| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
|
by the American Institute of Ultrasound in Medicine J Ultrasound Med 22:327-334 • 0278-4297 Correlation Between the Echogenicity of Dysplastic Nodules and Their Histopathologically Determined Fat ContentDepartment of Radiology and Center for Imaging Science (M.J.K., J.H.L., S.J.L., S.H.K., W.J.L., H.K.L., J.M.P.) and Department of Pathology (C.K.P.), Samsung Medical Center, Sungkyunkwan University School of Medicine, Seoul, South Korea. Address correspondence and reprint requests to Jae Hoon Lim, MD, Department of Radiology, Samsung Medical Center, 50 Ilwon-dong, Kangnam-ku, Seoul 135-710, South Korea.
Objective. To correlate the echogenicity of dysplastic nodules in cirrhotic liver with the difference in fat content between lesions and surrounding liver. Methods. This retrospective study involved 65 histopathologically proved dysplastic nodules (39 high grade and 26 low grade). Their echogenicity compared with that of surrounding parenchyma was determined sonographically, and differences in the proportions of fat globules contained in the nodules and in surrounding liver tissue were evaluated histopathologically. The sonographic and histopathologic findings were correlated. Results. Among the 65 dysplastic nodules, echogenicity was high in 30 (46%), equal in 5 (8%), and low in 30 (46%). In all cases, there was significant correlation between echogenicity on sonographic imaging and the difference in fat content between nodules and surrounding liver tissue (P < .01). There was, however, no significant correlation between the degree of dysplasia and sonographic echogenicity (P > .05). Conclusions. The echogenicity of dysplastic nodules correlated with their fat content. Echogenicity did not, however, predict whether the grade of a nodule was high or low.
Key Words: liver, cirrhosis • liver, neoplasm • liver, nodules • sonography
Hepatocellular carcinoma frequently occurs in association with liver cirrhosis.1,2 Resected specimens of cirrhotic liver sometimes contain small nodular lesions, although the presence of these does not provide definite evidence of hepatocellular carcinoma.1 Dysplastic nodules are, however, premalignant lesions, representing an intermediate stage along the pathway of hepatocarcinogenesis in cirrhotic liver.3,4 Thus a benign regenerative nodule may be the first step in the development of overt hepatocellular carcinoma, progressing in a multistep fashion through the intermediate phases of a low- and high-grade dysplastic nodule, and early hepatocellular carcinoma.1 Therefore, in clinical practice, the detection of dysplastic nodules has important implications. Because of its widespread availability, low cost, and lack of invasiveness, sonography is the imaging modality most commonly used to investigate cirrhotic patients at risk of hepatocellular carcinoma. In the imaging of dysplastic nodules, diverse echogenic patterns are observed; these may be hyperechoic, isoechoic, or hypoechoic, findings that are similar to those of regenerative nodules or small hepatocellular carcinomas.1,5–8 Some previous reports have described dysplastic nodules as hypoechoic, a feature probably attributable to their relatively homogeneous tissue features.5,7,8 Others, however, noted that in 43% of cases, echogenicity was high, a finding partly explained by the occurrence of fatty change.9,10 The echogenicity of hepatic nodules reflects their histologic composition, whereas their detectability on sonography depends on the difference in echogenicity between them and normal parenchyma, the presence of cellularity, fibrosis, fatty change, necrosis, or peliosis and the relative homogeneity or heterogeneity of their tissue. The purpose of this study was to correlate the echogenicity of dysplastic nodules in cirrhotic liver with the difference in fat content between nodules and background liver. We also investigated the possible relationship between the grade of dysplastic nodules and their echogenicity and fat content.
