Ultrasound of Musculoskeletal Soft-Tissue Tumors Superficial to the Investing Fascia

Ultrasound of Musculoskeletal Soft-Tissue Tumors Superficial to the Investing FasciaW532 AJR:202, June 2014
histologic type. Increasing experience with ultrasound has allowed radiologists to iden- tify particular soft-tissue tumors on the basis of their imaging characteristics. Recognition of a specific tumor type on ultrasound en- hances patient management and allows one to determine with greater clarity the need for percutaneous biopsy or additional imag- ing such as MRI. The aim of this study was to review our experience using ultrasound to assess superficial soft-tissue tumors first, to determine the accuracy of ultrasound in identifying the type of tumor; second, to rec- ognize the tumors that cause diagnostic dif- ficulty; and, third, to highlight the spectrum of superficial soft-tissue tumors encountered in clinical practice.
Ultrasound of Musculoskeletal Soft-Tissue Tumors Superficial to the Investing Fascia
Esther Hiu Yee Hung1
Alex Wing Hung Ng1
Ryan Ka Lok Lee1
Domily Ting Yi Lau1
Jason Chi Shun Leung2
Hung EHY, Griffith JF, Ng AWH, Lee RKL, Lau DTY, Leung JCS
1Department of Imaging and Interventional Radiology, Prince of Wales Hospital, The Chinese University of Hong Kong, 30-32 Ngan Shing St, Shatin, New Territories, Hong Kong. Address correspondence to E. H. Y. Hung ([email protected]).
2Jockey Club Centre for Osteoporosis Care and Control, The Chinese University of Hong Kong, Shatin, New Territories, Hong Kong.
Musculoskeleta l Imaging • Or ig ina l Research
WEB This is a web exclusive article.
AJR 2014; 202:W532–W540
© American Roentgen Ray Society
S uperficial soft-tissue tumors— that is, tumors arising in the skin and subcutaneous tissues—are a common clinical entity. Since the
mid 1990s, high-resolution ultrasound has been increasingly used as the first-line investi- gation to evaluate soft-tissue tumors [1–3]. The benefits of ultrasound include its ready applicability and availability, high affinity at discriminating tissue layers, good spatial and contrast resolution, ability to evaluate tissue texture through compression, real-time imag- ing capability, and ability to assess tissue vas- cularity and the fact that ultrasound findings directly relate to clinical symptoms.
The treatment and prognosis of subcuta- neous soft-tissue tumors heavily depend on
Keywords: soft-tissue tumors, subcutaneous tumors, superficial tumors, ultrasound
DOI:10.2214/AJR.13.11457
Received June 19, 2013; accepted after revision September 27, 2013.
OBJECTIVE. The objective of our study was to evaluate the diagnostic accuracy of ul- trasound in assessing musculoskeletal soft-tissue tumors superficial to the investing fascia.
MATERIALS AND METHODS. Seven hundred fourteen superficial soft-tissue tumors evaluated with ultrasound by two musculoskeletal radiologists were retrospectively reviewed. In all ultrasound reports, the reporting radiologists provided one, two, or three diagnoses de- pending on their perceived level of diagnostic certainty. Two hundred forty-seven tumors had subsequent histologic correlation, thus allowing the accuracy of the ultrasound diagnosis to be determined. Images of the lesions with a discordant ultrasound diagnosis and histologic di- agnosis were reviewed, and the ultrasound features were further classified as concordant with the known histologic diagnosis, concordant with the known histologic diagnosis with atypical features present, or discordant with the known histologic diagnosis. Four hundred sixty-seven tumors without pathologic confirmation were followed up clinically.
RESULTS. Overall the accuracy of ultrasound examination for assessing superficial soft- tissue masses was 79.0% when all differential diagnoses were considered and 77.0% when only the first differential diagnosis was considered. The sensitivity and specificity of the first ultrasound diagnosis were 95.2% and 94.3%, respectively, for lipoma; 73.0% and 97.7% for vascular malformation; 80.0% and 95.4% for epidermoid cyst; and 68.8% and 95.2% for nerve sheath tumor. Reduced observer awareness of specific tumor entities tended to con- tribute to underdiagnosis more than poor specificity of ultrasound findings. Most tumors (236/247, 96%) were benign. The sensitivity and specificity of ultrasound for identifying ma- lignant superficial soft-tissue tumors was 94.1% and 99.7%, respectively.
