Abstract
Purpose
This study aimed to develop a diagnostic algorithm for differentiating hepatic hemangiomas from hepatocellular carcinoma (HCC) using Sonazoid-enhanced ultrasonography (SEUS).
Methods
A retrospective analysis was conducted of SEUS examinations from 159 patients with 110 hemangiomas and 49 HCCs between January 2011 and December 2022. The evaluated imaging features included lesion hyperechogenicity, typical grayscale features of hemangioma, defined as a hyperechoic lesion or echogenic border without a hypoechoic rim or nodule-in-nodule appearance, and typical globular enhancement of hemangioma. In the Kupffer phase (KP), the lesion-to-liver ratio (KP ratio) was calculated, and receiver operating characteristic analysis was performed to determine the optimal KP ratio cutoff for predicting hemangioma. A two-step diagnostic algorithm was developed and compared with typical globular enhancement of hemangioma using the McNemar test.
Results
The optimal KP ratio cutoff for predicting hemangioma was 0.85. The two-step algorithm first identified hemangiomas based on typical globular enhancement and then applied typical grayscale features and a KP ratio ≥0.85 to lesions without typical globular enhancement. The algorithm showed higher sensitivity than typical globular enhancement alone (90.0% vs. 81.8%, P=0.008) while maintaining 100% specificity. In hyperechoic lesions, the algorithm also showed higher sensitivity than typical globular enhancement alone (91.9% vs. 77.4%, P=0.008).
Conclusion
The proposed two-step algorithm may improve sensitivity while maintaining specificity in clinical settings that require differentiation between hemangiomas and HCC.
Keywords: Hemangioma, Hepatocellular carcinoma, Liver, Ultrasonography, Contrast media
Graphical abstract
Introduction
Among benign hepatic tumors, hepatic hemangiomas are the most common and are frequently encountered in routine clinical practice. Because of its high prevalence, hemangioma is the first benign entity that should be reliably excluded when focal liver lesions are evaluated. Typical imaging features allow a confident diagnosis in most cases; however, small lesions and lesions with atypical characteristics may mimic hepatocellular carcinoma (HCC), creating substantial diagnostic challenges.
HCC has traditionally been regarded as a disease affecting high-risk patients, but its epidemiology is rapidly changing. The increasing prevalence of metabolic dysfunction-associated steatotic liver disease, which affects nearly 30% of the global population and accounts for up to 38% of HCC cases worldwide [1–3], means that HCC is now frequently diagnosed even in noncirrhotic patients and nontraditional risk groups. This expansion into the general population underscores the importance of distinguishing HCC from common benign lesions such as hemangiomas, both in high-risk patients and in broader clinical practice.
Although Sonazoid-enhanced ultrasonography (SEUS), which is recognized as a second-line imaging modality in several guidelines [4–6], offers several advantages for diagnosing hemangiomas and distinguishing them from HCC, diagnostic pitfalls remain. First, although SEUS provides higher temporal resolution than computed tomography (CT) or magnetic resonance imaging (MRI), some high-flow hemangiomas may still be difficult to evaluate because their fill-in patterns can be misclassified as HCC [7,8]. Second, in some markedly hyperechoic hemangiomas, pseudoenhancement, characterized by persistent tissue harmonic signals on contrast-mode images, may obscure the fill-in pattern or increase Kupffer-phase (KP) echogenicity [9,10]. Pseudoenhancement can complicate differentiation from well-differentiated HCC because 9.6%–10.8% of HCCs demonstrate isoenhancement during the KP [7,11]. Third, hemangiomas can show washout in the late phase because of reticuloendothelial system activity in the liver [12]. This phenomenon can be further accentuated in severe hepatic steatosis because of pseudoenhancement of the background liver [9].
Despite these issues, few studies have examined the role of KP images or the effect of grayscale hyperechogenicity on SEUS-based differentiation of hemangiomas from HCC. Therefore, the present study investigated the KP characteristics of hemangiomas and HCCs using SEUS and proposed a diagnostic algorithm that integrates ultrasound (US) imaging features with quantitative analysis to improve hemangioma identification. This study also evaluated the effects of hepatic steatosis and lesion hyperechogenicity on SEUS diagnosis to reduce the risk of misdiagnosis.
Materials and Methods
Compliance with Ethical Standards
This retrospective study was approved by the Institutional Review Board of Samsung Medical Center (No. SMC 2024-05-123), and the requirement for informed consent was waived.
Data Collection
At the authors’ affiliated institution, SEUS is indicated for the following purposes: (1) primary or secondary diagnostic testing for hepatic lesions, (2) follow-up examination of probably benign lesions such as hemangioma and focal nodular hyperplasia, and (3) supportive examination for US-guided procedures. Using a database of SEUS procedures performed between January 2011 and December 2022, 234 consecutive patients diagnosed with either hemangioma or HCC were initially identified. Among them, 75 patients were excluded for the following reasons: (a) lesion size too small for characterization, defined as a minimum diameter <10 mm (n=16); (b) insufficient SEUS images, such as absence of a KP image (n=46); and (c) suboptimal image quality that limited lesion assessment (n=13). Consequently, 159 patients were included in this study (Fig. 1). The US examination and scanning parameters are detailed in Supplementary Table 1.
Fig. 1. Flowchart of the study population.
SEUS, Sonazoid-enhanced ultrasonography; HCC, hepatocellular carcinoma.
Clinical Data Analysis
Age, sex, liver disease etiology, and the presence of liver cirrhosis were assessed. Patients with cirrhosis, chronic viral hepatitis B, or prior HCC were considered at high risk for HCC [13].
Imaging Analysis
Two radiologists (J.A.H. and T.H., with 15 and 3 years of experience in abdominal imaging, respectively) independently reviewed the SEUS images. The reviewers were informed that all included lesions were either hemangiomas or HCCs, but they were blinded to the final diagnosis of each individual case. Any disagreements were resolved by consensus after discussion with a third radiologist (W.K.J., with 26 years of experience in abdominal imaging).
The following US imaging features were evaluated: (1) liver parenchyma: presence of hepatic steatosis on grayscale images, defined as liver echogenicity greater than that of the renal cortex [14]; (2) lesion features on grayscale images: (a) maximal size (cm), (b) presence of hyperechogenicity, defined as whole or partial hyperechogenicity within the lesion, (c) hyperechoic lesion or presence of an echogenic border, (d) presence of a hypoechoic rim, and (e) presence of a nodule-in-nodule appearance [13]; (3) lesion features in the arterial phase (AP) of SEUS: (a) globular enhancement, (b) nonrim arterial phase hyperenhancement (APHE), defined as whole or partial lesion hyperenhancement relative to the liver in the AP without a rim-like or peripherally globular pattern, and (c) rim APHE, defined as the most pronounced AP enhancement at the lesion periphery [13]; (4) portal venous and vasculo-Kupffer phase features: (a) progressive fill-in, (b) persistent enhancement without washout, (c) late and mild washout, defined as lesion hypoenhancement relative to the liver at 60 seconds or later after contrast injection without marked hypoenhancement within 2 minutes, and (d) early washout within 60 seconds [13]; and (5) hypoechogenicity on KP images, defined as lesion hypoenhancement relative to the liver in the KP [4].
The following US imaging features for predicting hemangioma were evaluated: (a) typical grayscale features of hemangioma, defined as a hyperechoic lesion or lesion with an echogenic border without a hypoechoic rim or nodule-in-nodule appearance [10,15–17], and (b) typical globular enhancement of hemangioma, defined as discontinuous peripheral globular enhancement in the AP [13]. For the analysis of US imaging features predicting hemangioma, this study selected and integrated features known to characterize hemangioma in routine clinical practice. Because these US imaging features of hemangioma are already well established, this analysis was intended to compare their diagnostic performance and assess their relative contributions.
To identify KP characteristics that differed between hemangiomas and HCCs, the lesion-to-liver ratio in the KP (KP ratio) was evaluated using region-of-interest (ROI)–based analysis. KP ratio measurements were performed on stored Digital Imaging and Communications in Medicine images using a picture archiving and communication system workstation. On KP images, a circular ROI as large as possible was placed within the lesion, and another ROI was placed in the liver parenchyma at the same depth to minimize attenuation-related variability. The KP ratio was calculated as the lesion ROI value divided by the liver ROI value. To reduce measurement variability, each of the two reviewers independently repeated the ROI measurements three times for each lesion.
Reference Standards
Among the target lesions, 13.8% (22/159) were diagnosed pathologically, including three hemangiomas and 19 HCCs. The remaining lesions without pathological results were diagnosed according to CT or MRI criteria for HCC in the Korean Liver Cancer Association–National Cancer Center Korea Practice Guidelines [18]. Hemangiomas were diagnosed based on marked T2 hyperintensity with peripheral globular enhancement during the AP on MRI [19] and stability on follow-up images for at least 1.5 years.
Statistical Analysis
Categorical variables were compared using the chi-square test or Fisher’s exact test, and continuous variables were compared using Student’s t-test. For each reviewer, the mean KP ratio from the three repeated measurements was calculated and used for subsequent analysis. Receiver operating characteristic (ROC) curve analysis was performed separately for each reviewer to estimate the optimal KP ratio for predicting hemangioma using the Youden index. Logistic regression was performed to analyze imaging discriminators for diagnosing hemangioma based on US imaging features. Diagnostic performance, including sensitivity and specificity for US imaging features, was compared using the McNemar test.
Based on the multivariable analysis results and the diagnostic performance of imaging features, diagnostic criteria were proposed to distinguish hemangiomas from HCC. The diagnostic performance of the proposed algorithm and typical globular enhancement of hemangioma was compared in all participants and in subgroups stratified by hepatic steatosis and lesion hyperechogenicity, using the McNemar test for inter-method comparisons and the chi-square test for intra-method comparisons. For patients at high risk for HCC, diagnostic performance for HCC was calculated for lesions with nonrim APHE and late and mild washout or hypoechogenicity in the KP after hemangiomas were excluded according to imaging criteria.
Interobserver agreement for US imaging features between the two reviewers was calculated using κ-values and interpreted as follows: 0.00–0.20, poor agreement; 0.21–0.40, fair agreement; 0.41–0.60, moderate agreement; 0.61–0.80, good agreement; and 0.81–1.00, excellent agreement.
All statistical analyses were performed using SAS version 9.4 (SAS Institute Inc., Cary, NC, USA) or R version 4.0.3 (R Foundation for Statistical Computing, Vienna, Austria; http://www.R-project.org/). A P-value <0.05 was considered statistically significant.
Results
Baseline Characteristics of the Study Patients
The baseline characteristics and US imaging features are summarized in Table 1. The study included 159 patients (mean age±standard deviation, 52.8±13.5 years) with hemangioma (n=110) or HCC (n=49). The most common liver disease etiology was no liver disease (89.1%) among patients with hemangioma and chronic viral hepatitis B (51.0%) among patients with HCC. Overall, 30.2% of patients were at high risk for HCC.
Table 1.
