- Email: [email protected]
Acoustic Radiation Force Impulse Elastography for the Evaluation of Focal Solid Hepatic Lesions: Preliminary Findings
Ultrasound in Med. & Biol., Vol. 36, No. 2, pp. 202–208, 2010 Copyright Ó 2010 World Federation for Ultrasound in Medicine & Biology Printed in the USA. All rights reserved 0301-5629/09/$–see front matter
doi:10.1016/j.ultrasmedbio.2009.10.009
d
Original Contribution ACOUSTIC RADIATION FORCE IMPULSE ELASTOGRAPHY FOR THE EVALUATION OF FOCAL SOLID HEPATIC LESIONS: PRELIMINARY FINDINGS SEUNG HYUN CHO,*y JAE YOUNG LEE,* JOON KOO HAN,* and BYUNG IHN CHOI* * Department of Radiology and the Institute of Radiation Medicine, Seoul National University Hospital, Seoul, Korea; and y Department of Radiology, Catholic University of Daegu, School of Medicine, Daegu, Korea (Received 30 May 2009; revised 10 October 2009; in final form 15 October 2009)
Abstract—This study was designed to investigate the potential usefulness of acoustic radiation force impulse (ARFI) elastography to evaluate focal solid hepatic lesions. In total, 51 patients with 60 focal hepatic lesions, which included 17 hemangiomas, 25 hepatocellular carcinomas (HCCs), 15 metastases and three cholangiocarcinomas, underwent ARFI elastography. The lesions were classified into three groups: Group I consisted of metastatic liver tumors and cholangiocarcinomas, group II consisted of HCCs and group III consisted of hemangiomas. The stiffness and conspicuity of the tumors as depicted on ARFI elastography and the echogenicity and conspicuity of the tumors on corresponding B-mode images were analyzed. Shear wave velocity was obtained to quantify stiffness for 36 focal hepatic lesions: 11 hemangiomas, 17 HCCs and eight other malignant lesions. On ARFI elastography images, group I tumors (n 5 18) appeared stiffer than the background liver for 13 lesions (72%), softer for two lesions and had identical stiffness in three lesions compared with the background liver. For group II tumors (n 5 25), 13 lesions (52%) appeared stiffer than the liver, six lesions appeared softer than the liver and the remaining six lesions showed the same stiffness as the liver. For group III tumors (n 5 17), six lesions (35%) appeared stiffer than the liver, seven lesions appeared softer and the remaining four lesions showed the same stiffness as the liver. There were no statistical differences among the three groups in terms of tumor stiffness as seen on ARFI elastography images (p . 0.05). Of the 60 lesions, 41 (68%) displayed a clearer or equivalent margin on the ARFI elastography compared with that seen on B-mode images. The shear wave velocities were: Group I, 2.18 ± 0.96m/s (mean value ± SD); group II, 2.45 ± 0.81m/s; group III, 1.51 ± 0.71m/s (p 5 0.012). With a cut-off value of 2m/s for the shear wave velocity, the positive predictive value and specificity for malignancy were 89% and 81%, respectively. Images obtained with ARFI elastography provided additional qualitative information regarding the stiffness and tumor margin of liver tumors. By measuring shear wave velocity, quantification of stiffness was made possible and showed the potential to differentiate malignant hepatic tumors from hepatic hemangiomas. (E-mail: leejy@radiol. snu.ac.kr) Ó 2010 World Federation for Ultrasound in Medicine & Biology. Key Words: Acoustic radiation force impulse imaging, ARFI Imaging, Elastography, Ultrasonography, Liver neoplasm.
target tissue displacement, i.e., acoustic radiation force impulse (ARFI) elastography (Fahey et al. 2005, 2006; Nightingale et al. 2002; Walker et al. 2000). This technique has enabled in vivo liver elastography by transmission of the acoustic wave between the ribs or from the subcostal area (Fahey et al. 2008a, 2008b). Until now, only one in vivo study has characterized liver tumors in humans using ARFI elastography (Fahey et al. 2008a) and that study only included seven liver tumors. Additionally, there are no studies that have evaluated ARFI elastography’s usefulness in differentiating liver tumors and no studies have been reported on the quantification of tumor stiffness using shear wave velocity, which may help in the diagnosis of liver tumors.
INTRODUCTION Most elastographic imaging of ultrasonography to date has used manual or motorized compression with moving probes. Compression of the liver for elastographic imaging has always been difficult because it is surrounded by the rib cage, which limits using this procedure clinically. Recently, ultrasound elastography with a short, highintensity focused ultrasound beam has been introduced for
Address correspondence to: Jae Young Lee, M.D., Department of Radiology, Seoul National University Hospital, 28 Yeongon-dong, Jongno-gu, Seoul 110-744, Republic of Korea. E-mail: leejy@radiol. snu.ac.kr 202
Acoustic radiation force impulse elastography d S. H. CHO et al.