CONCLUSION. The diagnostic accuracy of ultrasound in the assessment of superficial musculoskeletal soft-tissue tumors is high and can be improved through increased radiologist awareness of less frequently encountered tumors. Ultrasound is accurate for differentiating benign from malignant superficial soft-tissue tumors.
Hung et al. Ultrasound of Musculoskeletal Soft-Tissue Tumors
Musculoskeletal Imaging Original Research
Ultrasound of Musculoskeletal Soft-Tissue Tumors
Materials and Methods Data Acquisition
Approval was granted by the institutional ethics committee to review patient databases and images for this retrospective review, and the need to ob- tain consent from individual patients was waived. To identify ultrasound examinations of superficial soft-tissue tumors, a single musculoskeletal radiol- ogist performed a database search of all ultrasound examination reports written during a 12-year inter- val (January 2001–November 2012) by one mus- culoskeletal radiologist (observer 1) and during a 3-year interval (May 2008–December 2010) by a second musculoskeletal radiologist (observer 2). Observer 1 had 5 years of experience in musculo- skeletal ultrasound before January 2001, and ob- server 2 had 4 years of experience in musculoskel- etal ultrasound before May 2008.
With regard to superficial soft-tissue tumors, the term “superficial tumor” referred to a tumor arising from the skin or subcutaneous tissue; thus, no tumors occurring deep to the investing fascia were included. For example, no ganglia or giant cell tumors of the tendon sheath were included be- cause these periarticular and peritendinous mass- es arise deep to the investing fascia. The term “tu- mor” referred to a discrete well-marginated mass present on ultrasound. Therefore, nondiscrete or poorly marginated soft-tissue masses such as li- pohypertrophy, soft-tissue edema, or focal inflam- mation were excluded. Other exclusion criteria were known histologic diagnosis at the time of the ultrasound examination and tumor recurrence at the site of a previous tumor resection.
Overall, 735 ultrasound examinations of dis- crete superficial soft-tissue tumors were identified. Of these 735 superficial soft-tissue tumors, 21 tu- mors were excluded because of a known histologic diagnosis at the time of the ultrasound examination (n = 3) or tumor recurrence at the site of a previ- ous tumor resection (n = 18); thus, 714 superficial soft-tissue tumors from 714 patients (417 females, and 297 male patients; mean age ± SD, 43.9 ± 20.3 years; age range, 1–96 years) were available for analysis. These 714 superficial soft-tissue tumors were located in the upper limb (n = 278, 39%), low- er limb (n = 200, 28%), chest wall (n = 34, 5%), ab- dominal wall (n = 39, 5%), back (n = 63, 9%), but- tock (n = 51, 7%), groin or perineum (n = 31, 4%), and head or neck region (n = 18, 3%).
Ultrasound Examination All ultrasound examinations were performed
using a high-resolution (9-17–MHz) linear trans- ducer on one of two ultrasound machines (Son- oline Elegra, Siemens Healthcare; or iU22, Phil- ips Healthcare). The ultrasound diagnosis as to the type of soft-tissue tumor was formulated, ac-
cording to usual clinical practice, not on the ba- sis of one or two ultrasound signs; instead, the ul- trasound diagnosis was formulated on the basis of the relative strength of multiple recognized ultra- sound signs interpreted in conjunction with clin- ical presentation [4–17]. Color Doppler imaging was routinely performed, and the type (arterial or venous), intensity, and pattern (chaotic or or- ganized) of vascularity were assessed. Spectral analysis and sonoelastography were not routinely performed. Lesion consistency was routinely as- sessed through application of transducer pressure.
A diagnosis of a malignant superficial soft-tissue tumor (including metastatic disease) was based on features such as rapidity of growth clinically, the presence of a known primary tumor, the presence of a medium-sized to large tumor, the presence of moderate to severe intratumoral hyperemia, and a lack of similarities with the recognized ultra- sound appearances of specific soft-tissue tumor types (Appendix 1).
