| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Gastrointestinal Imaging |
1 First Department of Medicine (M.E., H.F., Y.K., N.M., M. Yoshikawa, N.S., H.S.)
2 Second Department of Pathology (T.S., F.K.), Chiba University School of Medicine, 1-8-1 Inohana, Chuo-ku, Chiba 260, Japan
3 National Institute of Radiological Sciences, Chiba, Japan (M. Yukawa, T.M.).
 |
Abstract |
|---|
 TOP Abstract IntroductionMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
MATERIALS AND METHODS: The signal intensity patterns of 59 HCCs 3 cm in diameter or smaller were correlated with histologic findings and metal content.
RESULTS: HCCs with high signal intensity on T1-weighted images demonstrated more steatosis (P = .035) and higher copper content (P = .008) than did surrounding hepatic parenchyma. HCCs with high signal intensity on T2-weighted images demonstrated more clear cells (P = .001) than did surrounding hepatic parenchyma. The higher signal intensities on T1-weighted and T2-weighted images were related to a higher and a lower degree of histologic differentiation, respectively. Multivariate analysis showed that the contrast-to-noise ratio between the HCC and surrounding hepatic parenchyma was affected by intratumoral copper content (P = .0338), zinc content of surrounding hepatic parenchyma (P = .0379), and the degree of histologic differentiation on T1-weighted (P = .0031) and T2-weighted (P = .0062) images.
CONCLUSION: The signal intensity of HCC on T1-weighted images is related to the degree of histologic differentiation, intratumoral copper content, and zinc content of surrounding hepatic parenchyma, whereas the signal intensity on T2-weighted images is related to the degree of histologic differentiation.
Index terms: Copper • Liver neoplasms, 761.323 • Liver neoplasms, MR, 761.121411 • Zinc
|
Introduction |
|---|
|
TOPAbstractIntroductionMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
We investigated the relationship between MR imaging and histopathologic findings in HCC, particularly the content of various metals in the tumor and surrounding hepatic parenchyma, to clarify the relationship between the metal content and signal intensity of HCC.
|
MATERIALS AND METHODS |
|---|
|
TOPAbstractIntroductionMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
The patients consisted of 44 men and 15 women with a mean age of 62.2 years ± 8.1 (SD) (range, 41–79 years). The average tumor diameter was 21.3 mm ± 5.1 (range, 10–30 mm). The diagnosis of HCC was confirmed in each case by histologic examination of specimens obtained with US-guided biopsy. All cases of HCC were considered to be associated with liver cirrhosis according to the results of biopsy, US, MR imaging, and computed tomography (CT) and clinical findings. The cause of liver cirrhosis was considered to be hepatitis B surface antigen in one patient (2%); hepatitis C virus in 26 patients (44%); alcohol in three patients (5%); hepatitis B surface antigen and alcohol in three patients (5%); hepatitis C virus and alcohol in 22 patients (37%); hepatitis B surface antigen, hepatitis C virus, and alcohol in three patients (5%); and unknown in one patient (2%).
MR Imaging
A 1.5-T superconducting MR imaging system (Signa Advantage; GE Medical Systems, Milwaukee, Wis) with a body coil was used. Imaging parameters were as follows: 32 x 32-cm field of view, 16-kHz bandwidth, and 12 or 14 partitions, with an acquisition time of approximately 5 minutes for a T1-weighted image and approximately 8 minutes for a T2-weighted image. Axial spin-echo sequences were performed with the following parameters: 500/11 (repetition time msec/echo time msec) for T1-weighted images and 2,000/80 for T2-weighted images, 8-mm section thickness with a 2-mm gap, 256 x 192 matrix, and two signals acquired. Fat suppression was not used. Respiratory compensation, flow compensation, and the "no phase wrap" technique were used to reduce artifacts. The diameter of every tumor was first measured with US. All MR imaging examinations were performed after hepatic biopsy.
MR images were reviewed by three physicians (M.E., Y.K., N.M.) specializing in imaging diagnosis. The lesions were categorized as hyperintense, isointense, or hypointense relative to the hepatic parenchyma. Signal intensity was measured in regions of interest (mean area, 9.8 mm2 ± 2.5) in the liver and in the HCC. Noise was measured in a large, rectangular region of interest anterior to the abdomen; the SD of the noise was used in the quantitative analysis. The contrast-to-noise ratio (CNR) of the HCC relative to the surrounding hepatic parenchyma was calculated as follows: CNR = (SIt - SIp)/noise, where SIt = signal intensity of the tumor and SIp = signal intensity of the parenchyma.