The present study was performed to investigate the potential usefulness of ARFI elastography for evaluating focal solid hepatic lesions, assuming that different features can be shown on ARFI elastography images according to the type of liver tumor and that quantification of tumor stiffness may be helpful in the diagnosis. MATERIALS AND METHODS Our hospital Institutional Review Board approved this study (IRB number: 0902-039-272) and informed consent was obtained from all patients prior to the ARFI elastography examinations. Patients Over two periods of time, from May to July 2008 (55 patients) and May 2009 (14 patients), a total of 69 patients with focal solid hepatic lesions underwent ARFI elastography. Patient enrollment for these separate periods was dependent on the availability of the ARFI elastography machine in our hospital. Fifteen patients were excluded from the study due to either a technical failure such as patient motion, the presence of a deep-seated lesion or the patient’s inability to hold their breath properly. Another three patients were excluded from the study as they were pathologically proven, by follow-up biopsies, to have focal fatty deposition, liver abscess and fasciola hepatica, respectively. Thus, our study population consisted of 51 patients. There were 60 focal hepatic lesions included in this study because nine patients each had two focal hepatic lesions. The hepatic lesions included 17 hemangiomas in 14 patients (mean tumor diameter, 1.6 cm; range, 0.8–3.0 cm), 25 hepatocellular carcinomas (HCCs) in 20 patients (mean tumor diameter, 3.6 cm; range, 1.6–7.9 cm), 15 metastatic liver tumors in 14 patients and three cholangiocarcinomas in three patients (mean tumor diameter, 3.5 cm; range, 0.7–8.4 cm). The presence of a hemangioma was diagnosed in 13 of 14 patients based on a combination of the typical findings determined using either computed tomography (CT) or magnetic resonance (MR) imaging and the absence of an increase in tumor size for at least 12 months. The diagnosis was confirmed in the remaining patient by percutaneous biopsy performed immediately after ARFI elastography. The following imaging findings were considered for the diagnosis of hemangioma: (a) lesions showing early, peripheral nodular contrast enhancement during the hepatic arterial phase (HAP) and centripetal fill-in enhancement during the portal venous phase (PVP); or (b) lesions showing early strong homogeneous enhancement during the HAP and the same persistent, strong enhancement as that for enhanced intrahepatic vessels during the PVP (Jeong et al. 2000). There was
203
clinical and biochemical evidence of chronic liver disease in two patients with hemangiomas. In this study, the diagnosis of HCC lesions was confirmed by surgery in seven patients, by a biopsy following elastography in one patient and by clinical diagnosis in 12 patients. The clinical diagnosis for HCCs was made according to the American Association for the Study of Liver Disease (AASLD) 2005 recommendations (Bruix et al. 2005). According to the AASLD 2005 guidelines, a diagnosis of HCC can be made if a mass larger than 2 cm in diameter shows typical features of an HCC (i.e., hypervascularity in the arterial phase and washout in the venous phase as seen on contrast-enhanced CT or on MR imaging) or if a mass measuring 1–2 cm shows these features with the use of both modalities. All 20 patients with HCCs in our study had chronic liver disease: Child-Pugh classification A (15 patients), B (four patients) and C (one patient). In this study, the diagnosis of liver metastasis and cholangiocarcinoma was based on histologic confirmation following surgery (five patients) or following biopsy (12 patients). The origins of the 15 metastatic tumors in 14 patients were determined: five from colon cancers, three from neuroendocrine tumors, two from gallbladder cancers, one from rectal cancer, one from duodenal cancer, one from pancreatic cancer, one from renal cell carcinoma and one from adenocarcinoma of unknown origin. There was no clinical or biochemical evidence of chronic liver disease in patients with liver metastases or cholangiocarcinomas. All lesions were categorized into three groups based on the similarity of their pathology. Group I consisted of metastatic liver tumors and cholangiocarcinomas. Group II consisted of HCCs and group III consisted of hemangiomas. Imaging and analysis One of two examiners (S.H.C., J.Y.L.) performed ARFI elastography with an Acuson S2000 US unit (Siemens, Mountain View, CA, USA) using a 2–4 MHz curved array probe. The examiner scanned the liver to locate the lesion detected on prior imaging (e.g., ultrasonography, CT, MR). After fitting the ARFI image box to fully cover the lesion, an ARFI image was obtained with a corresponding B-mode image. Shear wave velocity (m/s) was measured three times for each tumor and was averaged for each of 36 tumors: eight group I tumors, 17 group II tumors and 11 group III tumors. Shear wave velocity was not obtained for the other 24 tumors due to either the depth limit or motion artifacts from the patient’s poor breath-hold or heartbeat. Two reviewers, who were blinded to the final diagnosis, independently reviewed all of the ARFI and B-mode images on picture archiving and communications system (PACS) workstation monitors (m-view,
204
Ultrasound in Medicine and Biology
Volume 36, Number 2, 2010
Table 1. Stiffness findings on acoustic radiation force impulse imaging Acoustic radiation force impulse imaging Brighter (Softer)
Group I Group II Group III
Same color (Equally stiff)
Consensus
R1
R2
Consensus
2 6 7
2 7 8
1 4 5
3 6 4
Darker (Stiffer)
R1
R2
Consensus
R1
R2
2 7 6
13 13 6
12 13 6
15 14 6
4 5 3 p . 0.05
The values are the number of tumors. Group I 5 metastases and cholangiocarcinomas; Group II 5 hepatocellular carcinomas; Group III 5 hemangiomas. R1: reviewer 1. R2: reviewer 2.
