| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Experimental Studies |
1 From the Department of Radiology, Division of Ultrasound, Thomas Jefferson University Hospital, Suite 763J, Main Bldg, 132 S 10th St, Philadelphia, PA 19107. Received November 10, 2000; revision requested December 24; final revision received August 9, 2001; accepted September 5. Supported in part by Nycomed Amersham, Oslo, Norway, and Siemens Medical Systems, Issaquah, Wash. Address correspondence to F.F. (e-mail: forsberg@esther.rad.tju.edu).
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
MATERIALS AND METHODS: Conventional gray-scale liver US was performed in 65 rabbits, 38 of which had VX-2 hepatic tumor implants. Twenty minutes after contrast agent injection, gray-scale pulse-inversion harmonic US images of the liver-specific phase were obtained. Following sacrifice of the animals, T1- and T2-weighted MR imaging was performed at 4-mm intervals. Pathologic analysis was performed as the reference standard. The capability of each imaging modality to correctly depict tumor presence or absence and the number of tumors was compared.
RESULTS: Conventional US correctly depicted the presence or absence of tumors in 54 rabbits, for an accuracy of 83%, sensitivity of 71%, and specificity of 100%. With contrast-enhanced US, accuracy increased to 92% (60 correct cases); sensitivity, to 87%; and specificity, to 100%. MR imaging facilitated 56 correct diagnoses, for an accuracy of 86%, sensitivity of 82%, and specificity of 93%. There was a marginally significant difference between US with and US without contrast agent (P = .07) but not between MR imaging and contrast-enhanced US (P 
.34). When the numbers of correctly detected tumors were compared, contrast-enhanced US performed significantly better than MR imaging (P = .02) and conventional US (P = .04).
CONCLUSION: There was no significant difference between contrast-enhanced US and MR imaging in the detection of hepatic tumors, whereas contrast-enhanced US had the highest accuracy (92%) of the three modalities studied.
© RSNA, 2002
Index terms: Animals • Liver neoplasms, 761.32 • Liver neoplasms, MR, 761.121411, 761.121415 • Liver neoplasms, US, 761.12988, 761.12989 • Ultrasound (US), harmonic study, 761.12988, 761.12989
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
To further suppress unwanted tissue signals and reduce the effect of receive filter characteristics, pulse- (or phase-) inversion harmonic US has been proposed (16,17). With this technique, linearly scattered signals are canceled by transmitting a pulse sequence in which each pulse is an inverted copy of the previous pulse. Thus, under linear scattering conditions, the sum of subsequent echoes will be zero and stationary tissue signals will be suppressed. However, the nonlinear echoes associated with contrast agent microbubbles will not cancel out and will therefore be displayed on the image. Initial in vivo evaluations of pulse-inversion harmonic US have been encouraging (18–21).
NC100100 (Sonazoid; Amersham Health, Oslo, Norway) is a recently introduced US contrast agent that is retained within the cells of the reticuloendothelial system (RES) of the liver and spleen 10–30 minutes after the normal blood pool phase. In this delayed RES-specific phase, the intact and stationary bubbles within the RES can be imaged to produce induced or stimulated acoustic emission (AE) signals (22). This phenomenon occurs at higher acoustic pressures (albeit still within the approved diagnostic range) when the acoustic field makes the contrast agent microbubbles oscillate and finally collapse. During this collapse, energy is released in the form of a localized, transient broadband signal that is seen with color Doppler US as a characteristic random pseudocolor and with gray-scale US as marked, transient enhancement (22–26). When AE signals are imaged, liver tumors are seen briefly as signal voids without enhancement. The purpose of this study was to compare conventional US and MR imaging with contrast-enhanced US during the delayed RES-specific phase for the detection of VX-2 liver tumors in rabbits.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
In each of the 38 rabbits, 0.5 mL of VX-2 tumor cells (approximately 3 million) (Bogden Laboratories, Worcester, Mass) was injected percutaneously into the liver in one or two locations. Following injection, a localized, avascular carcinoma-like mass develops at the site of injection after 10–15 days (27). In some cases, metastases occur. Twenty-seven rabbits without tumor implants served as control animals.
In all 65 rabbits, an 18-gauge angiocatheter was placed in the jugular vein for contrast agent administration by using the anesthesia protocol just described. Each animal was shaved in the appropriate areas to facilitate imaging, and one of two authors (N.M.R., D.A.M.) performed conventional gray-scale US of the liver (up to 4 cm in depth) by using a 7.5-MHz linear array with a US scanner (Sonoline Elegra; Siemens Medical Systems, Issaquah, Wash) and a dynamic range of 55 dB. All images were recorded on videotape, and the presence or absence of tumors was noted. Next, a single injection of NC100100 (dose, 0.0125 µL of microbubbles/kg; average volume, 0.036 mL) was administered at a rate of approximately 1 mL/sec, followed by a 5-mL saline flush to ensure that no residual agent remained in the catheter. NC100100 is a lipid-stabilized suspension of perfluorobutane microbubbles with a median diameter of 2.4–3.5 µm.
Because of the RES-specific uptake of NC100100, gray-scale pulse-inversion harmonic US data were recorded after a 20-minute delay to allow sufficient time for the agent to accumulate in the normal liver parenchyma. During this delayed phase, NC100100 emits the characteristic AE signals when it is imaged at moderate to high acoustic pressures. We assessed the pressure level by using the mechanical index (MI) displayed on the screen. The MI, which is defined as the peak rarefractional pressure divided by the square root of the imaging frequency (28), provides an indirect measure of the acoustic pressure and was kept at a value of approximately 0.6 (range, 0.5–0.7). These MI values were judged, on the basis of findings in our previous studies (22), to be sufficient to produce marked enhancement for improved tumor detection while limiting bubble destruction and thus ensure prolonged enhancement. Gray-scale US and pulse-inversion harmonic US were performed by one of two experienced sonographers (N.M.R., D.A.M.), who kept all imaging parameters constant before and after contrast agent administration in a given animal. In selected cases with multiple metastases, extended–field-of-view images also were acquired to provide an overview of the entire liver.
The number—that is, zero, one, or two or more—and size of tumors were determined in all animals on the basis of pulse-inversion harmonic US findings (excluding extended–field-of-view images). In the animals with multiple metastases, data on up to 10 lesions were collected. The contrast enhancement and AE signals of NC100100 are so characteristic that it was impossible to blind the observers (F.F. and the sonographers, each with more than 10 years experience, in consensus) with regard to the presence or absence of the agent. To minimize bias, the observers were not made aware of whether VX-2 tumors had been implanted in the rabbits; thus, they did not know the likelihood of finding tumors. Moreover, US imaging was performed in batches of six to eight randomly selected rabbits that included both normal and abnormal (ie, tumor-bearing) animals. After each US imaging study was completed, the rabbits were sacrificed with an overdose of pentobarbital sodium (Nembutal; Abbott Laboratories, North Chicago, Ill).
Following the sacrifice of the animals, T1- and T2-weighted MR imaging examinations were performed by using a 1.5-T unit (Signa Horizon; GE Medical Systems, Milwaukee, Wis) with a quadrature local volume coil designed for the extremities (17 cm in diameter and 22-cm long). This experimental design was chosen to eliminate all motion and breathing artifacts and to avoid keeping the animals sedated for more than 8 hours. However, it also meant that no MR imaging contrast agents could be used. Contiguous 4-mm sections were acquired by using T1-weighted spin-echo (533/9 [repetition time msec/echo time msec]) and T2-weighted fast spin-echo (3,000/96; echo train length, 12) MR sequences. For all images, the receive bandwidth was ±16 kHz, the field of view was 12 cm2, and the lipid signal intensity was suppressed by frequency-selective saturation. For the spin-echo images, the matrix was 256 x 160 and two signals were acquired, and for the fast spin-echo images, the matrix was 256 x 256 and three signals were acquired.
The MR images were read by one of two experienced observers (C.W.P., D.G.M.), each with more than 10 years experience, who were blinded to all US results and to which rabbits had implanted tumors. The observers noted the presence or absence of tumors, which were identified as focal lesions with low signal intensity on T1-weighted images and high signal intensity on T2-weighted images, as well as the numbers and sizes of the tumors, as described herein earlier.
Finally, gross pathologic analysis of the liver was performed to provide a reference standard for comparison with the three imaging modalities. After excision, one of the authors (J.B.L.) scanned the livers with conventional US to establish the orientation that best matched the US imaging planes. All of the livers were serially sectioned in their entirety (by J.B.L.), and the number and size of the masses were noted. Hence, the pathology results could be compared directly with the tumor numbers and positions established by using conventional US, pulse-inversion harmonic US, and MR imaging.
The capability of each imaging modality to enable correct identification of a rabbit with at least one implanted VX-2 tumor was determined and recorded as "yes" or "no." Sensitivity was calculated as the number of true-positive identifications divided by the total number of positive identifications, and specificity was calculated as the number of true-negative identifications divided by the total number of negative identifications. Overall accuracy was calculated as the number of correct identifications divided by the total number of cases. Hence, comparisons of sensitivity were conducted in the subset of animals with implanted tumors, and comparisons of specificity were conducted by using the control animals. For analysis of accuracy, all 65 rabbits were used. In addition, the capability of each modality to depict the correct number of implanted tumors was compared. Statistical analysis of tumor detection rates was performed by using the McNemar test for correlated proportions (29), with P values of less than .05 indicating a significant difference. The exact binomial probabilities were used in the McNemar test, because the number of matched pairs with different outcomes (ie, discordant pairs) was less than 20 in all analyses.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
|
|
|
|
|
|
|
|
|
|
Hence, the sensitivity and specificity of conventional US were 71% and 100%, respectively, and the overall accuracy was 83% (Table 1) for correct detection of the presence or absence of hepatic tumors in rabbits. After the administration of NC100100, accuracy increased to 92% (60 correct classifications); sensitivity, to 87%; and specificity, to 100%. MR imaging enabled 56 correct identifications, for an accuracy of 86%, sensitivity of 82%, and specificity of 93%. There was a marginally significant difference between the sensitivity and accuracy of US with contrast agent and those of US without contrast agent (P = .07; Table 1) but not between the specificity of conventional US and that of contrast-enhanced US (P > .999). There were no significant differences between MR imaging and US with or without contrast agent (P .29).
|
Finally, the sensitivity, specificity, and accuracy of conventional US, contrast-enhanced US, and MR imaging in the differentiation between rabbits with solitary lesions and those with two or more lesions were determined in the 38 tumor-bearing rabbits (Table 2). For conventional US, the sensitivity, specificity, and accuracy were 67%, 25%, and 61%, respectively, whereas the corresponding values achieved with contrast-enhanced US were 83%, 50%, and 79%. MR imaging had a sensitivity of 53% and a specificity of 63%, for an overall accuracy of 58%. The sensitivity and accuracy of contrast-enhanced US were significantly better than those of MR imaging (P = .004 and P = .06 for differences in sensitivity and accuracy, respectively) and conventional US (P = .07 and P = .04 for differences in sensitivity and accuracy, respectively). No other comparisons were significant at the .05 level.
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
After the contrast agent was deposited in the RES, the sensitivity and accuracy of VX-2 tumor detection with gray-scale pulse-inversion harmonic US improved compared with those achieved with conventional US (Table 1; Figs 2, 3). The sensitivity of US increased from 71% to 87% with the addition of contrast agent, without a decrease in specificity. The accuracy of conventional US was 83% before contrast agent administration and thus comparable to the reported 53%–85% accuracy of this modality for the detection of human hepatic metastases (3,4,30–32). After contrast agent administration, the accuracy of US increased marginally significantly to 92% (P = .07). The performance results for contrast-enhanced US were comparable to those for MR imaging (P .34) (Table 1). Thus, contrast-enhanced US of NC100100 during the RES-specific phase performed as well as MR imaging in the detection of hepatic tumors and had the highest accuracy (92%) of the three modalities studied.
One of the advantages in this study was that a definitive pathologic diagnosis was available in all cases. This enabled the reliable identification of identical lesions on US (with and without contrast agent) and MR images despite the somewhat different imaging planes used. The accuracy of conventional US, contrast-enhanced US, and MR imaging decreased to 77%, 88%, and 72%, respectively, when the number of tumors detected—that is, one lesion or two or more lesions—was considered. Contrast-enhanced US performed significantly better than MR imaging (P = .02) and conventional US (P = .04). Likewise, contrast-enhanced US had significantly better sensitivity and accuracy than MR imaging and conventional US (P 
.07) when rabbits with one versus those with multiple tumors were differentiated (Table 2). The improved detection rate of US with NC100100 was mainly due to the depiction of additional metastases (Table 2; Figs 2, 3). No significant differences between conventional US and MR imaging were found (P .13).
It was somewhat surprising that there was no significant difference between conventional US and MR imaging, because this is contrary to clinical experiences in diagnosing hepatic tumors (4,31,32). We hypothesize that this result may have been due to the shallow depths (<4 cm) imaged and thus the use of a high-spatial-resolution, high-frequency (7.5-MHz) US probe. It has been reported that in human liver evaluations, intraoperative and laparoscopic US depicts more and smaller lesions than CT and MR imaging because of the higher imaging frequencies used (4,32–34). Similar considerations may apply to our animal results.
The RES-specific phase of NC100100 was visualized in gray-scale pulse-inversion harmonic US mode in this study. However, the random mosaic pattern observed during color Doppler US of AE signals from other US contrast agents (24–26) can also be seen with NC100100 (35,36). The difference is that US imaging of AE signals alone requires bubble destruction, which limits the duration of enhancement to as little as a few frames (22,26), unlike the harmonic US performed in the current study, in which minutes of enhancement could be achieved by imaging a mixture of AE signals and nonlinear bubble oscillations. Moreover, gray-scale pulse-inversion harmonic US is less susceptible to motion artifacts and contrast agent–induced blooming (37) and has superior resolution (spatial and temporal) compared with color Doppler US.
Practical application: The finding of a significant number of additional hepatic VX-2 tumors in rabbits by using US after the administration of NC100100 indicates that gray-scale pulse-inversion harmonic US may improve the detection of liver masses in humans. A limitation of this study is that the statistical power of the McNemar test with this number of cases is relatively low. Moreover, results were obtained at shallow depths (<4 cm) with a high-frequency US probe, and no cirrhotic livers were evaluated. Thus, it is possible that lesions as small as 2 x 3 mm may not be detectable in the large livers of humans. However, other researchers (30) have reported observing marked AE signals from cirrhotic livers in humans using US with another contrast agent. Furthermore, it is estimated that tumors smaller than 5 mm cannot be consistently detected (1,2,31,32), even with contrast-enhanced CT and MR imaging, but the initial results of NC100100 trials with livers in humans are encouraging (36).
If the findings of future large clinical trials confirm our initial animal study results, this will have a major effect on the imaging of hepatic lesions and thus on patient treatment, especially if contrast-enhanced US also compares favorably with contrast-enhanced dual phase spiral CT and MR imaging, which were not a part of the current study. Early detection of small occult metastases with gray-scale pulse-inversion harmonic US of stationary bubbles—that is, in the RES-specific phase—could alter tumor staging. This in turn could permit earlier intervention, such as chemotherapy or surgical resection. In some instances, surgery might be avoided owing to the detection of additional unsuspected liver lesions.
Finally, the correct characterization of hepatic lesions requires knowledge of normal and abnormal hepatic vascularity (38). The results of initial studies (20,21) with harmonic imaging and pulse-inversion harmonic US suggest that imaging of US contrast agents during the vascular phase might enable the differentiation of a variety of hepatic tumors. Additional benefits should be possible with a contrast agent, such as NC100100, that has both a vascular and delayed RES-specific uptake. In particular, hemangiomas would not be expected to have substantial blood flow initially, but after a delay, one would expect them to show enhancement due to their very slow blood flow (38). The prolonged duration of enhancement due to the combination of moderate MI values, gray-scale pulse-inversion harmonic US, and NC100100 in the RES-specific phase also opens up the possibility of performing US-guided biopsy on the additional lesions found. This would be an advantage over the imaging of transient AE signals reported with other US contrast agents (18,23–26,30).
|
ACKNOWLEDGMENTS |
|---|
|
FOOTNOTES |
|---|
Author contributions: Guarantor of integrity of entire study, F.F.; study concepts, F.F., C.W.P., B.B.G., D.G.M.; study design, F.F., B.B.G.; literature research, F.F.; experimental studies, all authors; data acquisition and analysis/interpretation, all authors; statistical analysis, F.F.; manuscript preparation, F.F.; manuscript definition of intellectual content, editing, revision/review, and final version approval, all authors.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  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 Maruyama, S Matsutani, H Saisho, Y Mine, N Kamiyama, T Hirata, and M Sasamata Real-time blood-pool images of contrast enhanced ultrasound with Definity in the detection of tumour nodules in the liver Br. J. Radiol., June 1, 2005; 78(930): 512 - 518. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
F. Forsberg, W. T. Shi, C. R. B. Merritt, Q. Dai, M. Solcova, and B. B. Goldberg On the Usefulness of the Mechanical Index Displayed on Clinical Ultrasound Scanners for Predicting Contrast Microbubble Destruction J. Ultrasound Med., April 1, 2005; 24(4): 443 - 450. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 B. B. Goldberg, D. A. Merton, J.-B. Liu, M. Thakur, G. F. Murphy, L. Needleman, A. Tornes, and F. Forsberg Sentinel Lymph Nodes in a Swine Model with Melanoma: Contrast-enhanced Lymphatic US Radiology, March 1, 2004; 230(3): 727 - 734. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J Furuse, M Nagase, H Ishii, and M Yoshino Contrast enhancement patterns of hepatic tumours during the vascular phase using coded harmonic imaging and Levovist to differentiate hepatocellular carcinoma from other focal lesions Br. J. Radiol., June 1, 2003; 76(906): 385 - 392. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
W. T. Yang, G. M. K. Tse, P. K. W. Lam, C. Metreweli, and J. Chang Correlation Between Color Power Doppler Sonographic Measurement of Breast Tumor Vasculature and Immunohistochemical Analysis of Microvessel Density for the Quantitation of Angiogenesis J. Ultrasound Med., November 1, 2002; 21(11): 1227 - 1235. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| RADIOLOGY | RADIOGRAPHICS | RSNA JOURNALS ONLINE |