Based on the overall ultrasound appearances, each of the 714 superficial soft-tissue tumors was provided with a single diagnosis (n = 629, 88%),
TABLE 1: Number of Correct and Incorrect Ultrasound Diagnoses in 247 Superficial Soft-Tissue Masses With a Pathologic Diagnosis
Pathologic Diagnosis
Ultrasound Diagnosisa
Nerve sheath tumor 11 5 16
Abscess 6 2 8
Pilomatricoma 1 7 8
Organizing hematoma or thrombus 0 5 5
Fibroma or desmoid 6 0 6
Fibrocalcific nodule 4 0 4
Fibrous hamartoma of infancy 0 1 1
Dermatofibroma 1 0 1
Angiomatoid fibrous histiocytoma 0 1 1
Accessory breast tissue 0 1 1
Endometriosis 0 2 2
Connective tissue nevus 0 1 1
Ruptured cyst (unknown cause) 1 0 1
Eccrine spiradenoma 0 1 1
Malignant
Dermatofibrosarcoma protuberans 1 1 2
Note—A single ultrasound diagnosis was provided for 211 of these 218 masses (85%), two diagnoses were provided for 26 (11%) masses, and more than two diagnoses were provided for 10 (4%) masses.
aRefers to all differential diagnoses provided. The ultrasound diagnosis was considered as correct if one of the ultrasound diagnoses specifically matched the pathologic diagnosis.
D ow
nl oa
de d
fr om
w w
w .a
jr on
lin e.
or g
by 1
71 .2
43 .0
.1 61
o n
03 /1
1/ 23
f ro
m I
P ad
dr es
s 17
1. 24
3. 0.
16 1.
C op
yr ig
ht A
R R
S. F
or p
er so
Hung et al.
two diagnoses (n = 58, 8%), or three diagnoses (n = 27, 4%) on the initial ultrasound report. For all superficial soft-tissue tumors, at least one diag- nosis was provided in the report and no more than three differential diagnoses were provided. If the reporting radiologist had a high level of certain- ty regarding the likely tumor type, a single diag- nosis was offered. If the reporting radiologist had a lower level of certainty, two or three diagnoses were offered ranked in order of perceived likeli- ness. Over the study period, a minority of other musculoskeletal superficial soft-tissue tumors un- derwent ultrasound examination by other radiol- ogists in the same institution, but our analysis is confined to only two radiologists because either additional cases were not fully documented or it was not the radiologist’s reporting style to routine- ly provide one, two, or more than two differen- tial diagnoses based on his or her perceived level of certainty. It was imperative for our study that the same reporting style was used for all consecu- tive ultrasound reports to minimize selection bias regarding diagnostic accuracy. Also, only a small number of head and neck superficial tumors were analyzed because ultrasound examinations of these tumors were generally performed by radi- ologists who specialize in head and neck imaging.
Pathologic Diagnosis The 714 superficial soft-tissue tumors were di-
vided into two cohorts: tumors with a subsequent pathologic diagnosis (n = 247) and those without a subsequent pathologic diagnosis (n = 467) (Tables 1 and 2). Pathologic diagnosis was based on en bloc excision (n = 237), percutaneous biopsy followed by en bloc excision (n = 2), or percutaneous biopsy alone (n = 8). Most of these tumors were excised for cosmetic reasons and to obtain a histologic diagno- sis. Fine-needle aspiration for cytologic diagnosis alone was not regarded as a pathologic diagnosis. The pathology reports were used to categorize all superficial soft-tissue tumors according to the 2002 World Health Organization (WHO) classification of histologic typing for soft-tissue tumors [18] or the 2006 WHO classification of histologic typing for skin tumors [19]. All vascular anomalies were grouped under “vascular malformation” without distinctions being made between, for example, cap- illary vascular malformation, venous vascular mal- formation, and hemangioma. In addition, lipoma and angiolipoma were grouped under “lipoma” be- cause, according to the WHO classification [18], an angiolipoma is a “subtype of lipoma in which small capillary-sized vessels containing microthrombi transverse the tumor.”
