Superparamagnetic Iron Oxide Enhanced Magnetic Resonance Images of Hepatocellular Carcinoma: Correlation With Histological Grading

Transcription

1 Superparamagnetic Iron Oxide Enhanced Magnetic Resonance Images of Hepatocellular Carcinoma: Correlation With Histological Grading YASUHARU IMAI, 1 TAKAMICHI MURAKAMI, 5 SHIGEYUKI YOSHIDA, 2 MASAHIRO NISHIKAWA, 1 MASAHIKO OHSAWA, 3 KOH TOKUNAGA, 2 MASARU MURATA, 4 KUNITAKA SHIBATA, 4 SHINICHIRO ZUSHI, 1 MASANORI KUROKAWA, 1 TAKESHI YONEZAWA, 1 SUMIO KAWATA, 6 MANABU TAKAMURA, 5 HIROAKI NAGANO, 7 MASATO SAKON, 7 MORITO MONDEN, 7 KENICHI WAKASA, 8 AND HIRONOBU NAKAMURA 5 Superparamagnetic iron oxide (SPIO) enhanced magnetic resonance (MR) imaging has been used for the detection of hepatic tumors. However, little is known about this technique in relation to hepatocellular carcinoma (HCC). The aim of this study was to investigate whether SPIO enhanced MR imaging can be useful in assessing histological grades of HCC. The authors studied histologically proven tumors including 31 HCCs and 6 dysplastic nodules. The ratio of the Kupffer-cell count in the tumorous tissue relative to that in the nontumorous tissue (Kupffer-cell count ratio) decreased as HCCs became less well differentiated. The ratio of the intensity of the tumorous lesion to that of the nontumorous area on SPIO enhanced MR images (SPIO intensity ratio) correlated inversely with Kupffer-cell count ratio in HCCs and dysplastic nodules (r.826, P <.001) and increased as the degree of differentiation of HCCs decreased, indicating that the uptake of SPIO in HCCs decreased as the degree of differentiation of HCCs declined. All of the dysplastic nodules and some well-differentiated HCCs showed hypointense or isointense enhancement, relative to the surrounding liver parenchyma, indicating greater or similar uptake of SPIO in the tumor when compared with nontumorous areas. These results suggest that SPIO enhanced MR imaging reflects Kupffer-cell numbers in HCCs and dysplastic nodules, and is useful for estimation of histological grading in HCCs, although uncertainties persist in differentiating dysplastic nodules from well-differentiated HCCs. (HEPATOLOGY 2000;32: ) Abbreviations: HCC, hepatocellular carcinoma; CT, computed tomography; MR, magnetic resonance; SPIO, superparamagnetic iron oxide; fast SPGR, fast multiplanar spoiled gradient recalled acquisition in the steady state presession sequence; ROI, region of interest; CTHA, CT hepatic arteriography; CTAP, CT during arterial portography. From the 1 Departments of Internal Medicine, 2 Radiology, 3 Pathology, and 4 Surgery, Ikeda Municipal Hospital, Osaka, Japan; the 5 Departments of Radiology, 6 Internal Medicine and Molecular Science, and 7 Surgery II, Graduate School of Medicine, Osaka University, Osaka, Japan; and the 8 Department of Pathology, Osaka City University Hospital, Osaka Japan. Received November 24, 1999; accepted May 15, Address reprint requests to: Yasuharu Imai, M.D., Ph.D., Department of Internal Medicine, Ikeda Municipal Hospital, , Johnan, Ikeda, Osaka, Japan. imh@wombat.or.jp; fax: (81) Copyright 2000 by the American Association for the Study of Liver Diseases /00/ $3.00/0 doi: /jhep Hepatocellular carcinoma (HCC) is one of the most common cancers worldwide and a major cause of death in patients with cirrhosis. 1-4 One of the characteristics of HCC detected by imaging modalities such as computed tomography (CT), magnetic resonance (MR) imaging, and CT angiography, is increased arterial blood flow accompanied by decreased portal blood flow, although some well-differentiated HCCs do not show such blood supply characteristics. 5-9 Another characteristic of HCC, distinct from other metastatic tumors, has recently been clarified by Tanaka et al., 10 namely, that some well-differentiated HCCs have similar numbers of Kupffer cells as do the surrounding nontumorous regions of the liver. Superparamagnetic iron oxide (SPIO) is a tissue-specific MR imaging contrast that is taken up by Kupffer cells in the liver and macrophages in the spleen. 