HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS		QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

This Article Abstract Figures Only Full Text (PDF) Submit a response Alert me when this article is cited Alert me when eLetters are posted Alert me if a correction is posted Citation Map Services Email this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Ward, J. Articles by Robinson, P. J. Search for Related Content PubMed PubMed Citation Articles by Ward, J. Articles by Robinson, P. J.

Hepatocellular Carcinoma in the Cirrhotic Liver: Double-Contrast MR Imaging for Diagnosis¹

¹ From the Departments of Radiology, MRI Unit (J.W., J.A.G., D.J.S., J.A., P.J.R.), Medical Physics (D.W.), Hepatology (M.H.D.), and Pathology (J.I.W.), St James's University Hospital, Beckett Street, Leeds LS9 7TF, United Kingdom. From the 1999 RSNA scientific assembly. Received August 13, 1999; revision requested October 7; revision received November 1; accepted November 10. Address correspondence to J.W. (e-mail: Janice.Ward@gw.sjsuh.northy.nhs.uk).

	ABSTRACT

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

 
PURPOSE: To measure the sensitivity and accuracy of double-contrast magnetic resonance (MR) imaging for the diagnosis of hepatocellular carcinoma (HCC) in the cirrhotic liver.

MATERIALS AND METHODS: Twenty-seven patients with MR features of dysplastic nodules and/or HCC were examined. T2-weighted spin-echo and T1-weighted gradient-echo imaging was performed before and after superparamagnetic iron oxide (SPIO) administration and immediately followed by T1-weighted gradient-echo imaging at 10, 40, and 120 seconds after bolus injection of a gadolinium-based contrast material. Nonenhanced, nonenhanced plus SPIO-enhanced, and nonenhanced plus SPIO-enhanced plus gadolinium-enhanced images were reviewed. Alternative–free response receiver operating characteristic (ROC) methodology was used to analyze the results, which were correlated with histopathologic findings after transplantation in 15 patients and at biopsy in 12. Lesions visualized with all three techniques were characterized as a dysplastic nodule or HCC, and ROC analysis was performed.

RESULTS: For all observers, SPIO-enhanced MR imaging (mean accuracy, 0.76) was more accurate than nonenhanced MR imaging (mean accuracy, 0.64) (P < .04), and double-contrast MR imaging (mean accuracy, 0.86) was more accurate than SPIO-enhanced imaging (P < .05). Both types of lesions were correctly characterized with all three techniques, although observer confidence for lesion characterization was greatest with double-contrast MR imaging.

CONCLUSION: Double-contrast MR imaging significantly improves the diagnosis of HCC compared with SPIO-enhanced and nonenhanced imaging (P < .01).

Index terms: Gadolinium • Iron • Liver neoplasms, 761.31, 761.323 • Liver neoplasms, MR, 761.121411, 761.121412, 761.121416, 761.12143 • Magnetic resonance (MR), comparative studies, 761.121411, 761.121412, 761.121416, 761.12143 • Magnetic resonance (MR), contrast enhancement, 761.12143 • Magnetic resonance (MR), contrast media, 761.12143

	INTRODUCTION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

 
Cirrhosis is a diffuse liver disease that is characterized by fibrosis and nodular regeneration. Hepatocellular carcinoma (HCC) develops against a background of cirrhosis by means of a multistep dedifferentiation process that progresses from regenerative nodule to dysplastic nodule and then to HCC (1–4). Unlike in patients who have a normal liver and HCC, which may be treated successfully with hepatic resection or tumor ablation, in patients with cirrhosis, transplantation is the most effective treatment for HCC if the diagnosis is made at an early stage. The decision to proceed to transplantation is based on the number and size of lesions. Solitary HCCs that are smaller than 2 cm are usually treated successfully with transplantation, whereas patients with up to three discrete lesions that are smaller than 3 cm or solitary 2–5-cm lesions have a 75% 5-year survival rate (5). Transplantation is of no benefit when the lesions are diffuse or multiple because metastatic disease is usually present. Accurate preoperative imaging for the detection of HCC is therefore crucial, but because the cirrhotic liver is structurally abnormal, HCCs are often difficult to detect with conventional imaging methods. Furthermore, dysplastic nodules should be distinguished from HCCs for careful monitoring or for treatment with percutaneous alcohol ablation, which has been shown to inhibit the progression of dysplastic nodules to HCC (6–8).

Dual-phase helical computed tomography (CT) is a sensitive technique for the detection of HCC in patients with clinically suspected disease (9). However, in a study of 200 patients with cirrhosis in whom there was no clinical suspicion of HCC, the sensitivity of conventional dynamic CT was poor (10), although the sensitivity of dual-phase helical CT is likely to be higher. The use of magnetic resonance (MR) imaging enhanced with either superparamagnetic iron oxide (SPIO) or dynamic gadolinium-based contrast material has been shown to improve both the detection and characterization of HCC (11–14). Although the use of gadolinium-based contrast material results in increased signal intensity within the tissues in which it accumulates at T1-weighted imaging, SPIO enhances T2 relaxation by increasing local magnetic field inhomogeneity, and this results in decreased signal intensity on T1- and T2-weighted images. As lesions progress from regenerative nodules to dysplastic nodules and then to HCCs with increasing dedifferentiation, the uptake of SPIO is reduced, the portal venous blood supply decreases, and the arterial blood supply increases. In one study (15), with the use of an iron colloid preparation that causes less of a T2* shortening effect than does SPIO, the images obtained several minutes after the injection of gadopentetate dimeglumine and the iron colloid showed improved lesion characterization because the internal structure of the tumors was better seen, but there was no improvement in detection.

We hypothesized that by combining SPIO with rapid sequential imaging immediately after the administration of a gadolinium-based contrast material, we would further improve the detection of HCC. Using a double contrast material MR imaging technique—that is, with SPIO and bolus injection of a gadolinium-based contrast material—we undertook a prospective multiobserver study involving patients with cirrhosis who were considered to be at high risk for occult HCC.