This retrospective study involved 64 consecutive cirrhotic patients (44 male and 20 female; age range, 35–72 [mean, 56] years) for whom the results of previous sonographic examination, helical computed tomographic scanning, and gadolinium-enhanced magnetic resonance examination had suggested the existence of small nodular lesions. Between May 1997 and October 2000, the patients then underwent sonographically guided percutaneous liver biopsy, and the presence of 65 dysplastic nodules in 64 patients was histopathologically proved. Liver cirrhosis was hepatitis virus related in 60 patients, alcohol related in 3, and due to an unknown cause in 1. Tests proved positive for hepatitis B surface antigen in 51 patients, and for hepatitis C virus antibody (immunoglobulin G) in 9. The 65 dysplastic nodules were classified according to the criteria of the International Working Party11 on the terminology of nodular lesions of the liver: 39 were high grade and 26 were low grade. For all sonographic examinations, ATL HDI 5000 scanners (Philips Medical Systems, Bothell, WA) with C2-5 MHz transducers were used in accordance with the patients’ body habitus and the levels of the structures of interest. The examinations were performed by 4 gastrointestinal radiologists with 7 to 17 years of experience and 5 gastrointestinal radiology fellows with 4 to 6 years of experience. Sonographically guided percutaneous liver biopsies, for which a 19.5-gauge automated biopsy gun (Autovac; Angiomed GmbH, Karlsruhe, Germany) was used, were also carried out. Two radiologists (M.J.K. and J.M.P.) retrospectively reviewed the sonographic images on a consensus basis, comparing the echogenicity of the lesions that underwent biopsy with that of surrounding liver parenchyma and categorizing each as hypoechoic, isoechoic, or hyperechoic. Nodules showing heterogeneity were categorized according to the predominant pattern as hypoechoic or hyperechoic. When there was disagreement in assessing the echogenicity, the 2 radiologists discussed and reached an agreement. The location of each lesion was recorded in terms of the Couinaud segment involved. The greatest dimension of the nodule measured on initial sonographic examinations was also recorded. All nodule specimens including surrounding cirrhotic liver were stained with hematoxylin-eosin, and each histopathologic diagnosis was determined by a pathologist (C.K.P.) specializing in the liver. Using 2 high-power fields (x400), the fat globules in both the dysplastic nodules and surrounding liver were counted. Those biopsy materials that included only dysplastic nodules but not the surrounding background liver were excluded. The hepatocytes containing macrovesicular fat globules were also counted, but because microvesicular fat has little effect on the echogenicity observed on sonography, the number of these globules was not determined.12 Two representative areas were chosen, 1 from a nodule and 1 from surrounding parenchyma; those in which globules were either abundant or scarce were avoided. On the basis of the number of hepatocytes containing macrovesicular fat, the ratio of fat-containing cells to hepatocytes was established for both nodules and parenchyma, and the difference in fat content between the two was then determined. The relationship between the echogenicity of dysplastic nodules on sonography and differences in fat content between nodules and surrounding cirrhotic liver parenchyma was evaluated by the Spearman correlation analysis. Correlation between the grade of these nodules (high or low) and the proportion of fat globules they contained was determined by the Mann-Whitney test. The difference in fat content between hypoechoic and hyperechoic nodules was evaluated by the t test.
Thirty dysplastic nodules were hypoechoic (Figs. 1  and 2); another 30 dysplastic nodules were hyperechoic (Fig. 3); and 5 dysplastic nodules were isoechoic. Isoechoic nodules could be detected by the appearance of a perinodular hypoechoic halo and lateral shadowing (Fig. 4). Nodule diameter ranged from 0.8 to 4.5 (mean, 1.7) cm. Sixteen nodules (25%) were located in segment 5; 12 (19%) each in segments 7 and 8; 10 (15%) in the left lateral segment; 9 (14%) in segment 6; and 6 in segment 4. The correlation between the degree of echogenicity of the nodules and the ratio of fat content between a nodule and surrounding liver parenchyma appears in Table 1. There was a significant correlation between the echogenicity on sonographic imaging and the differences of fat content between a nodule and surrounding liver tissue according to Spearman correlation analysis (P < .01). There was no correlation between the echogenicity and the size of nodules.
There were 39 high-grade and 26 low-grade dysplastic nodules. There was no statistically significant correlation between nodular grade (high or low) and difference in echogenicity according to the Mann-Whitney test (P > .05; Table 2 ). Some low-grade nodules contained much more fat than surrounding liver, but there was no consistent correlation between fat content and echogenicity (Fig. 2 ). There was a significant difference for fat content differences between the hypoechoic and hyperechoic nodules according to the t test (P < .05).