11,12 Because metastatic liver cancers lack Kupffer cells, hepatic metastases do not take up SPIO and thus appear hyperintense relative to the surrounding liver, being useful for the detection of metastatic liver cancers In an animal model of HCC, Kawamori et al., 15 studied whether SPIO enhanced MR imaging could allow HCC to be distinguished from hyperplastic nodules. They concluded that SPIO enhanced MR imaging may be useful in differentiating HCC from hyperplastic nodules, although some HCC nodules were difficult to differentiate from hyperplastic nodules, even after SPIO enhanced MR imaging. In patients with HCC, Yamamoto et al., 16 reported that SPIO enhanced MR imaging was more accurate than unenhanced MR imaging for the detection of HCC. They showed a case of well-differentiated HCC that appeared hypointense relative to the surrounding liver tissue on MR images after administration of SPIO, indicating an accumulation of SPIO within the tumor. 16 Little is known about the relation of SPIO enhanced MR imaging and histological grading in HCCs. Also, heretofore, there has been no report on either the correlation between SPIO enhanced MR images and an immunohistological examination of Kupffer cells in HCC or in dysplastic nodules, which have been shown to have a potential of malignant transformation The purposes of this study were to investigate whether SPIO enhanced MR imaging correlates with Kupffer cell numbers in HCCs and dysplastic nodules, and to see whether SPIO enhanced MR imaging can be used to predict histological grading of HCC. The authors also examined the relationship between SPIO enhanced MR images and tumor vascularity in HCCs and in dysplastic nodules. 205

2 206 IMAI ET AL. HEPATOLOGY August 2000 PATIENTS AND METHODS Patients. The authors studied 27 patients with HCC and/or a dysplastic nodule who had been admitted to our institutes in Nineteen patients were men, and eight were women, with a mean age of 62 years (range years). Underlying liver diseases of these patients were histologically proven: 17 patients had HCV related cirrhosis, 7 had HCV related chronic hepatitis, 1 had HBV related cirrhosis, 1 had alcoholic cirrhosis, and 1 had cirrhosis of unknown cause. Diagnosis of HCCs and Dysplastic Nodules. In the 27 patients, the authors detected 37 tumors with a median diameter 1.6 cm (range from 0.8 to 12.5 cm) by ultrasound, dynamic CT, or dynamic MR imaging. All of the 37 tumors were histologically diagnosed as either HCCs or dysplastic nodules. Diagnosis of dysplastic nodules and HCCs was made histologically by 2 of the authors who specialized in liver pathology, in a blinded manner, in accordance with the histological criteria of the International Working Group. 21 Liver specimens of 23 tumors were obtained by 21-gauge needle core biopsy (Majima needle, Top, Japan) from 16 patients: 12 of these individuals had an HCC or a dysplastic nodule, 1 patient had an HCC and a dysplastic nodule, and 3 others each had 3 tumors consisting of histologically different HCCs and a dysplastic nodule. Of the 23 tumors obtained percutaneously, 3 samples taken from each tumor by 21-gauge needle showed the same histological grade, with the exception of 1 tumor, which contained a small area of moderately differentiated HCC within a larger area of poorly differentiated HCC. An additional 14 tumors were obtained from 11 patients who underwent partial hepatectomy: 8 of these individuals had a HCC, 2 had 2 HCCs, and 1 had an HCC and a dysplastic nodule. Therefore, complete tumors were examined histologically in these specimens: 1 dysplastic nodule and 13 HCCs. Of the 13 HCCs, 9 were histologically uniform and 4 were composed of 2 different histological grades of HCCs (well-differentiated HCC with a minor component of moderately-differentiated HCC in 2, moderately differentiated HCC with a minor component of poorly differentiated HCC in 1, and poorly differentiated HCC with a minor component of moderately differentiated HCC in 1). HCCs with regions of varying histological grades were classified as belonging to the predominating histological characteristic. The final diagnoses of the 37 tumors were as follows: 6 dysplastic nodules, 13 well-differentiated HCCs, 10 moderately differentiated HCCs, and 8 poorly differentiated HCCs. MR Imaging. MR imaging was performed with a 1.5-T whole-body imager (Signa Horizon, GE Medical Systems, Milwaukee, WI) with an Abdoflex single-coil. In this study, the authors employed breathhold fast multiplanar spoiled gradient recalled acquisition