	MATERIALS AND METHODS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

 
Patients
Between April 1997 and May 1999, 142 liver transplantation candidates who were either considered to be at risk of developing HCC or suspected of having HCC on the basis of prior imaging findings or elevated -fetoprotein levels underwent double-contrast MR imaging. Patients who had high levels of iron deposition within the liver due to hemochromatosis, according to liver biopsy results, and abnormally low signal intensity on nonenhanced images were excluded. From the 142 patients who underwent double-contrast MR imaging, patients were preselected for entry into this study on the basis of MR imaging features of HCC and/or dysplastic nodules (n = 28). One patient with small dysplastic nodules too numerous (>100) for accurate correlation and who underwent transplantation was later excluded. Therefore, the final study group consisted of 27 patients (16 men, 11 women; mean age, 60 years; age range, 29–79 years) with imaging features of HCC and/or dysplastic nodules at double-contrast MR imaging. The underlying cirrhosis was related to viral hepatitis type C in nine, viral hepatitis type B in five, granulomatous hepatitis in one, primary biliary cirrhosis in one, autoimmune chronic active hepatitis in three, alcoholic liver disease in four, and an unknown cause in four patients. Local ethical committee approval was granted, and written informed consent was obtained from each patient prior to entry into the study.

MR Imaging
All MR imaging was performed at a field strength of 1.0 T (Magnetom 42SP; Siemens, Erhlangen, Germany) with use of a body coil for transmission and reception of the signal. The nonenhanced images obtained before the injection of SPIO consisted of conventional, T2-weighted dual-echo images (repetition time [TR] msec/echo time [TE] msec, 2,000/45, 90; two signals acquired; matrix, 144 x 256) and T1-weighted spoiled gradient-echo (fast low-angle shot [FLASH]) in-phase (156/6; flip angle, 80°) and opposed-phase (135/4; flip angle, 80°) images (one signal acquired, 128 x 256 matrix). After SPIO administration, the same dual-echo and T1-weighted opposed-phase sequences were performed. For the SPIO injection, ferumoxides (Endorem; Laboratoire Guerbet, Roissy, France) was administered at a dose of 15 µmol of iron per kilogram of body weight diluted in 100 mL of 5% glucose and infused for 30 minutes; imaging commenced approximately 60 minutes after the end of the intravenous infusion.

Immediately after the acquisition of the SPIO-enhanced images, gadopentetate dimeglumine (Magnevist; Schering, Berlin, Germany) or gadodiamide (Omniscan; Nycomed Amersham, Birmingham, United Kingdom) was injected as a rapid bolus at a dose of 0.1 mmol per kilogram of body weight and immediately followed by a 20-mL saline flush. T1-weighted opposed-phase images were then obtained to coincide with the acquisition of the central lines of k space at 10, 40, and 120 seconds after the injection. For all sequences, an 8–10-mm section thickness with a 10%–30% intersection gap was used, and the field of view was rectangular and 35–40 cm, depending on the size of the liver. In each patient these factors were always the same for all sequences performed before and after contrast material administration. For the T1-weighted gradient-echo sequence, 15 sections were obtained during a 19-second breath hold to cover the entire liver.

Qualitative Analysis
Three separate sets of images were analyzed: nonenhanced images; nonenhanced and SPIO-enhanced images combined; and nonenhanced, SPIO-enhanced, and gadolinium-based contrast material–enhanced images combined (ie, double-contrast images). The analysis was undertaken in two parts. For lesion detection, each set of images was viewed independently by four observers (J.W., J.A.G., D.J.S., J.A.) who were blinded to the results of all the other imaging findings, the readings of the other observers, and the final diagnosis. Each observer recorded the presence and segmental location of one or more lesions on the basis of a four-point confidence rating scale on which 1 was defined as probably not a lesion; 2, a possible lesion; 3, a probable lesion; and 4, a definite lesion.

Alternative–free response receiver operating characteristic (ROC) analysis of all lesions was performed with each set of images and each observer (16). Conventional ROC methodology does not allow the recording of multiple responses per image, whereas alternative–free response ROC analysis allows positional information to be recorded and enables all of the observers' responses to be correlated with all of the lesions present. In addition, the sensitivity of each observer with each technique was assessed on the basis of lesions that were given confidence ratings of 3 or 4. These results were correlated with the histopathologic findings after transplantation in 15 patients and at lesion biopsy in 12 patients who were unsuitable for surgical management.

Biopsy was typically performed on only one lesion in each patient; when multiple lesions were present, those lesions that had the same imaging characteristics as the lesion on which biopsy was performed (on the basis of lesion signal intensity characteristics on nonenhanced images and enhancement features after SPIO and gadolinium-based contrast material administration) and that did not demonstrate the MR features of cysts, hemangiomas, or fibrosis were considered to have the same histopathologic features.

To achieve an accurate correlation between the scored lesions and those confirmed by using histopathologic analysis, at the time of scoring a grid reference was used, and each observer also recorded the individual image number, segment location, and size of each lesion. After transplantation and following dissection of the explanted liver in the transverse plane into 1-cm sections, pathologic correlation was performed by the same experienced hepatobiliary pathologist (J.I.W.). After macroscopic identification of any suspected lesion, a histologic examination with reticulin staining was performed to characterize the lesions. Transplantation was performed between 5 days and 35 weeks (mean, 10 weeks) after imaging. Before surgery, the pathologist was notified of all the patients who had undergone double-contrast MR so that any patient with lesions that were identified at histologic analysis but not at imaging could also be included in the study. In addition to the 15 patients in whom lesions were seen at imaging, 58 patients who had normal double-contrast MR imaging findings underwent transplantation and histologic examination, but no lesions were found in these cases. In 12 patients who did not undergo transplantation because of extensive disease, the results were correlated with the results of a four-panel consensus review combined with all the other imaging and clinical follow-up data.

