High-Resolution Sonography for Distinguishing Neoplastic Gallbladder Polyps and Staging Gallbladder Cancer

Transcription

1 Gastrointestinal Imaging Original Research Kim et al. Sonography of the Gallbladder Gastrointestinal Imaging Original Research Jung Hoon Kim 1 Jae Young Lee 1 Jee Hyun Baek 1 Hyo Won Eun 2 Young Jae Kim 3 Joon Koo Han 1 Byung Ihn Choi 1 Kim JH, Lee JY, Baek JH, et al. Keywords: gallbladder cancer, gallbladder polyp, high-resolution sonography, sonography, transducers DOI: /JR Received September 26, 2013; accepted after revision May 14, Department of Radiology and Institute of Radiation Medicine, Seoul National University College of Medicine, 101 Daehang-no, Chongno-gu, Seoul , Korea. ddress correspondence to J. H. Kim (jhkim2008@gmail.com). 2 Department of Radiology, Severance Hospital, Yonsei University College of Medicine, Seoul, Korea. 3 Department of Radiology, Soonchunhyang University Hospital, Seoul, Korea. This article is available for credit. WEB This is a web exclusive article. JR 2015; 204:W150 W X/15/2042 W150 merican Roentgen Ray Society High-Resolution Sonography for Distinguishing Neoplastic Gallbladder Polyps and Staging Gallbladder Cancer OBJECTIVE. The purposes of this study were to compare staging accuracy of high-resolution sonography (HRUS) with combined low- and high-mhz transducers with that of conventional sonography for gallbladder cancer and to investigate the differences in the imaging findings of neoplastic and nonneoplastic gallbladder polyps. MTERILS ND METHODS. Our study included 37 surgically proven gallbladder cancer (T1a = 7, T1b = 2, T2 = 22, T3 = 6), including 15 malignant neoplastic polyps and 73 surgically proven polyps (neoplastic = 31, nonneoplastic = 42) that underwent HRUS and conventional transabdominal sonography. Two radiologists assessed T-category and predefined polyp findings on HRUS and conventional transabdominal sonography. Statistical analyses were performed using chi-square and McNemar tests. RESULTS. The diagnostic accuracy for the T category was T1a = 92 95%, T1b = 89 95%, T2 = 78 86%, and T3 = 84 89%, all with good agreement (κ = 0.642) using HRUS. The diagnostic accuracy for differentiating T1 from T2 or greater than T2 was 92% and 89% on HRUS and 65% and 70% with conventional transabdominal sonography. Statistically common findings for neoplastic polyps included size greater than 1 cm, single lobular surface, vascular core, hypoechoic polyp, and hypoechoic foci (p < 0.05). The value of HRUS in the differential diagnosis of a gallbladder polyp was more clearly depicted internal echo foci than conventional transabdominal sonography (39 vs 21). polyp size greater than 1 cm was independently associated with a neoplastic polyp (odds ratio = 7.5, p = 0.02). The UC of a polyp size greater than 1 cm was The sensitivity and specificity were 66.67% and 89.13%, respectively. CONCLUSION. HRUS is a simple method that enables accurate T categorization of gallbladder carcinoma. It provides high-resolution images of gallbladder polyps and may have a role in stratifying the risk for malignancy. lthough gallbladder cancer is a relatively rarely detected neoplasm, it has a poor prognosis. The accurate preoperative staging of gallbladder carcinoma is crucial because T category is the most important factor for determining a patient s therapy. Tumors confined within the mucosa or muscularis have a relatively favorable prognosis and may be cured by simple or extended cholecystectomy. However, lesions spreading beyond the muscularis have been associated with a poor therapeutic outcome [1, 2]. Furthermore, because gallbladder polyps are commonly detected on sonography, accurate differentiation between neoplastic and nonneoplastic polyps is important because, although the vast majority of gallbladder polyps are nonneoplastic, neoplastic polyps can also be identified during routine abdominal ultrasound examinations [3, 4]. Transabdominal sonography is the preferred imaging modality initially used to evaluate the gallbladder. s ultrasound technology continues to improve and the use of ultrasound examination increases, more gallbladder lesions are being detected. lthough transabdominal sonography has been found to be useful for diagnosing gallstones, gallbladder sludge, acute cholecystitis, and chronic cholecystitis, it is of limited use not only for accurately staging gallbladder cancer but also for differentiating nonneoplastic from neoplastic polyps [5]. Thus, according to some reports, endoscopic sonography is considered superior to transabdominal sonography for imaging the gallbladder. During endoscopic sonography, the proximity of the W150 JR:204, February 2015

