| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Gastrointestinal Imaging |
1 From the Third Department of Internal Medicine (T.I., T. Kiba, K.H., M.M., T.S., H. Sekihara), Clinical Laboratory (K.N.), and Second Department of Surgery (T. Kubota, H. Shimada), Yokohama City University School of Medicine, Japan; Department of Medicine, Kanagawa Dental College, Yokosuka, Japan (T.M.); and Gastroenterological Center, Yokohama City University Medical Center, Japan (K.T.). Received July 11, 2002; revision requested September 5; final revision received May 2, 2003; accepted May 8. Address correspondence to K.N., Gastroenterological Center, Yokohama City University Medical Center, 4–57 Urafune-cho, Minami-ku, Yokohama 232-0024, Japan (e-mail: kz-numa@zero.ad.jp).
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODS RESULTS DISCUSSIONREFERENCES |
|---|
MATERIALS AND METHODS: A total of 183 hepatic lesions in 183 patients were studied; lesions included 116 hepatocellular carcinomas, 42 liver metastases, and 25 liver hemangiomas. After injection of a galactose–palmitic acid contrast agent, lesions were scanned with contrast-enhanced wideband harmonic gray-scale US in the arterial, portal venous, and late venous phases. The enhancement patterns were classified, and multiple logistic regression analysis was used to identify diagnostic patterns that enabled differentiation between hepatic tumors.
RESULTS: Five enhancement patterns were found to be significant in predicting different hepatic tumors. In hepatocellular carcinomas, the presence of intratumoral vessels in the arterial phase and homogeneous or heterogeneous enhancement in the portal phase were the most typical patterns. In metastases, the absence of intratumoral vessels in the arterial phase and ring enhancement or a perfusion defect in the portal phase were the most typical patterns. In hemangiomas, the absence of intratumoral vessels in the arterial phase and peripheral nodular enhancement in the portal phase were the most typical patterns. The sensitivity, specificity, and accuracy of diagnosis based on combinations of enhancement patterns were, respectively, 94.8%, 94.0%, and 94.5% for hepatocellular carcinoma; 90.5%, 94.3%, and 93.4% for metastasis; and 88.0%, 99.4%, and 97.8% for hemangioma.
CONCLUSION: Contrast-enhanced wideband harmonic gray-scale US is a useful tool for differentiating among the hepatic tumors studied.
© RSNA, 2003
Index terms: Liver neoplasms, diagnosis, 761.3194, 761.323, 761.332 • Liver neoplasms, US, 761.12988 • Microbubbles • Ultrasound (US), harmonic study
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTS DISCUSSIONREFERENCES |
|---|
Harmonic US imaging is a new technique that uses microbubble contrast agents and produces fewer artifacts (7,8). Two methods of harmonic US imaging are available: The first method is based on filtering the fundamental component of the received signal (9–12), while the second method is based on a phase cancellation of the fundamental frequency with phase-inverted wideband pulses (13–17). Contrast-enhanced phase-inverted wideband harmonic gray-scale US is a promising new technique that can enable detection of microbubbles with a high level of sensitivity. The advantages of using this wideband harmonic US imaging technique are the elimination of the fundamental component signal from normal tissue without the need for filtering and the high signal-to-noise ratio of the second harmonic signal obtained after the summation of both received signals, resulting in a high spatial and contrast resolution (18). The excellent sensitivity of this contrast-enhanced wideband harmonic US imaging technique can be used to visualize tumor blood perfusion (13–17), and authors of several earlier studies have assessed the diagnostic performance of contrast-enhanced wideband harmonic US imaging for differentiating among hepatic tumors (13,14).
Numata et al (16) used contrast-enhanced wideband harmonic gray-scale US with SH U 508A to examine hepatocellular carcinoma (HCC) for the presence of intratumoral vessels in the arterial phase and for enhancement patterns in the portal venous phase. They concluded that the presence of intratumoral vessels in the arterial phase and homogeneous or heterogeneous enhancement in the portal phase are helpful in differentiating HCC from other hepatic tumors.
The purpose of our study was to assess the accuracy of using a classification based on patterns of enhancement on wideband harmonic gray-scale US images in the differential diagnosis of hepatic tumors.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTS DISCUSSIONREFERENCES |
|---|
The final diagnosis of the lesions showed that there were 116 patients with HCC, 42 patients with liver metastasis (10 colon carcinomas, nine pancreatic carcinomas, five gastric carcinomas, four ovarian carcinomas, two malignant melanomas, two nasopharyngeal carcinomas, two esophageal carcinomas, three gallbladder carcinomas, two lung carcinomas, one rectal carcinoma, one laryngeal carcinoma, and one testicular carcinoma), and 25 patients with hemangioma. All HCCs and liver metastases were diagnosed by using histologic examination after surgical resection (four lesions), US-guided biopsy (151 lesions), or autopsy (three lesions). The diagnosis of hemangioma was confirmed by using contrast-enhanced CT and magnetic resonance imaging and by the absence of any changes on follow-up images more than 1 year later. Among those with HCC lesions, 107 of the 116 patients had cirrhosis; this diagnosis was made at histologic and/or clinical examination.
With regard to the age distribution by sex for each tumor type, there were no significant differences in the mean age between men and women among patients with HCC (men, 66.4 years ± 8.3; women, 68.4 years ± 5.4), patients with metastasis (men, 64.5 years ± 11.9; women, 59.3 years ± 8.0), or patients with hemangioma (men, 57.3 years ± 13.1; women, 52.6 years ± 10.6).
The size of each of the hepatic tumors was measured during conventional US by either one of the two operators (T.I., K.N.) who performed the contrast-enhanced harmonic US imaging. The mean maximum diameter was 28 mm ± 18 among HCCs, 34 mm ± 18 among metastases, and 41 mm ± 30 among hemangiomas. All patients gave full informed consent to our study, and approval from our institutional review board was obtained.
We defined early HCC lesions as well-differentiated lesions with no substantial destruction of the preexisting hepatic framework (19).
Procedures
To minimize variations between operators, the contrast-enhanced harmonic US studies were performed by either one of two operators (T.I., K.N.) using the same examination protocol. Neither operator was aware of the results obtained at helical CT and angiographic examinations or of the diagnosis made at histologic examination. Contrast-enhanced wideband harmonic gray-scale US was performed in all patients. The US examinations were performed with a Sonoline Elegra system (Siemens Medical Systems, Issaquah, Wash) with a 3.5-MHz convex probe. Each liver was imaged with fundamental gray-scale US (transmit, 3.4 MHz; receive, 3.4 MHz) before the patient was administered an intravenous bolus injection of the 300 mg/mL concentration of the galactose–palmitic acid mixture of SH U 508A (Levovist; Schering, Berlin, Germany).
After injection, the liver was scanned by using contrast-enhanced wideband harmonic gray-scale US (transmit, 2.8 or 2.5 MHz; receive, 5.6 or 5.0 MHz) at a frame rate of one to five images per second. The transmission power was 100%, and the mechanical index values were between 1.5 and 1.9. The focus position was set directly below the bottom of the tumor. SH U 508A is a suspension of galactose microparticles (99.9%) stabilized with 0.1% palmic acid. A 7-mL dose of this contrast agent was injected at 0.5 mL/sec via a 20–22-gauge cannula placed in an antecubital vein.
After bolus injection of the contrast agent, 5% glucose was continuously infused at a rate of 5.0 mL/min. The patients gently inspired and then underwent breath-hold imaging for about 30 seconds (20–50 seconds after injection of the contrast agent) while the tumor vessels were examined (observation of the arterial phase). After observing the arterial phase, we froze the recorded images and reviewed them frame by frame from cine loops and stored them on magneto-optical disks. This procedure took approximately 15–35 seconds (mean, 25 seconds), and we used this time to allow pooling of the contrast agent in the hepatic parenchyma. Approximately 80–100 seconds (mean, 90 seconds) after injection of the contrast agent, we observed the tumors for enhancement by using a sweep scan while patients underwent breath-hold imaging for a few seconds (observation of the portal venous phase). After observation of the portal phase, we froze the image again. Finally, 4 minutes after the injection of the contrast agent, we examined the tumor for presence or absence of contrast agent in a sweep scan (observation of the late venous phase). The entire examination was recorded on S-VHS videotape.
Image Evaluation
In the arterial phase, we evaluated the US images for the presence and shape of tumor vessels and classified the patterns of vessels into three categories (Fig 1): intratumoral vessels, peritumoral vessels, or no tumor vessels. In the portal phase, the enhancement patterns of the lesions were classified into four categories (Fig 1): homogeneous or heterogeneous enhancement, perfusion defect, ring enhancement, or peripheral nodular enhancement. In the late phase, the enhancement patterns of the lesion relative to the enhancement pattern in the surrounding liver parenchyma were classified into two categories (Fig 1): perfusion defect or residual enhancement.
|
Statistical Analysis
For nominal variables, such as the presence or absence of a certain imaging feature, the rates were compared among the groups by using the 
2 test. Two-tailed P values of less than .05 were accepted as showing statistical significance.
A multiple logistic regression analysis was performed to select independent variables of imaging features associated with the dependent variable (ie, the tumor type). Since only dichotomous variables can be used as the dependent variable in multiple logistic regression analysis, the dependent variables in our analysis were defined as one of the three tumor types versus the other two (HCC vs metastasis and hemangioma; metastatis vs HCC and hemangioma; hemangioma vs HCC and metastasis). The independent variables were different imaging features observed in the three phases after contrast enhancement: intratumoral vessels, peritumoral vessels, or no tumor vessels in the arterial phase; homogeneous or heterogeneous enhancement, ring enhancement, perfusion defect, or peripheral nodular enhancement in the portal phase; and residual enhancement or perfusion defect in the late phase. Each variable had a dichotomous value (not observed = 0, observed = 1). We selected independent variables with P values of less than .05 in the multiple logistic regression analysis.
Next, we constructed a multivariable model by using these significant independent variables. All tumors exhibited one of the defined combinations of statistically significant independent variables, and these defined combinations were made by using statistically significant independent variables. To evaluate the diagnostic performance of our multivariable model, we calculated the positive predictive values. The positive predictive value of each combined statistically significant independent variables was calculated according to Bayes theorem (20). Where PP is prior probability, SE is sensitivity, and SP is specificity, positive predictive value was defined as PP x SE/[SE x PP + (1 - SP) x (1 - PP)]. Prior probability was calculated by dividing the numbers of HCCs, metastases, and hemangiomas by the total number of tumors (183 tumors). The diagnosis of each hepatic tumor was made on the basis of the largest positive predictive value of each combined enhancement pattern. Next we summed the number of tumors diagnosed according to their largest positive predictive value and calculated the sensitivity, specificity, and accuracy of each tumor diagnosis based on the results of the enhancement pattern–based classification system already described. We also calculated the sensitivities for differentiating each hepatic tumor by using our multivariable model in the 138 patients with solitary focal lesions (100 patients with HCC, 17 with metastasis, and 21 with hemangioma) and in the 45 patients with multiple focal lesions. The data analysis was performed by using SPSS software (version 10.0J; SPSS, Tokyo, Japan).
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTS DISCUSSIONREFERENCES |
|---|
|
|
|
|
|
|
|
|
|
|
|
|
Factors Predicting Diagnosis of Hepatic Tumors
On the basis of the results of contrast-enhanced wideband harmonic gray-scale US, multiple logistic regression analysis was performed for nine parameters to determine predictors of the diagnosis of hepatic tumors. Only five parameters were selected as independent variables associated with a type of hepatic tumor (Table 2). Intratumoral vessels (odds ratio, 8.742; P < .01) and homogeneous or heterogeneous enhancement (odds ratio, 16.54; P < .01) were selected as statistically significant variables to differentiate HCC from the other two types of hepatic lesions combined. Ring enhancement (odds ratio, 103.2; P < .01) and perfusion defect (odds ratio, 17.92; P < .01) were selected as statistically significant variables to differentiate metastasis from the other two types of lesions combined. Peripheral nodular enhancement (odds ratio, 43.77; P < .05) was selected as a statistically significant variable to differentiate hemangioma from the other two types of lesions combined. Two parameters of the late phase were not statistically significant variables.
|
|
The respective sensitivity, specificity, and accuracy of the enhancement pattern–based diagnosis were 94.8%, 94.0%, and 94.5% for HCC; 90.5%, 94.3%, and 93.4% for metastasis; and 88.0%, 99.4%, and 97.8% for hemangioma. When only the solitary focal lesions were analyzed, the sensitivity was 94.0% for HCC, 87.5% for metastasis, and 85.0% for hemangioma. When only the multiple focal lesions were analyzed, the sensitivity was 94.3% for HCC, 94.1% for metastasis, and 100% for hemangioma. No significant differences in sensitivity were observed between solitary lesions and multiple lesions.
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTS DISCUSSIONREFERENCES |
|---|
In our study, the results of multiple logistic regression analysis and the positive predictive values calculated from the results of pattern combinations for each hepatic lesion demonstrated that enhancement pattern–based classification of contrast-enhanced wideband harmonic gray-scale US findings is useful in making the differential diagnosis of hepatic tumors.
In the arterial phase, tumors were classified as having intratumoral, peritumoral, or no tumor vessels (Fig 1), and they were observed mainly in HCCs, metastases, and hemangiomas, respectively. The results in HCCs in which intratumoral vessels were observed in the arterial phase were almost the same as those in our previous study (16). Our results are similar to those of Burns et al (15), who reported marginal or absent tumor vessels but reported no intratumoral vessels in hepatic metastases and hemangiomas with their imaging method.
In our study, in the portal phase more than 90% of HCCs showed homogeneous or heterogeneous enhancement, and more than 90% of metastases showed ring enhancement or perfusion defect. Eighty-five (73%) of the 116 HCCs showed a homogeneous pattern of enhancement in the portal phase; these lesions were visualized as isoechoic and were difficult to identify. Numata et al (25) reported that contrast-enhanced wideband harmonic gray-scale US was useful for diagnosing the viability of HCC after treatment with transcatheter arterial embolization and that it was superior to helical CT for evaluating residual HCC because the perfusion images are not limited by iodized oil deposition. Since viable HCC had the same enhancement as liver parenchyma in the portal phase, the investigators could decide where to perform percutaneous ethanol injection after transcatheter arterial embolization. Observation of the portal phase was important to detect viable portions of HCCs after transcatheter arterial embolization or percutaneous ethanol injection therapy.
Of the 116 HCC lesions in our study, four exhibited arterioportal shunting at angiography, and these lesions showed a homogeneous or heterogeneous pattern of enhancement in the arterial phase but not in the portal phase. During the portal phase, these lesions appeared hypovascular compared with the surrounding liver parenchyma, and we believe that the arterioportal shunting caused the contrast agent to be flushed through the area more quickly than usual.
Twenty (48%) of 42 metastases showed ring enhancement in the portal phase and peripheral vessels in the arterial phase. Peripheral ring enhancement in metastases may reflect viable tumor tissue at the periphery of the lesion and fibrosis or necrosis in the center (26,27). The perfusion defect observed in 18 (43%) of 42 metastases may reflect hypovascularity of the lesions.
Yamashita et al (28) reported that the dynamic CT enhancement patterns of cavernous hemangiomas are related to the collective size of their constituent vascular space. They also reported that hemangiomas smaller than 30 mm in diameter were homogeneously enhanced in the arterial-dominant phase (28). In our study, the only hemangioma that showed a homogeneous pattern of enhancement was 20 mm in maximum diameter. However, flow velocity in the capillary tissue of hemangiomas larger than 20 mm in maximum diameter, on the other hand, was very slow (28,29); therefore, it took a long time for the microbubbles to perfuse into every vascular space of the hemangioma.
A multiple logistic regression analysis was used to assess the enhancement pattern classification of hepatic tumors. The enhancement patterns were defined according to the presence of five parameters from the arterial and portal phases that were shown to be statistically significant predictors of tumor diagnosis: The presence of intratumoral vessels and homogeneous or heterogeneous enhancement suggested HCC, ring enhancement and perfusion defect suggested metastasis, and peripheral nodular enhancement suggested hemangioma. The combinations of these statistically significant predictors from the arterial and portal phases enabled a highly accurate differential diagnosis of hepatic tumors. Our results are similar to those in a study performed by Tanaka et al (23), in which US findings for the vascular phase (arterial and portal phases) and the liver parenchymal phase (late phase) were combined without using multivariate logistic regression analysis. In our study, however, the multivariate analysis showed that the parameters in the late phase were not significant predictors of a differential diagnosis. This finding suggests that observation of the vascular phase (arterial and portal phases) is important for obtaining a differential diagnosis of hepatic tumors.
Several investigators have reported that hepatic metastases showed a perfusion defect in the late phase of contrast-enhanced phase-inversion US and that this finding markedly improved the detection of liver metastases identified as isoechoic or subtle lesions with conventional US (30,31). We think that the late phase may be useful for detection of metastases or relatively small HCC lesions because all small HCCs (
20-mm diameter) exhibited a perfusion defect in the late phase. However, this phase is not of value for differentiating hepatic tumors because the perfusion defect and residual enhancement are not significant predictors for the differential diagnoses of hepatic tumors.