For characterization, only those lesions that were seen with all three MR imaging techniques and histologically proved were included and identified to the observers. All four observers then viewed the same three sets of images separately and independently characterized the lesions as HCC or dysplastic nodules by using the criteria listed in Table 1 (17,18). A score of 5 indicated high confidence that the lesion was HCC; 4, that the lesion was probably HCC; 3, that the lesion was possibly HCC or a dysplastic nodule; 2, that the lesion was probably a dysplastic nodule; and 1, high confidence that the lesion was a dysplastic nodule.

The interobserver variability for lesion detection with each technique was assessed by using statistic analysis. The values below 0.40 were considered to be indicative of poor correlation; of 0.41–0.75, good correlation; and above 0.75, excellent correlation.

Quantitative Analysis
By using user-defined regions of interest in the lesions, adjacent liver parenchyma, and background noise (anterior to and in line with the lesion in the phase-encoding direction), lesion-to-liver contrast-to-noise measurements for each sequence were calculated as follows: (signal intensity lesion - signal intensity liver)/signal intensity noise.

On the basis of values recorded on T2-weighted images obtained before and after SPIO administration and on T1-weighted images obtained before and after gadolinium-based contrast material administration, the percentage of signal intensity change in the background liver parenchyma and in the hepatic lesions was calculated as follows: [(postcontrast signal intensity - precontrast signal intensity) x 100]/precontrast signal intensity.

Identical regions of interest were used for each sequence and placed as accurately as possible when slight changes in position were encountered owing to variations in respiration or patient movement; all regions of interest were placed by the same observer. Measurements were performed in all the histopathologically confirmed dysplastic nodules and in the largest HCC when multiple lesions were present. The regions of interest were placed to encompass as much of the lesion as possible while avoiding necrosis or scarring within the larger lesions. The regions of interest larger than 2 cm were used to measure the signal intensity of the liver parenchyma and background noise. Data were expressed as the mean ± SD, and the observed differences were assessed for statistical significance by using the Mann-Whitney U test.

	RESULTS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

 
Eighty-six lesions—76 HCCs and 10 dysplastic nodules—were found in 25 patients. Thirty-four were confirmed at histopathologic analysis following transplantation, and 52 were confirmed at biopsy and with associated imaging findings. Twenty of the 86 lesions were smaller than 10 mm (Table 2). Of the 76 HCCs, 11 were well differentiated and 65 were moderately to poorly differentiated. The lesions were focal in 21 patients and diffuse in four. In an additional two patients who were thought to have HCCs at imaging and who underwent liver transplantation, lesions were not found at histopathologic analysis. The maximum number of lesions in any one patient was nine.

View larger version (32K): [in this window] [in a new window] [Download PPT slide]   Figure 1. Histogram illustrates the mean alternative-free response ROC values and the area under the alternative-free response ROC curve for each observer with each technique, for analysis of all 86 lesions. The mean difference in these values between nonenhanced and SPIO-enhanced MR imaging was 0.13 (95% CI: 0.09, 0.17; P = .002). The mean difference in these values between SPIO-enhanced and double-contrast MR imaging was 0.09 (95% CI: 0.02, 0.17; P = .03). The gray bars represent values for nonenhanced imaging; black bars, values for SPIO-enhanced imaging; and white bars, values for double-contrast MR imaging. Obs = observer.

Sensitivity.—The mean sensitivities, sensitivities of each observer with each technique, and 95% CIs for the differences in the mean sensitivities of each technique are shown in Figure 2. For all observers, at confidence thresholds of 3 and 4, SPIO-enhanced MR imaging was significantly more sensitive than nonenhanced imaging (P < .01), and double-contrast MR imaging was significantly more sensitive than SPIO-enhanced MR imaging (P < .01). In the subgroup of patients who underwent transplantation, the mean sensitivity of SPIO-enhanced MR imaging (53% [18 of 34 lesions]) was significantly greater than that of nonenhanced imaging (41% [14 of 34 lesions]); the mean difference between nonenhanced and SPIO-enhanced imaging was 14.725 (95% CI: 8.1, 21.3; P = .006). Although double-contrast MR imaging was more sensitive than SPIO-enhanced MR imaging for all four observers, the improvement was not statistically significant when the mean sensitivity of the four observers was considered (65% [22 of 34 lesions]).

View larger version (32K): [in this window] [in a new window] [Download PPT slide]   Figure 2. Histogram illustrates the mean sensitivities and the sensitivity of each observer with each technique, for analysis of all 86 lesions. The mean number of lesions detected and the number of lesions detected by each observer are indicated above the histogram bars. The mean difference in these values between nonenhanced and SPIO-enhanced MR imaging was 18% (95% CI: 10.8%, 24.7%; P = .004). The mean difference in these values between SPIO-enhanced and double-contrast MR imaging was 16% (95% CI: 7.8%, 23.5%; P = .008). The gray bars represent values for nonenhanced imaging; black bars, values for SPIO-enhanced imaging; and white bars, values for double-contrast MR imaging. Obs. = observer.

False-positive cases.—A total of 668 true-positive interpretations and 46 false-positive interpretations (13 at nonenhanced imaging, 18 at SPIO-enhanced imaging, 15 at double-contrast imaging) were recorded at a confidence level of 3 or 4, to give a false-positive rate of 6.8%. At retrospective review, 22 of the false-positive interpretations were attributed to fibrosis (size range, 1.0–3.0 cm); 16, to regenerative nodules (1.0–1.5 cm); four, to partial volume averaging (1.0–2.0 cm); and four, to vessels (<1 cm). Twelve of the 22 false-positive interpretations attributed to fibrosis and 14 of the 16 attributed to regenerative nodules arose from single lesions in four patients (two patients in each category). Those attributed to fibrosis were scored by multiple observers on only the nonenhanced and SPIO-enhanced images, and those attributed to regenerative nodules were scored by multiple observers on all three types of images. The false-positive interpretations attributed to partial volume averaging or vessels were never scored by more than one observer on more than one type of image.

Lesion Characterization
Thirty-nine lesions—30 HCCs and nine dysplastic nodules—that were seen with all three MR imaging techniques were analyzed. For the 30 HCCs, the number of observer responses, from a total of 120 responses, with characterization scores of 4 or 5 were 86, 108, 116 for the nonenhanced images, SPIO-enhanced images, and double-contrast MR images, respectively. For the nine dysplastic nodules, there were 6, 24, and 34 responses, out of a total of 36 responses, with scores of 1 for the nonenhanced images, SPIO-enhanced images, and double-contrast MR images, respectively. The mean areas under the ROC curves for dysplastic nodules and HCCs, respectively, were 0.95 and 0.95 for the nonenhanced images, 0.98 and 0.99 for the SPIO-enhanced images, and 0.99 and 0.99 for the double-contrast MR images. There was no statistically significant difference between the techniques for lesion characterization.

The contrast-to-noise ratios for each sequence are shown in Table 5. For HCCs, the contrast-to-noise ratio obtained at SPIO-enhanced imaging with a long TR and short TE was significantly greater than that with any other sequence (P < .01); among the images obtained after gadolinium-based contrast material administration, the highest contrast-to-noise ratio was obtained on those images acquired 10 seconds after injection. The highest contrast-to-noise ratio for dysplastic nodules was obtained on the images acquired 120 seconds after gadolinium-based contrast material administration (significantly greater than the ratios on the long TR/short TE precontrast, pre–in-phase, long TR/long TE postcontrast, T1-weighted postcontrast, and 10-second post–gadolinium enhancement images; P < .05).

	DISCUSSION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

 
Our study results demonstrate a statistically significant improvement in the detection of HCC with double-contrast MR imaging compared with the improvement with SPIO-enhanced imaging alone. Even when only the lesions 1 cm or larger were analyzed, double-contrast MR imaging was significantly more sensitive than SPIO-enhanced MR imaging. For the detection of lesions smaller than 1 cm, although the mean sensitivity of double-contrast MR imaging was somewhat disappointing, this technique depicted substantially more lesions than did SPIO-enhanced imaging, which had a sensitivity that was only slightly higher than that of nonenhanced imaging (Fig 3). The detection of small lesions was observer dependent, and the two more experienced observers detected more lesions.

View larger version (144K): [in this window] [in a new window] [Download PPT slide]   Figure 3a. Multiple HCCs in a 59-year-old man. Transverse (a) SPIO-enhanced long TR/long TE MR image (2,000/90), (b) T1-weighted FLASH image (135/4) obtained 10 seconds after gadolinium-based contrast material administration, and (c) T1-weighted FLASH image obtained 30 seconds after b clearly show a large, well-differentiated HCC (large solid arrow) with a poorly differentiated focus (small solid arrow). An additional 1-cm HCC (curved arrow in b) is clearly seen on the image obtained during the arterial phase after gadolinium enhancement, but it becomes isointense with the background liver during the portal phase (c). This lesion is not seen in a. Note the focus of high signal intensity in b (open arrow) is a false-positive lesion attributed to partial volume averaging of the stomach.

View larger version (137K): [in this window] [in a new window] [Download PPT slide]   Figure 3b. Multiple HCCs in a 59-year-old man. Transverse (a) SPIO-enhanced long TR/long TE MR image (2,000/90), (b) T1-weighted FLASH image (135/4) obtained 10 seconds after gadolinium-based contrast material administration, and (c) T1-weighted FLASH image obtained 30 seconds after b clearly show a large, well-differentiated HCC (large solid arrow) with a poorly differentiated focus (small solid arrow). An additional 1-cm HCC (curved arrow in b) is clearly seen on the image obtained during the arterial phase after gadolinium enhancement, but it becomes isointense with the background liver during the portal phase (c). This lesion is not seen in a. Note the focus of high signal intensity in b (open arrow) is a false-positive lesion attributed to partial volume averaging of the stomach.

View larger version (128K): [in this window] [in a new window] [Download PPT slide]   Figure 3c. Multiple HCCs in a 59-year-old man. Transverse (a) SPIO-enhanced long TR/long TE MR image (2,000/90), (b) T1-weighted FLASH image (135/4) obtained 10 seconds after gadolinium-based contrast material administration, and (c) T1-weighted FLASH image obtained 30 seconds after b clearly show a large, well-differentiated HCC (large solid arrow) with a poorly differentiated focus (small solid arrow). An additional 1-cm HCC (curved arrow in b) is clearly seen on the image obtained during the arterial phase after gadolinium enhancement, but it becomes isointense with the background liver during the portal phase (c). This lesion is not seen in a. Note the focus of high signal intensity in b (open arrow) is a false-positive lesion attributed to partial volume averaging of the stomach.

All the patients in our study had late-stage cirrhosis, which may have decreased lesion conspicuity, particularly on the SPIO-enhanced images, but of more importance is the fact that we had definitive proof of tumor burden after histopathologic-imaging correlation of the findings in the explanted liver in 15 of 27 patients. We also attempted to minimize any bias that may have been introduced in our group of patients without explant correlation by performing a separate analysis in only those patients who underwent transplantation. The results obtained in this subgroup were similar to those obtained in our entire patient population. In addition, by using alternative–free response ROC methodology, we were able to include every lesion in our analysis, and, consequently, we found that a substantial number (20 of 86) of the lesions were smaller than 1 cm.

Because our double-contrast MR technique includes the acquisition of images after SPIO administration but before gadolinium-based contrast material administration, the detection of hypovascular HCCs, which may not be detected on dynamic gadolinium-enhanced images, is likely to be improved compared with that at dynamic gadolinium-enhanced imaging alone. However, further investigation is required to assess the relative sensitivities of the two techniques. In this study, five lesions in five different patients were detected only on the SPIO-enhanced images, and patient treatment was influenced in three of these five patients (Fig 4). We also found that although focal lesions were typically more conspicuous on the early arterial phase, post–gadolinium enhancement images, diffuse lesions and their extent were better assessed on the SPIO-enhanced images.

View larger version (125K): [in this window] [in a new window] [Download PPT slide]   Figure 4a. Solitary HCC in a 44-year-old man. Transverse (a) nonenhanced long TR/long TE image (2,000/90), (b) SPIO-enhanced long TR/long TE image (2,000/90), (c) T1-weighted FLASH image (135/4) obtained 10 seconds after gadolinium-based contrast material administration, and (d) T1-weighted FLASH image (135/4) obtained 40 seconds after gadolinium enhancement. The solitary, poorly differentiated HCC (arrow in b) is clearly seen on the SPIO-enhanced image, but not in a, c, or d.

View larger version (135K): [in this window] [in a new window] [Download PPT slide]   Figure 4b. Solitary HCC in a 44-year-old man. Transverse (a) nonenhanced long TR/long TE image (2,000/90), (b) SPIO-enhanced long TR/long TE image (2,000/90), (c) T1-weighted FLASH image (135/4) obtained 10 seconds after gadolinium-based contrast material administration, and (d) T1-weighted FLASH image (135/4) obtained 40 seconds after gadolinium enhancement. The solitary, poorly differentiated HCC (arrow in b) is clearly seen on the SPIO-enhanced image, but not in a, c, or d.

View larger version (114K): [in this window] [in a new window] [Download PPT slide]   Figure 4c. Solitary HCC in a 44-year-old man. Transverse (a) nonenhanced long TR/long TE image (2,000/90), (b) SPIO-enhanced long TR/long TE image (2,000/90), (c) T1-weighted FLASH image (135/4) obtained 10 seconds after gadolinium-based contrast material administration, and (d) T1-weighted FLASH image (135/4) obtained 40 seconds after gadolinium enhancement. The solitary, poorly differentiated HCC (arrow in b) is clearly seen on the SPIO-enhanced image, but not in a, c, or d.

View larger version (122K): [in this window] [in a new window] [Download PPT slide]   Figure 4d. Solitary HCC in a 44-year-old man. Transverse (a) nonenhanced long TR/long TE image (2,000/90), (b) SPIO-enhanced long TR/long TE image (2,000/90), (c) T1-weighted FLASH image (135/4) obtained 10 seconds after gadolinium-based contrast material administration, and (d) T1-weighted FLASH image (135/4) obtained 40 seconds after gadolinium enhancement. The solitary, poorly differentiated HCC (arrow in b) is clearly seen on the SPIO-enhanced image, but not in a, c, or d.

All the dysplastic nodules in our study were hypovascular and in some cases difficult to identify on the images obtained after gadolinium enhancement. The nonenhanced images were necessary to visualize the dysplastic nodules, which were isointense or mildly hypointense relative to the background liver on both the SPIO-enhanced and double-contrast MR images. In accordance with the findings in other studies (20–25), all of the dysplastic nodules in our study were hyperintense on nonenhanced T1-weighted images and hypointense on T2-weighted images. Earls et al (25) conducted a study of the signal intensity characteristics of HCC and dysplastic nodules by performing thin-section MR imaging in fresh explanted cirrhotic livers. Although all the dysplastic nodules in their study were hyperintense and hypointense on T1- and T2-weighted images, respectively, most of the HCCs exhibited the same degree of signal intensity, and the authors therefore concluded that accurate differentiation between the two lesions was impossible.

In our study, there was no overlap in signal intensity between dysplastic nodules and HCCs; the HCCs had variable signal intensity on T1-weighted images, but they were all either isointense or hyperintense relative to the background liver on T2-weighted images. Although the dysplastic nodules were typically more conspicuous on the nonenhanced images, careful correlation of the nonenhanced, SPIO-enhanced, and double-contrast MR image findings facilitated the identification of the lesions on all the postcontrast images, and both dysplastic nodules and HCCs could be characterized with greater confidence on the basis of their enhancement characteristics. After SPIO administration, the percentage of signal intensity loss in the dysplastic nodules was significantly greater than that in the HCCs (P < .01), and all the lesions were hypovascular.

Although there may be overlap in the enhancement characteristics of each lesion—that is, well-differentiated HCCs may take up SPIO (12), dysplastic nodules may be hypervascular (26), and HCCs may be hypovascular (27,28)—in our patient population we were able to differentiate the two lesions by using the combined effects of SPIO and gadolinium-based contrast material. The accurate diagnosis of dysplastic nodules is necessary because it is now widely accepted that they represent an intermediate step in the pathogenesis of HCC, and the development of HCC within a dysplastic nodule may occur in as little as 4 months (29–31) (Fig 5). It must be noted also that only the lesions that were visible with all three MR techniques were included in our analysis of characterization. A substantial number of additional lesions that were subsequently identified as HCC were detected on the double-contrast MR images and on a minority of the SPIO-enhanced images.

View larger version (133K): [in this window] [in a new window] [Download PPT slide]   Figure 5a. Focus of HCC within a dysplastic nodule in a 50-year-old man. Transverse (a) nonenhanced long TR/short TE image (2,000/45), (b) SPIO-enhanced long TR/short TE image (2,000/90), and (c) T1-weighted FLASH image (135/4) obtained 10 seconds after gadolinium-based contrast material administration. In a, a high-signal-intensity focus of HCC (small arrow) is seen within a low-signal-intensity dysplastic nodule (large arrow). In b, SPIO is taken up by the dysplastic nodule (large arrow) but not by the focus of HCC (small arrow). In c, the focus of HCC (small arrow) becomes hyperintense relative to the background dysplastic nodule (large arrow) during the early arterial phase of enhancement.