in the steady state presession sequence (fast SPGR) imaging with a repetition time of 150 ms, an echo time of 10 ms, and a flip angle of 70. The authors obtained the fast SPGR images before and after SPIO enhancement with a matrix, 1 excitation, a field of view of 32 cm, section thickness of 8 mm, and an interscan gap of 2 mm. Ferumoxides (Feridex, Eiken Chemical, Tokyo, Japan) in a 5 ml vial containing 0.2 mol/l of iron (11.2 mg of Fe/mL) was used as the SPIO. Ten mol/kg of body weight of this agent were diluted with 100 ml of 5% glucose solution and administered intravenously for 30 minutes. Sixty minutes after the end of the infusion, the same fast SPGR images were carried out again. SPIO Intensity Ratio. The authors read signal intensities of the tumorous and nontumorous areas directly from a monitor by using an operator-defined region of interest (ROI), which was specified as a round area up to 10 mm in diameter. The authors took ROI readings from 1 area of the tumorous lesion and from 3 areas of adjacent nontumorous regions devoid of vascular structures. The mean of the latter 3 points was used as the signal intensity of the nontumorous liver. To evaluate the uptake of SPIO by the tumors accurately, the authors calculated a relative intensity ratio of the tumorous lesion to the nontumorous area on unenhanced MR images and on SPIO enhanced MR images as follows: Precontrast relative intensity ratio Precontrast signal intensity of tumorous lesion Precontrast signal intensity of nontumorous area Postcontrast relative intensity of ratio Postcontrast signal intensity of tumorous lesion Postcontrast signal intensity of nontumorous area and defined the ratio of these relative intensity ratios as the SPIO intensity ratio: Postcontrast relative intensity ratio SPIO intensity ratio Precontrast relative intensity ratio Using this calculation, the SPIO intensity ratio is above 1 when less SPIO accumulates in the tumorous tissue than in the nontumorous tissue, and the SPIO intensity ratio is below 1 when more SPIO accumulates in the tumorous tissue than in the nontumorous tissue. CT Hepatic Arteriography and CT During Arterial Portography. CT hepatic arteriography (CTHA) and CT during arterial portography (CTAP) were undertaken in 34 of the 37 tumors to examine the blood supply through the hepatic artery and portal vein to their tumors. 5,6,8 The authors performed CT imaging with a High Speed Advantage Scanner (General Electric Medical Systems, Milwaukee, WI), using the parameters of CTHA and CTAP as previously reported. 8 Immunohistochemistry of Kupffer Cells. The liver specimens were fixed in 10% formalin and embedded in paraffin. Four micrometerthick sections were deparaffinized and hydrated. The authors exposed sections to anti-cd68 (PGM1) antibody (Dakopatts, Glostrup, Denmark) 10 for 16 hours at 4 C after treatment with 0.1% trypsin for 30 minutes at 37 C and with 3% H 2 O 2 for 5 minutes to eliminate endogenous peroxidase activity. Then, sections were stained using the labeled streptoavidin-biotin methods (Dako LSAB2 kit; Dakopatts, Glostrup, Denmark). The antibody-positive cells, which were located in the sinusoids of nontumorous tissues or blood spaces in tumorous tissues and showed a stellate or spindle shape, were recognized as Kupffer cells. 10 Kupffer-Cell Count Ratio. The authors counted the number of Kupffer cells in three 0.12 mm 2 areas of both the tumorous and nontumorous tissues within each specimen. The authors then used these 3 values to calculate the mean for each type of tissue and used the means as the Kupffer cell counts of the tumorous and nontumorous tissues. The Kupffer-cell count ratio was calculated using the following formula: Kupffer-cell count ratio Kupffer cell count of tumorous tissue Kupffer cell count of nontumorous tissue Therefore, the Kupffer-cell count ratio is below 1 when the number of Kupffer cells in the tumorous tissue is smaller than in the nontumorous tissue, whereas the Kupffer-cell count ratio is above 1 when the number of Kupffer cells in the tumorous tissue is larger than in the nontumorous tissue. Statistical Analysis. The results were expressed as means SD. Statistical differences were determined by one-way ANOVA followed by Tukey s test or the Wilcoxon rank-sum test. The correlation coefficient was obtained by linear regression. A P value of less than.05 was considered statistically significant. RESULTS Kupffer-Cell Count Ratios of HCCs and Dysplastic Nodules. The Kupffer-cell count ratios of HCCs and dysplastic nodules are shown in Fig. 1. Kupffer-cell count ratios declined as the HCCs became less well-differentiated. Kupffer-cell count ratios of well-differentiated HCCs, moderately differentiated

