American Journal of Emergency Medicine (2012) 30, 1229-1234

Brief Report

Liver trauma diagnosis with Contrast-enhanced ultrasound: interobserver variability between radiologist and emergency physician in an animal study

Je Sung You MD ^a, Yong Eun Chung MD ^b,?, Hye-Jeong Lee MD ^b, Sung Phil Chung MD ^a, Incheol Park MD ^a, Myeong-Jin Kim MD ^b, Mi-Suk Park MD ^b, Joon Seok Lim MD ^b,

Jin-Young Choi MD ^b, Seungho Kim MD ^a, Ki Whang Kim MD ^b

^aDepartment of Emergency Medicine, Severance Hospital, Yonsei University College of Medicine, Seoul 120-752,

Republic of Korea

^bDepartment of Radiology, Research Institute of Radiological Science, Severance Hospital, Yonsei University College of Medicine, Seoul 120-752, Republic of Korea

Received 13 April 2011; revised 20 May 2011; accepted 14 June 2011

Abstract

Objective: The purpose of our study was to evaluate interobserver variability between the radiologist and emergency physician in detecting Blunt liver trauma by conventional and contrast-enhanced ultrasound (US) (CEUS).

Methods: We created 20 sites of blunt liver trauma in rabbits and performed conventional US and CEUS on the animals. A radiologist and an emergency physician independently evaluated the degree of liver trauma. Using contrast-enhanced computed tomography as a reference standard, the diagnostic performance of US and CEUS was calculated. Interobserver variability between radiologist and emergency physician was compared before and after contrast enhancement of US.

Results: Overall sensitivity and specificity for detecting liver trauma on conventional US, regardless of the degree of trauma, were 61.1% and 100% for the radiologist and 50% and 100% for the emergency physician. On CEUS, the sensitivity and specificity were 94.4% and 100% for both the radiologist and emergency physician. The Interobserver agreement between emergency physician and radiologist increased from 0.867 to 0.955 after contrast enhancement on US.

Conclusions: Contrast-enhanced US may permit a more accurate diagnosis for liver trauma than conventional US by both the radiologist and emergency physician. Contrast-enhanced US may also reduce interobserver variability for this diagnosis.

(C) 2012

* Corresponding author. Department of Radiology, Severance Hospital, Seodaemun-ku, Shinchon-dong 134, Seoul 120-752, Republic of Korea. Tel.: +82 2 2228 7400; fax: +82 2 393 3035.

E-mail address: [email protected] (Y.E. Chung).

Introduction

The liver, with its abundant Blood supply and superficial location, is the organ most frequently injured by blunt trauma to the abdomen [1]. Computed tomography (CT) is typically used to evaluate blunt abdominal trauma and shows high

0735-6757/$ – see front matter (C) 2012 doi:10.1016/j.ajem.2011.06.012

sensitivity in detecting the presence and extent of organ and vascular injuries [2]. Computed tomography cannot be performed, however, if the patient is unstable or has multiple bone fractures. Contrast enhancement, which is essential to diagnose active bleeding or blunt injury to a solid organ, cannot be performed in patients with history of severe adverse (eg, anaphylactoid) reaction to Contrast material. Computed tomography is also expensive and presents an excess of radiation exposure when repeated imaging is required during treatment or follow-up [3]. Bedside Ultrasonography , on the other hand, is easy to access and provides accurate evaluation of unstable patients in an emergency department (ED). Ultrasonography is shown to be a rapid, noninvasive, and effective tool to identify and assess blunt trauma injury [1], and it also performs well in a protocol to assess free fluid in the abdomen (focused assessment with sonography for trauma [FAST]) [4]. Conventional US has lower accuracy, however, in determin- ing the severity of parenchymal injuries and in the detection of active bleeding in abdominal organs [3]. Recently introduced US Contrast agents are reported to increase the accuracy of liver trauma diagnosis [1,3]. Although radiolo- gists may have more experience in contrast-enhanced US (CEUS) than emergency physicians, the radiologist does not always cover the ED. Emergency physicians must, therefore, perform the US to assist patients with blunt trauma who need immediate care. We performed this study to evaluate the interobserver variability in detecting and evaluating blunt trauma to the liver by a radiologist and an emergency physician using conventional and CEUS.

Methods

Preparation of the animal model

The Experimental Animal Ethical Committee of Yonsei University reviewed and approved all of our protocols according to “Guide for the Care and Use of Laboratory Animals.”