With recent advances in the modalities used for imaging patients with hepatitis C or B virus–associated chronic liver diseases, many sizable nodular lesions have been detected on sonography.13,14 A cirrhotic liver is often the background from which hepatocytic nodular lesions and 80% of hepatocellular carcinomas arise, and because of their possible link with hepatocellular carcinoma, the presence of nodular lesions is of particular importance.15 From a pathologic viewpoint, a benign regenerative nodule may be the first step in the development of overt hepatocellular carcinoma, progressing in a multistep fashion through the intermediate phases of ordinary and atypical adenomatous hyperplasia and early hepatocellular carcinoma. Alternatively, hepatocellular carcinoma may arise de novo.1,3,16 To clarify the nomenclature, the new term dysplastic nodule has been introduced, replacing several previously used terms.11,17 Dysplastic nodules are categorized as low or high grade. The former are commonly referred to as ordinary adenomatous hyperplasia or, occasionally, as "macroregenerative nodules type 1," whereas high-grade nodules are usually known as atypical adenomatous hyperplasia or, on occasion, "macroregenerative nodules type 2."11,17,18 Some previous studies have reported that most dysplastic nodules are hypoechoic,7,8,19 probably because of their relatively homogeneous tissue characteristics, but others have stated that most are hyperechoic.6,20 Research has also shown that dysplastic nodules have diverse echo patterns, which may be hyperechoic, isoechoic, or hypoechoic, and the sonographic findings were similar to those of regenerative nodules or small hepatocellular carcinomas.1,7,8,10,19 A varying degree of fatty change may be an important morphologic marker of the malignant transformation of a dysplastic nodule, as shown pathomorphologically21 and in experimental models.22 Both dysplastic nodules containing malignant foci and well-differentiated hepatocellular carcinomas contain high levels of fat, and fat deposition in the former may thus be related to malignancy. Eguchi et al9 reported that all dysplastic nodules containing cancerous foci and 75% of high-grade dysplastic nodules showed marked fatty changes. Kojiro23 observed that although the cytoplasm of high-grade dysplastic nodules usually contains fat globules in the cytoplasm, whereas low-grade dysplastic nodules usually do not contain fat globules, these are usually absent in low-grade nodules. Kojiro23 speculated that fat deposition is closely related to the developmental mechanism of hepatocellular carcinoma, possibly through nodular hypoxia caused by hepatic arterial degeneration. In our study, the echogenic patterns revealed at sonography were diverse, and other studies have shown that because the sonographic findings are similar, differentiation between dysplastic nodules and small hepatocellular carcinomas or regenerative nodules is difficult.1,5,6,8,9 Our study found a significant correlation between the echogenicity of dysplastic nodules and the differences in fat content between nodules and liver parenchyma. Previous reports24–29 have shown that the echogenicity of a nodule determined on sonography appears to reflect mainly fatty, clear-cell, and small-cell changes with increased nuclear crowding. Our study concludes that of all those factors, the echogenicity of dysplastic nodules is affected by the difference between a nodule and surrounding liver parenchyma in terms of fat content. Terasaki et al30 reported that the histologic features that predicted the malignant transformation of nonmalignant hepatocellular nodules included increased nuclear density, small-cell change, and fatty change. Therefore, it appears that hyperechoic dysplastic nodules that show abundant fatty change are likely to progress to hepatocellular carcinoma. In our study, there was no significant correlation between the grade of dysplastic nodules (high or low), fat content, and echogenicity. Various pathologic and morphologic factors, including clear- and small-cell change with increased nuclear crowding, affect echogenicity as revealed on sonography; fatty change alone might not contribute to the high echogenicity of dysplastic nodules. Our study has the following limitations. First, the echogenicity of the nodules on sonography was determined subjectively. Some nodules with heterogeneous echogenicity were classified on the basis of the predominant pattern. This might have caused some mismatches between the results of sonography and pathologic analysis. We did not assess the interobserver and intraobserver variability in assessing the echogenicity of nodules. However, we categorized the echogenicity simply as hyperechoic, isoechoic, and hypoechoic, and the 2 reviewers reached a consensus easily after discussion. Second, we did not count the microvesicular fat globules pathologically. A previous study12 reported that microvesicular fat has little effect on the echogenicity at sonography. Last, we could detect isoechoic nodules only when they had peripheral rims, which enabled us to differentiate the nodule from the background liver parenchyma. We might have missed some isoechoic nodules. We suggest that the difference in the degree of fatty change between a nodule and surrounding liver tissue is an important factor affecting echogenicity. However, we found no correlation between echogenicity and the degree of dysplasia. To determine which factor most affects echogenicity and the malignant potential of dysplastic nodules, further study is needed.
Received November 20, 2002 from the Department of Radiology and Center for Imaging Science (M.J.K., J.H.L., S.J.L., S.H.K., W.J.L., H.K.L., J.M.P.) and Department of Pathology (C.K.P.), Samsung Medical Center, Sungkyunkwan University School of Medicine, Seoul, South Korea. Revision requested November 27, 2002. Revised manuscript accepted for publication January 7, 2003. We thank Steve Downs, MA, for editing and Bonnie Hami (Department of Radiology, University Hospitals of Cleveland, Cleveland, OH) for editorial assistance.
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Top
Abstract
Introduction
