. The accuracy rate (“T”)
(n=359) (n=296) for spinal biopsies is 61% (57 of 94), which is consider-
ably less than the 75% (199 of 256) accuracy for the
T=256 (71%) T=242 (82%) P<0.004 non-spinal sites (P<0.008). Similarly the accuracy rate
N=49 (14%) N=10 (3%) –
F=54 (15%) F=44 (15%) NS for pelvic biopsies of 81% (66 of 81) is higher than the
68% (190 of 278) for non-pelvic biopsies (P<0.02). As
Bone (n=253) Soft tissue (n=106) – shown in Tables 2 and 3, the disease (diagnosis) has a
T=172 (68%) T=84 (79%) P<0.04 significant effect on the accuracy of the biopsy. Biopsies
N=34 (13%) N=15 (14%) – of the 333 benign or malignant tumors have a much bet-
F=47 (19%) F=7 (7%) P<0.003
ter “T” accuracy rate (74%) than the 50% for 26 cases of
FNA (n=101) Core biopsy (n=258) – infection (P<0.01). Additional analyses showed that the
T=64 (63%) T=192 (74%) P<0.04 lesion size, type of margin and gender did not influence
N=14 (14%) N=35 (14%) – the success or failure rates of the biopsies.
F=23 (23%) F=31 (12%) P<0.01 The normal concomitants of the needle biopsies in-
Spine/Sacrum (n=94) No spinal biopsies (n=265) – cluded moderate local pain following subsidence of the
T=57 (61%) T=199 (75%) P<0.008 anesthetic effect and, less frequently, a short-term de-
N=18 (19%) N=31 (12%) – crease in peripheral sensation which disappeared in sev-
F=19 (20%) F=35 (13%) NS eral hours. Local swellings and small hematomas were
considered to be standard features of the protocol and
Pelvic site (n=81) Non-pelvic site (n=278) –
none persisted or required treatment. No other complica-
T=66 (81%) T=190 (68%) P<0.02 tions were noted in the series.
N=11 (14%) N=38 (14%) –
F=4 (5%) F=50 (18%) P=0.003
Infection (n=26) No infection (n=333) – Discussion
T=13 (50%) T=243 (73%) P<0.01
N=10 (40%) N=39 (12%) – A series of successful aspiration biopsies was reported
F=3 (10%) F=51 (15%) NS
by Martin and Ellis in 1930 [25], and the first trephine
bone biopsy was performed by Ellis in 1947 [22]. With
the accumulation of clinical experience, improvement in
rate) or “N” (unknown or unreadable). This leaves 256 biopsy needles and techniques, advances in radiological
cases designated as “T”, which gives an overall accuracy imaging, and improvement in cytological and histologi-
rate of 71%. However, if the 39 “N” which are unknown cal diagnostic abilities, percutaneous needle biopsy of
cases are excluded, leaving the 10 unreadable and hence musculoskeletal lesions has advanced significantly [6,
less potentially harmful biopsies, the overall accuracy 26, 27].
rate is 82%, and the differences are also statistically sig- CT was used initially in pre-biopsy planning in the
nificant (P<0.004). early 1970s [6]. In the mid-1970s CT-guided biopsy in
As shown in Table 2, the 182 high-grade malignant oncology programs was begun [16, 27, 28, 29]. Since
tumors had a much greater accuracy of diagnosis (“T”) then, it has become increasingly popular, and is now
than the 134 benign bone or soft tissue tumors, infection widely accepted as the initial diagnostic tool in many
or lesions of unknown nature. The value for the latter cancer centers [3, 5, 14, 15, 17, 23]. The procedure is less
group was 79 of 134 cases (59%) while for the 182 ma- invasive and much safer (especially in older or debilitated
352

patients) than operative open biopsies [4]. The biopsies
are generally performed using local anesthesia, although
conscious sedation is usually used for children [5, 8]. The
complications of bleeding, infection and contamination of
the tract are minimal [4, 13, 18, 19, 20, 30]. The biopsy
tract can easily be included in the surgical resection and
is unlikely to be the site of a pathological fracture [2].
The procedure, which is associated with a high diagnostic
accuracy rate, is cost-effective [11, 12, 13]. The benefit is
more obvious when the lesion is deep-seated, located in
the pelvis or spine, or very closely related to neurovascu-
lar structures [18, 30, 31, 32, 33, 34]. Besides pain and
discomfort, the procedure has far fewer complications
compared with open biopsy [3, 4, 17, 19].
The better diagnostic accuracy with soft tissue is pre-
sumably because the lesions are more often uniform in
structure and do not require penetration of a cortex. It is
possible under these circumstances to obtain more sam-
ple cores from soft tissue lesions than from bony ones.