2 Sonography of the Gallbladder transducer to the target tissue and the use of higher-frequency ultrasound between 5 and 12 MHz have led to greater image resolution. However, endoscopic sonography is an invasive procedure, the success of which highly depends on the operator s skills. Routine transabdominal sonography usually uses a low-frequency transducer from Consecutive patients underwent HRUS (n = 1717) Main study cohort (n = 110) Included patients underwent both HRUS and cholecystectomy (n = 601) Excluded patients with: Chronic cholecystitis (n = 475) cute cholecystitis (n = 8) Metastasis (n = 3) Inadequate sonography (n = 5) Gallbladder cancer (n = 37) Gallbladder polyp (n = 73) Nonpolypoid-type gallbladder cancer (n = 22) Polypoid-type gallbladder cancer: malignant neoplastic polyp (n = 15) Gallbladder cancer (n = 37) Neoplastic polyp (n = 46) Benign neoplastic polyp (n = 31) Nonneoplastic polyp (n = 42) Nonneoplastic polyp (n = 42) Fig. 2 Flowchart shows study group inclusion process. HRUS = high-resolution ultra sonography. B C Fig. 1 Sonography findings in normal gallbladder wall in 45-year-old woman., Conventional transabdominal sonography image using low-frequency transducers shows single or two layers of gallbladder wall (arrows). B, High-resolution sonography (HRUS) using high-frequency transducers can better visualize gallbladder wall layers. HRUS image depicts three layers of gallbladder wall including innermost hyperechoic layer (white arrow), middle thin hypoechoic layer (arrowheads), and outermost hyperechoic layer (black arrows). These three layers correspond to mucosa, muscle layer, and subserosa plus serosa, respectively. C, Histopathologic image shows layers of gallbladder wall. Black, red, yellow, and blue correspond to mucosa, muscle layer, perimuscular connective tissue, and serosa, respectively. (H and E, original magnification 20) 2 to 5 MHz because most abdominal organs are located deep within the human body. Low-frequency probes can be used to penetrate deep into large anatomic areas, although at the cost of somewhat lower resolution. lternatively, high-frequency transducers have better resolution, although they are less deep in penetration. High-resolution sonography (HRUS) use for the gallbladder is a technology that uses both lowand high-frequency transducers during the evaluation of the gallbladder. Because highfrequency, 5- or 7-MHz transducers have better resolution, HRUS can better visualize the gallbladder wall layers and can more accurately depict the internal echo change of polyps than can conventional transabdominal sonography using low-frequency transducers (Fig. 1). ccording to our previous studies, HRUS provides successful performance when evaluating gallbladder cancer and differentiating gallbladder cancer from adenomyomatosis [6, 7]. To our knowledge, there have only been a few studies that assessed the diagnostic value of HRUS for the evaluation of gallbladder cancer or polyps [6, 7], and there are no studies that evaluated and compared the diagnos- JR:204, February 2015 W151

3 Kim et al. tic value of HRUS with that of conventional transabdominal sonography, pathology, and the published literature reports on endoscopic sonography for detecting gallbladder cancer or polyps. Therefore, the purposes of our study were to evaluate and compare the staging accuracy of HRUS with that of conventional transabdominal sonography, pathology, and the published literature reports on endoscopic sonography for detecting gallbladder cancer and to investigate the differences in the imaging findings of neoplastic gallbladder polyps from those of nonneoplastic gallbladder polyps. Materials and Methods Patients Our institutional review board approved this retrospective study and did not require patient approval or informed consent for review of the patient images and medical records. Using computerized searches of our pathology and radiology information systems for the time period between December 2010 and September 2012, we identified 1717 consecutive patients who had undergone HRUS for gallbladder examination that combined the use of low- and high-mhz transducers. Of these patients, 1116 were excluded because they had not undergone cholecystectomy. Of the 601 patients who underwent both HRUS and cholecystectomy, 491 were excluded for one of the following reasons: chronic cholecystitis including xanthogranulomatous cholecystitis and adenomyomatosis (n = 475), acute cholecystitis (n = 8), metastasis (n = 3), and inadequate sonography images (n = 5). The gallbladder diseases that met the criteria on pathology reports included gallbladder cancer; benign neoplastic polyps, including tubular adenoma, tubulopapillary adenoma, and intracystic papillary neoplasm; and nonneoplastic polyps, including cholesterol polyps and inflammatory polyps. Finally, 110 consecutive patients (mean age [± SD], 56.1 ± 13.4 years; range, years) were included in this study (Fig. 2); these patients included 51 men (mean age, 58.7 ± 13.5 years; range, years) and 59 women (mean age, 58.8 ± 13.4 years; range, years). The patients with gallbladder cancer (n = 37) included 15 men and 22 women with a mean age of 66.6 ± 8.8 years (range, years). The patients with benign neoplastic polyps (n = 31) included 18 men and 13 women with a mean age of 60.5 ± 11.7 years (range, years). The patients with nonneoplastic polyps (n = 42) included 18 men and 24 women with a mean age of 52.7 ± 13.4 years (range, years). These patients underwent laparoscopic cholecystectomy (n = 69), extended cholecystectomy (n = 29), or open cholecystectomy (n = 12). B C Fig. 3 T1a category polypoid-type cancer involving fundus of gallbladder in 74-year-old man., Conventional sonography image shows polypoid lesion in fundus that causes irregularity of outer hyperechoic layer of gallbladder wall (arrows). Both reviewers considered this lesion to be T2 category gallbladder