3 HEPATOLOGY Vol. 32, No. 2, 2000 IMAI ET AL. 207 FIG. 1. Kupffer-cell count ratios in HCCs and dysplastic nodules. Kupffer-cell count ratios of well-differentiated HCCs were significantly higher than those of both moderately and poorly differentiated HCCs (P.001 and P.001, respectively). Kupffer-cell count ratios of dysplastic nodules were also significantly higher than those of both moderately and poorly differentiated HCCs (P.001 and P.001, respectively), though there was no significant difference in Kupffer-cell count ratios between well-differentiated HCCs and dysplastic nodules. HCCs, and poorly differentiated HCCs were , , , respectively. Kupffer-cell count ratios of well-differentiated HCCs were significantly higher than those of both moderately and poorly differentiated HCCs (P.001 and P.001, respectively). In all of the moderately and poorly differentiated HCCs, Kupffer-cell count ratios were lower than 1, indicating that the numbers of Kupffer cells in the tumorous tissue were smaller than that of the nontumorous tissue. In contrast, Kupffer-cell count ratios in 6 of the 13 well-differentiated HCCs were higher than 1. The Kupffer-cell count ratio of dysplastic nodules was Kupffer-cell count ratios of dysplastic nodules were significantly higher than those of both moderately and poorly differentiated HCCs (P.001 and P.001, respectively). However, there was no significant difference in the Kupffercell count ratios between well-differentiated HCCs and dysplastic nodules (Fig. 1). Correlation Between SPIO Intensity Ratios and Kupffer-Cell Count Ratios in HCCs and Dysplastic Nodules. The authors also investigated the correlation between Kupffer-cell count ratios and SPIO intensity ratios. The authors found an inverse correlation between the SPIO intensity ratios and the Kupffercell count ratios (r.826, P.001, Fig. 2), suggesting that SPIO-enhanced MR images reflected Kupffer-cell counts very well in HCCs and dysplastic nodules. SPIO Intensity Ratios of HCCs and Dysplastic Nodules. As the degree of histological differentiation of HCCs decreased, the SPIO intensity ratio increased, indicating that the accumulation of SPIO in HCCs had declined (Fig. 3). SPIO intensity ratios of well-differentiated HCCs, moderately differentiated HCCs, and poorly differentiated HCCs were , , , respectively. SPIO intensity ratios of well-differentiated HCCs were significantly lower than those of both moderately and poorly differentiated HCCs (P.001 and P.001, respectively). In all of the moderately and FIG. 2. Correlation between Kupffer-cell count ratios and SPIO intensity ratios on fast multiplanar spoiled gradient recalled acquisition in the steady state presession sequence (fast SPGR) images in HCCs (F) and dysplastic nodules (E). There was an inverse correlation between the SPIO intensity ratio and Kupffer-cell count ratios (r.826, P.001). poorly differentiated HCCs, SPIO intensity ratios were above 1, indicating less uptake of SPIO in the tumorous tissue than in the nontumorous tissue (Fig. 4). Six of the thirteen welldifferentiated HCCs showed SPIO intensity ratios of less than 1 (Fig. 5), indicating more uptake of SPIO relative to the nontumorous tissue. SPIO intensity ratio of dysplastic nodules was Two of the six dysplastic nodules showed low intensities (Fig. 6), whereas the others were FIG. 3. SPIO intensity ratios of HCC and dysplastic nodules on fast SPGR images. The SPIO intensity ratios of well-differentiated HCCs were significantly lower than those of both moderately and poorly differentiated HCCs (P.001 and P.001, respectively). SPIO intensity ratios of dysplastic nodules were also significantly lower than those of both moderately and poorly differentiated HCCs (P.001 and P.001, respectively). However, there was no statistical difference in SPIO intensity ratios between well-differentiated HCCs and dysplastic nodules.