Ten male New Zealand white rabbits weighing 3.3 to 3.7 kg (DooYeol Biotech, Seoul, Korea) were housed in metal cages with access to food and water ad libitum until 1 day before the experiment. The rabbits were fasted overnight because the stomach distended with food may interfere with administration of the trauma. Rabbits were anesthetized with a mixture of tiletamine-zolazepam (15 mg/kg; Zoletil; Virbac Laboratories, Carros, France) and xylazine (5 mg/kg; Rompun; Bayer Korea, Seoul, Korea) given as an intramus- cular injection at the thigh. After anesthesia, the hair on the abdomen was shaved to provide a clear sonic window for US. An initial US was performed, and 2 target points were marked over the right and left lobes of the liver. After randomization, weights of 1, 3, and 5 kg were dropped on the animals from heights of 30 cm, 50 cm, and 1 m to induce

liver trauma of various degrees. Liver trauma was inflicted by a physician who did not participate in US examination and imaging analysis. During the experiment, fluid was administered through an intravenous line at the ear vein to maintain the animals’ blood pressure.

Ultrasound examination

Immediately after the liver trauma, the rabbit was fixed in a supine position at the table, and US was performed with Acuson Sequoia (Siemens Medical Solutions, Erlangen, Germany). A conventional US imaging was performed by a radiologist with 8 years of experience in abdominal US and by an emergency physician with 8 years of experience in emergency US using a 1- to 4-MHz convex or 8- to 15-MHz Linear probe.

After the conventional US, the mode was changed to contrast pulse sequence, and the mechanical index was adjusted to 0.16 to 0.18 to evaluate the CEUS. The US contrast medium (Sonovue; Bracco imaging Korea, Seoul, Korea) was injected through the ear vein, 0.05 mL/kg, and flushed with 10-mL saline. The CEUS was performed immediately after contrast medium injection and lasted for 5 minutes. In each US session, the radiologist and emergency physician performed the imaging procedure independently, in random order, and each was blinded to the results that the other physician obtained. High mechanical index mode US was performed between CEUS sessions for at least 5 minutes by the physician not participating in the examination to break up residual microbubbles within the liver.

Computed tomographic image acquisition

All CT scans were performed immediately after US examination with a conventional 64-channel multidetector CT (LightSpeed VCT; GE Healthcare, Milwaukee, WI). The CT parameters were as follows: rotation time, 0.6 seconds; 120 kV; 160 mA; pitch, 0.531:1; slice thickness, 0.625 mm; and slice gap, 0 mm. Contrast medium (Iopamiro 370; Phamaceuticals, Seoul, Korea) was administered via the ear vein by hand as a Bolus injection of 600 mg I/kg. Contrast- enhanced CT (CECT) was performed 70 seconds after contrast medium injection from the upper border of the heart to the level of the umbilicus to include the whole liver.

Image analysis

Blunt liver trauma is usually graded by the organ injury severity scale of the American Association for the Surgery of Trauma [5]. Because the rabbit liver is smaller than human liver and differs in lobar anatomy [6], we used the following simplified grading system: grade 0, no trauma; grade 1, minor trauma involving less than 10% of the one half of liver involved (right lobe or left lobe), either laceration or hematoma; grade 2, moderate liver trauma involving more

than 10% but less than 50% of the one half of the liver; and grade 3, severe liver trauma involving more than 50% of the one half of the liver. Based on previous studies, areas that were ill defined or discretely hypoechoic or hyperechoic as compared with adjacent normal liver on conventional US and areas that were hypoenhancing on CEUS were evaluated as trauma [1,3,7]. During conventional US and CEUS, the radiologist and emergency physician independently evaluat- ed the presence or absence of liver trauma and the grade of trauma if present. Computed tomographic scans were reviewed by another radiologist who did not participate in the Animal experiment or US based on the simplified grading system.

Statistical analysis

A medical statistician performed all statistical analyses using SAS (version 9.1.3; SAS Institute Inc, Cary, NC). Sensitivity and specificity for detecting liver trauma of any degree by conventional and CEUS were calculated for both the radiologist and emergency physician. The agreement between CT and US with and without contrast enhancement and between radiologist and emergency physician were calculated by weighted ? statistics [8]. Interobserver variability between radiologist and emergency physician

was evaluated using the same statistical methods [8]. Weighted ? coefficients with a goodness-of-fit procedure were used to compare data obtained by the radiologist and emergency physician and data obtained before and after contrast enhancement of US. A difference with a P value less than .05 was considered statistically significant.