INFINITT; Seoul, Korea). The echogenicity or stiffness, and the conspicuity of the tumors were analyzed on both ARFI and corresponding B-mode images. The tumors were categorized as stiffer (darker), equally stiff (of the same color) or softer (brighter) based on the lesion’s brightness relative to the liver on the ARFI images and the tumors were classified as either hypoechoic, isoechoic or hyperechoic based on the lesion’s echogenicity relative to the liver on B-mode images. The conspicuity of a tumor was classified as having either a clear or an unclear margin. Discrepancies between the two reviewers were resolved through consensus. Statistical analysis The Kruskal-Wallis test was used to calculate the statistical difference in the shear wave velocity values in the three groups. A Tukey post-hoc test was used to resolve any differences found on the Kruskal-Wallis test. Fisher’s exact test was used to calculate the statistical difference of the imaging findings regarding the stiffness in the three groups. A difference was determined to be significant if p , 0.05. RESULTS Tables 1 and 2 summarize the echogenicity, stiffness and conspicuity of the lesions in the three groups. Of the elastographic images of group I (metastatic tumors and cholangiocarcinomas), 13 of 18 lesions (72%) appeared darker, which indicated that they were stiffer than the surrounding liver parenchyma (Figs. 1 and 2). The stiffer tumors included three cholangiocarcinomas, four metastatic tumors from colon cancer, one metastatic tumor from duodenal cancer, three from neuroendocrine tumors and one from pancreatic cancer. The remaining lesion was an adenocarcinoma of an unknown origin. When compared with the liver background, three of the 18 lesions (17%) had the same brightness, which indicated that the lesions were as stiff as the surrounding liver. Gallbladder cancer caused two of the three lesions and the remaining lesion was a metastatic tumor from rectal
cancer. Finally, two of the 18 lesions (11%) appeared brighter than the background liver, which indicated that the lesions were softer than the surrounding liver parenchyma. The origins of the lesions were a colon cancer and a renal cell carcinoma. On the elastographic images of group II (HCCs) and group III (hemangiomas) (Figs. 3 and 4), group III (hemangiomas) had a similar number of stiffer tumors and softer tumors relative to the background liver but group II (HCCs) had a higher number of stiffer tumors, as was also found in group I (Table 1). However, there was no statistical difference in the three groups in terms of tumor stiffness suggested by ARFI elastography images (p . 0.05). Of the lesions in the study, 11 of 60 lesions (18%) displayed a clearer margin with ARFI elastography than with B-mode imaging and 30 lesions (50%) showed an equivalent margin with ARFI elastography to that seen with B-mode imaging (Table 2). For group I (metastases and cholangiocarcinomas), ARFI imaging showed the presence of a clear margin in five of the seven lesions that showed an unclear margin on B-mode images (Fig. 2). For group II (HCCs), ARFI imaging showed the presence of a clear margin in three of five lesions that had an unclear margin with B-mode imaging and for group III (hemangiomas), ARFI imaging showed the Table 2. Conspicuity findings on acoustic radiation force impulse imaging Conventional US/ARFI imaging
Group I Group II Group III Total
Clear/ clear
Clear/ unclear
Unclear/ clear
Unclear/ unclear
7 11 5 23
4 9 6 19
5 3 3 11
2 2 3 7
The values are the number of tumors. US 5 ultrasonography; ARFI 5 acoustic radiation force impulse; Group I 5 metastases and cholangiocarcinomas; Group II 5 hepatocellular carcinomas; Group III 5 hemangiomas.
Acoustic radiation force impulse elastography d S. H. CHO et al.
205
Fig. 1. Findings for a 65-year-old man with duodenal cancer and liver metastasis. (a) Contrast-enhanced computed tomography (CT) scan shows a small, low-attenuating lesion in segment 8 of the liver (arrow). (b) On a conventional B-mode image (left figure), this metastatic lesion is seen as an ovoid, clearly marginated, hyperechoic lesion and the lesion is seen as an ovoid, clearly marginated, homogeneously darker (stiffer) lesion on acoustic radiation force impulse (ARFI) imaging (right figure).
presence of a clear margin in three of six lesions that had an unclear margin with B-mode imaging. The shear wave measured velocities are shown in Table 3 and Figures 3 and 4. There was a statistically significant difference among the three groups (p 5 0.012). Tukey post-hoc tests in the individual groups revealed a significantly lower shear wave velocity in group III (hemangiomas) compared with that in groups I (metastases and cholangiocarcinomas) and II (HCCs) (p , 0.05). With a cut-off value of 2 m/s for the shear wave velocity, the sensitivity, specificity, positive predictive value and negative predictive value for malignancies were 74%, 82%, 89% and 60%, respectively. DISCUSSION For the abdominal application of ARFI elastography, Fahey et al. first reported the usefulness of ARFI elastography for improving both the conspicuity and contrast of
hepatic and renal tumors relative to the use of conventional B-mode images (Fahey et al. 2008a). According to this report, for all seven hepatic tumors that were analyzed including HCCs and metastases, the boundary definition on ARFI images was improved or equivalent to that on B-mode images. HCCs were softer than the regional cirrhotic liver parenchyma but the metastases were stiffer than the regional non-cirrhotic liver parenchyma. Our results that show most metastatic tumors (13/18, 72%) were stiffer than the surrounding liver and are consistent with previously reported stiffness of metastases, as visually measured on ARFI images, that were reported by Fahey et al. Further, we also found that ARFI images of most metastatic tumors (14/18, 78%) and most HCCs (16/25, 64%) showed a border definition equal to or clearer than that on conventional B-mode images, which is also consistent with their study. Our results seem to be inconsistent with the results of Fahey’s study in regards to the stiffness of HCCs that were
Fig. 2. Findings for a 41-year-old man with pancreatic cancer and liver metastasis. (a) A contrast-enhanced computed tomography (CT) scan shows an ill-defined, large, low-attenuating mass in the tail of the pancreas (asterisk) and a small, low-attenuating lesion in segment 8 of the liver (arrow). (b) On a conventional B-mode image (left figure), a metastatic lesion in the liver appears as an ovoid, unclearly marginated, slightly hypoechoic lesion (arrow), which appears as an ovoid, clearly marginated, homogeneously darker (stiffer) lesion on the acoustic radiation force impulse (ARFI) image (right figure).
206
Ultrasound in Medicine and Biology
Volume 36, Number 2, 2010
Fig. 3. Findings for a 52-year-old man with a hepatocellular carcinoma. (a) A dynamic liver T1-weighted magnetic resonance (MR) image obtained during arterial phase shows a round, enhancing mass (arrow) in segment 6 of the liver. (b) On a dynamic liver T1-weighted MR image obtained 3 min after gadolinium injection, the tumor (arrow) demonstrates washout of contrast enhancement, which is typical of a hepatocellular carcinoma. (c) On a conventional B-mode image (left figure), this tumor appears as a round, clearly marginated, hyperechoic mass and the lesion appears as a round, unclearly marginated, slightly brighter (softer) mass (arrow) on the acoustic radiation force impulse (ARFI) image (right figure). (d) The shear wave velocity (arrow) measured when the region of interest was placed within the tumor was 4.3 m/s.
