Practical approach to ultrasound of soft tissue tumors and the added value of MRI: how I do it

In about three-quarters of soft tissue tumors, the US features are specific enough for an experienced operator to be confident regarding the type of tumor present (Tab. 3) (Fig. 2, Fig. 3, Fig. 8, Fig. 9)^(4,5). Tumors which tend to have more distinctive US features are listed in Tab. 4 along with an approximation of the perceived confidence with which the diagnosis can usually be made based on US alone^(4,5,8). When one is confident as to the nature of the tumor, a single specific diagnosis can be provided in the US report without requiring a differential diagnosis (Tab. 2). Based on data drawn from references (Hung et al.⁽⁴⁾ and Griffith et al.⁽⁵⁾), this single confident diagnosis will be correct in 95% of cases compared to histology^(4,5) (Tab. 3). Most of the 5% incorrect diagnoses are benign tumors considered on US assessment to be other types of benign tumor (Tab. 3). Also, for masses in which the radiologist was confident regarding the diagnosis, based on data from references Hung et al.⁽⁴⁾ and Griffith et al.⁽⁵⁾, malignancy was overlooked in <0.1% of soft tissue masses overall and in <0.3% of those with histology^(4,5) (Tab. 3).

In most of the remaining cases, one will not be completely confident regarding the nature of the tumor following US examination, though one is still quite confident, based on the US appearances and clinical history, that the tumor is benign (Fig. 11). In such situations, the three most likely diagnoses in order of perceived likelihood can be listed, and, if the initial clinical suspicion was that of a malignant lesion, the US examiner should emphasize that the appearances favor a benign rather than a malignant tumor. Depending on location, the clinical context, and the patient’s response to being informed of the US findings, percutaneous biopsy, excisional biopsy, or MRI examination can be performed. If the patient refuses a biopsy or surgery, and MRI examination is not feasible, follow-up US examination is usually recommended at an appropriate time, usually in 3–6 months.

Fig. 11.

Occasionally, based on US appearances and clinical history, there is a possibility that the tumor may be malignant (Tab. 1) (Fig. 7). In this scenario, it is best to arrange an US-guided biopsy and/or MRI examination and proceed accordingly. The value of MRI is to provide a roadmap for surgical excision and also, in some instances, to supply more information on the type of tumor and consistency, such as whether there is necrosis or de-differentiation, which will influence the approach (Fig. 1). The threshold for undertaking percutaneous biopsy in our institution is low and recommended for any soft tissue mass not confidently considered to be benign, based on US and clinical features (Fig. 12). It is not necessary to perform MRI examination prior to percutaneous biopsy, as there is little or no consequent artifact related to biopsy. MRI should be performed, however, prior to open biopsy, which is seldom performed nowadays.

Fig. 12.

We generally do not recommend ultrasound follow-up for tumors suspected of being malignant, though this may be done in uncommon situations, where the initial percutaneous biopsy (of, for example, a lymph node) is inconclusive. One could also consider followup in situations where the likelihood of malignancy is low and the patient is not keen on percutaneous biopsy. Ultrasound follow-up in 1–3 months, depending on initial tumor growth pattern, will help to assess changes in tumor size or appearance.

Cautionary note: US is accurate at defining the presence and location of soft tissue masses and is quite accurate, though not 100% accurate, at defining the nature of many soft tissue masses. One is, after all, making a judgement call based on the clinical history and US appearances alone. US characterization of tumors is not an exact science. Exceptions do occur and incorrect diagnoses will occasionally be made, even by experienced operators (Fig. 10). One should use all available clinical information, as well as a thorough ultrasound assessment, to provide the most likely diagnosis. If uncertainty exists, listing the three most likely diagnoses based on experience is useful for further management rather than merely describing the tumor as ‘indeterminate’. Providing non-committal reports is not informative to the referring clinician, adds to patient and clinician anxiety/uncertainty, potentiates unnecessary investigation and possibly intervention, and is of little value in guiding further clinical management. It is best to be as definite as possible, though clearly not committing to a single specific diagnosis, unless one is sure. If there is a possibility of malignancy, percutaneous biopsy or excision should be arranged. The report should be tempered to reflect the level of any uncertainty. For example, one can use terminology like “probable giant cell tendon sheath” or “most likely a metastatic deposit” to reflect such findings (Fig. 11). Other tips on the ultrasound reporting of soft tissue masses are:

The location (subcutaneous, subfascial, etc.) of mass and relationship to adjacent structures should be noted.

Additional techniques: Elastography and contrast-enhanced US (CEUS) are two US techniques that interrogate aspects other tissue reflectivity and may be helpful in soft tissue tumor characterization.

Elastography measures tumor stiffness. The two types of US elastography used in musculoskeletal imaging are compression-based and shear wave. For compression-based elastography, the transducer is manually compressed and relaxed gently over the tumor, inducing a variable strain which indirectly assesses tissue strain. Tissue stiffness is conventionally color-coded from red (soft) to yellow, green, blue (hard), though the color range can be adjusted and inverted by the operator. The color pattern in the tumor can be subjectively assigned to a numbered category (‘elasticity score’) (Fig. 13). Tumor stiffness can also be compared with adjacent fat for superficial masses and muscle for deep masses, providing a semi-quantitative ‘strain ratio’ measurement. Strain ratio is calculated automatically by the US machine, comparing regions of interest in the tumor with peritumoral tissue. Each region of interest should be as large as possible.

Fig. 13.

In shear wave elastography, an US pulse causes horizontal displacement of the tumor tissues inducing a shear wave. Shear wave elastography is less operator-dependent than strain wave elastography and allows both qualitative and quantitative measurements. Stiffness is assessed qualitatively using a machine-dependent elasticity score or quantitatively using ‘shear wave velocity’ (m/s) of the induced shear wave or ‘Young’s modulus’ (kilopascals, kPa), based on certain assumptions⁽¹⁵⁾ (Fig. 13). Velocity measurements should be made on the most solid non-calcified part of the tumor, with sufficient gel to minimize any tumor compression. Elastography has issues related to inter-operator variability and imperfect uniformity across US machines.

While elastography may serve as a helpful adjunct to standard US examination, it is not specific enough to act as a standalone technique in characterizing soft tissue masses or differentiating benign from malignant masses. The shear wave modulus of epidermoid cysts (23.7 ± 15.5) is higher than that of ganglion cysts (5.8 ± 5.2) or lipomatous tumors (9.2 ± 5.3 kPa)⁽¹⁶⁾. The strain ratio of epidermoid cysts (0.17 ± 2.1) is less than that of lipomas (0.83 ± 0.18), which is less than that of ganglion cysts (2.78 ± 0.48)⁽¹⁷⁾.

Rather than identifying the specific type of tumor, it would be helpful if elastography could broadly differentiate between benign and malignant soft tissues masses, and thus reduce the number of percutaneous biopsies. Most studies, however, have found that standard elastography assessment does not perform sufficiently well enough for it to be applied in clinical practice, at least for detecting sarcomas^(15,18–21). This modest discriminatory ability of elastography in part reflects the tissue heterogeneity of sarcomas. Conversely, carcinomas tend to have a more uniform tissue composition than sarcomas, and, as such, strain elastography is accurate at discriminating cutaneous carcinomas (squamous and basal cell carcinoma) from benign cutaneous lesions (such as seborrhoeic keratosis, actinic keratosis, keloid scar, pilar cyst, epidermoid cyst) with a strain ratio of value of >3.0 indicating malignancy^(22,23). An alternative strain elastography measure, known as the elasticity/ B-mode ratio (E/B ratio), may prove to be more helpful in discriminating malignant from benign soft tissue tumors⁽¹⁸⁾. E/B ratio incorporates peritumoral as well as tumoral tissue stiffness. Malignant tumors often seem larger on elastography than on greyscale US, probably due to a ‘desmoplastic effect’ (i.e. fibrous tissue reaction) around malignant soft tissue tumors. The E/B ratio is obtained by dividing elastography tumor length by greyscale tumor length, with a ratio of >1.0 indicating malignancy and <1.0 indicating benignity. In a study of 83 soft tissue tumors (36 malignant, 47 benign), 86% of malignant tumors had an E/B ratio of >1.0, while only 3% had a ratio of <1.0⁽¹⁸⁾. A total of 30% of benign tumors had an E/B ratio of >1.0, while 58% had a ratio of <1.0⁽¹⁸⁾. Thus, malignancy is more likely if tumor length is larger on elastography than on greyscale imaging.

Contrast-enhanced ultrasound (CEUS) enables real time assessment of tumor microvascularity. Vascular endothelial growth factor (VEGF) activation and neoangiogenesis are features of many malignant soft tissue tumors. A low volume (usually 4.8 ml) bolus of US contrast agent (e.g. SonoVue, Bracco, Milan, Italy), comprising gas microbubbles with a stable shell, is injected intravenously, followed by a 5 ml saline flush. The transducer is positioned stationary over the soft tissue tumor, with microbubbles appearing within seconds of injection, depicting arterial and capillary flow <5 um, with different enhancement patterns (Fig. 14). Microbubbles remain within the intravascular space and are completely cleared from the circulation within 10 minutes⁽²⁴⁾. Six perfusion patterns are generally recognized based on the shape of the time-intensity curve (TIC)^(25,26) (Fig. 15). P1 is almost invariably seen in benign tumors, P2, P3, and P4 occur quite commonly in both benign and malignant tumors, while P5 and P6 patterns tend to be seen more in malignant tumors. For discriminating between benign and malignant tumors, the pooled sensitivity (76%) and specificity (67%) of CEUS is only moderate⁽²⁵⁾.

Fig. 14.

Fig. 15.

Video recording of the perfusion characteristics for 2 minutes also enables a TIC to be drawn using in-built software. Empirical parameters of tissue perfusion are derived from this TIC, such as time-to-peak (TTP), peak intensity (PI), and 50% wash-out intensity. CEUS-derived TICs parameters, similar to dynamic contrast-enhanced MRI, show only limited ability to distinguish benign from malignant soft tissue tumors^(26,27). While 50% wash-out intensity (odds ratio (OR) 1.156, p = 0.016) and 50% washout time (OR 1.023, p = 0.0222) were independent risk factors for malignancy, the relative risk afforded is less than that of, for example, an irregular tumor margin (OR 4.490, p = 0.000) or high tumor vascularity (OR 2.307, p = 0.013)⁽²⁶⁾. Therefore, while perfusion patterns parameters could be used in conjunction with other discriminators of malignancy (Tab. 1), they are not sufficiently accurate to be used as a standalone measure to distinguish benign and malignant soft tissue tumors.

CEUS may, however, be helpful in assessing tumor activity and treatment response. In a study of 19 patients with desmoid-type fibromatosis, variable tumor enhancement was seen, with the most common pattern being hyperenhancement with rapid wash-in and slow washout, compared to surrounding muscle⁽²⁸⁾. CEUS can also be helpful in identifying areas of viable tissue for targeted biopsy, especially in heterogeneous, possibly necrotic soft tissue tumors⁽²⁹⁾.

US alone is sufficient to fully evaluate most soft tissue masses. If US assessment is indeterminate regarding the nature of a soft tissue tumor, MRI usually does not increase specificity regarding tumor type⁽³⁰⁾. Of 42 soft tissue tumors deemed indeterminate on ultrasound examination, subsequent MRI examination (most performed at the behest of the reporting radiologist) did not narrow the differential diagnosis in over two-thirds of cases⁽³⁰⁾. In the remaining one-third of cases, MRI helped with tumor characterization⁽³⁰⁾. Situations where MRI may be advantageous over US in the assessment of soft tissue tumors are outlined in Tab. 5 (Fig. 1, Fig. 16). When MRI is performed following US examination, it can be tailored to address specific unanswered questions⁽³¹⁾. For example, if the tumor is hypervascular on US, there is often no need to undertake contrast-enhanced MRI to assess tumor perfusion⁽³¹⁾. Large deep soft tissue sarcomas (STS) should be evaluated by MRI, primarily to define tumor location and NVB involvement (Fig. 17). Optimization of MR protocol can be helpful in this regard, utilizing, if necessary, small field-of-view imaging or microscopy coils (Fig. 18, Fig. 19). NVB invasion also depends on the nature of the STS. For example, well-differentiated liposarcomas may completely encase a NVB but still be fully resectable (Fig. 20). If the STS abuts bone, the periosteum can be resected en-bloc with the tumor, though cortical or medullary invasion requires bone resection. Contact with the bone cortex in the absence of cortical signal change does not indicate bone invasion (Fig. 1). Joint involvement is uncommon in STS. High-grade (grade 2 or 3) STS on MRI tend to be larger, deeper, with unclear boundaries, have more T2 signal heterogeneity, necrosis, peritumoral edema and peritumoral enhancement compared to low grade STS⁽³²⁾. Specialized MRI techniques can help in grading tumor severity, assessing tumor chemotherapy response, and detecting local recurrence^(33,34).

Fig. 16.

Fig. 17.

Fig. 18.

Fig. 19.

Fig. 20.

Sarcoma accounts for about 1% of all malignant tumors, with about 40% of sarcomas occurring in the limbs, especially the lower limbs. There are over fifty different types of musculoskeletal STS, though liposarcoma, undifferentiated pleomorphic sarcoma, leiomyosarcoma, myxofibrosarcoma, and synovial sarcoma account for three-quarters of these sarcomas⁽³⁵⁾. Rhabdomyosarcoma is more common in children. Synovial sarcomas are more widespread in young adults, usually in the extremities and often around large joints⁽³⁶⁾. 44% of synovial sarcomas show matrix calcification on CT⁽³⁶⁾. Absence of calcification on CT is a poor prognostic sign along with increased patient age, higher tumor grade, and larger tumor size⁽³⁶⁾. On imaging alone, it is usually neither possible nor necessary to characterize the type of sarcoma, and biopsy should be performed instead. Molecular testing for amplification of the MDM2 gene region via fluorescence in situ hybridization (FISH) is 100% sensitive and specific on core needle biopsies at distinguishing benign lipomatous tumors from well-differentiated liposarcoma⁽³⁷⁾.