cancer on conventional sonography. B, High-resolution Doppler sonography image with high-frequency transducers shows feeding vessels at neck of polyp (arrows). C, High-resolution sonography image shows polypoid lesion with preserved inner hypoechoic layers (arrows). Both reviewers considered this lesion to be T1a category gallbladder cancer on high-resolution sonography. Sonography Techniques Sonography examination of the gallbladder was performed using a LOGIQ 9 ultrasound unit (GE Healthcare) with a convex low-mhz transducer with bandwidth of MHz (4C, GE Healthcare) and a linear high-mhz transducer with bandwidth of MHz (7 L, GE Healthcare) in all patients to assess the entire gallbladder. ll sonography examinations were independently performed by one of our three clinically experienced radiologists with 17, 16, and 12 years of clinical experience in abdominal sonography. The gallbladder was first carefully investigated using the intercostal or subcostal approach with the patient in a supine or decubitus position and using a convex low-mhz transducer with and without harmonic imaging. fter careful investigation, we also applied color Doppler sonography using a low-mhz transducer. fter the conventional sonography examination, sonography using linear high-mhz transducers with and without harmonic imaging was routinely performed by the same radiologists who performed the conventional sonography. fter careful investigation, we also performed color Doppler sonography using linear transducers. We designated the sonography performed using a convex low-mhz transducer as conventional sonography and the sonography using both a convex low-mhz transducer and a linear high-mhz W152 JR:204, February 2015

4 Sonography of the Gallbladder transducer as high-resolution sonography. The low-mhz transducer settings were as follows: frequency, 4 MHz; gain, 27 33%; dynamic range, 69; and frame rate, per second. We used realtime spatial compound imaging techniques that is, three, compound beams per frame and milddegree speckle-reduction techniques with and without harmonic imaging. The high-mhz transducer settings were as follows: frequency, 6 7 MHz; gain, 27 30%; dynamic range, 66 72; and frame rate, 14 per second. To optimize the evaluation of the gallbladder using the linear probe in real time, we used the spatial compound imaging techniques that is, three and five compound beams per frame and the mild-degree specklereduction techniques with and without harmonic imaging. While scanning the whole gallbladder, we frequently froze and stored the images, which we then selectively stored as a cine clip, depending on the operators decision. ll of the images were transferred to our local PCS. Image Interpretation Two board-certified radiologists, each with 16 years of clinical experience in abdominal sonography, analyzed the sonography images retrospectively and independently. ll images were reviewed on a PCS workstation (M-view, Marotech). The conventional transabdominal sonography images were analyzed first and the HRUS images were reviewed 2 weeks later to minimize any learning bias. Both reviewers knew that the patients had undergone surgery for gallbladder cancer or polyps, although they were blinded to the radiology reports and histopathologic diagnoses. The two radiologists retrospectively and independently assessed the T category seen in each conventional transabdominal sonography and HRUS image set. We used the T categorization of gallbladder cancer provided by the merican Joint Committee on Cancer (JCC), 7th edition [8]. The T category of gallbladder cancer is defined according to the depth of tumor invasion. Category T1a is when tumor invades the laminar propria. In category T1b, tumor invades the muscular layer. In category T2, tumor invades perimuscular connective tissue. In category T3, tumor perforates the serosa. The sonography definitions of the T categories are as follows: T1a is a polypoid or focal wall-thickening lesion with preserved inner hypoechoic layers; T1b is a polypoid or focal wall-thickening lesion that causes irregularity of the inner hypoechoic layer of the gallbladder wall; T2 is a polypoid or focal wall-thickening lesion B B Fig. 4 T2 category polypoid-type cancer involving body of gallbladder in 64-year-old man., Conventional sonography image shows polypoid lesion in body with preserved outer hyperechoic layer (arrow). Both reviewers considered this lesion to be T1b category gallbladder cancer on conventional sonography. B, High-resolution sonography image with highfrequency transducers shows polypoid lesion that causes irregularity of outer hyperechoic layer of gallbladder wall (arrowheads). Both reviewers considered this lesion to be T2 category gallbladder cancer on high-resolution sonography. that causes irregularity of the outer hyperechoic layer of the gallbladder wall; and T3 is a group of tumors that disrupts the entire layer structure of the gallbladder wall or extends to the liver [9]. The following imaging variables for gallbladder polyps were also analyzed using HRUS images: multiplicity (single or multiple), maximum diameter of the largest polyp (mm), shape (pedunculated or sessile), surface contour (smooth or lobulated), presence or absence of a vascular core seen on color Doppler sonography, presence