The accuracy rates for core biopsies and FNA are 74%
and 63%, respectively. The reasons for the lower diag-
nostic accuracy of FNA are that the procedure is more
operator- and cytologist-dependent. Also more tissue is
obtained with a core biopsy and this allows other special
studies that often are necessary for making a diagnosis.
The reported diagnostic accuracy for FNA was 64–90%
in other series compared with 76–96% for core biopsy
[17, 21, 25, 31].
An interesting finding in our study is that anatomical
site has a significant influence on the accuracy of the bi-
opsy. Pelvic biopsy has the highest accuracy rate (81%)
whereas spine biopsy has the lowest diagnostic accuracy
(61%). Although not clearly defined by this study, sug-
gestions by other authors clearly indicate the concerns of
the radiologists performing spinal biopsies and hence ex-
plain the difficulty with this technique [18, 19, 30, 31,
32, 33, 34]. The possible explanation for the increased
accuracy of pelvic lesions is that they are often larger on
presentation than those in other more cloistered anatomi-
cal sites. Biopsies of non-pelvic and non-spine lesions
have an accuracy rate between the two. Therefore, based
on our result, we can expect low biopsy accuracy in any
bony lesion which is located in the spine, whereas signif-
icantly higher biopsy accuracy can be anticipated in any
soft tissue lesion which is located in the pelvis.
Another important finding of this study is that diagno-
sis and type of lesion also play a significant role in the
accuracy of the biopsy outcome. Biopsies of primary
malignant tumors have more than 90% accuracy and
metastatic lesions have a biopsy accuracy close to 90%.
Benign tumors have a biopsy accuracy rate of around
80%, infection a significantly lower rate of only 50%,
and, as expected, those lesions which have non-charac-
teristic radiological features or unknown diagnostic le-
sions have the lowest accuracy rate of only 47%. It is
therefore very clearly demonstrated in this study that the
accuracy of CT-guided biopsy of a musculoskeletal le-
sion is significantly influenced by the type of lesion
(bone or soft tissue) and its anatomical location, as well
as the underlying pathology of the lesion.
The overall diagnostic accuracy in our study is 71%,
with the accuracy for bone and soft tissue biopsy being
68% and 79%, respectively. Although these results are
comparable to those of other studies, they tend to be
lower [1, 8, 10, 14, 15, 16, 17, 20, 29]. The possible ex-
planation for this variation is the types of tumor included
in the study (Table 2), which may differ from those in
other series, particularly in terms of the groups of benign
bone and soft tissue tumors, infection and lesions which
are classified as unknown, which accounted for 134
cases with a group accuracy of 59%. This differed con-
siderably from the value of 90% accuracy for 182 malig-
nant tumors of bone and soft tissues (P<0.00001).
Conclusions
In conclusion, although CT-guided core needle biopsy of
musculoskeletal lesions in this study has an overall accu-
racy of 71%, it is still an effective and safe diagnostic
tool in the evaluation of musculoskeletal lesions. Fine
needle aspiration biopsy seems to be not quite as useful
in the diagnosis of musculoskeletal lesion as biopsies
performed with CT guidance.

References
1. Jaffe HL. Introduction: problems
of classification and diagnosis.
In: Tumors and tumorous conditions of
bone and joints. Philadelphia: Lea and
Febiger, 1958:9–17
2. Clark CL, Morgan C, Sonstegard D,
Mathews LS, Arbor A. The effect of
biopsy hole shape and size on bone
strength. J Bone Joint Surg Am 1977;
59: 213–217
3. Mankin HJ, Lange TA, Spanier SS.
The hazards of biopsy in patients with
malignant primary bone and soft tissue
tumors. J Bone Joint Surg Am 1982;
64:1121–1127
4. Mankin HJ, Mankin CJ, Simon MA.
The hazards of the biopsy revisited. J
Bone Joint Surg Am 1996; 78:656–663
5. Simon MA. Current concepts review:
biopsy of musculoskeletal tumors.
J Bone Joint Surg Am 1982;
64:1253–1257
6. Hardy DC, Murphy WA, Gilula LA.
Computed tomography in planning per-
cutaneous bone biopsy. Radiology
1980; 134:447–450
7. Logan PM, Connell DG, O’Connell
JX, Munk PL, Janzen DL. Image-guid-
ed percutaneous biopsy of musculo-
skeletal tumors: an algorithm for selec-
tion of specific biopsy techniques. AJR
Am J Roentgenol 1996; 166:137–141

8. Peabody TD, Simon MA. Making the
diagnosis: keys to a successful biopsy
in children with bone and soft tissue tu-
mors. Orthop Clin North Am 1996;
27:453–459
9. Den-Heeten GJ, Oldhoff J, Oosterhuis
JW, Schraffordt KH, Koops H. Biopsy
of bone tumours. J Surg Oncol 1985;
28: 247–251
10. Fraser-Hill MA, Renfrew DL. Percuta-
neous needle biopsy of musculoskele-
tal lesions. 1. Effective accuracy and
diagnostic utility. AJR Am J Roentg-
enol 1992; 158:809–812
11. Fraser-Hill MA, Renfrew DL,
Hilsenrath PE. Percutaneous needle bi-
opsy of musculoskeletal lesions.