Baseline clinical characteristics and ultrasound imaging features
| Total (n=159) | Hemangioma (n=110) | HCC (n=49) | P-value | |
|---|---|---|---|---|
| Clinical information | ||||
| Age (year), mean±SD (range) | 52.8±13.5 (20‒82) | 49.0±12.2 (20‒74) | 61.5±12.3 (36‒82) | <0.001 |
| Male sex | 94 (59.1) | 61 (55.5) | 33 (67.4) | 0.217 |
| Etiology of liver disease | <0.001 | |||
| None | 103 (64.8) | 98 (89.1) | 5 (10.2) | |
| HBV | 36 (22.6) | 11 (10.0) | 25 (51.0) | |
| HCV | 1 (0.6) | 1 (0.9) | 0 | |
| Alcohol | 9 (5.7) | 0 | 9 (18.4) | |
| Othera) | 10 (6.3) | 0 | 10 (20.4) | |
| Clinically diagnosed liver cirrhosis | 29 (18.2) | 3 (2.7) | 26 (53.1) | <0.001 |
| Patients at high risk for HCCb) | 48 (30.2) | 11 (10.0) | 37 (75.5) | <0.001 |
| Grayscale imaging features | ||||
| Liver parenchyma | ||||
| Presence of hepatic steatosis | 48 (30.2) | 31 (28.2) | 17 (34.7) | 0.523 |
| Lesion | ||||
| Maximal size (cm), median (IQR) | 1.6 (1.2‒2.4) | 1.5 (1.2‒2.1) | 1.8 (1.3‒2.7) | 0.136 |
| Lesion hyperechogenicity | 69 (43.4) | 62 (56.4) | 7 (14.3) | <0.001 |
| Hyperechoic lesion or echogenic border | 83 (52.2) | 76 (69.1) | 7 (14.3) | <0.001 |
| Hypoechoic rim | 17 (10.7) | 5 (4.6) | 12 (24.5) | <0.001 |
| Nodule-in-nodule appearance | 7 (4.4) | 2 (1.8) | 5 (10.2) | 0.029 |
| SEUS imaging features of the lesion | ||||
| Arterial phasec) | ||||
| Globular enhancement | 90 (56.6) | 90 (81.8) | 0 | <0.001 |
| Nonrim APHE | 48 (30.2) | 5 (4.6) | 43 (87.8) | <0.001 |
| Rim APHE | 2 (1.3) | 1 (0.9) | 1 (2.0) | 0.523 |
| Portal venous and vasculo-KPc) | ||||
| Progressive fill-in | 70 (44.0) | 69 (62.7) | 1 (2.0) | <0.001 |
| Persistent enhancement without washout | 33 (20.8) | 27 (24.6) | 6 (12.2) | 0.120 |
| Late and mild washout | 52 (32.7) | 13 (11.8) | 39 (79.6) | <0.001 |
| Early washout within 1 minute | 2 (1.3) | 0 | 2 (4.1) | 0.094 |
| KP | ||||
| Hypoechogenicity in the KP | 117 (73.6) | 69 (62.7) | 48 (98.0) | <0.001 |
| US imaging features predicting hemangioma | ||||
| Typical grayscale features of hemangioma | 79 (49.7) | 74 (67.3) | 5 (10.2) | <0.001 |
| Typical globular enhancement of hemangioma | 90 (56.6) | 90 (81.8) | 0 | <0.001 |
| KP ratio ≥0.85 | 76 (47.8) | 68 (61.8) | 8 (16.3) | <0.001 |
Values are presented as number (%) unless otherwise indicated.
HCC, hepatocellular carcinoma; SD, standard deviation; HBV, hepatitis B virus; HCV, hepatitis C virus; IQR, interquartile range; SEUS, Sonazoid-enhanced ultrasonography; APHE, arterial phase hyperenhancement; KP, Kupffer phase; KP ratio, lesion-to-liver ratio on KP.
a)Steatohepatitis (n=2) and unknown (n=8). b)Cirrhosis, chronic viral hepatitis B, or prior HCC. c)Lesions that did not fall into any of the classification categories were not included.
Hepatic steatosis was present in 30.2% of patients, with no significant difference between groups (P=0.523). Lesion hyperechogenicity was more common in patients with hemangioma than in those with HCC (56.4% vs. 14.3%, P<0.001). Among hemangiomas, 67.3% showed typical grayscale features, and 81.8% showed typical globular enhancement of hemangioma.
Fig. 2 shows the ROC curves for the KP ratio in predicting hemangioma, generated separately for each reviewer. The area under the ROC curve was 0.811 (95% confidence interval [CI], 0.737 to 0.879) for reviewer 1 and 0.693 (95% CI, 0.603 to 0.777) for reviewer 2. The optimal cutoff value was identical for both reviewers (KP ratio, 0.85), yielding sensitivities and specificities of 60.0% and 93.5%, respectively, for reviewer 1 and 59.2% and 71.7%, respectively, for reviewer 2. A KP ratio ≥0.85 was observed more frequently in hemangiomas than in HCCs (61.8% vs. 16.3%, P<0.001).
Fig. 2. Receiver operating characteristic curves for predicting hemangioma using the lesion-to-liver ratio (Kupffer phase ratio [KP ratio]) on KP images.
A. Receiver operating characteristic curve was generated using KP ratio measurements obtained by reviewer 1. B. Receiver operating characteristic curve was generated using KP ratio measurements obtained by reviewer 2. AUC, area under the receiver operating characteristic curve; CI, confidence interval; Sp, specificity; Se, sensitivity.