or absence of gallbladder stone, and internal echo level (hypoechoic or iso- to hyperechoic). Pedunculated polyp was defined as the maximum diameter of a polyp and one which is larger than the base of the polyp. The reference for the internal echo level was the hyperechoic layer of the gallbladder wall. We also analyzed the internal echo pattern of gallbladder polyps seen in each of the conventional transabdominal sonography and HRUS image sets. The internal echo pattern of gallbladder polyps included the presence or absence of hyperechoic foci or hypoechoic foci, and the internal echo pattern was determined by whether it was heterogeneous or homogeneous. The heterogeneous internal echo pattern was defined as that seen in a polyp containing hyper- Fig. 5 T3 category focal wall thickening type cancer involving fundus of gallbladder in 62-year-old woman. and B, Both conventional sonography () and highresolution sonography (B) show focal wall thickening in fundus of gallbladder that disrupts entire layer structure of gallbladder wall and extends to liver (arrows, and arrowheads, B). Both reviewers considered this lesion to be T3 category. JR:204, February 2015 W153

5 Kim et al. TBLE 1: Sensitivity, Specificity, and Diagnostic ccuracy for Determining T Category Using High-Resolution Sonography and Conventional Transabdominal Sonography Conventional Transabdominal Sonography High-Resolution Sonography Sensitivity 95% CI Specificity 95% CI Diagnostic ccuracy 95% CI Sensitivity 95% CI Specificity 95% CI Diagnostic ccuracy 95% CI Reader Category R1 T1a 0.43 (3/7) (22/30) (25/37) (5/7) (29/30) (34/37) T1b 0.5 (1/2) (29/35) (30/37) (1/2) (34/35) (35/37) T (9/22) (13/15) (22/37) (18/22) (14/15) (32/37) T (5/6) (28/31) (33/37) (5/6) (26/31) (31/37) R2 T1a 0.14 (1/7) (28/30) (28/37) (6/7) (29/30) (35/37) T1b 0.5 (1/2) (27/35) (28/37) (2/2) (31/35) (33/37) T (10/22) (11/15) (21/37) (15/22) (14/15) (29/37) T (5/6) (25/31) (30/37) (5/6) (28/31) (33/37) Note Data in parentheses are number/total. echoic or hypoechoic foci. If the patient had multiple polyps, we analyzed the image findings of the largest polyp. The maximum diameter of the largest polyp was measured three times, after which the mean value was obtained. When there was a discrepancy in the two radiologists diagnoses, the third reviewer with 12 years of clinical experience in abdominal sonography made the final decision. Statistical nalysis Comparison of the diagnostic accuracy of the T categorization was made using chi-square and McNemar tests. The chisquare test was used to assess the association of each sonography finding on the basis of the consensus interpretation and the histopathologic diagnosis of neoplastic or nonneoplastic polyps. Interobserver agreement was evaluated using weighted kappa statistics with linear weights and was interpreted as follows: poor, less than 0.20; fair, ; moderate, ; good, ; and excellent, 0.80 or greater. ccording to the statistically significant sonography findings, a multivariate stepwise logistic regression model was used to identify the most accurate diagnostic predictors. Variables with a p value less than 0.05, determined by univariate analysis, were included in the logistic regression model. Statistical analyses were performed using ROC analysis to assess the diagnostic performance of the polyp size. ll statistical analyses were performed using commercially available statistical software (SPSS, version 14.0, and MedCalc, version 6.15), and a p value less than 0.05 was considered to indicate a statistically significant difference. Results Pathologic Characteristics Histopathologic examination confirmed that 37 patients had gallbladder cancer. In terms of the categorization, seven patients (19%) had T1a tumors, two patients (5%) had T1b tumors, 22 patients (59%) had T2 tumors, and six patients (16%) had T3 tumors. The tumors were located at the fundus in seven patients, in the fundus and body in five patients, in the body in five patients, in the body and neck in six patients, in the neck in three patients, and in the entire gallbladder in 11 patients. The histopathologic analysis of the gallbladder polyps showed that there were 46 patients with a neoplastic polyp and 42 patients with a nonneoplastic polyp. The neoplastic polyps included not only malignant neoplastic polyps (n = 15), which is gallbladder cancer of the polypoid type counted in both the gallbladder cancer group and the neoplastic polyp group, but also benign neoplastic polyp, which includes high-grade tubular adenoma (n = 3), low-grade tubular adenoma (n = 17), high-grade tubulopapillary adenoma (n = 3), low-grade tubulopapillary adenoma (n = 5), and intracystic papillary neoplasm with a high-grade intraepithelial neoplasm (n = 3). The nonneoplastic polyps included cholesterol polyps (n = 40) and inflammatory polyps (n = 2). Staging ccuracy of Conventional Transabdominal Sonography and HRUS for Gallbladder Cancer In the 37 gallbladder cancer patients, the diagnostic accuracy for determining the T category using HRUS was higher than that using conventional transabdominal sonography, regardless of the T category. The sensitivity, specificity, and diagnostic accuracy for determining the T category using HRUS and conventional transabdominal sonography are shown in Table 1. The