2. Cost-effectiveness. AJR Am J
Roentgenol 1992; 158:813–818
12. 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 1996;
78:644–649
13. Yao L, Nelson SD, Seeger LL, Eckardt
JJ, Eilber FR. Primary musculoskeletal
neoplasms: effectiveness of core needle
biopsy. Radiology 1999; 212:682–686
14. Barth RJ Jr, Merino MJ, Solomon D,
Yang JC, Baker AR. A prospective
study of the value of core needle biop-
sy and fine needle aspiration in the
diagnosis of soft tissue masses.
Surgery 1992; 112:536–543
15. Bickels J, Jelinek JS, Schmookler BM,
Neff RS, Malawer MM. Biopsy of
musculoskeletal tumors: current con-
cepts. Clin Orthop 1999; 368:212–219
16. Desantos LA. Murray JA, Ayala AG.
The value of percutaneous needle biop-
sy in the management of primary bone
tumors. Cancer 1979; 43:735–744
17. Dupuy DE, Rosenberg AE,
Punyaratabandhu T, Tan MH, Mankin
HJ. Accuracy of CT guided needle bi-
opsy of musculoskeletal neoplasms.
AJR Am J Roentgenol 1998;
171:759–762
18. Kattapuram SV, Khurana JS,
Rosenthal DI. Percutaneous needle
biopsy of the spine. Spine 1992;
7:561–564
19. Olscamp A, Rollins J, Tao SS,
Ebraheim NA. Complication of CT
guided biopsy of spine and sacrum. Or-
thopedics 1997; 20:1149–1152
20. Schweitzer ME, Gannon FH, Dealy
PM, O’Hara BJ, Juneja V. Percutane-
ous skeletal aspiration and core biopsy:
complementary techniques. AJR Am J
Roentgenol 1996; 166:415–418
21. El-khoury GY, Terepka RH,
Mickelson MR, Rainville KL, Zaleski
MS. Fine needle aspiration biopsy of
bone. J. Bone Joint Surg Am 1983;
65:522–525
22. Ellis F. Needle biopsy on the clinical
diagnosis of tumors. Br J Surg 1947;
34:240–261
23. Leffler SG, Chew FS. CT guided per-
cutaneous biopsy of sclerotic bone le-
sions: diagnostic yield and accuracy.
AJR Am J Roentgenol 1999;
172:1389–1392
24. Tikkakoski T, Lahde S, Puranen J,
Apaja-Sarkkinen M. Combined CT
guided biopsy and cytology in diagno-
sis of bony lesions. Acta Radiol 1992;
33:225–229
25. Martin HE, Ellis EB. Biopsy by needle
puncture and aspiration. Ann Surg
1930; 92:169–181
353
26. Hajdu SI, Melamed MR. Needle biop-
sy of primary malignant bone tumors.
Surg Gynecol Obstet 1977;
133:829–832
27. Moore TM, Meyers MH, Patzakis MJ,
Terry R, Harvey JP Jr. Closed biopsy
of musculoskeletal lesions. J Bone
Joint Surg Am 1979; 61:375–380
28. Ferrucci JT, Wittenberg J. CT biopsy of
abdominal tumours: aids for lesion lo-
calisation. Radiology 1978;
129:739–744
29. Murphy WA, Destouet JM, Gilula LA.
Percutaneous skeletal biopsy 1981:
a procedure for radiologists. Result, re-
view, and recommendations.
Radiology 1981; 139:545–549
30. Kattapuram SV, Rosenthal DI. Percuta-
neous biopsy of the cervical spine us-
ing CT guidance. AJR Am J Roentg-
enol 1987; 149:539–541
31. Brugieres P, Revel MP, Dumas JL,
et al. CT guided vertebral biopsy: a re-
port of 89 cases. J Neuroradiol 1991;
18:351–359
32. Ghelman B, Lospinuso MF, O’Leary
PF, Burke SW. Percutaneous computed
tomography guided biopsy of the tho-
racic and lumbar spine. Spine 1991;
16:736–739
33. Kornblum MB, Wesolowski DP,
Fischgrund JJ, Herkowitz HN.
Computed tomography guided biopsy
of the spine. A review of 103 patients.
Spine 1998; 23:81–85
34. Stoker DJ, Kissin CM. Percutaneous
vertebral biopsy: a review of 135 cases.
Clin Radiol 1985; 36:569–577