4 208 IMAI ET AL. HEPATOLOGY August 2000 FIG. 4. A representative case of a moderately differentiated HCC. The tumor (arrow), 2.2 cm in diameter, showed hypointensity on unenhanced fast SPGR MR image (A) and hyperintensity on SPIO-enhanced fast SPGR MR image (B). The SPIO intensity ratio was 3.78; indicating lower uptake of SPIO in the HCC. There were fewer CD68-positive Kupffer cells in the tumorous tissue (C) than in the nontumorous tissue (D). (Counterstained with hematoxylin; original magnification 100.) The Kupffer-cell count ratio was isointense on SPIO-enhanced MR images. As shown in Fig. 3, there was no statistical difference in SPIO intensity ratios between well-differentiated HCCs and dysplastic nodules. Therefore, it was difficult to distinguish well-differentiated HCCs from dysplastic nodules by SPIO-enhanced MR imaging. SPIO intensity ratios of dysplastic nodules were significantly lower than those of both moderately and poorly differentiated HCCs (P.001 and P.001, respectively). SPIO Intensity Ratio and Tumor Vascularity. Of the 29 HCCs, 21 showed hyperenhancement on CTHA and hypoperfusion on CTAP, 6 showed isoenhancement relative to the surrounding liver parenchyma on both CTHA and CTAP, 1 showed hypoenhancement on CTHA and isoenhancement on CTAP, and 1 showed hypoenhancement on CTHA and hyperenhancement on CTAP. All of the HCCs showing isoenhancement or hypoenhancement on CTHA were well-differentiated HCCs. The SPIO intensity ratios of HCCs showing hyperenhancement on CTHA were significantly higher than those of HCCs showing isoenhancement or hypoenhancement on CTHA ( vs , P.001) (Fig. 7). All of the 6 dysplastic nodules showed isoenhancement on both CTHA and CTAP, and these tumors showed isointensity or hypointensity on SPIO-enhanced MR images. DISCUSSION In this study, the authors carried out immunohistochemical studies of Kupffer cells in HCCs and dysplastic nodules using the anti-cd68 antibody. The authors found that the ratio of the Kupffer-cell count of the tumorous tissue relative to that of the nontumorous tissue (Kupffer-cell count ratio) decreased as the degree of differentiation of HCCs declined, and that most of the well-differentiated HCCs had a similar number of Kupffer cells in the tumorous and nontumorous tissues. These findings are compatible with the previous report by Tanaka et al., 10 which revealed that two-thirds of well-differentiated HCCs contained Kupffer cells similar in number to

5 HEPATOLOGY Vol. 32, No. 2, 2000 IMAI ET AL. 209 FIG. 5. A representative case of a well-differentiated HCC that took up SPIO in the tumor. The tumor (arrow), 2.0 cm in diameter, showed isointensity on unenhanced fast SPGR MR image (A) and hypointensity on SPIO-enhanced fast SPGR MR image (B). The SPIO intensity ratio was 0.95; indicating a slightly greater uptake of SPIO in the HCC. There was almost the same number of CD68-positive Kupffer cells in the tumorous tissue (C) as in the nontumorous tissue (D). (Counterstained with hematoxylin; original magnification 100.) The Kupffercell count ratio was the surrounding nontumorous tissues, and with the study by Sakamoto et al., 19 which reported that very well-differentiated HCCs have higher cellularity than nontumorous tissue. 19 One of the purposes of this study was to investigate whether SPIO-enhanced MR imaging reflects the number of Kupffer cells in HCCs and dysplastic nodules. The authors have clearly shown, for the first time, that the ratio of the intensity of the tumorous lesion relative to that of the nontumorous area in SPIO-enhanced MR images (SPIO intensity ratio) correlated well with the ratio of the Kupffer cell count of the tumorous lesion to that of the nontumorous area (Kupffercell count ratio) in HCCs and dysplastic nodules. Moreover, the authors investigated the relation between SPIO intensity ratios and histological