diagnostic accuracy of HRUS for determining the T1a category was 92% for reader 1 and 95% for reader 2, the T1b category was 95% for reader 1 and 89% for reader 2, the T2 category was 86% for reader 1 and 78% for reader 2, and the T3 category was 84% for reader 1 and 89% for reader 2. On the other hand, the diagnostic accuracy of conventional transabdominal sonography for determining the T1a category was 68% for reader 1 and 78% for reader 2, the T1b category was 81% for reader 1 and 76% for reader 2, the T2 category was 59% for reader 1 and 57% for reader 2, and the T3 category was 89% for reader 1 and 81% for reader 2 (Fig. 3). The interobserver agreement was good (κ = 0.642) using HRUS and moderate (κ = 0.569) using conventional transabdominal sonography. Compared with the pathology assessment, HRUS correctly categorized 78.3% (n = 29/37) for reader 1 and 75.6% (n = 28/37) for reader 2. Table 2 summarizes the T category accura- W154 JR:204, February 2015

6 Sonography of the Gallbladder cy of HRUS and conventional transabdominal sonography compared with the pathologic category. With HRUS, the overstaging rate was 16.2% (n = 6/37) for reader 1 and 10.8% (n = 4/37) for reader 2. The downstaging rate was 5.4% (n = 2/37) for reader 1 and 13.5% (n = 5/37) for reader 2. On the other hand, with conventional transabdominal sonography, the overstaging rate was 21.6% (n = 8/37) for reader 1 and 32.4% (n = 12/37) for reader 2. The downstaging rate was 29.7% (n = 11/37) for reader 1 and 21.6% (n = 8/37) for reader 2. With HRUS, the overstaging of T2 to T3 was common for both readers 1 (n = 3) and 2 (n = 3), and the downstaging of T2 to T1 was also common for both readers 1 (n = 1) and 2 (n = 4). The diagnostic accuracy of HRUS for differentiating T1 from T2 or greater than T2 (92% for reader 1 and 89% for reader 2) was higher than that using conventional transabdominal sonography (65% for reader 1 and 70% for reader 2) (Fig. 4). However, the diagnostic accuracy for differentiating T3 from less than T3 was similar in HRUS (84% for reader 1 and 89% for reader 2) and conventional transabdominal sonography (89% for reader 1 and 81% for reader 2) (Fig. 5). The sensitivity, specificity, and diagnostic accuracy for differentiating the T1 or T3 category using HRUS and conventional transabdominal sonography are shown in Table 3. The sonography examination took approximately 613 ± 97.4 seconds using conventional transabdominal sonography and approximately 632 ± seconds using HRUS. Sonography Findings for Differentiating Neoplastic Polyps From Nonneoplastic Polyps The mean diameter (± SD) of the neoplastic polyps we measured was ± 7.38 mm, whereas the mean diameter of the nonneoplastic polyps was 9.7 ± 2.87 mm. Therefore, the neoplastic and nonneoplastic polyps differed statistically in size (p < 0.000). TBLE 2: T Category ccuracy of Conventional Transabdominal Sonography and High-Resolution Sonography Compared With Pathologic Category Parameter The sonography findings for both neoplastic and nonneoplastic polyps of the gallbladder are summarized in Table 4. Statistically common sonography findings for neoplastic polyps included size greater than 1 cm, single polyp, lobulated surface contour, presence of a vascular core seen on color Doppler sonography, hypoechoic internal echo of the polyp, and hypoechoic foci within the polyp (p < 0.05). Statistically common sonography findings for nonneoplastic polyps included size less than 1 cm, multiple polyps, T Category Pathologic Category pt1a pt1b pt2 pt3 Total Reader 1 with conventional transabdominal sonography T1a T1b T T Total Reader 1 with high-resolution sonography T1a T1b T T Total Reader 2 with conventional transabdominal sonography T1a T1b T T Total Reader 2 with high-resolution sonography T1a T1b T T Total Note Data are in bold along diagonal line that would reflect perfect performance. smooth surface contour, iso- to hyperechoic internal echo of the polyp, and hyperechoic foci within the polyp (p < 0.05). HRUS depicted the internal echo pattern of gallbladder polyps more clearly than conventional transabdominal sonography; HRUS detected 39 hypoechoic foci within neoplastic polyps (Fig. 6) and 29 hyperechoic foci within nonneoplastic polyps; however, conventional transabdominal sonography only detected 21 hypoechoic foci within neoplastic polyps and 21 hyperechoic foci within nonneoplas- TBLE 3: Sensitivity, Specificity, and Diagnostic ccuracy for Differentiating T1 or T3 Category Using Conventional Transabdominal Sonography and High-Resolution Sonography Reader R1 R2 T Category Conventional Transabdominal Sonography Sensitivity Specificity PPV NPV High-Resolution Sonography Diagnostic ccuracy Sensitivity Specificity PPV NPV Diagnostic ccuracy T1 vs > T < T2 vs T T1 vs > T < T2 vs T Note PPV = positive predictive value, NPV = negative predictive value. JR:204, February 2015 W155