Accuracy of the Ultrasound Diagnosis Tumors with a pathologic diagnosis—For 247
superficial soft-tissue tumors in 247 patients (136 females, and 111 male patients; mean age ± SD, 43.7 ± 18 years; age range, 1–83 years), there was an ultrasound examination with a subsequent pathologic diagnosis. A single ultrasound diagno- sis was provided for 211 of these 247 (85%) tu- mors, two diagnoses were provided for 26 (11%) tumors, and three diagnoses were provided for 10 (4%) tumors. For this subgroup of tumors, two separate analyses were undertaken. First, the sen- sitivity and specificity of the first ultrasound di- agnosis for the four most common superficial soft-tissue tumors (lipoma, vascular malforma- tion, epidermoid cyst, and nerve sheath tumor) were evaluated. The sensitivity and specificity of less common tumors were not evaluated because of the small sample size (< 10) [20]. Second, the overall accuracy of the ultrasound diagnosis com- pared with the pathologic diagnosis was evaluat- ed. The ultrasound diagnosis was considered cor- rect if one of the ultrasound diagnoses specifically matched the pathologic diagnosis. The accuracies of the first, second, and third ultrasound diagnoses for all tumors compared with the pathologic diag- nosis were also evaluated; an ultrasound diagnosis was considered correct if it specifically matched the pathologic diagnosis.
The images and available video clips of the tu- mors for which the ultrasound diagnosis was incor- rect were reviewed in unison by three radiologists (observers 1–3) who were unaware of the pathologic diagnosis. These radiologists collectively provided an extended list of potential differential diagnoses before being given the pathologic diagnosis. At this stage, the ultrasound appearances were further clas- sified into one of three groups. The first group was tumors with ultrasound features that were fully con- cordant with the pathologic diagnosis—that is, the ultrasound appearance was considered concordant with the pathologic diagnosis but this diagnosis had not been provided in the report. The second group was tumors with ultrasound features that were con- cordant with the histologic diagnosis but atypical features were also present. Examples of atypical fea- tures would be the absence of color Doppler signal in a vascular malformation and a nerve sheath tu- mor or the presence of heterogeneous echogenicity in a lipoma. The third group was tumors with ultra- sound features that were discordant with the histo- logic diagnosis—that is, the ultrasound appearance was considered incompatible with the described or expected ultrasound features of the tumor. In oth- er words, if faced with the same ultrasound appear- ance again, one could not specifically consider this tumor type because the ultrasound features were ei- ther atypical or nonspecific.
TABLE 2: Spectrum of Ultrasound Diagnoses in 467 Superficial Soft-Tissue Masses Without a Pathologic Diagnosis
Ultrasound Diagnosisa No. of Superficial Soft-Tissue Masses
Benign
Abscess 14
Pilomatricoma 2
D ow
nl oa
de d
fr om
w w
w .a
jr on
lin e.
or g
by 1
71 .2
43 .0
.1 61
o n
03 /1
1/ 23
f ro
m I
P ad
dr es
s 17
1. 24
3. 0.
16 1.
C op
yr ig
ht A
R R
S. F
or p
er so
Ultrasound of Musculoskeletal Soft-Tissue Tumors
Tumors without a pathologic diagnosis—Four hundred sixty-seven superficial soft-tissue tumors in 467 patients (281 females, and 186 male patients; mean age ± SD, 43.3 ± 22.6 years; age range, 1–96 years) had no pathologic diagnosis after ultrasound examination. A single ultrasound diagnosis was provided for 418 of these 467 (90%) tumors, two diagnoses for 32 (7%) tumors, and more than two diagnoses for 17 (4%) tumors. For a diagnosis of benignity or malignancy to be established, the elec- tronic databases containing the clinical, radiolog- ic, operative, and pathologic records of these 467 patients were analyzed until the most recent clini- cal entry regarding the outcome of these tumors. The mean follow-up period was 5 ± 3.4 (SD) years (range, 2.4 months to 11 years 3.6 months). A di- agnosis of benignity was based on nonprogression of the tumor either clinically or on serial imaging studies (or both). A diagnosis of malignancy was based on progressive enlargement of the tumor ei- ther clinically or on serial imaging studies together with histologic confirmation of disseminated ma- lignancy including lymphoma. In other words, the tumor was characterized as malignant if there was a confirmed disseminated malignancy and a pro- gressively enlarging superficial soft-tissue tumor consistent with a metastatic tumor. Other than the tumors in patients with known metastatic disease, none of the superficial soft-tissue tumors had pro- gressed on follow-up.