Histopathologic Findings
Specimens from liver masses suspected to be HCC at US were obtained with a 21-gauge biopsy needle (Sonopsy C1; Hakko, Tokyo, Japan) under sonographic guidance. In addition, control specimens of hepatic parenchyma were obtained from nontumor regions with a 17-gauge biopsy needle (Quick-Cut; Hakko) or the 21-gauge biopsy needle. Histologic slides were reviewed by two pathologists (T.S., F.K.), who were blinded to all other findings. Histopathologic parameters evaluated included the degree of histologic differentiation of the lesion and the extents of steatosis, clear cells, bleeding, and necrosis in the lesion and in nontumor regions.
To quantify the extent of steatosis, light microscopic images (magnification, x20) were captured on an image analyzer (ie, a high-definition television) (Personal Image Analysis System IV; Pias, Osaka, Japan) with a red-green-blue, digital, color, charge-coupled device camera; the percentage of adipose cells in five randomly selected fields was calculated with a threshold technique. To quantify the extent of clear cells, each slide was projected on a high-resolution monitor, and tumor cells and clear cells were counted in five randomly selected fields; values were expressed as percentages of the total cells.
Measurement of Metal Content
The metal content was measured in 59 HCCs and the surrounding hepatic parenchyma. Specimens from the HCCs and the surrounding hepatic parenchyma obtained with US-guided needle biopsy were allowed to dry. Each specimen was then fixed with carbon tape, and the metal content was analyzed with a particle-induced x-ray emission (PIXE) analyzer.
PIXE analysis is a method of supersensitive, nondestructive, polyelementary coincidence analysis in which every specimen is irradiated with proton beams accelerated by a Van de Graaff accelerator; the resultant characteristic x rays are measured with a crest analyzer (11–13). We performed this procedure with a device developed by the Japanese National Institute of Radiological Sciences (11). The metal content in tissue was measured with the assumption that the potassium concentration in the tissue was a constant. The measurements were converted into weights by using normal bovine liver (National Institute of Standards and Technology, Washington, DC) as a sample standard.
Statistical Analysis
Values were expressed as means ± SDs. To study the relationship of signal intensity and tumor size to histologic differentiation of HCC, the Kruskal-Wallis rank test or Tukey-Kramer test was used. Differences between HCC and surrounding hepatic parenchyma were analyzed with the Wilcoxon signed rank test together with the Tukey-Kramer test if necessary. Stepwise regression analysis was used to identify factors that affected the signal intensity of HCC relative to that of surrounding hepatic parenchyma (ie, the CNR) on T1- and T2-weighted images. Factors considered were tumor diameter; degree of histologic differentiation; extents of steatosis and clear cells in the HCC and in surrounding hepatic parenchyma; and levels of iron, copper, and zinc in the HCC and in surrounding hepatic parenchyma. A P value of less than .05 was considered to indicate a statistically significant difference.
|
RESULTS |
|---|
|
TOPAbstractIntroductionMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
The fibrous pseudocapsule was visualized as a ring sign (2) in 11 lesions (19%) on T1-weighted images.
Signal Intensity Group versus CNR
On T1-weighted images, the difference in signal intensity between the HCC and surrounding hepatic parenchyma was 8.73 ± 6.25 (range, 2.09–26.84) for hyperintense lesions (n = 36), -0.03 ± 1.12 (range, -1.90 to 2.14) for isointense lesions (n = 9), and -12.09 ± 11.07 (range, -35.67 to 1.37) for hypointense lesions (n = 14).
On T2-weighted images, the difference in signal intensity was 12.29 ± 11.49 (range, 1.27–69.07) for hyperintense lesions (n = 44), 0.43 ± 1.24 (range, -1.47 to 1.99) for isointense lesions (n = 6), and -6.35 ± 5.17 (range, -19.7 to 2.18) for hypointense lesions (n = 9).
Relationship between Signal Intensity and Histologic Differentiation
Of the 36 lesions that were hyperintense on T1-weighted images, 23 (64%) were well differentiated and 13 (36%) were moderately differentiated at histologic examination (Table 1). Therefore, there was a greater prevalence of well-differentiated lesions in this group. Of the nine lesions that were isointense on T1-weighted images, three (33%) were well differentiated, four (44%) were moderately differentiated, and two (22%) were poorly differentiated. Of the 14 lesions that were hypointense on T1-weighted images, three (21%) were well differentiated, 10 (71%) were moderately differentiated, and one (7%) was poorly differentiated. Therefore, there was a greater prevalence of moderately differentiated lesions in this group. The degree of histologic differentiation was significantly different between hyperintense lesions and hypointense lesions on T1-weighted images (P = .019).
|
Thus, lesions with higher signal intensity on T1-weighted images tended to have a higher degree of histologic differentiation. Conversely, lesions with higher signal intensity on T2-weighted images tended to have a lower degree of histologic differentiation.
Relationship between Signal Intensity and Histopathologic Findings
Histologic specimens were successfully obtained in all instances. The histopathologic parameters (steatosis, clear cells, bleeding, and necrosis) were evaluated according to the signal intensity groups. No statistically significant intergroup differences in the extent of steatosis were observed, but the extent of steatosis was significantly higher in hyperintense lesions on T1-weighted images than in the noncancerous region (P = .035) (Table 2). No analogous differences were seen in the other groups. No statistically significant intergroup differences in the extent of clear cells were observed, but the extent of clear cells was significantly higher in hyperintense lesions on T1-weighted images (P = .008), hypointense lesions on T1-weighted images (P = .008), and hyperintense lesions on T2-weighted images (P = .001) than in the respective noncancerous regions. No analogous differences were seen in the other groups. There was no bleeding in any of the lesions. Three lesions demonstrated some necrosis, but the necrosis had no effect on the signal intensity categories.
|
|
|
|
|
|
|
|
The copper content was significantly higher in hyperintense lesions on T1-weighted images (P = .008), hyperintense lesions on T2-weighted images (P = .015), and isointense lesions on T2-weighted images (P = .028) than in the respective noncancerous regions. There were no statistically significant intergroup differences in copper content. All five HCCs that contained more than 1,000 µg of copper per gram of dry weight were hyperintense on T1-weighted images.
The zinc content was significantly higher in the noncancerous region than in isointense lesions on T1-weighted images (P = .008). The zinc content in the respective noncancerous regions was significantly lower for hyperintense lesions on T1-weighted images (P = .001) and isointense lesions on T1-weighted images (P = .046) than for hypointense lesions on T1-weighted images.
The manganese content was significantly higher in hyperintense lesions on T1-weighted images than in the noncancerous region (P = .016). The manganese content was significantly higher in the noncancerous region than in isointense lesions on T1-weighted images (P = .008). Finally, the manganese content was significantly higher in hyperintense lesions on T1-weighted images than in isointense lesions on T1-weighted images (P = .046).
Multivariate Analysis of Factors That Affected Signal Intensity
Factors that significantly affected the CNR between any HCC and surrounding hepatic parenchyma were the copper content of the HCC on T1-weighted images (P = .0338), the zinc content of surrounding hepatic parenchyma on T1-weighted images (P = .0379), and the degree of histologic differentiation on both T1-weighted (P = .0031) and T2-weighted (P = .0062) images (Table 3). Iron content in the HCC and in surrounding hepatic parenchyma, extents of steatosis and clear cells in the HCC and in surrounding hepatic parenchyma, and tumor diameter did not significantly affect the CNR on either T1- or T2-weighted images.
|
|
DISCUSSION |
|---|
|
TOPAbstractIntroductionMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
High signal intensity on T1-weighted images was observed in 61% of the lesions, a result suggesting that this finding is important for a definite diagnosis of small HCC. However, a pattern of high signal intensity on T1-weighted images and low signal intensity on T2-weighted images is reported to be characteristic of dysplastic nodules (15,16). In our study, seven of 59 HCCs (12%) showed the same pattern. This discrepancy might have been due to differences in histologic criteria for HCC and dysplastic nodule or to sampling error. Furthermore, it has been shown that one-third of HCCs less than 2 cm in diameter display various combinations of histologic grades (17). The wide variation in metal content associated with each signal intensity group might have been due to uneven intratumoral distribution of the metal. Because histologic features and metal content were determined in small specimens obtained with a thin needle, variations in our results might have been expected. Nevertheless, our study showed statistically significant relationships between the signal intensity, histologic features, and metal content of HCC.
Differential diagnosis of minute HCCs versus dysplastic nodules and large regenerative nodules with imaging modalities is a difficult but unavoidable issue. At present, it is almost impossible to differentiate these lesions with conventional imaging modalities such as CT and US. Further investigation is needed to determine whether HCC can be distinguished from dysplastic nodules and large regenerative nodules with MR imaging. In our series, no particular predominance of signal intensity pattern was observed, but the number of lesions was limited.
The origin of the high signal intensity on T1-weighted images has not been established. Good histologic differentiation was significantly more common among hyperintense lesions on T1-weighted images than among hypointense lesions on T1-weighted images (P = .019), whereas good histologic differentiation was significantly more common among hypointense lesions on T2-weighted images than among hyperintense lesions on T2-weighted images (P = .027). In contrast to results of earlier studies (2,8), we found that signal intensity on T1- and T2-weighted images did not correlate significantly with tumor diameter; this result was probably due to the fact that all of the HCCs were 3 cm in diameter or smaller. Some investigators have suggested steatosis as a causative factor in the high signal intensity on T1-weighted images (2,18). In our series, there were no statistically significant differences in extent of steatosis between hyperintense lesions on T1-weighted images, isointense lesions on T1-weighted images, and hypointense lesions on T1-weighted images; however, the extent of steatosis was significantly higher in hyperintense lesions on T1-weighted images than in the noncancerous region (P = .035).
With respect to the extent of clear cells, there were no statistically significant differences between hyperintense lesions on T1-weighted images, isointense lesions on T1-weighted images, and hypointense lesions on T1-weighted images; however, the extent of clear cells was significantly higher in hyperintense lesions on T1-weighted images (P = .008) and hypointense lesions on T1-weighted images (P = .008) than in the respective noncancerous regions. Glycogen and fatty droplets have been reported to be related to clear cell formation in HCC (19), but macroscopic fat may also be involved. In our series, however, 17 of 36 hyperintense lesions on T1-weighted images (47%) demonstrated low extents of both steatosis and clear cells (<10% for each), a result suggesting that other factors may explain the T1 shortening.
The presence of heavy metals within HCCs has been thought to have a paramagnetic effect. In earlier studies, however, the sensitivity and specificity of measurements of heavy metal content were unreliable, particularly for the copper content of cancerous and noncancerous regions, because metal staining (eg, rhodamine staining, rubeanic acid staining) or metal-binding protein staining (eg, orcein staining) was used (8,9). To overcome this limitation, metal content in our study was quantified with PIXE analysis.
When the lesions were classified according to signal intensity group at T1-weighted imaging, the copper content was significantly higher than in the noncancerous region only in hyperintense lesions (P = .008). Although it has been reported that there is a substantial amount of iron in tumors and that it influences signal intensity patterns (20,21), we observed no statistically significant differences in iron content between the signal intensity groups or between cancerous and noncancerous regions within the groups. Because the manganese content in both cancerous and noncancerous regions was low (Fig 3), it did not seem to affect signal intensity. Zinc, which has no effect on signal intensity patterns, was significantly related to CNR on T1-weighted images. That is, the zinc content of noncancerous regions was a statistically significant factor in the signal intensity of HCC. There was also a correlation between the signal intensity of HCC and copper content. However, the fact that the degree of correlation was relatively low suggests that other unknown factors affect the signal intensity of HCC. It has been reported that zinc and copper often bind to metallothionein to form zinc-thionein and copper-thionein, respectively, in the hepatic parenchyma (22), but the levels of other forms of copper or the ratios of these forms, especially within the tumor, have not yet been fully elucidated.
In conclusion, the signal intensity of HCC on T1-weighted MR images is related to the degree of histologic differentiation, intratumoral copper content, and zinc content of surrounding hepatic parenchyma, whereas the signal intensity on T2-weighted images is related to the degree of histologic differentiation.
|
Footnotes |
|---|
Address reprint requests to M.E.
Abbreviations: CNR = contrast-to-noise ratio HCC = hepatocellular carcinoma PIXE = particle-induced x-ray emission
Author contributions: Guarantor of integrity of entire study, M.E.; study concepts and design, M.E.; definition of intellectual content, M.E.; literature research, M.E., H.F.; clinical studies, M. Yoshikawa, N.S., H.S.; data acquisition, M. Yukawa, H.F.; data analysis, Y.K., N.M., T.S., F.K.; statistical analysis, T.M.; manuscript preparation, editing, and review, M.E.
Received December 15, 1997;
revision requested March 18, 1998; revision received May 4, 1998;
accepted June 23, 1998.
|
References |
|---|
|
TOPAbstractIntroductionMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
This article has been cited by other articles:
|
|
 |
|
  J.-S. Yu, J.-J. Chung, J. H. Kim, and K. W. Kim Fat-Containing Nodules in the Cirrhotic Liver: Chemical Shift MRI Features and Clinical Implications Am. J. Roentgenol., April 1, 2007; 188(4): 1009 - 1016. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
Z.-H. Fan, M.-H. Chen, Y. Dai, Y.-B. Wang, K. Yan, W. Wu, W. Yang, and S.-S. Yin Evaluation of primary malignancies of the liver using contrast-enhanced sonography: correlation with pathology. Am. J. Roentgenol., June 1, 2006; 186(6): 1512 - 1519. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 R. Shinmura, O. Matsui, S. Kobayashi, N. Terayama, J. Sanada, K. Ueda, T. Gabata, M. Kadoya, and S. Miyayama Cirrhotic Nodules: Association between MR Imaging Signal Intensity and Intranodular Blood Supply Radiology, November 1, 2005; 237(2): 512 - 519. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
Y. Y. Jeong, N. Y. Yim, and H. K. Kang Hepatocellular Carcinoma in the Cirrhotic Liver with Helical CT and MRI: Imaging Spectrum and Pitfalls of Cirrhosis-Related Nodules Am. J. Roentgenol., October 1, 2005; 185(4): 1024 - 1032. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 S. R. Prasad, H. Wang, H. Rosas, C. O. Menias, V. R. Narra, W. D. Middleton, and J. P. Heiken Fat-containing Lesions of the Liver: Radiologic-Pathologic Correlation RadioGraphics, March 1, 2005; 25(2): 321 - 331. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
H. K. Hussain, I. Syed, H. V. Nghiem, T. D. Johnson, R. C. Carlos, W. J. Weadock, and I. R. Francis T2-weighted MR Imaging in the Assessment of Cirrhotic Liver Radiology, March 1, 2004; 230(3): 637 - 644. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M. Fujita, T. Horinouchi, S. Ishiguro, R. Ishihara, H. Kasugai, T. Yamada, Y. Sasaki, H. Maeda, E. Inoue, and C. Kuroda T2-Shortening Effect of Fibrinogen Inclusions on MRI of Hepatocellular Carcinoma: Case Report and Experimental Relaxation Measurement Am. J. Roentgenol., February 1, 2004; 182(2): 459 - 462. [Full Text] [PDF]
|
|
|
|
|
|
G. A. Krinsky and G. Israel Nondysplastic Nodules That Are Hyperintense on T1-Weighted Gradient-Echo MR Imaging: Frequency in Cirrhotic Patients Undergoing Transplantation Am. J. Roentgenol., April 1, 2003; 180(4): 1023 - 1027. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. P. Heiken Invited Commentary RadioGraphics, September 1, 2002; 22(5): 1037 - 1039. [Full Text] [PDF]
|
|
|
|
|
|
R. L. Baron and M. S. Peterson From the RSNA Refresher Courses: Screening the Cirrhotic Liver for Hepatocellular Carcinoma with CT and MR Imaging: Opportunities and Pitfalls RadioGraphics, October 1, 2001; 21(90001): S117 - 132. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
G. A. Krinsky, V. S. Lee, N. D. Theise, J. C. Weinreb, N. M. Rofsky, T. Diflo, and L. W. Teperman Hepatocellular Carcinoma and Dysplastic Nodules in Patients with Cirrhosis: Prospective Diagnosis with MR Imaging and Explantation Correlation Radiology, May 1, 2001; 219(2): 445 - 454. [Abstract] [Full Text]
|
|
|
|
|
|
T. Nakakoshi, M. Ebara, and H. Fukuda Copper and Hepatocellular Carcinoma • Drs Ebara and Fukuda respond: Radiology, January 1, 2000; 214(1): 304 - 306. [Full Text]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| RADIOLOGY | RADIOGRAPHICS | RSNA JOURNALS ONLINE |