7 Kim et al. TBLE 4: Sonography Findings for Neoplastic and Nonneoplastic Polyps of Gallbladder Parameter tic polyps. Each sonography finding of a neoplastic or nonneoplastic polyp showed moderate to excellent interobserver agreement (κ = ). In the multivariate logistic regression analysis using statistically common sonography findings for neoplastic polyps (p < 0.05), a polyp size greater than 1 cm was an independent positive predictor of a neoplastic polyp (odds ratio, 7.552; 95% CI, ). Regarding the diagnostic performance for determining polyp size, the area under the ROC curve was (95% CI, ; p < ). The sensitivity and specificity were 66.67% and 89.13% with a cutoff value of 1 cm. Findings Discussion HRUS is easy to use after our routine transabdominal sonography examination. The combined use of low- and high-frequency transducers improves the diagnostic performance of sonography for both staging gallbladder cancer and characterizing gallbladder polyps. In our study, the diagnostic accuracy of HRUS for determining the T category of gallbladder cancer compared with that of pathology was better than that of conventional transabdominal sonography, especially when differentiating category T1 from T2 or greater (92% and 89% vs 65% and 70%); however, the diagnostic accuracy for differentiating category T3 and lower was similar for HRUS and conventional transabdominal sonography (84% and 89% vs 89% and 81%). ccording to published reports, endoscopic sonography is considered to be superior to transabdominal sonography for imaging the gallbladder because of the higher operating frequency ( MHz) that can provide high-resolution images of small lesions. Fujita et al. [9] first proposed the endoscopic sonography criteria for the T categorization of gallbladder cancer: type is pedunculated, has a fine nodular surface, and preserves the outer hyperechoic and inner hypoechoic layers; type B consists of Neoplastic Polyp (n = 46) Nonneoplastic Polyp (n = 42) κ p Both conventional transabdominal sonography and high-resolution sonography Size (mm) ± ± 2.87 < Multiplicity Single Multiple Shape Pedunculated Sessile 6 2 Surface < Smooth Lobular Vascular core < Stone Internal echogenicity Conventional transabdominal sonography Hyperechoic foci Hypoechoic foci < Internal echo pattern Homogeneous Heterogeneous High-resolution sonography Hyperechoic foci < Hypoechoic foci < Internal echo pattern Homogeneous 2 8 Heterogeneous broad-based masses or areas of wall thickening, once again associated with the normal gallbladder wall structure; type C includes those seen in broad-based lesions that cause irregularity of the outer hyperechoic layer; and type D shows disruption of the entire layer structure. Those authors concluded that all type tumors are confined to the mucosa (Tis T1a) and that type B lesions invade at varying depths between the mucosa and the subserosa (T1 2). Type C tumors invade the subserosa or beyond (T2 T3), and type D tumors invade beyond the serosa (T3 4). fter their report, to assess the accuracy of each T category, Sadamoto et al. [10] retrospectively analyzed endoscopic sonography and the histopathologic findings in 41 patients with gallbladder cancer. The accuracies for the endoscopic sonography classification as type corresponding to Tis, type B to T1, type C to T2, and type D to T3 4 were, respectively, 100%, 75.6%, 85.3%, and 92.7%. ccording to a previous study performed at W156 JR:204, February 2015

8 Sonography of the Gallbladder B C Fig. 6 Low-grade tubular adenoma involving fundus of gallbladder in 66-year-old woman., Conventional sonography image shows polypoid lesion in fundus of gallbladder (arrows). Size of polyp is measured as 2.8 cm. Conventional sonography shows hypoechoic internal echo of polyp without any hypo- or hyperechoic foci within polyp. B, High-resolution Doppler sonography image with high-frequency transducers shows feeding vessels at center of polyp (arrowheads). C, High-resolution sonography image shows hypoechoic foci within polyp, suggesting vascular core (arrow and arrowheads). our institution, the diagnostic accuracy of HRUS is similar to that of endoscopic sonography [6]. We followed 144 patients who had 1-cm gallbladder polyps and who underwent cholecystectomy. Of these 144 patients, there were 29 (20.2%) with gallbladder cancer. When evaluating whether these modalities enabled prediction of the depth of the gallbladder cancer invasion, we could not detect any significant differences between the diagnostic accuracy of endoscopic sonography (55.5%) and that of HRUS (62.9%). In our study, the diagnostic accuracy of HRUS for determining the T1a category was 92% and 95%, for the T1b category it was 95% and 89%, for the T2 category it was 86% and 78%, and for the T3 category it was 84% and 89%. Compared with pathology, HRUS correctly categorized 78.3% (n = 29/37) and 75.6% (n = 28/37). On the basis of these data, HRUS was found to have diagnostic accuracy comparable to that of endoscopic sonography for obtaining a differential diagnosis of the T category of gallbladder cancer. In addition, the diagnostic accuracy of HRUS for differentiating T1 from T2 or greater was 92% and 89%. Therefore, it is expected that HRUS is a simple method that enables accurate T categorization of gallbladder carcinoma. During endoscopic sonography, the proximity of the transducer to the target tissue and the use of higher frequency ultrasound between 5 and 12 MHz have led to greater image resolution. However, endoscopic sonography is an invasive procedure, the success of which highly depends on the operator s skills. In addition, because the patient is sedated during endoscopic sonography, there is a risk for cardiac or pulmonary complications during the procedure. Furthermore, endoscopic sonography has risk for complications such as bleeding, perforation, and infection. lthough transabdominal sonography is safer and easier to use than endoscopic sonography, it has inherent limitations because of its use of low-frequency transducers ranging from 2 to 5 MHz. Because HRUS uses both lowand high-frequency transducers, it provides high-resolution images during gallbladder evaluation. ccording to the results of published reports [9 11], endoscopic sonography studies have shown the three layers of the gallbladder wall the innermost hyperechoic layer, the middle hypoechoic layer, and the outermost hyperechoic layer. These three layers correspond to the mucosa, the muscle layer, and the subserosa plus serosa, respectively [11, 12]. By using both low- and high-frequency transducers during transabdominal sonography of the gallbladder, we succeeded in delineating the layers of the gallbladder wall. On the basis of our clinical experience, we believe that it is important to apply the ultrasonic beam perpendicular to the plane of the gallbladder wall because the gallbladder wall is relatively thin. Otherwise, the wall of the gallbladder may not be clearly visible on the ultrasound images. Furthermore, it is easy to then apply the ultrasonic beam after a routine transabdominal sonography procedure. HRUS markedly decreased both overstaging (21.6% and 32.4% for conventional transabdominal sonography vs 16.2% and 10.8% for HRUS) and downstaging rates (29.7% and 21.6% for conventional transabdominal sonography vs 5.4% and 13.5% for HRUS). In our experience, harmonic imaging was found to be helpful for determining the depth of tumor invasion in most patients because this technique enables more detailed visualization of the layers of the gallbladder wall. However, because harmonic imaging is limited in terms of its ability to visualize deep-seated lesions, both harmonic and fundamental imaging should be used. JR:204, February 2015 W157

9 Kim et al. HRUS can be helpful in further distinguishing nonneoplastic from neoplastic gallbladder polyps. In our study, HRUS more clearly depicted the internal echo pattern of gallbladder polyps than conventional transabdominal sonography. HRUS detected 39 hypoechoic foci within a neoplastic polyp, whereas conventional transabdominal sonography detected only 21 hypoechoic foci. The primary goal of gallbladder polyp management is to differentiate nonneoplastic polyps from neoplastic polyps. In our study, statistically common sonography findings of neoplastic polyps included size greater than 1 cm, single polyp, lobulated surface contour, presence of a vascular core seen on color Doppler sonography, hypoechoic internal echo of the polyp, and hypoechoic foci within the polyp (p < 0.05). polyp size greater than 1 cm was an independent positive predictor for a neoplastic polyp. These results are similar to those of previous reports indicating that the common sonography findings of neoplastic polyps are sessile polyps with a hypoechoic to isoechoic internal echo and without echogenic spots, microcysts, or comet tail artifacts. Risk factors for neoplastic polyps include polyps larger than 1 cm, older patients, solitary lesions, coexisting gallstones, and presence of symptoms. mong these risk factors, polyp size is known as the usual predictor of potential malignancy, particularly among patients more than 50 years old [13 16]. Because HRUS can more accurately delineate the internal echo pattern of gallbladder polyps, it may have an important role in stratifying polyps as having a high or low risk for malignancy. In our previous study [6], we prospectively analyzed 115 (79.8%) patients with nonneoplastic gallbladder polypoid lesions and 29 (20.2%) patients with gallbladder cancers, all greater than 1 cm in size. In this study, the diagnostic sensitivity of HRUS for detecting malignancy was comparable to that of endoscopic sonography (90% vs 86%, respectively). Corwin et al. [17] retrospectively analyzed 146 incidentally detected gallbladder polyps. The risk of cancer in incidentally detected polyps is also extremely low, with none seen at diameters smaller than 7 mm. Those authors proposed that incidentally detected gallbladder polyps measuring 6 mm or less in diameter require no additional follow-up. In our study, a polyp size greater than 1 cm was an independent positive predictor of malignancy, and the sensitivity and specificity were 66.67% and 89.13% when used as the cutoff value for polyps 10 mm in size. Therefore, we propose that gallbladder polyps measuring 1 cm in size require close follow-up because of the increasing risk of cancer. Our current study has some limitations. First, it was a retrospective study, which therefore has the possibility of selection bias. In addition, because we could only include patients with surgically resected gallbladder polyps, the study population did not reflect the entire spectrum of gallbladder cancer and polyps. For example, most small benign-appearing polyps and advanced gallbladder cancers are usually not treated using surgery. Second, because only a small number of our study patients underwent endoscopic sonography, we could not make a clear comparison between HRUS with endoscopic sonography. Third, in