Results

Based on CT, 2 liver traumas of a minor degree, 3 of a moderate degree, and 13 of a severe degree were diagnosed in the rabbits. There was no trauma in 2 sites. Using conventional US, the emergency physician did not detect any of the minor-to-moderate liver traumas, and the radiologist missed all of the mild-to-moderate traumas except for 1 minor trauma. The emergency physician missed 30.8% (4/13); and the radiologist, 23.1% (3/13) of the major degree traumas on conventional US. Using CEUS, both the radiologist and the emergency physician diagnosed 1 of 2 minor traumas (Fig. 1). All traumas of moderate degree were detected, but 66.7% (2/3) and 33.3% (1/3) of them were diagnosed as minor liver traumas by the emergency physician and radiologist, respectively. All of 13 severe liver traumas were detected by both the radiologist and emergency

Fig. 1 A, On conventional US, a faint, ill-defined hypoechoic lesion was noted (arrows). The radiologist detected this lesion as a minor liver trauma, whereas the emergency physician did not detect this lesion. B, After contrast enhancement, the lesion showed more clearly (arrows), and the radiologist and emergency physician both detected this lesion as a minor liver trauma. C, On a CT scan, a wedge-shaped, low- attenuating lesion involving less than 10% of the left lobe of the liver was seen (arrowheads). D, A photograph of the gross specimen showed a minor Liver laceration in the left lobe (arrow).

physician using CEUS (Fig. 2 and Table 1). Overall sensitivity and specificity for detecting liver trauma regard- less of degree were 61.1% and 100% for the radiologist and 50% and 100% for the emergency physician on conventional US. On CEUS, sensitivity and specificity were 94.4% and 100% for both the radiologist and emergency physician.

Agreement between CT and US evaluations increased significantly after contrast enhancement of US, from 0.452 (95% confidence interval [CI], 0.184-0.721) to 0.902 (95% CI, 0.783-1.000) for the radiologist (P = .003) and from

0.369 (95% CI, 0.127-0.611) to 0.858 (95% CI, 0.720-0.995)

for the emergency physician (P = .001). Although the agreement between CT and US was higher for the radiologist than for the emergency physician, both before and after contrast enhancement of US, the difference did not reach statistical significance (P = .653 for conventional US and P =

.631 for CEUS). As an index of interobserver agreement, the ? statistic for the emergency physician and radiologist on US increased from 0.867 (95% CI, 0.671-1.000) to 0.955 (95% CI, 0.869-1.000) after contrast enhancement.

Discussion

Ultrasonography is increasingly available in emergency medicine departments, and emergency physicians have gained

extensive experience in performing US. In a patient with blunt trauma, a conventional US is frequently used to detect free intraperitoneal fluid [3]. Although the positive predictive value of an FAST examination is very high, a negative FAST cannot exclude injury. In previous studies, the sensitivity of US in detecting abdominal free fluid ranged from 63% to 99% and was significantly lower for parenchymal lesions. After all, 29% to 34% of patients with blunt trauma incur solid organ injury without hemoperitoneum [9].

A US contrast agent contains microbubbles of various gases, which generate contrast by scattering US from volume oscillations in the sound beam [10]. Recently developed US contrast agents contain a low-solubility gas such as sulfur hexafluoride, which forms more stable microspheres than first-generation agents composed of air within a rigid outer shell [10]. As compared with conventional US and CT, CEUS presents several advantages, including a greatly reduced rate of severe contrast agent-related adverse reactions, no nephrotoxocity, and higher temporal resolution than other Imaging techniques. In a large scale study, severe adverse reactions such as dyspnea, bronchospasm, and clouding of consciousness have been reported in only 0.0086% of abdominal applications, which are much less than the CT contrast media and comparable with magnetic resonance contrast agents. Nonserious adverse reactions, which were defined symptoms that were completely resolved without medical intervention, were reported in approximately

Fig. 2 A, Conventional US showed no definitely abnormal lesions in this liver. Both the radiologist and emergency physician found no trauma. B, On CEUS, diffuse, poorly enhanced areas were noted in the left lobe of the liver (arrowheads). This lesion was diagnosed as trauma of severe degree by both radiologist and emergency physician. C, Computed tomography showed ill-defined low-attenuating lesions involving more than half of the left lobe of the liver, suggesting severe liver trauma. D, Photograph of the gross specimen showed diffuse blunt trauma in the left lobe of the liver (arrowheads).

CT		Conventional US				Underassessment	Overassessment
Grade of liver trauma		No trauma	Minor	Moderate	Major
Emergency physician No trauma		2	0	0	0	0	0
Minor		2	0	0	0	2	0
Moderate		3	0	0	0	3	0
Major Radiologist No trauma		4 2	0 0	0 0	9 0	4 0	0 0
Minor		1	1	0	0	1	0
Moderate		3	0	0	0	3	0
Major		3	0	0	10	3	0
CEUS Emergency physician
No trauma	2		0	0	0	0	0
Minor	1		1	0	0	1	0
Moderate	0		2	1	0	2	0
Major Radiologist No trauma	0 2		0 0	0 0	13 0	0 0	0 0
Minor	1		1	0	0	1	0
Moderate	0		1	2	0	1	0
Major	0		0	0	13	0	0

0.12% of cases. These reactions included itching sensation, mild dizziness, moderate hypotension, and headache [4,11]. Because CEUS can be performed with a conventional US machine by just adding CEUS-specific sequence contrast pulse sequence, it is used increasingly to diagnose and characterize focal lesions in several solid organs (liver, kidney, spleen, and adrenal glands). As an added advantage, CEUS can be used in emergency medicine to evaluate blunt abdominal trauma [1,4]. Real-time CEUS can accurately determine the degree of injury by evaluating the extension of lesions and rapidly diagnose the active bleeding site by confirming the extravasation of contrast. Contrast-enhanced US can also be used for follow-up without problems from both ionizing radiation and the nephrotoxic effects of the CT contrast [12].

Table 1 Diagnostic performance of conventional US and CEUS

Previous studies show that FAST and conventional US can diagnose intraparenchymal hepatic trauma in approxi- mately 40% and 57 to 60% of cases, respectively. With CEUS, the detection rate for liver trauma increases significantly, to 80% to 100% [7,13]. These data are consistent with our results showing that sensitivity in detecting liver trauma increased from 61.1% to 94.4% for the radiologist and from 50% to 94.4% for the emergency physician after contrast enhancement of US. The US performance improved with contrast enhancement because the contrast material, interacting with the sound beam and the tissues, made the liver injury more conspicuous [14]. On conventional US, liver trauma may present a variable appearance due to hyperechogenicity, mixed echogenicity, and hypoechogenicity. As a result, minor trauma may be

overlooked through mixing and masking of the signal by adjacent normal liver parenchyma. After contrast enhance- ment, not only the echogenicity but also vascular perfusion can be evaluated. Areas of trauma show minimal or no enhancement compared with adjacent normal liver paren- chyma, which is strongly enhanced. The increase in contrast leads to greater sensitivity for detecting liver trauma [15]. Diffuse parenchymal injury is especially hard to detect on conventional US because an area of diffuse mixed echogenicity can be confused with liver parenchyma itself or with a diffuse underlying liver disease such as chronic hepatitis. With contrast enhancement, an area of Hepatic injury may appear as an area of nonperfusion or of relatively low enhancement. Diffuse liver injury can, thus, be diagnosed more easily. In our experiments, one diffuse grade 3 liver lesion was missed on conventional US but was accurately diagnosed on CEUS.

In an animal study, Tang et al [3] found 93.3% agreement between CEUS and CECT on the degree of hepatic injury, and CEUS grading showed a high level of concordance with CECT. Using CECT as a reference standard, these results suggested that CEUS could diagnose the degree of liver injury more accurately than conventional US. In our study, contrast enhancement of US significantly increased the agreement between CT and US, from 0.452 to 0.902 for the radiologist (P = .003) and from 0.369 to 0.858 for the emergency physician (P = .001). The ? statistic between the emergency physician and radiologist also increased from 0.867 to 0.955 after contrast enhancement of US. This result suggested that CEUS enables a more accurate evaluation of

liver trauma than conventional US by both the radiologist and the emergency physician.

Limitations

Our study encountered several limitations. First, the numbers of minor and moderate degree liver traumas induced were relatively small. Second, the human and rabbit livers differ significantly in their anatomy and size. Because the human liver is larger and anatomically more familiar to most physicians than the rabbit liver, CEUS may be more accurate in real patients with blunt liver trauma than data from this study indicate. This point will require further study. Third, although diagnosis of active bleeding is important in patients with blunt trauma, we could not evaluate the diagnostic performance and interobserver variability of CEUS for active bleeding because the rabbit liver and trauma sites were too small. Fourth, we used CT as a reference standard rather than the pathologic examination; thus, minute traumas, which are not seen on CT, could be missed in the evaluation. However, the minute traumas that even CT does not detect are not likely to affect a patient’s prognosis. Fifth, conventional US and CEUS were per- formed with interval of only several minutes, so there was a possibility that conventional US findings likely biased the assessment of the CEUS. Finally, only a single observer from each field (radiology and emergency medicine) participated in the experiment. A comparison of the diagnostic performance between multiple observers from the same or different fields should be performed in the future.

Conclusions

Contrast-enhanced US gave a more accurate diagnosis of liver trauma than conventional US by both the radiologist and the emergency physician and reduced interobserver variability between the 2.

Conflict of interest

The authors have no relevant financial interests or sources of financial support to disclose.

References

1. Tang J, Li W, Lv F, et al. Comparison of gray-scale contrast- enhanced ultrasonography with contrast-enhanced computed to- mography in different grading of blunt hepatic and splenic trauma: an animal experiment. Ultrasound Med Biol 2009;35: 566-75.
2. Valentino M, Serra C, Pavlica P, et al. Blunt abdominal trauma: diagnostic performance of contrast-enhanced US in children–initial experience. Radiology 2008;246:903-9.
3. Tang J, Wang Y, Mei X, et al. The value of contrast-enhanced gray- scale ultrasound in the diagnosis of hepatic trauma: an animal experiment. J Trauma 2007;62:1468-72.
4. Claudon M, Cosgrove D, Albrecht T, et al. Guidelines and good clinical practice recommendations for contrast enhanced ultrasound (CEUS)–update 2008. Ultraschall Med 2008;29:

28-44.

1. Moore EE, Cogbill TH, Jurkovich GJ, et al. Organ injury scaling: spleen and liver (1994 revision). J Trauma 1995;38: 323-4.
2. Seo TS, Oh JH, Lee DH, et al. Radiologic anatomy of the rabbit liver on hepatic venography, arteriography, portography, and cholangiography. Invest Radiol 2001;36:186-92.
3. Yekuo L, Shasha W, Xiansheng Z, et al. Contrast-enhanced ultrasound for Blunt hepatic trauma: an animal experiment. Am J Emerg Med 2010;28:828-33.
4. Cicchetti DV, Allison T. A new procedure for assessing reliability of scoring EEG sleep recordings. Am J EEG Technol 1971;11:101-9.
5. Valentino M, Serra C, Zironi G, et al. Blunt abdominal trauma: emergency contrast-enhanced sonography for detection of solid Organ injuries. AJR Am J Roentgenol 2006;186:1361-7.
6. Sontum PC. Physicochemical characteristics of Sonazoid, a new contrast agent for Ultrasound imaging. Ultrasound Med Biol 2008;34: 824-33.
7. Piscaglia F, Bolondi L. The safety of Sonovue in abdominal applications: retrospective analysis of 23188 investigations. Ultra- sound Med Biol 2006;32:1369-75.
8. Manetta R, Pistoia ML, Bultrini C, et al. Ultrasound enhanced with sulphur-hexafluoride-filled microbubbles agent (SonoVue) in the follow-up of mild liver and spleen trauma. Radiol Med 2009;114: 771-9.
9. Poletti PA, Platon A, Becker CD, et al. Blunt abdominal trauma: does the use of a second-generation sonographic contrast agent help to detect Solid organ injuries? AJR Am J Roentgenol 2004;183: 1293-301.
10. McGahan JP, Horton S, Gerscovich EO, et al. Appearance of solid organ injury with contrast-enhanced sonography in blunt abdominal trauma: preliminary experience. AJR Am J Roentgenol 2006;187: 658-66.
11. Bruce M, Averkiou M, Tiemann K, et al. Vascular flow and perfusion imaging with ultrasound contrast agents. Ultrasound Med Biol 2004;30:735-43.