visually inspected on the ARFI images. In our study, only 24% of the HCCs (6/25) were softer than the surrounding liver, i.e., 76% of the HCCs (19/25) showed equal or greater stiffness than the surrounding liver. The discrepancy with Fahey’s report might be explained by the difference in the severity of the cirrhosis of the background liver in each study population. In our study, the degree of liver cirrhosis in patients with HCCs was likely to be less severe compared with that seen in the prior study, assuming that the liver is stiffer with more severe liver cirrhosis. These findings are supported by the fact that 15 of 20 patients in our study with an HCC or HCCs had chronic liver disease of Child-Pugh classification A. When HCCs are excluded, seven of nine soft lesions depicted on ARFI images (78%) were hemangiomas. This finding might suggest that a soft tissue lesion seen on
ARFI elastography imaging is more likely to be benign in a patient with a normal liver. Shear wave velocity, also known as virtual touch tissue quantification, which is generated by acoustic radiation force, is an objective method for evaluating mechanical tissue properties (Palmeri et al. 2008). Shear wave velocity depends on tissue elasticity as the speed of the shear wave traveling through a region-of-interest (ROI) increases as tissue stiffness increases. Since our study revealed a significantly lower shear wave velocity in hemangiomas compared with those in common malignant hepatic tumors, the shear wave velocity obtained during ARFI elastography may be useful as a complementary tool to differentiate malignant hepatic lesions from benign hepatic lesions, regardless of the presence of cirrhosis in the surrounding liver.
Acoustic radiation force impulse elastography d S. H. CHO et al.
207
Fig. 4. Findings for a 60-year-old man with a hemangioma. (a) A dynamic liver T1-weighted magnetic resonance (MR) image obtained during portal venous phase shows a round mass with peripheral nodular strong enhancement in the posterior segment of the liver (arrow). (b) A dynamic liver T1-weighted MR image obtained five minutes after gadolinium injection demonstrates more centripetal gadolinium enhancement (arrow), which is typical for a hemangioma. (c) A conventional B-mode image (left figure) shows a clearly marginated, homogeneous, hyperechoic tumor but the acoustic radiation force impulse (ARFI) elastography (right figure) shows a clearly marginated, darker (stiffer) tumor (arrows) with multifocal brighter (softer) spots relative to the background liver. (d)The shear wave velocity measured when the region of interest was placed in the tumor was 1.64 m/s.
This study has several limitations. First, we did not include other benign hepatic tumors such as focal nodular hyperplasia, adenoma or abscess. Also, our study included many heterogeneous types of metastatic tumors. Additional studies with other benign tumors and more homogeneous metastatic tumors in a larger series will be needed to establish the clinical value of ARFI elastography in the Table 3. Shear wave velocity
Group I Group II Group III
n
Mean 1 SD(m/s)
8 17 11
2.18 6 0.96 2.45 6 0.81 1.51 6 0.71
Kruskal-Wallis Test
Post-hoc Analysis
p 5 0.012
n 5 number of tumors; SD 5 standard deviation; Group I 5 metastases and cholangiocarcinomas; Group II 5 hepatocellular carcinomas; Group III 5 hemangiomas. Asterisks denote that there is a statistical difference with a p , 0.05 between the two indicated groups.
differential diagnosis of diverse focal solid hepatic lesions. Second, we did not correlate the measured stiffness on ARFI imaging with real stiffness, cellularity or the amount of necrosis, as determined on pathologic specimens but because the aim of this study was to investigate the clinical value of ARFI imaging, such a pathologic correlation was beyond the scope of the study. Third, we were not able to measure the mean shear wave velocity of all tumors because some tumors were located deeper than 6 cm from the skin or near the heart and several patients had poor breath-holds. The US unit we used for ARFI elastography allows transmission of an impulse up to 6 cm from the skin for measurement of shear wave velocity and 10 cm from the skin for imaging of ARFI elastography, which is mainly due to safety concerns. To use a stronger impulse to measure more deeply seated tumors, it must be demonstrated that the impulse does not injure tissue, or a new transmission technique should be developed to avoid tissue injury. Additionally, the current ARFI
208
Ultrasound in Medicine and Biology
imaging is sensitive to motion during measurement. These are technical challenges that remain for ARFI imaging. Despite these study limitations, we found that images obtained through ARFI elastography provided additional qualitative information regarding the stiffness and tumor margin of liver tumors. By measuring shear wave velocity, quantification of tumor stiffness was made possible and showed the potential to differentiate malignant hepatic tumors from hepatic hemangiomas Acknowledgments—The authors wish to thank Chris Woo (Korea) and Bonnie Hami, M.A. (USA) for their assistance in preparing and editing this manuscript.
REFERENCES Bruix J, Sherman M. Practice Guidelines Committee, American Association for the Study of Liver disease. Management of hepatocellular carcinoma. Hepatology 2005;42:1208–1236. Fahey BJ, Hsu SJ, Wolf PD, Nelson RC, Trahey GE. Liver ablation guidance with acoustic radiation force impulse imaging: Challenges and opportunities. Phys Med Biol 2006;51:3785–3808.
Volume 36, Number 2, 2010 Fahey BJ, Nelson RC, Bradway DP, Hsu SJ, Dumont DM, Trahey GE. In vivo visualization of abdominal malignancies with acoustic radiation force elastography. Phys Med Biol 2008a;53:279–293. Fahey BJ, Nelson RC, Hsu SJ, Bradway DP, Dumont DM, Trahey GE. In vivo guidance and assessment of liver radio-frequency ablation with acoustic radiation force elastography. Ultrasound Med Biol 2008b;34:1590–1603. Fahley BJ, Nightingale KR, Nelson RC, Palmeri ML, Trahey GE. Acoustic radiation force impulse imaging of the abdomen: Demonstration of feasibility and utility. Ultrasound Med Biol 2005;31: 1185–1198. Jeong MG, Yu JS, Kim KW. Hepatic cavernous hemangioma: Temporal peritumoral enhancement during multiphase dynamic MR imaging. Radiology 2000;216:692–697. Nightingale K, Soo MS, Nightingale R, Trahey G. Acoustic radiation force impulse imaging: In vivo demonstration of clinical feasibility. Ultrasound Med Biol 2002;28:227–235. Palmeri ML, Wang MH, Dahl JJ, Frinkley KD, Nightingale KR. Quantifying hepatic shear modulus in vivo using acoustic radiation force. Ultrasound Med Biol 2008;34:546–558. Walker WF, Fernandez FJ, Negron LA. A method of imaging viscoelastic parameters with acoustic radiation force. Phys Med Biol 2000;45:1437–1447.
doi:10.1016/j.ultrasmedbio.2009.10.009
d
Original Contribution ACOUSTIC RADIATION FORCE IMPULSE ELASTOGRAPHY FOR THE EVALUATION OF FOCAL SOLID HEPATIC LESIONS: PRELIMINARY FINDINGS SEUNG HYUN CHO,*y JAE YOUNG LEE,* JOON KOO HAN,* and BYUNG IHN CHOI* * Department of Radiology and the Institute of Radiation Medicine, Seoul National University Hospital, Seoul, Korea; and y Department of Radiology, Catholic University of Daegu, School of Medicine, Daegu, Korea (Received 30 May 2009; revised 10 October 2009; in final form 15 October 2009)
Abstract—This study was designed to investigate the potential usefulness of acoustic radiation force impulse (ARFI) elastography to evaluate focal solid hepatic lesions. In total, 51 patients with 60 focal hepatic lesions, which included 17 hemangiomas, 25 hepatocellular carcinomas (HCCs), 15 metastases and three cholangiocarcinomas, underwent ARFI elastography. The lesions were classified into three groups: Group I consisted of metastatic liver tumors and cholangiocarcinomas, group II consisted of HCCs and group III consisted of hemangiomas. The stiffness and conspicuity of the tumors as depicted on ARFI elastography and the echogenicity and conspicuity of the tumors on corresponding B-mode images were analyzed. Shear wave velocity was obtained to quantify stiffness for 36 focal hepatic lesions: 11 hemangiomas, 17 HCCs and eight other malignant lesions. On ARFI elastography images, group I tumors (n 5 18) appeared stiffer than the background liver for 13 lesions (72%), softer for two lesions and had identical stiffness in three lesions compared with the background liver. For group II tumors (n 5 25), 13 lesions (52%) appeared stiffer than the liver, six lesions appeared softer than the liver and the remaining six lesions showed the same stiffness as the liver. For group III tumors (n 5 17), six lesions (35%) appeared stiffer than the liver, seven lesions appeared softer and the remaining four lesions showed the same stiffness as the liver. There were no statistical differences among the three groups in terms of tumor stiffness as seen on ARFI elastography images (p . 0.05). Of the 60 lesions, 41 (68%) displayed a clearer or equivalent margin on the ARFI elastography compared with that seen on B-mode images. The shear wave velocities were: Group I, 2.18 ± 0.96m/s (mean value ± SD); group II, 2.45 ± 0.81m/s; group III, 1.51 ± 0.71m/s (p 5 0.012). With a cut-off value of 2m/s for the shear wave velocity, the positive predictive value and specificity for malignancy were 89% and 81%, respectively. Images obtained with ARFI elastography provided additional qualitative information regarding the stiffness and tumor margin of liver tumors. By measuring shear wave velocity, quantification of stiffness was made possible and showed the potential to differentiate malignant hepatic tumors from hepatic hemangiomas. (E-mail: leejy@radiol. snu.ac.kr) Ó 2010 World Federation for Ultrasound in Medicine & Biology. Key Words: Acoustic radiation force impulse imaging, ARFI Imaging, Elastography, Ultrasonography, Liver neoplasm.
target tissue displacement, i.e., acoustic radiation force impulse (ARFI) elastography (Fahey et al. 2005, 2006; Nightingale et al. 2002; Walker et al. 2000). This technique has enabled in vivo liver elastography by transmission of the acoustic wave between the ribs or from the subcostal area (Fahey et al. 2008a, 2008b). Until now, only one in vivo study has characterized liver tumors in humans using ARFI elastography (Fahey et al. 2008a) and that study only included seven liver tumors. Additionally, there are no studies that have evaluated ARFI elastography’s usefulness in differentiating liver tumors and no studies have been reported on the quantification of tumor stiffness using shear wave velocity, which may help in the diagnosis of liver tumors.
INTRODUCTION Most elastographic imaging of ultrasonography to date has used manual or motorized compression with moving probes. Compression of the liver for elastographic imaging has always been difficult because it is surrounded by the rib cage, which limits using this procedure clinically. Recently, ultrasound elastography with a short, highintensity focused ultrasound beam has been introduced for
Address correspondence to: Jae Young Lee, M.D., Department of Radiology, Seoul National University Hospital, 28 Yeongon-dong, Jongno-gu, Seoul 110-744, Republic of Korea. E-mail: leejy@radiol. snu.ac.kr 202
Acoustic radiation force impulse elastography d S. H. CHO et al.
The present study was performed to investigate the potential usefulness of ARFI elastography for evaluating focal solid hepatic lesions, assuming that different features can be shown on ARFI elastography images according to the type of liver tumor and that quantification of tumor stiffness may be helpful in the diagnosis. MATERIALS AND METHODS Our hospital Institutional Review Board approved this study (IRB number: 0902-039-272) and informed consent was obtained from all patients prior to the ARFI elastography examinations. Patients Over two periods of time, from May to July 2008 (55 patients) and May 2009 (14 patients), a total of 69 patients with focal solid hepatic lesions underwent ARFI elastography. Patient enrollment for these separate periods was dependent on the availability of the ARFI elastography machine in our hospital. Fifteen patients were excluded from the study due to either a technical failure such as patient motion, the presence of a deep-seated lesion or the patient’s inability to hold their breath properly. Another three patients were excluded from the study as they were pathologically proven, by follow-up biopsies, to have focal fatty deposition, liver abscess and fasciola hepatica, respectively. Thus, our study population consisted of 51 patients. There were 60 focal hepatic lesions included in this study because nine patients each had two focal hepatic lesions. The hepatic lesions included 17 hemangiomas in 14 patients (mean tumor diameter, 1.6 cm; range, 0.8–3.0 cm), 25 hepatocellular carcinomas (HCCs) in 20 patients (mean tumor diameter, 3.6 cm; range, 1.6–7.9 cm), 15 metastatic liver tumors in 14 patients and three cholangiocarcinomas in three patients (mean tumor diameter, 3.5 cm; range, 0.7–8.4 cm). The presence of a hemangioma was diagnosed in 13 of 14 patients based on a combination of the typical findings determined using either computed tomography (CT) or magnetic resonance (MR) imaging and the absence of an increase in tumor size for at least 12 months. The diagnosis was confirmed in the remaining patient by percutaneous biopsy performed immediately after ARFI elastography. The following imaging findings were considered for the diagnosis of hemangioma: (a) lesions showing early, peripheral nodular contrast enhancement during the hepatic arterial phase (HAP) and centripetal fill-in enhancement during the portal venous phase (PVP); or (b) lesions showing early strong homogeneous enhancement during the HAP and the same persistent, strong enhancement as that for enhanced intrahepatic vessels during the PVP (Jeong et al. 2000). There was
203
clinical and biochemical evidence of chronic liver disease in two patients with hemangiomas. In this study, the diagnosis of HCC lesions was confirmed by surgery in seven patients, by a biopsy following elastography in one patient and by clinical diagnosis in 12 patients. The clinical diagnosis for HCCs was made according to the American Association for the Study of Liver Disease (AASLD) 2005 recommendations (Bruix et al. 2005). According to the AASLD 2005 guidelines, a diagnosis of HCC can be made if a mass larger than 2 cm in diameter shows typical features of an HCC (i.e., hypervascularity in the arterial phase and washout in the venous phase as seen on contrast-enhanced CT or on MR imaging) or if a mass measuring 1–2 cm shows these features with the use of both modalities. All 20 patients with HCCs in our study had chronic liver disease: Child-Pugh classification A (15 patients), B (four patients) and C (one patient). In this study, the diagnosis of liver metastasis and cholangiocarcinoma was based on histologic confirmation following surgery (five patients) or following biopsy (12 patients). The origins of the 15 metastatic tumors in 14 patients were determined: five from colon cancers, three from neuroendocrine tumors, two from gallbladder cancers, one from rectal cancer, one from duodenal cancer, one from pancreatic cancer, one from renal cell carcinoma and one from adenocarcinoma of unknown origin. There was no clinical or biochemical evidence of chronic liver disease in patients with liver metastases or cholangiocarcinomas. All lesions were categorized into three groups based on the similarity of their pathology. Group I consisted of metastatic liver tumors and cholangiocarcinomas. Group II consisted of HCCs and group III consisted of hemangiomas. Imaging and analysis One of two examiners (S.H.C., J.Y.L.) performed ARFI elastography with an Acuson S2000 US unit (Siemens, Mountain View, CA, USA) using a 2–4 MHz curved array probe. The examiner scanned the liver to locate the lesion detected on prior imaging (e.g., ultrasonography, CT, MR). After fitting the ARFI image box to fully cover the lesion, an ARFI image was obtained with a corresponding B-mode image. Shear wave velocity (m/s) was measured three times for each tumor and was averaged for each of 36 tumors: eight group I tumors, 17 group II tumors and 11 group III tumors. Shear wave velocity was not obtained for the other 24 tumors due to either the depth limit or motion artifacts from the patient’s poor breath-hold or heartbeat. Two reviewers, who were blinded to the final diagnosis, independently reviewed all of the ARFI and B-mode images on picture archiving and communications system (PACS) workstation monitors (m-view,
204
Ultrasound in Medicine and Biology
Volume 36, Number 2, 2010
Table 1. Stiffness findings on acoustic radiation force impulse imaging Acoustic radiation force impulse imaging Brighter (Softer)
Group I Group II Group III
Same color (Equally stiff)
Consensus
R1
R2
Consensus
2 6 7
2 7 8
1 4 5
3 6 4
Darker (Stiffer)
R1
R2
Consensus
R1
R2
2 7 6
13 13 6
12 13 6
15 14 6
4 5 3 p . 0.05
The values are the number of tumors. Group I 5 metastases and cholangiocarcinomas; Group II 5 hepatocellular carcinomas; Group III 5 hemangiomas. R1: reviewer 1. R2: reviewer 2.
INFINITT; Seoul, Korea). The echogenicity or stiffness, and the conspicuity of the tumors were analyzed on both ARFI and corresponding B-mode images. The tumors were categorized as stiffer (darker), equally stiff (of the same color) or softer (brighter) based on the lesion’s brightness relative to the liver on the ARFI images and the tumors were classified as either hypoechoic, isoechoic or hyperechoic based on the lesion’s echogenicity relative to the liver on B-mode images. The conspicuity of a tumor was classified as having either a clear or an unclear margin. Discrepancies between the two reviewers were resolved through consensus. Statistical analysis The Kruskal-Wallis test was used to calculate the statistical difference in the shear wave velocity values in the three groups. A Tukey post-hoc test was used to resolve any differences found on the Kruskal-Wallis test. Fisher’s exact test was used to calculate the statistical difference of the imaging findings regarding the stiffness in the three groups. A difference was determined to be significant if p , 0.05. RESULTS Tables 1 and 2 summarize the echogenicity, stiffness and conspicuity of the lesions in the three groups. Of the elastographic images of group I (metastatic tumors and cholangiocarcinomas), 13 of 18 lesions (72%) appeared darker, which indicated that they were stiffer than the surrounding liver parenchyma (Figs. 1 and 2). The stiffer tumors included three cholangiocarcinomas, four metastatic tumors from colon cancer, one metastatic tumor from duodenal cancer, three from neuroendocrine tumors and one from pancreatic cancer. The remaining lesion was an adenocarcinoma of an unknown origin. When compared with the liver background, three of the 18 lesions (17%) had the same brightness, which indicated that the lesions were as stiff as the surrounding liver. Gallbladder cancer caused two of the three lesions and the remaining lesion was a metastatic tumor from rectal
cancer. Finally, two of the 18 lesions (11%) appeared brighter than the background liver, which indicated that the lesions were softer than the surrounding liver parenchyma. The origins of the lesions were a colon cancer and a renal cell carcinoma. On the elastographic images of group II (HCCs) and group III (hemangiomas) (Figs. 3 and 4), group III (hemangiomas) had a similar number of stiffer tumors and softer tumors relative to the background liver but group II (HCCs) had a higher number of stiffer tumors, as was also found in group I (Table 1). However, there was no statistical difference in the three groups in terms of tumor stiffness suggested by ARFI elastography images (p . 0.05). Of the lesions in the study, 11 of 60 lesions (18%) displayed a clearer margin with ARFI elastography than with B-mode imaging and 30 lesions (50%) showed an equivalent margin with ARFI elastography to that seen with B-mode imaging (Table 2). For group I (metastases and cholangiocarcinomas), ARFI imaging showed the presence of a clear margin in five of the seven lesions that showed an unclear margin on B-mode images (Fig. 2). For group II (HCCs), ARFI imaging showed the presence of a clear margin in three of five lesions that had an unclear margin with B-mode imaging and for group III (hemangiomas), ARFI imaging showed the Table 2. Conspicuity findings on acoustic radiation force impulse imaging Conventional US/ARFI imaging
Group I Group II Group III Total
Clear/ clear
Clear/ unclear
Unclear/ clear
Unclear/ unclear
7 11 5 23
4 9 6 19
5 3 3 11
2 2 3 7
The values are the number of tumors. US 5 ultrasonography; ARFI 5 acoustic radiation force impulse; Group I 5 metastases and cholangiocarcinomas; Group II 5 hepatocellular carcinomas; Group III 5 hemangiomas.