grading of HCCs and found that SPIO intensity ratios increased as the degree of differentiation of HCCs declined. This is in agreement with our other results showing that the Kupffer-cell count ratio decreased as the degree of differentiation of HCCs declined. Accordingly, the histological grade of HCC could be predicted by using SPIOenhanced MR imaging. However, to see whether SPIOenhanced MR imaging truly reflects the functions of Kupffer cells in HCCs and dysplastic nodules, the ability of Kupffer cells to take up SPIO, as well as the numbers of Kupffer cells, has to be considered. In this respect, further studies are needed. Heretofore, there has been no report on SPIO-enhanced MR imaging in patients with dysplastic nodules. In 1995, the International Working Group on terminology of nodular hepatocellular lesions designated formerly adenomatous hyperplasia or atypical adenomatous hyperplasia as dysplastic nodules. 21 In our study, we used this definition and investigated MR images of dysplastic nodules before and after SPIO administration. As has been reported by Tanaka et al., 10 we also found that dysplastic nodules had a similar or higher numbers of Kupffer cells when compared with the surrounding tissue. Consistent with this immunohistochemical study, dysplastic nodules showed hypointensity or isointensity relative to the surrounding liver on SPIO-enhanced MR images. One case of

6 210 IMAI ET AL. HEPATOLOGY August 2000 FIG. 6. A representative case of a dysplastic nodule. The tumor (arrow), 0.8 cm in diameter, showed isointensity on unenhanced fast SPGR MR images (A) and hypointensity on SPIO-enhanced fast SPGR MR image (B). SPIO intensity ratio was 0.74; clearly indicating a greater uptake of SPIO in the dysplastic nodule. There were more CD68-positive Kupffer cells in the tumorous tissue (C) than in the nontumorous tissue (D). (Counterstained with hematoxylin; original magnification 100.) The Kupffercell count ratio was well-differentiated HCC, which appeared hypointense relative to the surrounding liver tissue on MR images after administration of SPIO, was reported by Yamamoto et al., 16 although most of the HCCs in their study appeared hyperintense relative to the surrounding liver tissue on SPIO-enhanced MR images. 16 We also studied SPIO-enhanced MR images of histologically proven well-differentiated HCCs. Interestingly, 6 of 13 well-differentiated HCCs were found to be hypointense relative to the surrounding liver on SPIO-enhanced MR images. As shown in Fig. 3, there was considerable overlap between SPIO intensity ratios of dysplastic nodules and those of well-differentiated HCCs. Thus, dysplastic nodules were not distinguishable from well-differentiated HCCs by SPIOenhanced MR imaging. The authors investigated the relation between SPIOenhanced MR images and tumor vascularity assessed by CT angiography in HCCs, as well as in dysplastic nodules. HCCs usually lack the normal dual hepatic blood supply of the normal liver. Rather, they have a single neovascular arterial blood supply. However, some well-differentiated HCCs, especially those of small size, do not show this characteristic of blood supply to the tumor. Kudo et al., 6 investigated arterial blood flows in HCCs smaller than 3 cm in diameter with US angiography and reported that almost all of the HCCs showing isovascularity or hypovascularity on US angiography were histologically well-differentiated HCCs. Toyoda et al. 9 evaluated blood supply of HCC less than 2 cm in diameter and concluded that portal perfusion to the tumor decreased and that this decrease was followed by an increase in arterial blood flow as HCC grew. Similarly, all of the HCCs in our study that showed isoenhancement or hypoenhancement relative to the liver parenchyma on CTHA were well-differentiated HCCs. The SPIO intensity ratios of HCCs, showing isoenhancement or hypoenhancement on CTHA, were significantly lower than those of HCCs showing hyperenhancement on CTHA. These results may suggest that a decrease in Kupffer-cell numbers occurred in parallel with a change in tumor vascularity in HCCs. In dysplastic nodules, both arterial and portal flows