Imaging Discriminators for Differentiating Hemangioma from HCC
Table 2 summarizes the logistic regression analysis and diagnostic performance of imaging features for differentiating hemangioma from HCC. In the multivariable analysis, typical grayscale features of hemangioma (odds ratio [OR], 17.4; 95% CI, 4.2 to 72.5), typical globular enhancement of hemangioma (OR, 509.1; 95% CI, 26.5 to 9795.2), and a KP ratio ≥0.85 (OR, 7.1; 95% CI, 1.7 to 29.5) were independently associated with the prediction of hemangioma. Among the evaluated imaging features, typical globular enhancement of hemangioma showed the highest diagnostic performance, with significantly higher sensitivity than typical grayscale features and a KP ratio ≥0.85 and higher specificity than a KP ratio ≥0.85.
Table 2.
Imaging discriminators for differentiating hemangioma from HCC
| Logistic regression analysis for differentiation of hemangioma from HCC | Diagnostic performance of US imaging features for hemangioma | |||||
|---|---|---|---|---|---|---|
| Univariable analysis | Multivariable analysis | Sensitivity (%, 95% CI) | Specificity (%, 95% CI) | |||
| OR (95% CI) | P-value | OR (95% CI) | P-value | |||
| Imaging feature | ||||||
| 1. Typical grayscale features of hemangioma | 18.1 (6.6‒49.5) | <0.001 | 17.4 (4.2‒72.5) | <0.001 | 67.3 (57.7‒75.9) | 89.8 (77.8‒96.6) |
| 2. Typical globular enhancement of hemangioma | 436.7 (25.5‒7,484.7) | <0.001 | 509.1 (26.5‒9,795.2) | <0.001 | 81.8 (73.3‒88.5) | 100.0 (92.8‒100) |
| 3. KP ratio ≥0.85 | 8.3 (3.6‒19.4) | <0.001 | 7.1 (1.7‒29.5) | 0.007 | 61.8 (52.1‒70.9) | 83.7 (70.3‒92.7) |
| P-value (pairwise comparison) | ||||||
| Imaging feature 1 vs. Imaging feature 2 | - | - | - | - | 0.024 | 0.074 |
| Imaging feature 1 vs. Imaging feature 3 | - | - | - | - | 0.429 | 0.579 |
| Imaging feature 2 vs. Imaging feature 3 | - | - | - | - | 0.002 | 0.013 |
HCC, hepatocellular carcinoma; US, ultrasonography; OR, odds ratio; CI, confidence interval; KP ratio, lesion-to-liver ratio on Kupffer phase.
Two-Step Diagnostic Algorithm for Differentiating Hemangioma from HCC Using US Imaging Features
Based on the multivariable analysis and comparison of diagnostic performance among imaging features, a two-step diagnostic algorithm was established (Fig. 3). The first step involved assessing typical globular enhancement of hemangioma. Lesions demonstrating typical globular enhancement were diagnosed as hemangiomas. In cases without typical globular enhancement, the second step consisted of evaluating typical grayscale features of hemangioma and a KP ratio ≥0.85.
Fig. 3. Proposed two-step diagnostic algorithm for differentiating hemangioma from hepatocellular carcinoma (HCC) using ultrasonography imaging features.
Step 1 assesses the presence of typical globular enhancement and confirms hemangioma if this feature is present. Step 2 evaluates grayscale features, defined as a hyperechoic lesion or echogenic border without a hypoechoic rim or nodule-in-nodule appearance, and a lesion-to-liver ratio on Kupffer phase (KP ratio) ≥0.85. The KP ratio is calculated as the region-of-interest value of the lesion divided by that of the liver on the Kupffer phase image. SEUS, Sonazoid-enhanced ultrasonography.
In this study, 20 hemangiomas were not identified based on typical globular enhancement, of which 14 (70.0%) were slow-filling hemangiomas. Application of the second step additionally identified nine hemangiomas (Figs. 4, 5, Supplementary Tables 2, 3). No HCC was incorrectly classified as hemangioma by the proposed algorithm.
Fig. 4. Sonazoid-enhanced ultrasonography in a 52-year-old woman with chronic hepatitis C infection who underwent evaluation of a focal hepatic lesion detected on computed tomography.
A. On the grayscale image, the lesion appears as a 1-cm hyperechoic lesion (arrow) in the right liver, representing a typical grayscale feature of hemangioma. B. The lesion (arrow) remains hyperechoic on the contrast-mode image. C. After intravenous Sonazoid injection, the lesion (arrow) shows suspected weak rim arterial phase hyperenhancement at 17 seconds but is not clearly delineated from the surrounding hepatic parenchyma. D, E. At 1 minute (D), the lesion shows near isoenhancement, and at 2 minutes (E), it exhibits mild and late washout (arrows). F. On the Kupffer-phase (KP) image, the lesion (arrow) appears as a subtle hypoenhancing lesion with a lesion-to-liver ratio on Kupffer phase (KP ratio) of 0.88. Although typical globular enhancement is not observed, the lesion satisfies the typical grayscale-feature and KP ratio ≥0.85 criteria, leading to a diagnosis of hemangioma according to the two-step diagnostic algorithm. The lesion remained stable for 1.5 years.
Fig. 5. Sonazoid-enhanced ultrasonography (US) in a 42-year-old woman undergoing imaging follow-up for hepatic hemangioma.
A. On the grayscale image, the lesion appears as a 1.5-cm hyperechoic lesion (arrow), representing a typical grayscale feature of hemangioma. B. The lesion (arrow) remains hyperechoic on the contrast-mode image. C. After intravenous Sonazoid injection, the lesion (arrow) shows poor arterial phase enhancement at 26 seconds. D. At 1 minute, the lesion shows near isoenhancement (arrow). E. On the Kupffer-phase (KP) image, the lesion appears isoechoic (arrow), with a lesion-to-liver ratio on KP (KP ratio) of 1.4. Although typical globular enhancement is not observed, the lesion satisfies the typical grayscale-feature and KP ratio ≥0.85 criteria, leading to a diagnosis of hemangioma according to the two-step diagnostic algorithm. The lesion remained stable for 6 years.
Diagnostic Performance of US Imaging Criteria for Hemangioma with Subgroup Analysis
Table 3 shows the inter-method comparison of diagnostic performance between typical globular enhancement of hemangioma and the two-step diagnostic algorithm. The two-step diagnostic algorithm showed significantly higher sensitivity than typical globular enhancement of hemangioma (90.0% vs. 81.8%, P=0.008) while maintaining the same specificity (100%).
Table 3.
Inter-method comparisons of diagnostic performance of ultrasound imaging criteria for hemangioma
| Typical globular enhancement of hemangioma | Two-step diagnostic algorithm | P-value | |
|---|---|---|---|
| Overall analysis (n=159) | |||
| Sensitivity (%, 95% CI) | 81.8 (73.3–88.5) | 90.0 (82.8–94.9) | 0.008 |
| Specificity (%, 95% CI) | 100 (92.8–100) | 100 (92.8–100) | N/A |
| Subgroup analysis according to hepatic steatosis | |||
| No hepatic steatosis (n=111) | |||
| Sensitivity (%, 95% CI) | 83.5 (73.9–90.1) | 89.9 (81.3–94.8) | 0.074 |
| Specificity (%, 95% CI) | 100 (89.3–100) | 100 (89.3–100) | N/A |
| Hepatic steatosis (n=48) | |||
| Sensitivity (%, 95% CI) | 77.4 (60.2–88.6) | 90.3 (75.1–96.7) | 0.134 |
| Specificity (%, 95% CI) | 100 (81.6–100) | 100 (81.6–100) | N/A |
| Subgroup analysis according to lesion hyperechogenicity | |||
| No lesion hyperechogenicity (n=90) | |||
| Sensitivity (%, 95% CI) | 87.5 (75.3–94.1) | 87.5 (75.3–94.1) | N/A |
| Specificity (%, 95% CI) | 100 (91.6–100) | 100 (91.6–100) | N/A |
| Lesion hyperechogenicity (n=69) | |||
| Sensitivity (%, 95% CI) | 77.4 (65.6–86.0) | 91.9 (82.5–96.5) | 0.008 |
| Specificity (%, 95% CI) | 100 (64.6–100) | 100 (64.6–100) | N/A |
CI, confidence interval.
In the subgroup analysis, hepatic steatosis did not contribute to differences in diagnostic performance between the two criteria. In contrast, lesion hyperechogenicity affected the difference in diagnostic performance: typical globular enhancement showed reduced sensitivity in lesions with hyperechogenicity, whereas the two-step diagnostic algorithm maintained its diagnostic performance (91.9% vs. 77.4%, P=0.008).
Table 4 shows the intra-method comparison of the diagnostic performance of each criterion. Neither hepatic steatosis nor lesion hyperechogenicity significantly affected the diagnostic performance of either criterion, although typical globular enhancement of hemangioma showed a nonsignificant trend toward lower sensitivity in these subgroups.
Table 4.
Intra-method comparison of diagnostic performance of ultrasound imaging criteria according to hepatic steatosis and lesion hyperechogenicity
| Measure | Typical globular enhancement of hemangioma | Two-step diagnostic algorithm |
|---|---|---|
| Subgroup analysis according to hepatic steatosis | ||
| Sensitivity (%, 95% CI) | ||
| No hepatic steatosis (n=111) | 83.5 (73.9–90.1) | 89.9 (81.3–94.8) |
| Hepatic steatosis (n=48) | 77.4 (60.2–88.6) | 90.3 (75.1–96.7) |
| P-value | 0.454 | 0.944 |
| Specificity (%, 95% CI) | ||
| No hepatic steatosis (n=111) | 100 (89.3–100) | 100 (89.3–100) |
| Hepatic steatosis (n=48) | 100 (81.6–100) | 100 (81.6–100) |
| P-value | >0.99 | >0.99 |
| Subgroup analysis according to lesion hyperechogenicity | ||
| Sensitivity (%, 95% CI) | ||
| No lesion hyperechogenicity (n=90) | 87.5 (75.3–94.1) | 87.5 (75.3–94.1) |
| Lesion hyperechogenicity (n=69) | 77.4 (65.6–86.0) | 91.9 (82.5–96.5) |
| P-value | 0.174 | 0.442 |
| Specificity (%, 95% CI) | ||
| No lesion hyperechogenicity (n=90) | 100 (91.6–100) | 100 (91.6–100) |
| Lesion hyperechogenicity (n=69) | 100 (64.6–100) | 100 (64.6–100) |
| P-value | >0.99 | >0.99 |
CI, confidence interval.
Diagnostic Performance of SEUS for HCC in Patients at High Risk
After hemangiomas were excluded using each criterion, the diagnostic performance of SEUS for HCC based on nonrim APHE and late and mild washout or hypoechogenicity in the KP was evaluated in 48 patients at high risk for HCC (Table 5). The diagnostic performance for HCC was identical after hemangioma exclusion using either typical globular enhancement or the two-step diagnostic algorithm, with a sensitivity of 83.8% and specificity of 100.0%.