Statistical Analysis Statistical analyses were performed using statis-
tics software (SPSS, version 13.0, SPSS) for Micro- soft Windows. Data are presented as means ± SDs unless otherwise stated. The accuracy, sensitivity, and specificity are presented to the 10th decimal place whereas the other percentages are rounded up to the nearest whole number. The accuracy of the first, second, and third ultrasound diagnoses and the overall accuracy were analyzed. The sensitivity and specificity of the first ultrasound diagnosis for lipoma, vascular malformation, epidermoid cyst, and nerve sheath tumor and of malignant tumors were analyzed and 95% CIs are provided.
Results Soft-Tissue Tumors With a Pathologic Diagnosis
For the cohort of 247 tumors with a patho- logic diagnosis, 27 different superficial soft- tissue tumors were encountered (Table 1). Of these tumor types, lipoma, vascular malfor- mation, epidermoid cyst, and nerve sheath tu- mor comprised nearly three quarters (181/247, 73%) of all pathologic diagnoses (Table 1). Most (236/247, 96%) superficial soft-tissue tumors were benign, with a minority (11/247, 4%) being malignant (Table 1).
Accuracy of First Ultrasound Diagnosis Lipomas—Lipomas accounted for 105 of
the 247 (43%) superficial soft-tissue tumors with a pathologic diagnosis. The first ultra- sound diagnosis was correct in 100 (95%) and incorrect in five (5%) of the 105 cases. Ul- trasound had a sensitivity of 95.2% (95% CI, 89.0–98.0%) and a specificity of 94.3% (95% CI, 88.0–97.0%) for lipoma. False-positive ul- trasound diagnoses of lipoma were assigned to vascular malformation (n = 2), epidermoid cyst (n = 2), myxoid neurofibroma (n = 1), ac- cessory breast tissue (n = 1), and fibrous ham- artoma of infancy (n = 1). False-negative ul- trasound diagnoses of lipoma were assigned to epidermoid cyst (n = 3), nerve sheath tumor (n = 1), and fibroma (n = 1).
Vascular malformations—Vascular mal- formations accounted for 30 of the 247 (12%) superficial soft-tissue tumors with a pathologic diagnosis. The first ultrasound diagnosis was correct in 22 (73%) and in- correct in eight (27%) of the 30 cases. Ul- trasound had a sensitivity of 73.0% (95% CI, 58.0–87.0%) and a specificity of 97.7% (95% CI, 95.0–99.0%) for vascular malfor- mation. False-positive ultrasound diagnoses of vascular malformation were assigned to epidermoid cyst (n = 2), organizing throm- bus (n = 1), pilomatricoma (n = 1), and neu- rofibroma (n = 1). False-negative ultrasound diagnoses of vascular malformation were as- signed to lipoma (n = 2), epidermoid cyst (n = 1), fibroma (n = 1), organizing hema- toma (n = 1), enlarged lymph nodes (n = 2), and connective tissue nevus (n = 1).
Epidermoid cysts—Epidermoid cysts ac- counted for 30 of the 247 (12%) superficial soft-tissue tumors with a pathologic diagno- sis. The first ultrasound diagnosis was cor- rect in 24 (80%) and incorrect in six (20%) of the 30 cases. Ultrasound had a sensitiv- ity of 80.0% (95% CI, 63.0–90.0%) and a specificity of 95.4% (95% CI, 92.0–97.0%) for the diagnosis of epidermoid cyst. False- positive ultrasound diagnoses of epidermoid cyst were assigned to lipoma (n = 3), abscess (n = 1), pilomatricoma (n = 2), vascular mal- formation (n = 1), vascular leiomyoma (n = 1), organizing hematoma (n = 1), and neu- rofibroma (n = 1). False-negative ultrasound diagnoses were assigned to vascular mal- formation (n = 2), lipoma (n = 2), abscess (n = 1), and granulation tissue (n = 1).
Nerve sheath tumors—Nerve sheath tu- mors accounted for 16 of the 247 (6%) su- perficial soft-tissue tumors with a patholog- ic diagnosis. The first ultrasound diagnosis
was correct in 11 (69%) and incorrect in five (31%) of the 16 cases. Ultrasound had a sen- sitivity of 68.8% (95% CI, 44.0–86.0%) and a specificity of 95.2% (95% CI, 92.0–97.0%) for the diagnosis of nerve sheath tumor. False-positive ultrasound diagnoses of…