Professional Documents
Culture Documents
Liver Masses
Liver Masses
Liver Masses
CME Article
Objective. The purpose of this study was to compare the diagnostic accuracy, confidence level, and rec-
ommended management of focal liver masses after contrast-enhanced ultrasonography (CEUS) com-
pared with unenhanced ultrasonography alone. Methods. One hundred sixty-seven patients were
referred for CEUS to characterize a focal liver mass. A 2-person blind read determined benignancy or
malignancy, comparative diagnosis, and accuracy on both ultrasonography and CEUS. Results were
compared with the final diagnoses. Results. The 2 readers could not determine benignancy or malig-
nancy in 77 (46.1%) and 46 (27.5%) of 167 unenhanced scans compared with 2 (1.2%) and 1 (0.6%)
of 167 CEUS scans. The confidence level increased from 0 responses in the 2 highest grades (4 and 5)
on the unenhanced scans to 135 (81.8%) and 132 (79.5%) of 167 at level 5 for CEUS. Regarding the
diagnosis, the confidence level was lowest (grade 1) on the unenhanced scans in 128 (82.1%) and 79
(65.3%) of 167 for the 2 readers and improved to the highest (grade 5) in 110 (65.9%) and 113
(68.1%) of 167. Regarding diagnostic accuracy, the unenhanced scans agreed with the correct diagno-
sis in 85 (50.9%) and 63 (37.7%) of 167, and CEUS agreed with the correct diagnosis in 133 (79.6%)
and 142 (85%) of 167 for readers 1 and 2, respectively. Recommendations for further imaging
decreased from 166 (99.4%) and 147 (88%) of 167 on the unenhanced scans to 30 (18%) and 5 (3%)
of 167 on CEUS for readers 1 and 2. Conclusions. Contrast-enhanced ultrasonography improves the
accuracy and confidence of diagnosis of focal liver lesions and reduces recommendations for further
investigations. Key words: contrast agents; liver neoplasms; microbubbles; ultrasonography.
U
Abbreviations nenhanced ultrasonography has excellent spa-
CEUS, contrast-enhanced ultrasonography; CT, comput- tial and contrast resolution and may therefore
ed tomographic; FNH, focal nodular hyperplasia; HCC,
hepatocellular carcinoma; MI, mechanical index; MR, provide useful information regarding the liver
magnetic resonance and liver masses without the use of contrast
agents. Liver cysts can be identified and confidently diag-
Received February 5, 2007, from the Department of nosed, and a variety of appearances of solid masses may
Medical Imaging, Toronto General Hospital and suggest a specific diagnosis.1 Recognition of a hypoe-
University of Toronto, Toronto, Ontario, Canada
(S.R.W., H.-J.J., T.K.K.); and Departments of Medical choic halo or rim surrounding an echogenic or isoechoic
Biophysics and Medical Imaging, University of liver mass, for example, suggests probable malignancy,2
Toronto, and Imaging Research, Sunnybrook Health
Sciences Centre, Toronto, Ontario, Canada (P.N.B.). and masses with this morphologic characteristic would
Revision requested February 20, 2007. Revised always provoke confirmatory imaging with either com-
manuscript accepted for publication March 1, 2007. puted tomographic (CT) or magnetic resonance (MR)
This work was supported in part by the Terry Fox
Program of the National Cancer Institute of Canada. scans. Multiple hypoechoic masses in the liver most often
Address correspondence to Stephanie R. Wilson, suggest metastases.3 By comparison, the common
MD, Toronto General Hospital, 585 University Ave,
Toronto ON M5G 2N2, Canada. appearance of hemangioma as a solid, uniformly
E-mail: stephanie.wilson@uhn.on.ca echogenic mass, possibly showing increased enhance-
ment deep to the mass, is so well recognized that in a
CME Article includes CME test patient without risk of hepatocellular carcinoma (HCC)
© 2007 by the American Institute of Ultrasound in Medicine • J Ultrasound Med 2007; 26:775–787 • 0278-4297/07/$3.50
26.6.jum.online.q 5/16/07 2:15 PM Page 776
or metastases, identification of such a mass pre- patient’s history and demographics, therefore,
cludes the need for further imaging.4 However, in two different interpretations may result from an
patients at risk for HCC or for a variety of metas- identical ultrasonographic appearance. In the
tases, there is recognition that small, uniformly case of a hemangiomalike mass, this type of
echogenic hemangiomalike masses may repre- interpretation tends to work relatively well in
sent malignant liver tumors,5 and confirmation clinical practice, although it illustrates the lack of
of all such masses in high-risk patients with a methodological basis on which the interpreta-
either CT or MR scans is recommended. This tions can be made in the absence of clinical
extreme reliance on clinical information has information (Figure 1). In many other cases, a
become part of our practice standard but mass seen on ultrasonography is referred for
emphasizes the lack of specificity of convention- contrast-enhanced CT or MR imaging for a con-
al ultrasonography. With knowledge of the fident diagnosis.
B C
Wilson et al
The introduction of contrast-enhanced ultra- the liver, lung, and bones with disease progres-
sonography (CEUS) has provided the opportu- sion over a 3-year follow-up. Three patients with
nity for ultrasonography to play a role in the HCC had an α-fetoprotein level of greater than
noninvasive diagnosis of focal liver masses, an 4000 ng/mL, and 1 patient with HCC had diag-
area previously dominated by contrast- nosis only on medical imaging.
enhanced CT and MR imaging. Studies have The mean size of the benign lesions was 4.3 cm
shown accurate diagnosis of commonly encoun- (range, 0.7–11 cm). Pathologic confirmation was
tered liver masses with CEUS6–10 and also a high obtained in 12 of 97 benign lesions including the
level of agreement of enhancement patterns single lipoma, 3 of 4 regenerative nodules, 2 of 3
between CEUS and contrast-enhanced CT.11,12 adenomas, 5 of 47 FNHs, and 1 hemangioma. All
Nonetheless, the actual impact of the perfor- other benign lesions were confirmed by CT or
mance of CEUS in terms of patient treatment MR imaging or technetium-labeled red blood
remains unclear. In this study, we used a blind cell scintigraphy. Benign diagnoses were estab-
read to compare the diagnosis, confidence level, lished by continued imaging and clinical obser-
and recommended management of focal liver vation over a mean period of more than 4 years
masses after a standard unenhanced sonogram (mean, 50 months; range, 31–70 months).
with examinations of the same patients using
CEUS. These were judged against a final, inde- Imaging Technique
pendent diagnosis for each studied lesion. Ultrasound scans were performed by a single
physician and one of the authors (S.R.W.) with
Materials and Methods more than 25 years of experience in liver ultra-
sonography. Contrast-specific imaging was used:
Patients pulse inversion imaging on an HDI 5000 system
This prospective study had approval of the (Philips Medical Systems, Bothell, WA; n = 145)
Institutional Research and Ethics Board. Patients and contrast pulse sequence imaging on an
gave informed consent. Nonconsecutive patients Acuson Sequoia system (Siemens Medical
with solid liver masses visible on routine ultra- Solutions, Santa Clara, CA; n = 22). All examina-
sonography were recruited for CEUS at the time tions included a thorough scan before liver
of their ultrasound examination. The study pop- enhancement, including identification of the
ulation comprised 156 patients with 167 liver lesion and determination of its size and mor-
masses referred to our department for character- phologic characteristics. Curvilinear transduc-
ization. Two masses were evaluated in 11 ers were used with a center frequency of 1.1 to
patients: in each case, the masses were in differ- 2.2 MHz. The ultrasound contrast agent com-
ent locations in the liver and were reviewed in a prised gas-filled perflutren lipid microspheres
blind read format without visualization of both (Definity; Bristol-Myers Squibb, Billerica, MA)
lesions at the same time. There were 68 men and administered intravenously by multiple small
88 women (age range, 19–86 years; mean age, 51 boluses (range, 3–6; median, 4; aliquot dose
years). Confirmed liver masses comprised 50 range, 0.1–0.4 mL; median, 0.2 mL), followed by
HCCs, 16 metastases, 42 hemangiomas, 47 focal a saline flush, to a maximum total dose of 10
nodular hyperplasias (FNH), and 12 others (3 µL/kg. Injections were separated by an interval
cholangiocarcinomas, 3 adenomas, 4 regenera- of greater than 5 minutes to minimize any cumu-
tive nodules, 1 lymphoma, and 1 lipoma). lative effect of the contrast agent. The scanning
The mean size of the malignant lesions was 5.5 technique used was low-mechanical index (MI
cm (range, 1–12 cm). Pathologic confirmation of <0.1), real-time imaging, which preserves the
the diagnosis was obtained in 62 of 70 malignant microbubble population and allows evaluation
lesions: 45 HCCs, 13 metastases, 3 cholangiocar- of lesional vessels as well as lesion and liver
cinomas, and 1 lymphoma, the last 2 lesions cat- enhancement. The transmit focal zone was
egorized with “other lesions.” Three patients always positioned distal to the lesion of interest.
with known extrahepatic primary cancer had The image frame rate was typically 10 to 15 Hz.
widespread metastases on medical imaging in The region of interest was observed continuous-
ly from the time of injection for about 5 minutes. tion of CEUS, respectively. The readers were
The arterial phase was timed for 45 seconds after blinded to all clinical and demographic informa-
the completion of the flush, after which we tion. The image file included a single unen-
defined an “extended portal venous phase,” hanced image followed by several images from
encompassing the commonly described inter- the CEUS scan, including the baseline image, an
val from 45 to 70 seconds after injection, as well arterial phase image, a portal venous phase
as the remainder of the observation period. image, and a cine clip of the wash-in of the con-
Because the contrast microbubbles are purely trast agent to the peak of arterial phase enhance-
intravascular, there is no interstitial or equilibri- ment. The arterial and portal venous phase
um phase: enhancement in the extended portal images were chosen to show the maximum dif-
phase shows progressive decay over about 3 min- ference between the enhancement of the mass
utes until the baseline appearance is again and the enhancement of the adjacent liver.
observed.
Reader Training
Blind Read Readers relied on their own experience and the
A 2-person blind read was conducted to docu- commonly acknowledged features for determi-
ment the interpretation of the lesion as benign or nation of malignancy and the diagnosis for
malignant, as well as the diagnosis, together with unenhanced ultrasonography and CEUS, sum-
a confidence level and recommended manage- marized below and in Tables 1 and 2.
ment. The procedure was carried out on the On unenhanced ultrasonography, malignancy
ultrasonographic and CEUS images for each was suggested by identification of a hypoechoic
patient. Both readers were authors (H.-J.J. and halo around a solid mass as well as the presence
T.K.K.) who had no prior exposure to the images of hypoechoic or multiple masses.
before the blind read. They had 5 and 3 years Criteria for the differential diagnosis of a focal
experience with the performance and interpreta- mass on unenhanced ultrasonography, as much
Table 1. Diagnostic Criteria of Benignancy and Malignancy on Unenhanced Ultrasonography and CEUS
Tumor Type Imaging Findings
Benign Unenhanced Homogeneous hyperechogenicity
Hypoechogenicity with hyperechoic rind
Posterior sonic enhancement
CEUS No wash-out during portal phase
Malignant Unenhanced Hypoechoic halo
Target appearance
Hypoechoic
CEUS Wash-out during portal venous phase regardless of arterial vascularity
Table 2. Suggestive Features of Common Focal Liver Lesions Based on CEUS Enhancement Patterns
Diagnosis Unenhanced Ultrasonography CEUS
HCC Varied Diffuse arterial enhancement
Portal phase wash-out
Hemangioma Homogeneous hyperechogenicity Peripheral nodular enhancement
Hypoechoic with hyperechoic rind Centripetal progression
Posterior sonic enhancement Sustained enhancement
FNH Subtle lesion with mass effect Brisk homogeneous enhancement
Central spoke wheel artery
Nonenhancing scar
Sustained enhancement
Metastasis Hypoechoic Rim enhancement
Nonhomogeneous echogenicity Rapid and complete wash-out
Hypoechoic halo
Wilson et al
as they were available,1,2 included a uniformly third question addressed differential diagnos-
echogenic4 or a hypoechoic mass with an tic possibilities, with the same choice of
echogenic rim for hemangioma,13 a subtle isoe- answers as for the diagnosis. The last question
choic mass often with a mass effect or contour was the recommendation for further manage-
bulge for FNH,14 single, diffuse, or multifocal ment: nothing, biopsy, nuclear medicine scan,
masses of varying echogenicity for HCC, and a CT, MR imaging, or follow-up ultrasonography.
hypoechoic halo surrounding a solid mass2 or a A single answer was required for each of these
hypoechoic mass for metastases.3 questions except the last, where multiple selec-
On CEUS, criteria for malignancy were based tions were permitted.
on published guidelines,15 combined with other
reports in the literature.8,16,17 A major require- Analysis
ment for the determination of malignancy or The blinded read end points comprised an
benignancy was the appearance of the mass in assessment of benign/malignant, specific diag-
the portal venous phase relative to the liver.8,16,18 nosis, diagnostic confidence, differential diagno-
A mass that appeared more enhanced or of sis, and recommendation for management.
greater echogenicity was interpreted as showing Descriptive statistics were calculated for each
sustained enhancement, commonly encoun- question showing the number and percentage
tered with benign lesions. Malignant lesions, in of cases with specific question responses.
comparison, tend to show wash-out or hypoe- Inferential statistics were used to compare the
chogenicity relative to the enhanced liver. performance of the CEUS and unenhanced
Additional arterial phase enhancement charac- imaging, performed for individual readers and
teristics used to make a diagnosis included with a combined analysis to evaluate an overall
identification of peripheral nodular enhance- effect and reader effect. For benign/malignant
ment with centripetal progression for heman- classification, the evaluations were compared
gioma and stellate vascularity for FNH. by the overall proportion of correct assessments.
Hepatocellular carcinoma shows more variabili- The analysis of the benign/malignant classifica-
ty in its enhancement pattern, with arterial tion was completed assuming cases in which no
phase hypervascularity and wash-out that may assessment was possible as nonagreement and
be less complete and later than that seen for for only cases in which an assessment was made
nonhepatocyte malignancy.19,20 Metastases may for both imaging sets. Similarly, the ability to
show some variation of their arterial phase make a correct diagnosis was compared assum-
enhancement but are characterized by rapid ing the cases in which no diagnosis could be
wash-out within the time frame traditionally made as failure and using only the cases in
defined as the arterial phase.21 which a diagnosis was made for both imaging
sessions. The need for additional procedures
Reader Questions was analyzed as a binary variable, using the
The same questions were asked after assessment Fisher exact test for comparisons within a read-
of the unenhanced ultrasonographic images er and a repeated measures model to incorpo-
and the CEUS images. The responses based on rate both readers. In the case of specific
the unenhanced images were completed diagnoses, the overall number of diagnoses, dif-
before evaluation of the CEUS. The first ques- ferential diagnosis, and cases in which no diag-
tion was the determination of benignancy or nosis was possible were summarized. The
malignancy, with 3 possible responses: benign, homogeneity of the 3 levels and the dichoto-
malignant, or do not know. The confidence mous outcome of single versus multiple or no
level for this interpretation was graded from 1 diagnosis cases were also compared with a
(lowest confidence) to 5 (highest confidence). Fisher exact test for each reader and with a
The second question addressed the most likely repeated measures logistic model to incorporate
diagnosis, with choices comprising FNH, both readers. P < .05 was regarded as significant.
hemangioma, HCC, metastases, other, and do Analysis was performed with SAS statistical soft-
not know, also with a confidence level. The ware (SAS Institute Inc, Cary, NC).
Wilson et al
(37.7%) correct diagnoses for the unenhanced case, neither reader was confident enough in the
scans and 133 (79.6%) and 142 (85%) correct unenhanced diagnosis to not recommend an
diagnoses for the CEUS scans for readers 1 and 2 additional diagnostic procedure (Figures 3 and
(Figure 2). 5). For the CEUS studies, the 2 readers were con-
The readers were then asked what they would fident enough in their diagnosis to recommend
recommend as the next step in the patient’s diag- no additional diagnostic testing in 118 (70.7%)
nostic workup. As can be seen in Table 4, in every and 113 (67.7%) cases, respectively (Figure 6).
Figure 2. Agreement of unenhanced imaging and CEUS for determination of malignancy in a young woman with hepatitis B virus
and HCC. A, Unenhanced sonogram of the right lobe of the liver shows an expansive, somewhat exophytic mass with a faint hypoe-
choic halo. The image suggests a malignant mass, either metastasis or HCC. B, Arterial phase image shows extensive dysmorphic
vascularity. C, At the peak of arterial phase enhancement, the lesion is enhanced more than the liver. D, In the portal venous phase,
the lesion has washed out. The diagnosis was HCC on CEUS.
A B
C D
Figure 3. Unenhanced scan suggests malignancy; CEUS definitively shows benign FNH. A, Unenhanced sagittal scan of the left lobe
shows a grossly fatty liver with a focal hypoechoic mass. With or without clinical information, this is suggestive of malignancy. Further
imaging would be recommended on the basis of this scan. B, Arterial phase image shows faint evidence of a stellate pattern to
enhancement. C, At peak arterial phase enhancement, the lesion is homogeneous and hypervascular. D, In the portal venous phase,
the lesion is completely isovascular, and there is a nonenhancing scar.
A B
C D
Wilson et al
B C
B C
Wilson et al
graphers, aware of the patient’s sex, age, clinical siderable improvement in both diagnostic accu-
presentation, and results of other imaging tests, racy and specificity provided by the addition of
would perform much better than did the readers contrast to an ultrasound examination per-
in this study. The addition of CEUS, however, formed for the characterization of a focal liver
improved all aspects of the liver mass diagnosis mass. Furthermore, recommendations for fur-
despite the continued absence of clinical and ther imaging and biopsy were also significantly
demographic information. This attests to a con- reduced on CEUS when compared with the
Figure 6. Giant hemangioma is indeterminate on an unenhanced scan but is definitively diagnosed on CEUS. A, Unenhanced sagit-
tal sonogram of the right lobe of the liver shows a very large lobulated and somewhat heterogeneous mass, suggestive of possible
malignancy. Such a large mass would always be investigated further with either CT or MR imaging. B, Early arterial phase image shows
multiple peripheral puddles of contrast. C and D, Over subsequent frames, there is progressive centripetal progression of enhance-
ment, classic for hemangioma.
A B
C D
unenhanced scans. The use of CEUS to provide an Although many of the patients in our study did
immediate diagnosis of a mass detected on unen- have chronic liver disease, we have not addressed
hanced ultrasonography would, therefore, have a that specific problem in this present investiga-
substantial benefit to management in terms of tion.
reduction of hospital visits, a shorter time to diag- Contrast-enhanced ultrasonography has the
nosis, and also the initiation of proper treatment. advantages of safety, with no radiation exposure
For diagnosis of benign lesions, such as heman- and a very low incidence of adverse events,27 ease
giomas, the reduction in referral for further imag- of performance, and relatively low cost. The
ing such as CT or MR imaging is also important for examination is performed in real time and shows
both radiation risk and reduction in costs. lesional enhancement similar to that of contrast-
Clinical information is so imperative for liver enhanced CT and MR scans together with vessel
mass diagnosis that competent imagers will not morphologic characteristics. A clear improve-
interpret images without it.22 Knowledge of risk ment associated with CEUS for determination of
factors for HCC, such as positive serologic results malignancy and provision of the correct diagno-
for hepatitis B or C or known cirrhosis, has a sis is evident. As experienced sonographers
strong impact on the interpretation of imaging working with CEUS, we think that the addition of
studies. Hepatocellular carcinoma may be con- contrast enhancement allows ultrasound to
sidered in preference to other diagnoses, even in occupy a reliable role in the noninvasive diagno-
the absence of the classic diagnostic enhance- sis of liver masses.
ment features. Similarly, a history of malignancy
makes every identified liver mass a possible References
metastasis until proven otherwise.
Previous investigators have compared the per- 1. Harvey CJ, Albrecht T. Ultrasound of focal liver lesions. Eur
Radiol 2001; 11:1578–1593.
formance of CEUS with that of unenhanced
ultrasonography. Bryant et al23 documented the 2. Wernecke K, Vassallo P, Bick U, Diederich S, Peters PE. The
distinction between benign and malignant liver tumors on
positive impact of CEUS performed with Levovist sonography: value of a hypoechoic halo. AJR Am J
(SH U 508A; Schering AG, Berlin, Germany), an Roentgenol 1992; 159:1005–1009.
air-filled microbubble contrast agent, scanned
3. Paulson EK. Evaluation of the liver for metastatic disease.
with high MI imaging. Although their results Semin Liver Dis 2001; 21:225–236.
were positive, the necessity to use a high MI with
4. Bree RL, Schwab RE, Neiman HL. Solitary echogenic spot in
this agent precluded real-time imaging, the cur- the liver: is it diagnostic of a hemangioma? AJR Am J
rently preferred method. Subsequent investiga- Roentgenol 1983; 140:41–45.
tors24,25 described the more significant impact of 5. Caturelli E, Pompili M, Bartolucci F, et al. Hemangioma-like
CEUS with the second-generation perfluorocar- lesions in chronic liver disease: diagnostic evaluation in
bon agent SonoVue (Bracco SpA, Milan, Italy), patients. Radiology 2001; 220:337–342.
similar to that used in this current study. Leen et 6. Wilson SR, Burns PN, Muradali D, Wilson JA, Lai X.
al24 showed a reduction in indeterminate scans Harmonic hepatic US with microbubble contrast agent: ini-
tial experience showing improved characterization of
in 127 focal liver masses and also a reduced need
hemangioma, hepatocellular carcinoma, and metastasis.
for further imaging, in agreement with our Radiology 2000; 215:153–161.
results. In a larger series of 452 liver lesions,
7. Brannigan M, Burns PN, Wilson SR. Blood flow patterns in
Quaia et al25 also showed results similar to our focal liver lesions at microbubble-enhanced US.
own, with improved differentiation of benign Radiographics 2004; 24:921–935.
from malignant lesions and a similar improve- 8. Nicolau C, Vilana R, Catala V, et al. Importance of evaluat-
ment in both diagnostic accuracy and confi- ing all vascular phases on contrast-enhanced sonography
dence level. Xu et al,26 evaluating the difficult in the differentiation of benign from malignant focal liver
lesions. AJR Am J Roentgenol 2006; 186:158–167.
problem of determining benignancy or malig-
nancy in 200 liver lesions of less than 3 cm in 9. Ricci P, Laghi A, Cantisani V, et al. Contrast-enhanced
diameter, many in patients with chronic liver dis- sonography with SonoVue: enhancement patterns of
benign focal liver lesions and correlation with dynamic
ease, also showed a considerable improvement gadobenate dimeglumine-enhanced MRI. AJR Am J
of CEUS over unenhanced ultrasonography. Roentgenol 2005; 184:821–827.
Wilson et al
10. Dietrich CF. Characterisation of focal liver lesions with con- 25. Quaia E, Calliada F, Bertolotto M, et al. Characterization of
trast enhanced ultrasonography. Eur J Radiol 2004; focal liver lesions with contrast-specific US modes and a
51(suppl):S9–S17. sulfur hexafluoride-filled microbubble contrast agent: diag-
nostic performance and confidence. Radiology 2004;
11. Burns PN, Wilson SR. Focal liver masses: enhancement pat- 232:420–430.
terns on contrast-enhanced images—concordance of US
scans with CT scans and MR images. Radiology 2007; 26. Xu HX, Liu GJ, Lu MD, et al. Characterization of small focal
242:162–174. liver lesions using real-time contrast-enhanced sonogra-
phy: diagnostic performance analysis in 200 patients.
12. Catala V, Nicolau C, Vilana R, et al. Characterization of J Ultrasound Med 2006; 25:349–361.
focal liver lesions: comparative study of contrast-enhanced
ultrasound versus spiral computed tomography. Eur Radiol 27. Piscaglia F, Bolondi L. The safety of SonoVue in abdominal
2007; 17:1066–1073. applications: retrospective analysis of 23,188 investiga-
tions. Ultrasound Med Biol 2006; 32:1369–1375.
13. Moody AR, Wilson SR. Atypical hepatic hemangioma: a
suggestive sonographic morphology. Radiology 1993; 188:
413–417.
20. Jang HJ, Kim TK, Burns PN, Wilson SR. Enhancement pat-
terns of hepatocellular carcinoma on contrast-enhanced
ultrasound: correlation with pathologic differentiation.
Radiology. In press.