View larger version (139K): [in this window] [in a new window] [Download PPT slide]   Figure 5b. Focus of HCC within a dysplastic nodule in a 50-year-old man. Transverse (a) nonenhanced long TR/short TE image (2,000/45), (b) SPIO-enhanced long TR/short TE image (2,000/90), and (c) T1-weighted FLASH image (135/4) obtained 10 seconds after gadolinium-based contrast material administration. In a, a high-signal-intensity focus of HCC (small arrow) is seen within a low-signal-intensity dysplastic nodule (large arrow). In b, SPIO is taken up by the dysplastic nodule (large arrow) but not by the focus of HCC (small arrow). In c, the focus of HCC (small arrow) becomes hyperintense relative to the background dysplastic nodule (large arrow) during the early arterial phase of enhancement.

View larger version (145K): [in this window] [in a new window] [Download PPT slide]   Figure 5c. Focus of HCC within a dysplastic nodule in a 50-year-old man. Transverse (a) nonenhanced long TR/short TE image (2,000/45), (b) SPIO-enhanced long TR/short TE image (2,000/90), and (c) T1-weighted FLASH image (135/4) obtained 10 seconds after gadolinium-based contrast material administration. In a, a high-signal-intensity focus of HCC (small arrow) is seen within a low-signal-intensity dysplastic nodule (large arrow). In b, SPIO is taken up by the dysplastic nodule (large arrow) but not by the focus of HCC (small arrow). In c, the focus of HCC (small arrow) becomes hyperintense relative to the background dysplastic nodule (large arrow) during the early arterial phase of enhancement.

View larger version (129K): [in this window] [in a new window] [Download PPT slide]   Figure 6a. Solitary HCC in a 67-year-old woman. Transverse (a) SPIO-enhanced long TR/long TE image (2,000/90), (b) T1-weighted FLASH image (135/4) obtained 10 seconds after gadolinium-based contrast material administration, and (c) T1-weighted FLASH image (135/4) obtained 30 seconds after b. In a, the solitary HCC (large arrow in a-c) is indistinguishable from the adjacent fibrosis (small arrows in a and c); this suggests that the lesion is extensive and larger than 5 cm, which is large enough to preclude transplantation. On the image obtained during the arterial phase of enhancement (b), the lesion is distinct from the adjacent fibrosis, which is enhanced during the later portal phase (c).

View larger version (121K): [in this window] [in a new window] [Download PPT slide]   Figure 6b. Solitary HCC in a 67-year-old woman. Transverse (a) SPIO-enhanced long TR/long TE image (2,000/90), (b) T1-weighted FLASH image (135/4) obtained 10 seconds after gadolinium-based contrast material administration, and (c) T1-weighted FLASH image (135/4) obtained 30 seconds after b. In a, the solitary HCC (large arrow in a-c) is indistinguishable from the adjacent fibrosis (small arrows in a and c); this suggests that the lesion is extensive and larger than 5 cm, which is large enough to preclude transplantation. On the image obtained during the arterial phase of enhancement (b), the lesion is distinct from the adjacent fibrosis, which is enhanced during the later portal phase (c).

View larger version (117K): [in this window] [in a new window] [Download PPT slide]   Figure 6c. Solitary HCC in a 67-year-old woman. Transverse (a) SPIO-enhanced long TR/long TE image (2,000/90), (b) T1-weighted FLASH image (135/4) obtained 10 seconds after gadolinium-based contrast material administration, and (c) T1-weighted FLASH image (135/4) obtained 30 seconds after b. In a, the solitary HCC (large arrow in a-c) is indistinguishable from the adjacent fibrosis (small arrows in a and c); this suggests that the lesion is extensive and larger than 5 cm, which is large enough to preclude transplantation. On the image obtained during the arterial phase of enhancement (b), the lesion is distinct from the adjacent fibrosis, which is enhanced during the later portal phase (c).

Only four false-positive interpretations each were attributed to vessels and to partial volume averaging, and none of these cases was scored by more than one observer with more than one type of MR technique. This is in contrast to previous studies (32,33) in which we used SPIO-enhanced MR imaging to detect colorectal liver metastases: We found that the high signal intensity of vascular structures relative to the signal intensity of the liver parenchyma was the most frequent cause of false-positive results. The different findings in the current study are probably reflective of the lower uptake of SPIO in the liver parenchyma in the patients owing to the more advanced stage of their liver disease and of the fact that vessels are less problematic with double-contrast MR imaging because they are reliably distinguished from lesions at dynamic imaging after gadolinium-based contrast material administration.

The potential criticisms of our study require consideration. First, we did not obtain T2-weighted gradient-echo, FLASH, SPIO-enhanced images, which have been shown to be sensitive for the detection of HCC in the cirrhotic liver (11,17). Our current study was commenced at a time when the provisional results of an ongoing study to establish the optimum pulse sequence after SPIO administration in patients with colorectal liver metastases suggested that conventional dual-echo images were superior to FLASH images. The results after the subsequent completion of that study, however, showed FLASH images to be as accurate as conventional dual-echo images (32), and it may be that the inclusion of a FLASH sequence in our current study protocol would have improved the accuracy of SPIO-enhanced imaging.

We now routinely obtain SPIO-enhanced FLASH images as part of our double-contrast MR protocol. Even so, we subjectively prefer long TR/short TE images after SPIO administration, and the results of our quantitative analysis showed a significantly greater contrast-to-noise ratio with long TR/short TE imaging than with all the other sequences (P < .01). However, we now regard FLASH as a valuable complementary sequence because lesions that may be equivocal on conventional spin-echo images owing to respiratory artifacts are frequently seen with greater clarity on breath-hold FLASH images.

Second, in this study, MR imaging was not performed with high-performance gradients or phased-array coils, which may have improved the detection of small lesions owing to an improved contrast-to-noise ratio and improved spatial resolution. However, although improvements in the accuracy of MR are likely with the most recent technology, we do not expect to see a difference in the relative performance of each technique.

In conclusion, in the detection of HCC and dysplastic nodules in the cirrhotic liver, double-contrast MR imaging is significantly more accurate than SPIO-enhanced MR imaging (P < .04), which is more accurate than nonenhanced imaging (P < .01). The sensitivity for detecting lesions smaller than 1 cm is greater with double-contrast MR imaging than with SPIO-enhanced or nonenhanced MR imaging. Post–gadolinium-enhanced images obtained during the arterial and portal phases of enhancement are essential for differentiating HCCs from adjacent fibrosis, which is the most frequent cause of false-positive findings on SPIO-enhanced and nonenhanced images. The accuracy and observer confidence in the characterization of dysplastic nodules and HCCs were greatest with double-contrast MR imaging, although the advantage of this technique over SPIO-enhanced and nonenhanced imaging was not statistically significant.

	ACKNOWLEDGMENTS

 
The authors thank Jane Howard for assistance with the preparation of the manuscript and statistical help, the medical illustration department, and the staff at the MRI unit.

	FOOTNOTES

 
Abbreviations: FLASH = fast low-angle shot, HCC = hepatocellular carcinoma, ROC = receiver operating characteristic, SPIO = superparamagnetic iron oxide, TE = echo time, TR = repetition time

Author contributions: Guarantors of integrity of entire study, J.W., P.J.R.; study concepts, J.W., J.A.G., P.J.R.; study design, J.W.; definition of intellectual content, J.W., P.J.R.; literature research, J.W.; clinical studies, J.W., J.A.G., D.J.S., J.A., M.H.D., P.J.R., J.I.W.; data acquisition, J.W., J.A.G., D.J.S., J.A., J.I.W.; data analysis, J.W.; statistical analysis, D.W.; manuscript preparation, J.W.; manuscript editing, P.J.R., J.A.G., M.H.D., J.W., J.I.W.; manuscript review, J.W.

	REFERENCES

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

 