Table 5.
Diagnostic performance for HCC in high-risk patients after exclusion of hemangiomas using different ultrasound imaging criteria
| Measure | Typical globular enhancement | Two-step diagnostic algorithm |
|---|---|---|
| Patients at high risk for HCC (n=48) | ||
| Sensitivity (%, 95% CI) | 83.8 (68.0–93.8) | 83.8 (68.0–93.8) |
| Specificity (%, 95% CI) | 100.0 (71.5–100.0) | 100.0 (71.5–100.0) |
| Subgroup analysis according to hepatic steatosis | ||
| Sensitivity (%, 95% CI) | ||
| No hepatic steatosis (n=35) | 88.5 (71.0–96.0) | 88.5 (71.0–96.0) |
| Hepatic steatosis (n=13) | 63.6 (35.4–84.8) | 63.6 (35.4–84.8) |
| P-value | 0.078 | 0.078 |
| Specificity (%, 95% CI) | ||
| No hepatic steatosis (n=35) | 100.0 (70.1–100.0) | 100.0 (70.1–100.0) |
| Hepatic steatosis (n=13) | 100.0 (34.2–100.0) | 100.0 (34.2–100.0) |
| P-value | >0.99 | >0.99 |
| Subgroup analysis according to lesion hyperechogenicity | ||
| Sensitivity (%, 95% CI) | ||
| No lesion hyperechogenicity (n=36) | 81.3 (64.7–91.1) | 81.3 (64.7–91.1) |
| Lesion hyperechogenicity (n=12) | 80.0 (37.6–96.4) | 80.0 (37.6–96.4) |
| P-value | 0.947 | 0.947 |
| Specificity (%, 95% CI) | ||
| No lesion hyperechogenicity (n=36) | 100.0 (51.0–100.0) | 100.0 (51.0–100.0) |
| Lesion hyperechogenicity (n=12) | 100.0 (64.6–100.0) | 100.0 (64.6–100.0) |
| P-value | >0.99 | >0.99 |
HCC, hepatocellular carcinoma; CI, confidence interval.
In the subgroup analysis according to hepatic steatosis and lesion hyperechogenicity, no significant difference in diagnostic performance was observed.
Interobserver Agreement
Interobserver agreement for grayscale imaging features and SEUS imaging features ranged from moderate to excellent (κ=0.43–0.91), except for persistent enhancement without washout, which showed fair agreement (κ=0.36). The three US imaging features predicting hemangioma, namely typical grayscale features, typical globular enhancement of hemangioma, and a KP ratio ≥0.85, showed good to excellent agreement (κ=0.71–0.87) (Supplementary Table 4).
Discussion
This study proposed a two-step diagnostic algorithm for differentiating hemangioma from HCC using typical globular enhancement, typical grayscale features of hemangioma, and a KP ratio ≥0.85. The proposed algorithm significantly increased sensitivity compared with typical globular enhancement alone while maintaining 100% specificity. In lesions with hyperechogenicity, the two-step algorithm demonstrated significantly higher sensitivity than typical globular enhancement alone.
The KP ratio was used based on the expectation that hemangiomas would show relatively higher KP echogenicity than HCC because of differences in vascular space size, interstitial characteristics, and Kupffer cell activity and distribution [20–22]. The application of typical grayscale features also used characteristic hemangioma findings while excluding findings suggestive of malignancy or HCC, thereby increasing specificity. By identifying features such as a nodule-in-nodule appearance or hypoechoic rim, the algorithm could prevent HCCs with these features from being misdiagnosed as hemangiomas, thereby reducing false-positive hemangioma diagnoses. In this study, incorporating typical grayscale features of hemangioma and the KP ratio into the two-step diagnostic algorithm additionally diagnosed 45% of the hemangiomas among the 18.2% that did not exhibit globular enhancement. Hemangiomas that are not diagnosed as such on imaging are often classified as indeterminate because they also frequently do not meet the diagnostic criteria for HCC. Indeterminate lesions may require invasive procedures, such as biopsy or surgery. This diagnostic algorithm improved the diagnostic performance of US for hemangiomas, potentially reducing unnecessary diagnostic steps and minimizing the need for invasive procedures and their associated complications.
In the subgroup analysis, hepatic steatosis did not affect the diagnostic performance of either typical globular enhancement or the two-step diagnostic algorithm. Contrast-mode imaging with US equipment was thought to effectively suppress background liver echogenicity. In contrast, lesion hyperechogenicity decreased diagnostic performance when typical globular enhancement alone was used, presumably because of persistent tissue harmonic signals at a low mechanical index during SEUS [11]. Interference with contrast-enhancement assessment in hyperechoic lesions was expected to have a greater effect in slow-filling hemangiomas because their contrast enhancement is slow and weak. Among hemangiomas missed by typical globular enhancement in this study, 70% were slow-filling hemangiomas. Because grayscale features and the KP ratio are less dependent on vascular characteristics, they can be applied to both slow-filling and rapid-filling hemangiomas, thereby improving the overall diagnostic performance for hemangioma.
This study has several limitations. First, its retrospective design may have introduced selection bias and limited causal inference. Second, the study population was limited to hemangiomas and HCCs, creating a simplified diagnostic setting that does not fully reflect real-world clinical practice, in which a broader spectrum of hepatic lesions must be considered. Nevertheless, this study was designed to focus specifically on differentiation between hemangioma and HCC because hemangioma is highly prevalent, HCC is increasingly encountered in patients without traditional risk factors, and differentiation between these entities may occasionally remain challenging on SEUS. In addition, SEUS is generally used as a second-line imaging modality after CT or MRI, when other hepatic lesions have often already been evaluated. Therefore, the proposed algorithm may be most applicable in selected clinical scenarios in which differentiation between hemangioma and HCC remains problematic despite prior imaging assessment. The relatively high proportion of hemangiomas in the study cohort may also have influenced the estimated diagnostic performance. Further studies including a broader spectrum of hepatic lesions are warranted to validate the generalizability and clinical applicability of the algorithm. Third, the reviewers were aware that all lesions were either hemangiomas or HCCs, which may have introduced expectation bias and potentially overestimated diagnostic performance. Fourth, only a minority of lesions were histologically confirmed, and imaging-based diagnoses were used for most cases; this may affect diagnostic certainty despite adherence to established guidelines and follow-up criteria. Finally, US-based assessment is inherently operator-dependent, which may affect reproducibility across institutions. Future prospective studies with larger and more diverse patient populations, including external validation cohorts, are required to confirm the generalizability and clinical utility of the proposed algorithm.
In conclusion, the present two-step diagnostic algorithm using typical globular enhancement, grayscale features, and the KP ratio may improve sensitivity while maintaining specificity in selected clinical scenarios requiring differentiation between hemangioma and HCC.
Key points
Sonazoid-enhanced ultrasonography has limitations differentiating hemangiomas from hepatocellular carcinoma in hyperechoic lesions. The two-step algorithm improves diagnostic performance compared to traditional methods. Enhanced accuracy aids in better clinical differentiation of hepatic hemangioma.
Footnotes
Author Contributions
Conceptualization: Han T, Hwang JA, Jeong WK. Data acquisition: Han T, Hwang JA, Jeong WK, Han S, Choi SY, Gu K. Data analysis or interpretation: Han T, Hwang JA. Drafting of the manuscript: Han T, Hwang JA. Critical revision of the manuscript: Han T, Hwang JA, Jeong WK, Han S, Choi SY, Gu K. Approval of the final version of the manuscript: all authors.
Conflict of Interest
Woo Kyoung Jeong serves as Editor for the Ultrasonography, but has no role in the decision to publish this article. All remaining authors have declared no conflicts of interest.
Acknowledgments
Among the subjects included in this study, six individuals were included in publications in European Radiology (2022;32:8507–8517. "Perfluorobutane-enhanced ultrasonography with a Kupffer phase: improved diagnostic sensitivity for hepatocellular carcinoma"), while five were included in Ultrasonography (2021;40:486-498. "Sonazoid-enhanced ultrasonography: comparison with CT/MRI Liver Imaging Reporting and Data System in patients with suspected hepatocellular carcinoma"), with overlap. However, the overlap constitutes less than 4% of the total subjects. It is important to note that this study and the prior publications were conducted for different research purposes. While this study focuses on distinguishing hepatic hemangiomas from hepatocellular carcinoma (HCC) using Sonazoid CEUS, the previous research aimed to compare Sonazoid CEUS findings with CT/MRI LIRADS findings, or to develop diagnostic criteria for HCC using Sonazoid CEUS in at-risk patients.
Supplementary Material
Contrast-enhanced ultrasonography performed using Sonazoid examination protocol (https://doi.org/10.14366/usg.26140).
Application of the two-step diagnostic algorithm for cases with negative typical globular enhancement of hemangioma (https://doi.org/10.14366/usg.26140).
Characteristics of hemangiomas additionally identified by the two-step diagnostic algorithm (https://doi.org/10.14366/usg.26140).
Interobserver agreement for US imaging features (https://doi.org/10.14366/usg.26140).
References
- 1.Younossi ZM, Golabi P, Paik JM, Henry A, Van Dongen C, Henry L. The global epidemiology of nonalcoholic fatty liver disease (NAFLD) and nonalcoholic steatohepatitis (NASH): a systematic review. Hepatology. 2023;77:1335–1347. doi: 10.1097/hep.0000000000000004. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Le MH, Le DM, Baez TC, Dang H, Nguyen VH, Lee K, et al. Global incidence of adverse clinical events in non-alcoholic fatty liver disease: a systematic review and meta-analysis. Clin Mol Hepatol. 2024;30:235–246. doi: 10.3350/cmh.2023.0485. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Huang DQ, El-Serag HB, Loomba R. Global epidemiology of NAFLD-related HCC: trends, predictions, risk factors and prevention. Nat Rev Gastroenterol Hepatol. 2021;18:223–238. doi: 10.1038/s41575-020-00381-6. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Jeong WK, Kang HJ, Choi SH, Park MS, Yu MH, Kim B, et al. Diagnosing hepatocellular carcinoma using Sonazoid contrast-enhanced ultrasonography: 2023 Guidelines from the Korean Society of Radiology and the Korean Society of Abdominal Radiology. Korean J Radiol. 