our study, although HRUS more clearly depicted internal echo foci than conventional transabdominal sonography, HRUS still had a limitation for the differentiation of neoplastic polyps. Fourth, the sonography techniques we used depend on the operator s skill and the patient s body habitus. In our study, although only five patients were excluded because of inadequate sonography images, not all of our study patients could undergo HRUS using a limited sonic window. Even with experienced radiologists, proper body habitus of the patients is needed to obtain high-quality images with better diagnostic value when using HRUS, which is a diagnostic tool capable of making accurate staging of gallbladder cancer and a differential diagnosis of gallbladder polypoid lesions compared with endoscopic sonography. In our hospital, HRUS has been performed routinely on all patients with suspected neoplastic gallbladder polyps as the firstline diagnostic workup for patients planning cholecystectomy. HRUS shows better performance than conventional transabdominal sonography for accurately depicting the early gallbladder carcinoma, and because tumors confined within the mucosa have a favorable prognosis, they may be cured by simple cholecystectomy. HRUS plays an important role in diagnosing early gallbladder carcinoma. However, we need further prospective studies for the evaluation of the clinical impact of HRUS. We expect that HRUS will become an invaluable diagnostic modality for making differential diagnoses of gallbladder lesions. In conclusion, to our knowledge, there are no dedicated studies regarding the diagnostic value of the combined use of both low- and high-frequency transducers during the evaluation of gallbladder cancer or polyps. ccording to our results, HRUS is a simple method that enables accurate T categorization of gallbladder carcinoma. It provides high-resolution images of gallbladder polyps and may have a role in stratifying the risk for malignancy. Therefore, we expect that HRUS will become an invaluable diagnostic modality for making accurate diagnoses of gallbladder carcinoma and gallbladder polyps. cknowledgments We thank Bonnie Hami and Jeehyun Kim for their editorial assistance in the preparation of this article. References 1. Dutta U. Gallbladder cancer: can newer insights improve the outcome? J Gastroenterol Hepatol 2012; 27: Jayaraman S, Jarnagin WR. Management of gallbladder cancer. Gastroenterol Clin North m 2010; 39: Gallahan WC, Conway JD. Diagnosis and management of gallbladder polyps. Gastroenterol Clin North m 2010; 39: Myers RP, Shaffer E, Beck PL. Gallbladder polyps: epidemiology, natural history and management. Can J Gastroenterol 2002; 16: Gore RM, Yaghmai V, Newmark GM, Berlin JW, Miller FH. Imaging benign and malignant disease of the gallbladder. Radiol Clin North m 2002; 40: Jang JY, Kim SW, Lee SE, et al. Differential diagnostic and staging accuracies of high resolution ultrasonography, endoscopic ultrasonography, and multidetector computed tomography for gallbladder polypoid lesions and gallbladder cancer. nn Surg 2009; 250: Joo I, Lee JY, Kim JH, et al. Differentiation of adenomyomatosis of the gallbladder from early-stage, wall-thickening-type gallbladder cancer using highresolution ultrasound. Eur Radiol 2013; 23: Edge SB, Compton CC. The merican Joint Committee on Cancer: the 7th edition of the JCC cancer staging manual and the future of TNM. nn Surg Oncol 2010; 17: Fujita N, Noda Y, Kobayashi G, Kimura K, Yago. Diagnosis of the depth of invasion of gallbladder carcinoma by EUS. Gastrointest Endosc 1999; 50: Sadamoto Y, Kubo H, Harada N, Tanaka M, Eguchi T, Nawata H. Preoperative diagnosis and staging of gallbladder carcinoma by EUS. Gastroin- W158 JR:204, February 2015

10 Sonography of the Gallbladder test Endosc 2003; 58: Fujimoto T, Kato Y, Kitamura T, Hiratsuka T. Hypoechoic area as an ultrasound finding suggesting subserosal invasion in polypoid carcinoma of the gall bladder. Br J Radiol 2001; 74: Lu MD, Hirata T, Nishihara K, Yamasaki K, Nakayama F. Improved delineation of the gallbladder wall with ultrasonography: its value in assessment of the depth of carcinoma invasion. J Clin Ultrasound 1991; 19: ldouri Q, Malik HZ, Waytt J, et al. The risk of gallbladder cancer from polyps in a large multiethnic series. Eur J Surg Oncol 2009; 35: Chattopadhyay D, Lochan R, Balupuri S, Gopinath B, Wynne K. Outcome of gall bladder polypoidal lesions detected by transabdominal ultrasound scanning: a nine year experience. World J Gastroenterol 2005; 11: Cho JH, Park JY, Kim YJ, et al. Hypoechoic foci on EUS are simple and strong predictive factors for neoplastic gallbladder polyps. Gastrointest Endosc 2009; 69: Sugiyama M, Xie X-Y, tomi Y, Saito M. Differential diagnosis of small polypoid lesions of the gallbladder: the value of endoscopic ultrasonography. nn Surg 1999; 229: Corwin MT, Siewert B, Sheiman RG, Kane R. Incidentally detected gallbladder polyps: is followup necessary? Long-term clinical and US analysis of 346 patients. Radiology 2011; 258: FOR YOUR INFORMTION This article is available for CME and Self-ssessment (S-CME) credit that satisfies Part II requirements for maintenance of certification (MOC). To access the examination for this article, follow the prompts. JR:204, February 2015 W159