Acoustic radiation force impulse elastography d S. H. CHO et al.
205
Fig. 1. Findings for a 65-year-old man with duodenal cancer and liver metastasis. (a) Contrast-enhanced computed tomography (CT) scan shows a small, low-attenuating lesion in segment 8 of the liver (arrow). (b) On a conventional B-mode image (left figure), this metastatic lesion is seen as an ovoid, clearly marginated, hyperechoic lesion and the lesion is seen as an ovoid, clearly marginated, homogeneously darker (stiffer) lesion on acoustic radiation force impulse (ARFI) imaging (right figure).
presence of a clear margin in three of six lesions that had an unclear margin with B-mode imaging. The shear wave measured velocities are shown in Table 3 and Figures 3 and 4. There was a statistically significant difference among the three groups (p 5 0.012). Tukey post-hoc tests in the individual groups revealed a significantly lower shear wave velocity in group III (hemangiomas) compared with that in groups I (metastases and cholangiocarcinomas) and II (HCCs) (p , 0.05). With a cut-off value of 2 m/s for the shear wave velocity, the sensitivity, specificity, positive predictive value and negative predictive value for malignancies were 74%, 82%, 89% and 60%, respectively. DISCUSSION For the abdominal application of ARFI elastography, Fahey et al. first reported the usefulness of ARFI elastography for improving both the conspicuity and contrast of
hepatic and renal tumors relative to the use of conventional B-mode images (Fahey et al. 2008a). According to this report, for all seven hepatic tumors that were analyzed including HCCs and metastases, the boundary definition on ARFI images was improved or equivalent to that on B-mode images. HCCs were softer than the regional cirrhotic liver parenchyma but the metastases were stiffer than the regional non-cirrhotic liver parenchyma. Our results that show most metastatic tumors (13/18, 72%) were stiffer than the surrounding liver and are consistent with previously reported stiffness of metastases, as visually measured on ARFI images, that were reported by Fahey et al. Further, we also found that ARFI images of most metastatic tumors (14/18, 78%) and most HCCs (16/25, 64%) showed a border definition equal to or clearer than that on conventional B-mode images, which is also consistent with their study. Our results seem to be inconsistent with the results of Fahey’s study in regards to the stiffness of HCCs that were
Fig. 2. Findings for a 41-year-old man with pancreatic cancer and liver metastasis. (a) A contrast-enhanced computed tomography (CT) scan shows an ill-defined, large, low-attenuating mass in the tail of the pancreas (asterisk) and a small, low-attenuating lesion in segment 8 of the liver (arrow). (b) On a conventional B-mode image (left figure), a metastatic lesion in the liver appears as an ovoid, unclearly marginated, slightly hypoechoic lesion (arrow), which appears as an ovoid, clearly marginated, homogeneously darker (stiffer) lesion on the acoustic radiation force impulse (ARFI) image (right figure).
206
Ultrasound in Medicine and Biology
Volume 36, Number 2, 2010
Fig. 3. Findings for a 52-year-old man with a hepatocellular carcinoma. (a) A dynamic liver T1-weighted magnetic resonance (MR) image obtained during arterial phase shows a round, enhancing mass (arrow) in segment 6 of the liver. (b) On a dynamic liver T1-weighted MR image obtained 3 min after gadolinium injection, the tumor (arrow) demonstrates washout of contrast enhancement, which is typical of a hepatocellular carcinoma. (c) On a conventional B-mode image (left figure), this tumor appears as a round, clearly marginated, hyperechoic mass and the lesion appears as a round, unclearly marginated, slightly brighter (softer) mass (arrow) on the acoustic radiation force impulse (ARFI) image (right figure). (d) The shear wave velocity (arrow) measured when the region of interest was placed within the tumor was 4.3 m/s.
visually inspected on the ARFI images. In our study, only 24% of the HCCs (6/25) were softer than the surrounding liver, i.e., 76% of the HCCs (19/25) showed equal or greater stiffness than the surrounding liver. The discrepancy with Fahey’s report might be explained by the difference in the severity of the cirrhosis of the background liver in each study population. In our study, the degree of liver cirrhosis in patients with HCCs was likely to be less severe compared with that seen in the prior study, assuming that the liver is stiffer with more severe liver cirrhosis. These findings are supported by the fact that 15 of 20 patients in our study with an HCC or HCCs had chronic liver disease of Child-Pugh classification A. When HCCs are excluded, seven of nine soft lesions depicted on ARFI images (78%) were hemangiomas. This finding might suggest that a soft tissue lesion seen on
ARFI elastography imaging is more likely to be benign in a patient with a normal liver. Shear wave velocity, also known as virtual touch tissue quantification, which is generated by acoustic radiation force, is an objective method for evaluating mechanical tissue properties (Palmeri et al. 2008). Shear wave velocity depends on tissue elasticity as the speed of the shear wave traveling through a region-of-interest (ROI) increases as tissue stiffness increases. Since our study revealed a significantly lower shear wave velocity in hemangiomas compared with those in common malignant hepatic tumors, the shear wave velocity obtained during ARFI elastography may be useful as a complementary tool to differentiate malignant hepatic lesions from benign hepatic lesions, regardless of the presence of cirrhosis in the surrounding liver.