2023;24:482–497. doi: 10.3348/kjr.2023.0324. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Kudo M, Kawamura Y, Hasegawa K, Tateishi R, Kariyama K, Shiina S, et al. Management of hepatocellular carcinoma in Japan: JSH consensus statements and recommendations 2021 update. Liver Cancer. 2021;10:181–223. doi: 10.1159/000514174. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Omata M, Cheng AL, Kokudo N, Kudo M, Lee JM, Jia J, et al. Asia-Pacific clinical practice guidelines on the management of hepatocellular carcinoma: a 2017 update. Hepatol Int. 2017;11:317–370. doi: 10.1007/s12072-017-9799-9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Sugimoto K, Kamiyama N, Kakegawa T, Takahashi H, Wada T, Abe M, et al. Modified CEUS LI-RADS using Sonazoid for the diagnosis of hepatocellular carcinoma. Ultrasonography. 2023;42:388–399. doi: 10.14366/usg.23065. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Sugimoto K, Kakegawa T, Takahashi H, Tomita Y, Abe M, Yoshimasu Y, et al. Usefulness of modified CEUS LI-RADS for the diagnosis of hepatocellular carcinoma using Sonazoid. Diagnostics (Basel) 2020;10:828. doi: 10.3390/diagnostics10100828. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Jo PC, Jang HJ, Burns PN, Burak KW, Kim TK, Wilson SR. Integration of contrast-enhanced US into a multimodality approach to imaging of nodules in a cirrhotic liver: how I do it. Radiology. 2017;282:317–331. doi: 10.1148/radiol.2016151732. [DOI] [PubMed] [Google Scholar]
- 10.Lee JY, Minami Y, Choi BI, Lee WJ, Chou YH, Jeong WK, et al. The AFSUMB consensus statements and recommendations for the clinical practice of contrast-enhanced ultrasound using Sonazoid. J Med Ultrasound. 2020;28:59–82. doi: 10.4103/jmu.jmu_124_19. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Takahashi H, Sugimoto K, Kamiyama N, Sakamaki K, Kakegawa T, Wada T, et al. Noninvasive diagnosis of hepatocellular carcinoma on Sonazoid-enhanced US: value of the Kupffer phase. Diagnostics (Basel) 2022;12:141. doi: 10.3390/diagnostics12010141. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Khalili K, Atri M, Kim TK, Jang HJ. Recognizing the role of the reticuloendothelial system in the late phase of US contrast agents. Radiology. 2021;298:287–291. doi: 10.1148/radiol.2020203245. [DOI] [PubMed] [Google Scholar]
- 13.American College of Radiology CEUS LI-RADS® v2017 [Internet] Reston, VA: American College of Radiology, 2017 [cited 2024 Dec 17]. Available from: https://www.acr.org/-/media/ACR/Files/RADS/LI-RADS/CEUS-LI-RADS-2017-Core.pdf.
- 14.Hamer OW, Aguirre DA, Casola G, Lavine JE, Woenckhaus M, Sirlin CB. Fatty liver: imaging patterns and pitfalls. Radiographics. 2006;26:1637–1653. doi: 10.1148/rg.266065004. [DOI] [PubMed] [Google Scholar]
- 15.Wilson SR, Feinstein SB. Introduction: 4th guidelines and good clinical practice recommendations for contrast enhanced ultrasound (CEUS) in the liver: update 2020 WFUMB in cooperation with EFSUMB, AFSUMB, AIUM and FLAUS. Ultrasound Med Biol. 2020;46:3483–3484. doi: 10.1016/j.ultrasmedbio.2020.08.015. [DOI] [PubMed] [Google Scholar]
- 16.Kim KW, Kim TK, Han JK, Kim AY, Lee HJ, Park SH, et al. Hepatic hemangiomas: spectrum of US appearances on gray-scale, power Doppler, and contrast-enhanced US. Korean J Radiol. 2000;1:191–197. doi: 10.3348/kjr.2000.1.4.191. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.European Association for the Study of the Liver (EASL) EASL Clinical Practice Guidelines on the management of benign liver tumours. J Hepatol. 2016;65:386–398. doi: 10.1016/j.jhep.2016.04.001. [DOI] [PubMed] [Google Scholar]
- 18.Korean Liver Cancer Association (KLCA) National Cancer Center (NCC) 2022 KLCA-NCC Korea practice guidelines for the management of hepatocellular carcinoma. Clin Mol Hepatol. 2022;28:583–705. doi: 10.3350/cmh.2022.0294. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Korean Liver Cancer Association (KLCA) National Cancer Center (NCC) 2018 Korean Liver Cancer Association-National Cancer Center Korea practice guidelines for the management of hepatocellular carcinoma. Korean J Radiol. 2019;20:1042–1113. doi: 10.3348/kjr.2019.0140. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Giannetti A, Franci L, Grechi C, Giangregorio F. Contrast-enhanced sonography in the diagnosis of hepatic hemangiomas: atypical appearance due to the washout of microbubbles. J Clin Ultrasound. 2013;41:361–365. doi: 10.1002/jcu.21939. [DOI] [PubMed] [Google Scholar]
- 21.Hatanaka K, Kudo M, Minami Y, Ueda T, Tatsumi C, Kitai S, et al. Differential diagnosis of hepatic tumors: value of contrast-enhanced harmonic sonography using the newly developed contrast agent, Sonazoid. Intervirology. 2008;51 Suppl 1:61–69. doi: 10.1159/000122600. [DOI] [PubMed] [Google Scholar]
- 22.Sugimoto K, Moriyasu F, Saito K, Yoshiara H, Imai Y. Kupffer-phase findings of hepatic hemangiomas in contrast-enhanced ultrasound with Sonazoid. Ultrasound Med Biol. 2014;40:1089–1095. doi: 10.1016/j.ultrasmedbio.2013.12.019. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Contrast-enhanced ultrasonography performed using Sonazoid examination protocol (https://doi.org/10.14366/usg.26140).
Application of the two-step diagnostic algorithm for cases with negative typical globular enhancement of hemangioma (https://doi.org/10.14366/usg.26140).
Characteristics of hemangiomas additionally identified by the two-step diagnostic algorithm (https://doi.org/10.14366/usg.26140).
Interobserver agreement for US imaging features (https://doi.org/10.14366/usg.26140).