1. Kondo F, Ebara M, Sugiura N, et al. Histological features and clinical course of large regenerative nodules: evaluation of their precancerous potential. Hepatology 1990; 12:592-598.[Medline]
2. Furuya K, Nakamura M, Yamamoto Y, Togei K, Otsuka H. Macroregenerative nodule of the liver: a clinical pathologic study of 345 autopsy cases of chronic liver disease. Cancer 1988; 61:99-105.[Medline]
3. Sakamoto M, Horihashi S, Shimosato Y. Early stages of multistep hepatocarcinogenesis: adenomatous hyperplasia and early hepatocellular carcinoma. Hum Pathol 1991; 22:172-178.[Medline]
4. Ferrell L, Wright T, Lake J, Roberts J, Ascher N. Incidence and diagnostic features of macroregenerative nodules vs small hepatocellular carcinoma in cirrhotic livers. Hepatology 1992; 16:1372-1381.[Medline]
5. Mazzaferro V, Regalia E, Doci R, et al. Liver transplantation for the treatment of small hepatocellular carcinomas in patients with cirrhosis. N Engl J Med 1996; 334:693-699.[Abstract/Free Full Text]
6. Lencioni R, Bartolozzi C, Caramella D, Di Coscio G. Management of adenomatous hyperplastic nodules in the cirrhotic liver: US follow-up or percutaneous alcohol ablation?. Abdom Imaging 1993; 18:50-55.[Medline]
7. Livraghi T, Sangalli G, Vettori C. Adenomatous hyperplastic nodules in the cirrhotic liver: a therapeutic approach. Radiology 1989; 170:155-157.[Abstract/Free Full Text]
8. Lencioni R, Caramella D, Bartolozzi C, et al. Percutaneous ethanol injection therapy of adenomatous hyperplastic nodules in cirrhotic liver disease. Acta Radiol 1994; 35:138-142.[Medline]
9. Oliver JH, Baron RL, Federle MP, Rockette HE. Detecting hepatocellular carcinoma: value of unenhanced or arterial phase CT imaging or both used in conjunction with conventional portal venous phase contrast enhanced CT imaging. AJR Am J Roentgenol 1996; 167:71-77.[Abstract/Free Full Text]
10. Miller WJ, Baron RL, Dodd , IIIJD, Federle MP. Malignancies in patients with cirrhosis: CT sensitivity and specificity in 200 consecutive transplant patients. Radiology 1994; 193:645-650.[Abstract/Free Full Text]
11. Yamamoto H, Yamashita Y, Yoshimatsu S, et al. Hepatocellular carcinoma in cirrhotic livers: detection with unenhanced and iron oxide–enhanced MR imaging. Radiology 1995; 195:106-112.[Abstract/Free Full Text]
12. Vogl TJ, Hammerstingl R, Schwarz W, et al. Superparamagnetic iron oxide–enhanced versus gadolinium-enhanced MR imaging for differential diagnosis of focal liver lesions. Radiology 1996; 198:881-887.[Abstract/Free Full Text]
13. Peterson MS, Baron RL, Murakami T. Hepatic malignancies: usefulness of acquisition of multiple arterial and portal venous phase images at dynamic gadolinium-enhanced MR imaging. Radiology 1996; 201:337-345.[Abstract/Free Full Text]
14. Yamashita Y, Mitsuzaki K, Yi T, et al. Small hepatocellular carcinoma in patients with chronic liver damage: prospective comparison of detection with dynamic MR imaging and helical CT of the whole liver. Radiology 1996; 200:79-84.[Abstract/Free Full Text]
15. Suto Y, Sminatani Y. Dual contrast magnetic resonance imaging with combined use of positive and negative contrast agent in human hepatocellular carcinoma. Br J Radiol 1995; 68:116-120.[Abstract]
16. Chakraborty DP, Winter LHL. Free-response methodology: alternate analysis of a new observer-performance experiment. Radiology 1990; 174:873-881.[Abstract/Free Full Text]
17. Brown JJ, Naylor MJ, Yagan N. Imaging of hepatic cirrhosis. Radiology 1997; 202:1-16.[Free Full Text]
18. Ohtoma K. Cirrhosis and premalignant nodules. Magn Reson Imaging Clin N Am 1997; 5:331-346.[Medline]
19. Tang Y, Yamashita Y, Arakawa A, et al. Detection of hepatocellular carcinoma arising in cirrhotic livers: comparison of gadolinium- and ferumoxides-enhanced MR imaging. AJR Am J Roentgenol 1999; 172:1547-1554.[Abstract/Free Full Text]
20. Matsui O, Kadoya M, Kameyama T, et al. Adenomatous hyperplastic nodules in the cirrhotic liver: differentiation from hepatocellular carcinoma with MR imaging. Radiology 1989; 173:123-126.[Abstract/Free Full Text]
21. Choi BI, Takayasu K, Han MC. Small hepatocellular carcinomas and associated nodular lesions of the liver: pathology, pathogenesis and imaging findings. AJR Am J Roentgenol 1993; 160:1177-1187.[Abstract/Free Full Text]
22. Murakami T, Kuroda C, Karukawa T, et al. Regenerating nodules in hepatic cirrhosis: MR findings with pathological correlation. AJR Am J Roentgenol 1990; 155:1227-1231.[Abstract/Free Full Text]
23. Itai Y, Ohnishi S, Ohtomo K, et al. Regenerating nodules of liver cirrhosis: MR imaging. Radiology 1987; 165:419-423.[Abstract/Free Full Text]
24. Ohtomo K, Itai Y, Ohtomo Y, Shiga J, Iio M. Regenerating nodules of liver cirrhosis: MR imaging with pathological correlation. AJR Am J Roentgenol 1990; 154:505-507.[Abstract/Free Full Text]
25. Earls JP, Theise ND, Weinreb JC, et al. Dysplastic nodules and hepatocellular carcinoma: thin-section MR imaging of explanted cirrhotic livers with pathologic correlation. Radiology 1996; 201:207-214.[Abstract/Free Full Text]
26. Krinsky GA, Theise ND, Rofsky NM, Mizrachi H, Tepperman LW, Weinreb JC. Dysplastic nodules in cirrhotic liver: arterial phase enhancement at CT and MR imaging—a case report. Radiology 1998; 209:461-464.[Abstract/Free Full Text]
27. Lencioni R, Macalchi M, Caramella D, Bartolozzi C. Small hepatocellular carcinoma: differentiation from adenomatous hyperplasia with color Doppler US and dynamic Gd-DTPA–enhanced MR imaging. Abdom Imaging 1996; 21:41-48.[Medline]
28. Matsui O, Kadoya M, Kameyama T, et al. Benign and malignant nodules in cirrhotic livers: distinction based on blood supply. Radiology 1991; 178:493-497.[Abstract/Free Full Text]
29. Rapaccini G, Pompili M, Caturelli E, et al. Focal ultrasound lesions in liver cirrhosis diagnosed as regenerating nodules by fine-needle biopsy: follow-up of 12 cases. Dig Dis Sci 1990; 35:422-427.[Medline]
30. Takayama T, Makuuchi M, Hirohashi S, et al. Malignant transformation of adenomatous hyperplasia to hepatocellular carcinoma. Lancet 1990; 336:1150-1153.[Medline]
31. Sadek AG, Mitchell DG, Siegelman ES, et al. Early hepatocellular carcinoma that develops within macroregenerative nodules: growth rate depicted at serial MR imaging. Radiology 1995; 195:753-756.[Abstract/Free Full Text]
32. Ward J, Naik KS, Guthrie JA, Wilson D, Robinson PJ. Hepatic lesion detection: comparison of SPIO-enhanced MRI and dual-phase CT with AFROC analysis. Radiology 1999; 210:459-466.[Abstract/Free Full Text]
33. Ward J, Chen F, Guthrie JA, et al. Hepatic lesion detection after superparamagnetic iron oxide enhancement: comparison of five T2-weighted sequences at 1.0 T by using alternative–free response receiver operating characteristic analysis. Radiology 2000; 214:159-166.[Abstract/Free Full Text]

This article has been cited by other articles:

				J. M. Willatt, H. K. Hussain, S. Adusumilli, and J. A. Marrero MR Imaging of Hepatocellular Carcinoma in the Cirrhotic Liver: Challenges and Controversies Radiology, May 1, 2008; 247(2): 311 - 330. [Abstract] [Full Text] [PDF]

				R. F. Hanna, D. A. Aguirre, N. Kased, S. C. Emery, M. R. Peterson, and C. B. Sirlin Cirrhosis-associated Hepatocellular Nodules: Correlation of Histopathologic and MR Imaging Features RadioGraphics, May 1, 2008; 28(3): 747 - 769. [Abstract] [Full Text] [PDF]

				R. F. Hanna, N. Kased, S. W. Kwan, A. C. Gamst, A. C. Santosa, T. Hassanein, and C. B. Sirlin Double-Contrast MRI for Accurate Staging of Hepatocellular Carcinoma in Patients with Cirrhosis Am. J. Roentgenol., January 1, 2008; 190(1): 47 - 57. [Abstract] [Full Text] [PDF]

				S. N. Gandhi, M. A. Brown, J. G. Wong, D. A. Aguirre, and C. B. Sirlin MR Contrast Agents for Liver Imaging: What, When, How RadioGraphics, November 1, 2006; 26(6): 1621 - 1636. [Abstract] [Full Text] [PDF]