A limitation of this study was that this criterion for diagnosis was not applied prospectively. A prospective study of this enhancement pattern–based classification is needed to confirm its accuracy for the differential diagnosis for hepatic tumors. Furthermore, the proportion of HCC tumors in our study appears to be relatively higher than that in Western countries (32).
In summary, we classified contrast-enhanced wideband harmonic gray-scale US findings in hepatic tumors into combinations of contrast enhancement patterns. The enhancement patterns of hepatic tumors can be accurately identified through the examination of the arterial and portal phase. We conclude that contrast-enhanced wideband harmonic gray-scale US is a useful modality for differentiating among the types of hepatic tumors we studied.
|
FOOTNOTES |
|---|
Author contributions: Guarantor of integrity of entire study, K.T.; study concepts, T.I., K.T.; study design, K.T.; literature research, K.N., T.I.; clinical studies, T.I., T.S., T. Kubota, M.M.; data acquisition, T.I., K.N.; data analysis/interpretation, K.T., K.H.; statistical analysis, T.M.; manuscript preparation, T.I.; manuscript definition of intellectual content, T. Kiba, T.S.; manuscript editing, H. Sekihara, H. Shimada; manuscript revision/review, K.T., K.N.; manuscript final version approval, K.T.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTS DISCUSSIONREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  N.-Y. Cheng, R.-C. Chen, T. Y.-C. Chen, and H.-Y. Tu Contrast-Enhanced Ultrasonography of Hepatic Metastasis of Hemangiopericytoma J. Ultrasound Med., April 1, 2008; 27(4): 667 - 671. [Full Text] [PDF]
|
|
|
|
|
|
K. Numata, M. Morimoto, T. Ogura, K. Sugimori, S. Takebayashi, M. Okada, and K. Tanaka Ablation Therapy Guided by Contrast-Enhanced Sonography with Sonazoid for Hepatocellular Carcinoma Lesions Not Detected by Conventional Sonography J. Ultrasound Med., March 1, 2008; 27(3): 395 - 406. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 H.-J. Jang, T. K. Kim, P. N. Burns, and S. R. Wilson Enhancement Patterns of Hepatocellular Carcinoma at Contrast-enhanced US: Comparison with Histologic Differentiation Radiology, September 1, 2007; 244(3): 898 - 906. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
O. Catalano, V. Migaleddu, E. Quaia, and G. Caruso Terminology for Contrast-Enhanced Sonography: A Practical Glossary J. Ultrasound Med., June 1, 2007; 26(6): 717 - 730. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
K. Numata, H. Oka, M. Morimoto, K. Sugimori, R. Kunisaki, H. Nihonmatsu, K. Matsuo, Y. Nagano, A. Nozawa, and K. Tanaka Differential Diagnosis of Gallbladder Diseases With Contrast-Enhanced Harmonic Gray Scale Ultrasonography J. Ultrasound Med., June 1, 2007; 26(6): 763 - 774. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
H. Yang, G.-J. Liu, M.-D. Lu, H.-X. Xu, and X.-Y. Xie Evaluation of the Vascular Architecture of Hepatocellular Carcinoma by Micro Flow Imaging: Pathologic Correlation J. Ultrasound Med., April 1, 2007; 26(4): 461 - 467. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
P. N. Burns and S. R. Wilson Focal Liver Masses: Enhancement Patterns on Contrast-enhanced Images--Concordance of US Scans with CT Scans and MR Images Radiology, December 1, 2006; 242(1): 162 - 174. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 E. Quaia, A. Palumbo, S. Rossi, F. Degobbis, S. Cernic, G. Tona, and M. Cova Comparison of visual and quantitative analysis for characterization of insonated liver tumors after microbubble contrast injection. Am. J. Roentgenol., June 1, 2006; 186(6): 1560 - 1570. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
C. Nicolau, R. Vilana, V. Catala, L. Bianchi, R. Gilabert, A. Garcia, and C. Bru Importance of Evaluating All Vascular Phases on Contrast-Enhanced Sonography in the Differentiation of Benign from Malignant Focal Liver Lesions Am. J. Roentgenol., January 1, 2006; 186(1): 158 - 167. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 O. Catalano, A. Nunziata, R. Lobianco, and A. Siani Real-Time Harmonic Contrast Material-specific US of Focal Liver Lesions RadioGraphics, March 1, 2005; 25(2): 333 - 349. [Abstract] [Full Text] [PDF]
|
|
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