Acoustic radiation force impulse elastography d S. H. CHO et al.
207
Fig. 4. Findings for a 60-year-old man with a hemangioma. (a) A dynamic liver T1-weighted magnetic resonance (MR) image obtained during portal venous phase shows a round mass with peripheral nodular strong enhancement in the posterior segment of the liver (arrow). (b) A dynamic liver T1-weighted MR image obtained five minutes after gadolinium injection demonstrates more centripetal gadolinium enhancement (arrow), which is typical for a hemangioma. (c) A conventional B-mode image (left figure) shows a clearly marginated, homogeneous, hyperechoic tumor but the acoustic radiation force impulse (ARFI) elastography (right figure) shows a clearly marginated, darker (stiffer) tumor (arrows) with multifocal brighter (softer) spots relative to the background liver. (d)The shear wave velocity measured when the region of interest was placed in the tumor was 1.64 m/s.
This study has several limitations. First, we did not include other benign hepatic tumors such as focal nodular hyperplasia, adenoma or abscess. Also, our study included many heterogeneous types of metastatic tumors. Additional studies with other benign tumors and more homogeneous metastatic tumors in a larger series will be needed to establish the clinical value of ARFI elastography in the Table 3. Shear wave velocity
Group I Group II Group III
n
Mean 1 SD(m/s)
8 17 11
2.18 6 0.96 2.45 6 0.81 1.51 6 0.71
Kruskal-Wallis Test
Post-hoc Analysis
p 5 0.012
n 5 number of tumors; SD 5 standard deviation; Group I 5 metastases and cholangiocarcinomas; Group II 5 hepatocellular carcinomas; Group III 5 hemangiomas. Asterisks denote that there is a statistical difference with a p , 0.05 between the two indicated groups.
differential diagnosis of diverse focal solid hepatic lesions. Second, we did not correlate the measured stiffness on ARFI imaging with real stiffness, cellularity or the amount of necrosis, as determined on pathologic specimens but because the aim of this study was to investigate the clinical value of ARFI imaging, such a pathologic correlation was beyond the scope of the study. Third, we were not able to measure the mean shear wave velocity of all tumors because some tumors were located deeper than 6 cm from the skin or near the heart and several patients had poor breath-holds. The US unit we used for ARFI elastography allows transmission of an impulse up to 6 cm from the skin for measurement of shear wave velocity and 10 cm from the skin for imaging of ARFI elastography, which is mainly due to safety concerns. To use a stronger impulse to measure more deeply seated tumors, it must be demonstrated that the impulse does not injure tissue, or a new transmission technique should be developed to avoid tissue injury. Additionally, the current ARFI
208
Ultrasound in Medicine and Biology
imaging is sensitive to motion during measurement. These are technical challenges that remain for ARFI imaging. Despite these study limitations, we found that images obtained through ARFI elastography provided additional qualitative information regarding the stiffness and tumor margin of liver tumors. By measuring shear wave velocity, quantification of tumor stiffness was made possible and showed the potential to differentiate malignant hepatic tumors from hepatic hemangiomas Acknowledgments—The authors wish to thank Chris Woo (Korea) and Bonnie Hami, M.A. (USA) for their assistance in preparing and editing this manuscript.
REFERENCES Bruix J, Sherman M. Practice Guidelines Committee, American Association for the Study of Liver disease. Management of hepatocellular carcinoma. Hepatology 2005;42:1208–1236. Fahey BJ, Hsu SJ, Wolf PD, Nelson RC, Trahey GE. Liver ablation guidance with acoustic radiation force impulse imaging: Challenges and opportunities. Phys Med Biol 2006;51:3785–3808.
Volume 36, Number 2, 2010 Fahey BJ, Nelson RC, Bradway DP, Hsu SJ, Dumont DM, Trahey GE. In vivo visualization of abdominal malignancies with acoustic radiation force elastography. Phys Med Biol 2008a;53:279–293. Fahey BJ, Nelson RC, Hsu SJ, Bradway DP, Dumont DM, Trahey GE. In vivo guidance and assessment of liver radio-frequency ablation with acoustic radiation force elastography. Ultrasound Med Biol 2008b;34:1590–1603. Fahley BJ, Nightingale KR, Nelson RC, Palmeri ML, Trahey GE. Acoustic radiation force impulse imaging of the abdomen: Demonstration of feasibility and utility. Ultrasound Med Biol 2005;31: 1185–1198. Jeong MG, Yu JS, Kim KW. Hepatic cavernous hemangioma: Temporal peritumoral enhancement during multiphase dynamic MR imaging. Radiology 2000;216:692–697. Nightingale K, Soo MS, Nightingale R, Trahey G. Acoustic radiation force impulse imaging: In vivo demonstration of clinical feasibility. Ultrasound Med Biol 2002;28:227–235. Palmeri ML, Wang MH, Dahl JJ, Frinkley KD, Nightingale KR. Quantifying hepatic shear modulus in vivo using acoustic radiation force. Ultrasound Med Biol 2008;34:546–558. Walker WF, Fernandez FJ, Negron LA. A method of imaging viscoelastic parameters with acoustic radiation force. Phys Med Biol 2000;45:1437–1447.