7 HEPATOLOGY Vol. 32, No. 2, 2000 IMAI ET AL. 211 with SPIO-enhanced MR images appearing nearly homogeneous in each case. Two of these HCCs consisted of poorly and moderately differentiated HCCs, in which Kupffer-cell count ratios of the 2 components were similar enough to account for the homogeneous appearance on the MR images. In the 2 remaining HCCs which consisted of well-differentiated HCC as a predominant component and moderately differentiated HCC as a very minor component, the area of moderately differentiated HCC was most likely too small for SPIO-MR imaging to detect, in spite of the fact that Kupffer-cell count ratios were different. In conclusion, SPIO-enhanced MR imaging accurately reflects Kupffer-cell numbers in HCCs and dysplastic nodules, and is useful in predicting histological grading of HCCs. Dysplastic nodules and some well-differentiated HCCs took up more or similar amounts of SPIO in comparison with the surrounding liver tissue on SPIO-enhanced MR images, thus making it difficult to distinguish dysplastic nodules from welldifferentiated HCCs by these methods. FIG. 7. Comparison of SPIO intensity ratios of HCCs showing hyperenhancement on CTHA and those of HCCs showing isoenhancement or hypoenhancement on CTHA. The SPIO intensity ratios of HCCs showing hyperenhancement on CTHA ( ) were significantly higher than those of HCCs showing isoenhancement or hypoenhancement in CTHA ( ) (P.001). were preserved and Kupffer-cell counts were maintained, as estimated by SPIO-enhanced MR imaging. To accurately evaluate the change in intensities of the tumors relative to the surrounding liver on MR images before and after administration of SPIO, a SPIO intensity ratio was formulated based on the fast SPGR images used in this study. For the detection of HCCs by SPIO-enhanced MR imaging, T1-,T2-, and proton-density-weighted spin-echo sequences and gradient-echo sequences, such as fast low-angle shot imaging, have been employed. 16 The authors also studied 4 pulse sequences, including T1-, T2-, and proton-density-weighted spin-echo, and fast SPGR imagings. SPIO intensity ratios were calculated for each tumor (data not shown). Of these pulse sequences, fast SPGR imaging was found to correlate best with Kupffer-cell count ratios. Therefore, the authors used fast SPGR images for the calculation of SPIO intensity ratios. The main advantage of fast SPGR imaging is that fast SPGR images can be obtained while the patients hold their breaths. This eliminates respiratory artifact, which degrades image quality. There was a limitation on diagnosis of HCCs and dysplastic nodules in our study because the diagnoses of about 60% of the tumors were made on the basis of needle biopsy specimens. To ensure accurate histological diagnosis, we performed at least 3 needle-core biopsies for each tumor in the 23 tumors. However, it was not possible to carry out precise correlation between the biopsy sites and the related areas on MR images in the 23 tumors as the specimens were obtained by ultrasound-guided needle biopsy, making determination of the exact location difficult. On the other hand, liver specimens from the 14 tumors were obtained by partial hepatectomy. Of these tumors, 10 appeared histologically uniform. The other 4 tumors were each composed of 2 histological grades of HCCs, REFERENCES 1. Di Bisceglie AM, Rustgi VK, Hoofnagle JH, Dusheiko GM, Lotze MT. Hepatocellular carcinoma. Ann Intern Med 1988;108: Okuda K. Hepatocellular carcinoma: recent progress. HEPATOLOGY 1992; 15: Simonetti RG, Camma C, Fiorello