Gastrointestinal Imaging
Original Research
Diagnostic Performance and Confidence of Contrast-Enhanced Ultrasound in the Differential Diagnosis of Cystic and Cysticlike Liver Lesions
Audio Available | 
Share
OBJECTIVE. The aims of this study were to assess the diagnostic performance of contrast-enhanced ultrasound (CEUS) in the characterization of atypical cystic and cysticlike focal liver lesions in comparison with conventional US and to determine whether the use of CEUS can reduce the need for further diagnostic workup.
SUBJECTS AND METHODS. In a 3-year period 48 patients with 50 atypical cystic and cysticlike lesions found at conventional US underwent CEUS. Diagnostic confirmation was obtained in cytohistopathologic examinations, with other imaging modalities, and in follow-up. Overall, there were 24 cystic lesions and 26 cysticlike solid lesions, specifically 32 benign and 18 malignant lesions. The conventional US and CEUS images and cine loops were reviewed by two blinded readers independently. Sensitivity, specificity, area under the ROC curve (Az), and interobserver agreement were calculated.
RESULTS. Diagnostic performance improved after review of CEUS examinations by both readers (conventional US Az = 0.781 vs 0.972; CEUS Az = 0.734 vs 0.957). Interreader agreement increased, although slightly (conventional US weighted κ = 0.894; CEUS weighted κ = 0.953). In terms of differential diagnosis, the occurrence of correctly characterized lesions increased after CEUS for both readers (reader 1, 62% to 98%; reader 2, 56% to 96%).
CONCLUSION. The development of low-acoustic-power CEUS has made it possible to identify several imaging features of cystic and cysticlike focal liver lesions that, in association with history and clinical findings, may help to correctly characterize them. Our data indicate the usefulness of CEUS in the evaluation of patients with these lesions.
Keywords: contrast-enhanced ultrasound, cystic lesions, cysticlike lesions, focal liver lesions, microbubbles, ultrasound
Cystic and cysticlike focal liver lesions (FLLs) encompass a spectrum of disorders ranging from nonneoplastic conditions to benign and malignant tumors. They can create a diagnostic dilemma if one is not aware of the potential diagnoses and imaging pitfalls [1, 2].
Because of the widespread use of ultrasound (US) in routine examinations of the abdomen, FLLs are increasingly found. Most are simple cysts and have a benign course and do not cause much clinical concern. However, although the sonographic diagnosis of a simple cyst is straightforward, there are several situations in which misdiagnosis is probable [3]. Diagnostic controversies arise when the lesion encountered has internal echoes without a distinctly visible solid appearance or when the lesion is hypoechoic but lacks posterior wall enhancement. In this regard, the term “atypical,” as used in the literature and in this study, refers to lesions that differ sonographically from simple cysts in the following US features: internal echoes, absence of posterior enhancement, or both [4, 5].
Because FLLs that appear cysticlike at US may turn out to be solid, the radiologist must carefully assess such US features to determine whether further imaging workup is indicated [5]. The development of low-mechanical-index real-time contrast-enhanced ultrasound (CEUS) techniques performed with the second generation of US contrast agents has led to accurate depiction of macrovasculature and microvasculature. The technique yields information about contrast enhancement of the liver and FLLs almost as CT and MRI do but in real time and without the use of ionizing radiation [3].
CEUS has proved to be extremely useful in the evaluation of both solid liver lesions and complex renal cysts [6, 7]. CEUS is also highly accurate in the categorization of complex cystic FLLs as benign or malignant. Sensitivity ranges from 87% to 93% and specificity from 88% to 98% depending on operator experience [8]. However, little experience exists in the CEUS evaluation of atypical cystic and cysticlike FLLs. To our knowledge, no reports have focused on the diagnostic performance of CEUS in the characterization of these lesions.
The aim of our study was to assess the diagnostic performance of CEUS in the characterization of atypical cystic and cysticlike lesions in comparison with that of conventional US and to determine whether the use of CEUS can reduce the need for further diagnostic workup.
| Subjects and Methods |   |
|---|
Between January 2013 and April 2016 we prospectively enrolled 48 patients (23 men, 25 women; age range, 18–78 years; mean, 65 years) imaged at our cancer institute. These patients had 50 atypical cystic and cysticlike FLLs at conventional US. Fifteen patients had a concomitant extrahepatic malignancy diagnosed within 3 months, and 13 patients were undergoing staging evaluation after a new diagnosis of extrahepatic malignancy. The other 20 patients had no history of malignancy, and the lesion represented an incidental finding during an abdominal US examination performed as part of different screening programs. No patient enrolled had previously undergone chemoradiotherapy or other treatment. Written informed consent was obtained from all patients, and the study was approved by the ethics committee of the cancer institute.
We included all lesions that at conventional US appeared anechoic lacking posterior enhancement or were markedly hypoechoic and exhibited internal echoes with or without posterior enhancement (Fig. 1). We excluded simple cysts appearing as well-defined anechoic lesions with posterior acoustic enhancement and imperceptible walls, cystic clusters with thin and regular septa (indicating coalescence of simple cysts), and complex cysts with thick or irregular walls, internal septa, and mural or septal nodules.
 View larger version (27K) | Fig. 1 —Diagram shows ultrasound (US) appearance of atypical cystic and cysticlike focal liver lesions (Illustration by Corvino A). A, Simple cysts (not included in series) are defined as well-circumscribed round or ovoid anechoic lesions with increased through-transmission of US waves. B–D, Included in series were all lesions that at conventional US appeared anechoic lacking posterior enhancement (B) or were markedly hypoechoic and exhibited internal echoes with (C) or without (D) posterior enhancement. |
Among the 48 patients enrolled in this study, 46 had a single liver lesion, and two had multiple lesions. In 14 of the 50 FLLs, diagnostic confirmation was obtained by pathologic examination of specimens obtained at resection (n = 7) and US-guided percutaneous fine-needle aspiration cytologic analysis or biopsy (n = 7). Final diagnosis without histologic or cytologic confirmation was made for patients for whom cytologic analysis or biopsy was not justified. In these cases, another imaging modality (CT [n = 11] or MRI [n = 5]) or clinical follow-up of at least 12 months (n = 20) was judged the reference standard. CT and MRI confirmation was based on diagnostic criteria for each entity according to the most recent literature [9–12]. In the patients who underwent clinical follow-up, the final diagnosis was based on all available clinical and laboratory data, including serologic test results for hydatid disease, and follow-up information.
In the end, there were 24 cystic lesions and 26 cysticlike lesions, specifically 22 hemorrhagic cysts, two hydatid cysts, eight hemangiomas, and 18 metastases (Table 1).
US and CEUS were performed with a MyLab 70 Twice scanner (Esaote) and multifrequency (2.5–5 MHz) convex probes. The conventional US examination included assessment of the lesions with B-mode gray-scale US and color and power Doppler imaging. For Doppler imaging, the lowest pulse repetition frequency and Doppler gain settings that did not cause aliasing were used. The pulse repetition frequency was set at 350–1000 Hz with 70–80% color gain, and a 25- to 50-kHz wall filter was used. With a restricted FOV, the color box was maintained in the area of interest and as small as possible to keep the frame rate high. Vessels were identified on the basis of color flow signals.
CEUS studies were performed with contrast-specific low-mechanical-index software (CnTI, Contrast Tuned Imaging). The agent used was sulfur hexafluoride–filled microbubble contrast medium (SonoVue, Bracco). A bolus injection into an antecubital vein was followed by a flush of 10 mL of 0.9% normal saline solution. The volume of agent used was 2.4 or 4.8 mL, depending on radiologist preference, patient type, or number and distribution of lesions.
In patients with multiple lesions, each lesion was evaluated. In these cases, the single bolus was split into two injections, one for each liver lobe. We waited at least 5 minutes between the two injections. For both injections we scanned the largest or the most suspicious lesion during the arterial phase and explored the other lesions in the same lobe for up to 5 minutes. We used a dual-frame real-time mode with the fundamental-mode US image on the right part of the monitor and the contrast-enhanced image on the left.
The US beam was focused at the deeper aspect of the lesion examined. After contrast injection, continuous scanning was begun immediately and lasted 4–5 minutes. A low acoustic power setting was used (40–45 kPa derated pressure, expressing a mechanical index of approximately 0.06). Digital cine loops were registered during both baseline and CEUS scanning in the arterial (10, 15, 25, 35 seconds from the beginning of contrast bolus injection), portal venous (30, 45, 120 seconds from the beginning of injection), and sinusoidal (> 120 seconds to the disappearance of microbubbles) phases. All cine loops were digitally stored as raw data at a PC-based workstation connected to the US units via a standard Ethernet link.
All conventional US and CEUS on-site examinations were performed by four radiologists, who had at least 2 years of experience in liver CEUS. The examiners were blinded to other imaging results but not to the clinical history. For each patient, we subjectively evaluated the echogenicity of the liver parenchyma, which was categorized according to the bright echo pattern of the liver-to-kidney gradient as normal (grade 0), mildly fat (grade 1), moderately fat (grade 2), and markedly fat (grade 3).
The conventional US features of the FLLs were analyzed as follows: anatomic location, size (largest diameter measured), depth from bottom of lesion to abdominal wall (≤ 5 cm, 5.1–10 cm, > 10 cm), margins, echogenicity (classified as anechoic, hypoechoic, isoechoic, or hyperechoic by comparing the echogenicity of the lesion with that of the surrounding liver), internal echotexture (homogeneous or heterogeneous), and presence or absence of posterior acoustic enhancement. To complement conventional US, color and power Doppler imaging were performed to assess the presence or absence of intralesional vascularity.
The following CEUS parameters were evaluated: time of initial lesion enhancement, degree of enhancement, and pattern of each lesion at different phases of CEUS. According to World Federation for Ultrasound in Medicine and Biology and European Federation of Societies for Ultrasound in Medicine and Biology guidelines [13], contrast enhancement patterns after contrast injection were classified as follows: 1, absent (no difference in echogenicity); 2, peripheral globular (enhancing peripheral nodular areas); 3, rimlike (continuous ring of peripheral enhancement); 4, dotted (tiny hyperechoic spots of enhancement distributed throughout the lesion); 5, spoke wheel (enhancement of the central portion of the lesion with a central vessel branching from the center toward the periphery of the lesion); 6, diffuse homogeneous (uniform enhancement of entire lesion); 7, diffuse inhomogeneous (heterogeneous enhancement of entire lesion). The diagnostic criteria used to define the cystic or solid nature of an FLL were based on our experience with complex cystic FLLs [14, 15] and on our previous experience with liver abscesses [16]. Complete absence of internal enhancement was considered a CEUS finding suggestive of a cystic FLL. On the contrary, if the lesion had evidence of contrast enhancement at CEUS, it was interpreted a solid FLL. In that case, enhancement pattern type suggested the specific diagnosis.
Conventional US and CEUS images and cine loops were randomly reviewed from digital files on a computer screen and evaluated by two investigators with at least 5 years of experience in liver CEUS who were not involved in the US scanning and were unaware of the clinical and other imaging information on the patients. The examiners could not be blinded to the liver parenchymal echo structure.
Six consecutive interpretation sessions with 2-day intervals were conducted to complete the review of all patient baseline and CEUS examinations. To assess the diagnostic performance of conventional US and CEUS, for each lesion the two radiologists were asked first to provide a diagnosis of benign or malignant using a 5-point confidence rating score (1, definitely benign; 2, probably benign; 3, indeterminate; 4, probably malignant; 5, definitely malignant) and second to characterize, if possible, the lesion and provide a specific diagnosis.
Benign or malignant diagnoses were judged to be true-positive, the lesion was correctly assessed as malignant with an assigned confidence grade of 4 or 5; false-negative, the lesion was assigned a confidence grade of 1, 2, or 3 and incorrectly assessed as benign; true-negative, the lesion was assigned a confidence grade of 1 or 2 and correctly assessed as benign; or false-positive, the lesion was assigned a confidence grade of 3, 4, or 5 and incorrectly assessed as malignant. Sensitivity, specificity, positive predictive value, and negative predictive value were calculated. Differences in sensitivity and specificity were tested by McNemar test.
Interobserver agreement between radiologists for each conventional US and CEUS interpretation was assessed with weighted kappa statistics. The agreement was graded in the linear set with five categories: 1, 0.75, 0.50, 0.25, and 0 when there was a difference of 0 (total agreement) or as 1, 2, 3, and 4.
ROC curves were plotted with MedCalc for Microsoft Windows (version 13.1.2.0, MedCalc) for evaluating the diagnostic performance of conventional US and CEUS in regard to discrimination between benign and malignant lesions. Diagnostic performance was expressed as area under the ROC curve (Az). Differences between ROC curves were compared by univariate z score test. The ROC curve is a plot of test sensitivity (plotted on the y-axis) versus its false-positive rate or 1 minus specificity (plotted on the x-axis). All possible combinations of sensitivity and specificity that can be achieved by changing the cutoff value of the test can be summarized with the single parameter area under the ROC curve. Diagnostic tests with perfect discrimination between negative and positive reference groups have ROC plots passing through the coordinates (0, 1), corresponding to 100% sensitivity and specificity (in this case, Az = 1).
| Results | |
|---|
According to the bright liver echo pattern liver-to-kidney gradient, liver background was categorized as normal (grade 0) in 11 patients, mildly fat (grade 1) in 24 patients, moderately fat (grade 2) in nine patients, and markedly fat (grade 3) in four patients. FLLs were located in the right lobe in 39 cases and in the left lobe in the other 11 cases. The largest diameter of the lesions ranged from 9 to 39 mm (mean, 20 mm). A mean depth of 4.8 cm (range, 2.4–14.2 cm) was calculated. Specifically, 27 lesions were found at a depth of less than 5 cm, 20 lesions were between 5.1 and 10 cm deep, and three lesions were more than 10 cm deep. All lesions in our series had well-defined margins: smooth in 48 cases (96%) and lobulated in the other two (4%).
In terms of echogenicity, 18 lesions (36%) were anechoic lacking posterior enhancement, and 32 lesions (64%) were markedly hypoechoic. Among the markedly hypoechoic lesions, eight lesions (25%) exhibited internal echoes with posterior enhancement, and 24 (75%) had internal echoes without posterior enhancement. At color Doppler and power Doppler evaluation, intralesional signal was not identified in 44 lesions (88%). The other six FLLs (12%) had Doppler signal within the lesions.
After injection of the US contrast agent, complete nonenhancement throughout all vascular phases was observed in all benign cysts, including 22 hemorrhagic cysts (Figs. 2 and 3) and two hydatid cysts.
 View larger version (444K) | Fig. 2A —38-year-old woman with lung adenocarcinoma and hemorrhagic liver cyst correctly diagnosed with contrast-enhanced ultrasound. Diagnostic confirmation was obtained with MRI. A, Gray-scale ultrasound image shows dirty cystic lesion with low-level internal echoes (arrows) and posterior enhancement (arrowheads). |
 View larger version (287K) | Fig. 2B —38-year-old woman with lung adenocarcinoma and hemorrhagic liver cyst correctly diagnosed with contrast-enhanced ultrasound. Diagnostic confirmation was obtained with MRI. B, Contrast-enhanced ultrasound images obtained 30 (B) and 180 (C) seconds after contrast injection show complete absence of internal enhancement of lesion (arrows). |
 View larger version (450K) | Fig. 2C —38-year-old woman with lung adenocarcinoma and hemorrhagic liver cyst correctly diagnosed with contrast-enhanced ultrasound. Diagnostic confirmation was obtained with MRI. C, Contrast-enhanced ultrasound images obtained 30 (B) and 180 (C) seconds after contrast injection show complete absence of internal enhancement of lesion (arrows). |
 View larger version (145K) | Fig. 2D —38-year-old woman with lung adenocarcinoma and hemorrhagic liver cyst correctly diagnosed with contrast-enhanced ultrasound. Diagnostic confirmation was obtained with MRI. D, Axial T1-weighted (D) and T2-weighted (E) fat-suppressed MR images show fluid content of hepatic cyst (arrows) to be heterogeneously isointense on T1-weighted and hyperintense on T2-weighted images. |
 View larger version (143K) | Fig. 2E —38-year-old woman with lung adenocarcinoma and hemorrhagic liver cyst correctly diagnosed with contrast-enhanced ultrasound. Diagnostic confirmation was obtained with MRI. E, Axial T1-weighted (D) and T2-weighted (E) fat-suppressed MR images show fluid content of hepatic cyst (arrows) to be heterogeneously isointense on T1-weighted and hyperintense on T2-weighted images. |
 View larger version (355K) | Fig. 3 —68-year-old man with hemorrhagic liver cyst and history of colon cancer. Correct diagnosis was made at contrast-enhanced ultrasound examination. Diagnostic confirmation was obtained with MRI. Split-screen modality images (left, gray-scale ultrasound; right, contrast-enhanced ultrasound) show liver cyst with dirty internal echoes at gray-scale ultrasound that are constantly avascular during contrast-enhanced ultrasound (88 seconds after contrast injection). Absence of any enhancement within lesion excludes presence of vital tissue in lesion, suggesting complicated cyst rather than solid neoplasm. |
Six hemangiomas (of eight) exhibited peripheral hyperechoic globules during the arterial phase that progressed in a centripetal direction to partial or complete filling in the portal or sinusoidal phases. In total, three of eight hemangiomas (38%) were completely filled-in, although the time required for filling varied widely from 35 to 145 seconds. One of eight hemangiomas smaller than 2 cm in diameter exhibited immediate and homogeneous hyperenhancement in the arterial phase and sustained enhancement in the portal and delayed phases (Fig. 4). Finally, another hemangioma in the series exhibited peripheral rimlike hyperenhancement in the arterial phase that was not followed by centripetal fill-in in the portal and sinusoidal phases. An extensive nonenhancing area was present in the center throughout the three phases (Fig. 5). For this lesion, pathologic examination of the specimens revealed a rare diagnosis of thrombosed hemangioma.
 View larger version (150K) | Fig. 4A —69-year-old woman with high-flow hemangioma and history of colorectal cancer 3 years previously. Example of lesion not visualized with CT or MRI but correctly diagnosed with contrast-enhanced ultrasound. Diagnostic confirmation was obtained by sonographic follow-up. A, Split-screen-mode gray-scale sonograms (left) show small cysticlike lesion (arrowheads, A and B). Serial contrast-enhanced ultrasound images (right) obtained 15 (A), 43 (B), and 184 (C) seconds after contrast injection show bright enhancement of lesion in early contrast-enhanced phase that remains hyperechoic to rest of parenchyma in portal and sinusoidal contrast-enhanced phases. Feeding and draining vessel also are evident. |
 View larger version (119K) | Fig. 4B —69-year-old woman with high-flow hemangioma and history of colorectal cancer 3 years previously. Example of lesion not visualized with CT or MRI but correctly diagnosed with contrast-enhanced ultrasound. Diagnostic confirmation was obtained by sonographic follow-up. B, Split-screen-mode gray-scale sonograms (left) show small cysticlike lesion (arrowheads, A and B). Serial contrast-enhanced ultrasound images (right) obtained 15 (A), 43 (B), and 184 (C) seconds after contrast injection show bright enhancement of lesion in early contrast-enhanced phase that remains hyperechoic to rest of parenchyma in portal and sinusoidal contrast-enhanced phases. Feeding and draining vessel also are evident. |
 View larger version (181K) | Fig. 4C —69-year-old woman with high-flow hemangioma and history of colorectal cancer 3 years previously. Example of lesion not visualized with CT or MRI but correctly diagnosed with contrast-enhanced ultrasound. Diagnostic confirmation was obtained by sonographic follow-up. C, Split-screen-mode gray-scale sonograms (left) show small cysticlike lesion (arrowheads, A and B). Serial contrast-enhanced ultrasound images (right) obtained 15 (A), 43 (B), and 184 (C) seconds after contrast injection show bright enhancement of lesion in early contrast-enhanced phase that remains hyperechoic to rest of parenchyma in portal and sinusoidal contrast-enhanced phases. Feeding and draining vessel also are evident. |
 View larger version (143K) | Fig. 5A —62-year-old man with thrombosed hemangioma incidentally detected during abdominal ultrasound examination. Lesion was inappropriately assessed as malignant by reader 2 at contrast-enhanced ultrasound. Diagnostic confirmation was obtained at pathologic examination of specimens obtained from ultrasound-guided percutaneous biopsy. A, Split-screen-mode gray-scale ultrasound images of right liver lobe (left) show heterogeneous hypoechoic lesion with internal echoes. Contrast-enhanced images (right) show peripheral rim of enhancement in arterial phase (28 seconds after contrast injection) (A) not followed by centripetal fill-in in portal and sinusoidal phases (106 seconds after contrast injection) (B). Extensive nonenhancing area is present in center. |
 View larger version (170K) | Fig. 5B —62-year-old man with thrombosed hemangioma incidentally detected during abdominal ultrasound examination. Lesion was inappropriately assessed as malignant by reader 2 at contrast-enhanced ultrasound. Diagnostic confirmation was obtained at pathologic examination of specimens obtained from ultrasound-guided percutaneous biopsy. B, Split-screen-mode gray-scale ultrasound images of right liver lobe (left) show heterogeneous hypoechoic lesion with internal echoes. Contrast-enhanced images (right) show peripheral rim of enhancement in arterial phase (28 seconds after contrast injection) (A) not followed by centripetal fill-in in portal and sinusoidal phases (106 seconds after contrast injection) (B). Extensive nonenhancing area is present in center. |
Peripheral rimlike hyperenhancement in the arterial phase was seen in 10 of the 18 metastatic lesions. Seven of the 18 metastases had diffuse (homogeneous or heterogeneous) enhancement during the arterial phase. During the portal and sinusoidal phases, all of these lesions washed out and became hypoenhancing (Fig. 6). Finally, 1 of 18 metastases (measuring 3.5 × 3 cm and deeply located in segment VII in the subcapsular region) exhibited apparent lack of vascularization at CEUS, being hypoechoic throughout all the vascular phases, whereas CT correctly depicted a typical contrast enhancement pattern, that is, arterial hyperenhancement and portal washout. The CT diagnosis was confirmed at US-guided biopsy.
 View larger version (138K) | Fig. 6A —54-year-old man with liver metastasis from retroperitoneal sarcoma confirmed by pathologic examination of specimens obtained at surgical resection. A, Split-screen mode gray-scale ultrasound images (left) show cystlike liver lesion in fatty liver background with uncertain diagnosis. Internal echoes and subtle posterior enhancement are evident. Contrast-enhanced ultrasound images (right) show hyperenhancement of solid lesion in arterial phase (20 seconds after contrast injection) (A) and hypoenhancement in sinusoidal phase (145 seconds after contrast injection) (B). |
 View larger version (211K) | Fig. 6B —54-year-old man with liver metastasis from retroperitoneal sarcoma confirmed by pathologic examination of specimens obtained at surgical resection. B, Split-screen mode gray-scale ultrasound images (left) show cystlike liver lesion in fatty liver background with uncertain diagnosis. Internal echoes and subtle posterior enhancement are evident. Contrast-enhanced ultrasound images (right) show hyperenhancement of solid lesion in arterial phase (20 seconds after contrast injection) (A) and hypoenhancement in sinusoidal phase (145 seconds after contrast injection) (B). |
The confidence levels for both radiologists are shown in Table 2. After CEUS, the number of indeterminate lesions (assigned confidence rating score, 3) decreased, whereas the number of definite lesions (assigned confidence rating score, 1 or 5) increased for both radiologists. For both readers, sensitivity, specificity, positive predictive value, and negative predictive value improved after CEUS (all p < 0.05) (Table 3). The Az values were 0.781 before versus 0.972 after CEUS for radiologist 1 and 0.734 versus 0.957 for radiologist 2. Interobserver agreement also increased, although slightly, from a kappa value of 0.894 at conventional US (95% CI, 0.808–0.974) to 0.953 at CEUS (95% CI, 0.909–0.998).
Overall, in terms of differential diagnosis between benignity and malignancy, the total number of correctly characterized lesions increased from 31 of 50 (62%) to 49 of 50 (98%) for reader 1 and from 28 of 50 (56%) to 48 of 50 (96%) for reader 2 in the comparison of CEUS with conventional US. In particular, for malignant lesions, reader 1 correctly characterized 13 of 18 (72.2%) lesions at conventional US and 17 of 18 (94.4%) lesions at CEUS. Reader 2 also correctly characterized 13 of 18 (72.2%) and 17 of 18 (94.4%) lesions. Specifically, both readers failed to characterized one metastasis using CEUS. For benign lesions, reader 1 correctly characterized 18 of 32 (56.2%) lesions at conventional US and 32 of 32 (100%) lesions at CEUS. Reader 2 correctly characterized 15 of 32 (46.8%) and 31 of 32 (96.8%) lesions. Specifically, reader 2 diagnosed as malignant one thrombosed hemangioma.
| Discussion | |
|---|
In the US diagnosis of an FLL, the initial evaluation begins with determining whether the lesion is solid or cystic, which narrows the differential diagnosis [4]. A simple cyst is defined as a well-circumscribed lesion with imperceptible walls, homogeneous internal anechogenicity, and posterior acoustic enhancement. In clinical practice, however, cystic lesions fulfilling most but not all of these requirements and solid tumors exhibiting some of these US features are often encountered [17, 18].
Infected or hemorrhagic cysts can be partly or completely filled with dirty internal echoes that make it difficult to differentiate these cysts from solid lesions. Similarly, solid FLLs, including both benign and malignant processes, can be extremely hypoechoic and have poor low-amplitude internal echoes, causing confusion differentiating them from hepatic cysts [3, 4].
Some true cysts do not exhibit posterior enhancement because of scattering and attenuation from tissues lying between them and the transducer. Conversely, some solid FLLs can exhibit posterior acoustic enhancement, and this has been reported as correlating with their histologic structure. Thus, any solid tumor with a high cell-to-matrix ratio and homogeneous cellular composition can appear to be homogeneous and hypoechoic with posterior acoustic enhancement, simulating a cystic lesion [19, 20].
In clinical practice, a frequent further cause of US misdiagnosis is fatty liver. Hepatic steatosis may largely alter the conventional US appearance of liver tumors, which more frequently appear as hypoechoic lesions regardless of their nature. Thus, an FLL may simulate a cyst and make correct US diagnosis challenging, particularly if the lesion is deep or very small in the parenchyma. This issue is extremely important in the care of cancer patients with hepatic steatosis who undergo chemotherapy and may have liver metastases [20, 21].
Doppler techniques are complementary to conventional US because they yield information about the presence of internal blood flow to confirm the solid nature of a lesion with a cystic appearance [22] regardless of the pattern and degree of vascularity [19, 20]. According to Bartolotta et al. [17], Doppler signal is not easily identified. In our series, 6 of the 50 FLLs had intralesional signal at color and power Doppler ultrasound, whereas 46 had no vascularity.
In the current study, satisfactory diagnostic performance in characterization of FLLs (Az = 0.781 vs 0.972 for conventional US and 0.734 vs 0.957 for CEUS) and better interreader agreement in a comparison of CEUS and conventional US (weighted κ = 0.894 at baseline US vs κ = 0.953 at CEUS) have been reported. Nevertheless, in our experience, a small number of lesions have not been correctly characterized because of a nonspecific or atypical contrast enhancement pattern.
In our series, CEUS correctly depicted all true cysts by showing their internal avascularity and, thus, the absence of proliferating intralesional vital tissue [23]. This finding confirmed that CEUS may be helpful in differentiating cysts from solid FLLs, even those that are minimally enhancing.
CEUS also proved to play an important role in the characterization of the other benign lesions, such as hemangiomas. Specifically, except for the thrombosed hemangioma in our series, which was inappropriately assessed as malignant by reader 2, both reviewers correctly diagnosed all hemangiomas using CEUS. In this regard, making the correct differential diagnosis can be challenging because thrombosed hemangiomas can be confused with malignancies, as in our case (Fig. 5). Peripheral enhancement in the arterial phase and lack of enhancement in the thrombosed portions can be misinterpreted as washout [19, 24].
In our series, all but one metastasis (94%) had arterial hypervascularity associated with various contrast enhancement patterns (rimlike or diffuse), but they invariably presented as hypoechoic lesions in the portal and sinusoidal phases. Other authors [25–28] have reported similar results in the CEUS evaluation of liver metastases. Murphy-Lavallee et al. [28], for example, found arterial hyperenhancement in 86% of metastases, and rim enhancement was the most common arterial pattern. As do other reports in the literature [27, 28], our data confirm that the distinctive feature in the appropriate differential diagnosis is hypoenhancement in the portal and sinusoidal phases, with or without arterial enhancement.
Although all CEUS diagnoses of metastatic lesions with a cysticlike US appearance were correct, an unexpected finding was that both readers made incorrect diagnoses of a deeply located metastatic lesion on a background of fatty changes, misdiagnosing it as a cystic lesion. Our results confirm in part data from other authors reporting that CEUS may not show contrast enhancement in deeply located lesions, especially in patients with liver steatosis [29]. This finding may be correlated, in fatty liver, with the decrease of US beam energy delivered to tissues in the deepest planes leading to a less effective signal-to-noise ratio, even when a microbubble contrast agent was administered [29].
Even though the finding of atypical cystic and cysticlike lesions at US is not common and the differential diagnosis is not easy, the solid nature should always be always suspected, especially in oncology patients. Therefore, we have developed an evidence-based diagnostic algorithm to guide clinicians and radiologists in characterizing these lesions (Fig. 7).
 View larger version (72K) | Fig. 7 —Flowchart shows algorithm for evaluation of cystic and cysticlike focal liver lesions detected at conventional ultrasound. CEUS = contrast-enhanced ultrasound. |
Our study had limitations. The first was selection bias, because the patients enrolled were selected from a population of patients entering a cancer institute. It was therefore obvious that these patients had a high prevalence of malignant FLLs. Furthermore, depending on the established inclusion criteria, the overall number of lesions evaluated was low relative to the long observation period (3 years). Another limitation was that in some cases the presence of fatty liver and the final diagnosis were established without pathologic evaluation because of ethical concerns. However, all of the lesions not examined by histologic analysis were well characterized at multiphase contrast-enhanced CT or MRI on the basis of typical contrast enhancement patterns or at follow-up evaluations.
| Conclusion | |
|---|
The possibility of noninvasive differentiation of cystic and cysticlike solid lesions of the liver is extremely important because the clinical implications and therapeutic strategies vary considerably. The development of low-acoustic-power CEUS has made it possible to identify several imaging features of these lesions that, in association with history and clinical findings, may help to correctly characterize them. Our data indicate the usefulness of CEUS in the evaluation of patients with these lesions. It has added value in countries where US is regarded as the first-choice modality for liver surveys.
WEB
This is a web exclusive article.
| Acknowledgment | |
|---|
We thank Andrea Soricelli, Department of Radiology, University of Naples Parthenope, Naples, Italy, for contributing in this article.
| References | |
|---|