F, Politi F, D Amico G, Pagliaro L. Hepatocellular carcinoma: a worldwide problem and the major risk factors. Dig Dis Sci 1991;36: Imai Y, Kawata S, Tamura S, Yabuuchi I, Noda S, Inada M, Maeda Y, et al. Relation of interferon therapy and hepatocellular carcinoma in patients with chronic hepatitis C. Ann Intern Med 1998;129: Matsui O, Kadoya M, Kameyama T, Yoshikawa J, Takashima T, Nakanuma Y, Unoura M, et al. Benign and malignant nodules in cirrhotic livers: distinction based on blood supply. Radiology 1991;178: Kudo M, Tomita S, Tochino H, Mimura J, Okabe Y, Kashida H, Hirasa M, et al. Small hepatocellular carcinoma: diagnosis with US angiography with intraarterial CO2 microbubbles. Radiology 1992;182: Takayasu K, Muramatsu Y, Furukawa H, Wakao F, Moriyama N, Takayama T, Yamasaki S, et al. Early hepatocellular carcinoma: appearance at CT during arterial portography and CT arteriography with pathologic correlation. Radiology 1995;194: Murakami T, Oi H, Hori M, Kim T, Takahashi S, Tomoda K, Narumi Y, et al. Helical CT during arterial portography and hepatic arteriography for detecting hypervascular hepatocellular carcinoma. AJR 1997;169: Toyoda H, Fukuda Y, Hayakawa T, Kumada T, Nakano S. Changes in blood supply in small hepatocellular carcinoma: correlation of angiographic images and immunohistochemical findings. J Hepatol 1997;27: Tanaka M, Nakashima O, Wada Y, Kage M, Kojiro M. Pathomorphological study of Kupffer cells in hepatocellular carcinoma and hyperplastic nodular lesions in the liver. HEPATOLOGY 1996;24: Saini S, Stark DD, Hahn PF, Bousquet J, Introcasso J, Wittenberg J, Brady TJ, et al. Ferrite particles: a superparamagnetic MR contrast agent for enhanced detection of liver carcinoma. Radiology 1987;162: Stark DD, Weissleder R, Elizondo G, Hahn PF, Saini S, Todd LE, Wittenberg J, et al. Superparamagnetic iron oxide: clinical application as a contrast agent for MR imaging of the liver. Radiology 1988;168: Fretz CJ, Elizondo G, Weissleder R, Hahn PF, Stark DD, Ferrucci Jr JT. Superparamagnetic iron oxide-enhanced MR imaging: pulse sequence optimization for detection of liver cancer. Radiology 1989;172: Fretz CJ, Stark DD, Metz CE, Elizondo G, Weissleder R, Shen J, Wittenberg J, et al. Detection of hepatic metastases: comparison of contrastenhanced CT, unenhanced MR imaging, and iron oxide-enhanced MR imaging. AJR 1990;155: Kawamori Y, Matsui O, Kadoya M, Yoshikawa J, Demachi H, Takashima T. Differentiation of hepatocellular carcinomas from hyperplastic nodules induced in rat liver with ferrite-enhanced MR imaging. Radiology 1992;183:65-72.

8 212 IMAI ET AL. HEPATOLOGY August Yamamoto H, Yamashita Y, Yoshimatsu S, Baba Y, Hatanaka Y, Murakami R, Nishiharu T, et al. Hepatocellular carcinoma in cirrhotic livers: detection with enhanced and iron oxide-enhanced MR imaging. Radiology 1995;195: Nakanuma Y, Terada T, Terasaki S, Ueda K, Nonomura A, Kawahara E, Matsui O. Atypical adenomatous hyperplasia in liver cirrhosis: lowgrade hepatocellular carcinoma or borderline lesion? Histopathology 1990;17: Takayama T, Makuuchi M, Hirohashi S, Hirohashi S, Sakamoto M, Okazaki N, Takayasu K, et al. Malignant transformation of adenomatous hyperplasia to hepatocellular carcinoma. Lancet 1990;336: Sakamoto M, Hirohashi S, Shimosato Y. Early stages of multistep hepatocarcinogenesis: adenomatous hyperplasia and early hepatocellular carcinoma. Hum Pathol 1991;22: Eguchi A, Nakashima O, Okudaira S, Sugihara S, Kojiro M. Adenomatous hyperplasia in the vicinity of small hepatocellular carcinoma. HEPATOL- OGY 1992;15: International Working Party. Terminology of nodular hepatocellular lesions. HEPATOLOGY 1995;22: Terasaki S, Kaneko S, Kobayashi K, Nonomura A, Nakanuma Y. Histological features predicting malignant transformation of nonmalignant hepatocellular nodules: A prospective study. Gastroenterology 1998; 115: