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) Alert me when this article is cited 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 Mori, K. Articles by Reiser, M. Search for Related Content PubMed PubMed Citation Articles by Mori, K. Articles by Reiser, M. Social Bookmarking                      What's this? Hotlight (NEW!) What's Hotlight?

Detection of Malignant Hepatic Lesions Before Orthotopic Liver Transplantation: Accuracy of Ferumoxides-Enhanced MR Imaging

¹ Department of Clinical Radiology, Klinikum Grosshadern, Ludwig Maximilias University of Munich, Marchioninistr. 15, D-81377 Munich, Germany.
² Department of Radiology, Radiologisches Zentrum Munchen-Pasing, Pippingerstr. 25, D-81245 Munich, Germany.
³ Department of Surgery, Klinikum Grosshadern, Ludwig Maximilias University of Munich, D-81377 Munich, Germany.
⁴ Department of Medicine II, Klinikum Grosshadern, Ludwig Maximilias University of Munich, D-81377 Munich, Germany.
⁵ Institute of Pathology, Ludwig Maximilias University of Munich, D-81377 Munich, Germany.
⁶ Department of Medical Informatics, Biometrics, and Epidemiology, Klinikum Grosshadern, Ludwig Maximilias University of Munich, D-81377 Munich, Germany.

Received November 5, 2001; accepted after revision March 29, 2002.

 
K. Mori was supported by a grant from the German academic exchange service (Deutscher Akademischer Austauschdienst).

Address correspondence to K. Mori.

Abstract

Top Abstract Introduction Materials and Methods Results Discussion References

 
OBJECTIVE. The purpose of this study was to evaluate the diagnostic accuracy of ferumoxides-enhanced MR imaging for screening malignant hepatic lesions before orthotopic liver transplantation.

MATERIALS AND METHODS. The study comprised 48 patients who underwent MR imaging within 6 months before transplantation. Imaging techniques included unenhanced and ferumoxides-enhanced T1-weighted gradient-echo and T2-weighted fast spin-echo sequences and ferumoxides-enhanced T2^*-weighted gradient-echo sequences. Qualitative and quantitative analyses were performed; the gold standard was the histopathologic reports of explanted livers.

RESULTS. Twenty patients had malignant hepatic lesions, and 24 hepatocellular carcinomas were histopathologically proven. The mean area under the receiver operating characteristic curve and the mean sensitivity were significantly greater for the image sets with ferumoxides-enhanced gradient-echo sequences than for those without these sequences. The mean sensitivity and specificity of all sequences were 85% and 74% on a per-patient basis, respectively. The mean contrast-to-noise ratio was significantly greater for the ferumoxides-enhanced T2^*-weighted gradient-echo sequences than for any other sequences and for the ferumoxides-enhanced T1-weighted gradient-echo sequences than for unenhanced sequences and the ferumoxides-enhanced T2-weighted fast spin-echo sequences.

CONCLUSION. Ferumoxides-enhanced gradient-echo sequences improved the diagnostic accuracy and the sensitivity for detecting malignant hepatic lesions in patients with end-stage cirrhosis of the liver. However, the specificity was not improved even after the administration of ferumoxides because of the false-positive lesions that were mainly the result of fibrotic changes.

Introduction

Top Abstract Introduction Materials and Methods Results Discussion References

 
Orthotopic liver transplantation is a promising treatment for unresectable hepatocellular carcinoma and for end-stage cirrhosis. In patients with hepatocellular carcinoma, the contraindications for orthotopic liver transplantation are lymph node spread, vascular invasion, and distant metastasis. Additionally, the indication criteria should be limited to a single tumor of 5 cm or less in diameter or to no more than three tumors, none of which is more than 3 cm in diameter [1]. Pretransplantation screening for malignant hepatic lesions is therefore crucial to selecting appropriate patients.

The measurement of serum -fetoprotein, sonography, and CT are the common methods of screening. On a per-patient basis, the sensitivities for detecting malignant hepatic lesions are 58-62% for the measurement of serum -fetoprotein, 58-67% for sonography, and 53-68% for CT [2,3,4,5]. When (single-detector) helical CT scanners were used, sensitivity reached 68-91% [2, 6, 7]. CT with iodized oil and CT arterial portography have been also reported as screening methods. On a per-lesion basis, sensitivities were 37-58% for CT with iodized oil [8,9,10] and 85% for CT arterial portography [10]. CT arterial portography was significantly more sensitive than CT with iodized oil [10].

Ferumoxides-enhanced MR imaging is equivalent in sensitivity to CT arterial portography or to a combination of CT arterial portography and CT hepatic arteriography for detecting malignant hepatic lesions [11, 12]. The sensitivity of ferumoxides-enhanced MR imaging has been reported to be approximately 80-93% on a per-lesion basis [11,12,13,14]. To our knowledge, however, no article has described the diagnostic accuracy of ferumoxides-enhanced MR imaging with histopathologic correlation of explanted livers. Theoretically, well-differentiated hepatocellular carcinoma could be difficult to detect on ferumoxides-enhanced MR imaging because such carcinomas contain a similar amount of Kupffer's cells to surrounding liver parenchyma and hence may lose signal on T2-weighted sequences after the administration of the contrast material [15]. Therefore, dynamic gadolinium-enhanced MR imaging might be a better choice for screening than ferumoxides-enhanced MR imaging, as was previously reported [16]. However, according to a study by Krinsky et al. [17] published in 2001, the sensitivity of dynamic gadolinium-enhanced MR imaging was only 54% in patients who had undergone transplantation because the patient cohort was limited to only those who had no known hepatic tumors at the time of MR imaging.

The purpose of our study was to evaluate the efficacy of ferumoxides-enhanced MR imaging as a screening method before orthotopic liver transplantation.

Materials and Methods

Top Abstract Introduction Materials and Methods Results Discussion References

 
Patients
From January 1996 to October 2000, 172 consecutive patients received orthotopic liver transplantation at our hospital. Of these patients, 22 undergoing retransplantation, 15 with acute liver failure, and 13 younger than 18 years were excluded. The remaining 122 patients underwent the following radiologic examinations at our hospital within 6 months before their primary transplantation: ferumoxides-enhanced MR imaging (n=27), CT (n=27), ferumoxides-enhanced MR imaging and CT (n = 21), gadolinium-enhanced MR imaging and CT (n = 3), gadolinium-enhanced MR imaging (n=2), and neither MR imaging nor CT (n=42). Of the 122 patients, 48 had undergone ferumoxides-enhanced MR imaging; those 48 patients constituted our study. The time between MR imaging and transplantation was 11-184 days (mean, 99.7 days). The 48 patients consisted of 40 men and eight women who were 24-62 years old (mean, 49.8 years). Ten of the 48 patients had radiologically proven malignant hepatic lesions at the time of MR imaging. Histopathologically proven diagnoses in 47 of the 48 patients were cirrhosis of the liver caused by viral hepatitis type C (n=19), alcoholic liver disease (n=13), viral hepatitis type B (n = 7), autoimmune hepatitis (n=3), ₁-antitrypsin deficiency (n = 2), viral hepatitis types B and D (n=1), primary sclerosing cholangitis (n = 1), and primary biliary cirrhosis (n = 1). Written informed consent was obtained from each patient before MR imaging.

Imaging Technique
All patients were examined using a 1.5-T unit (Magnetom Vision; Siemens, Erlangen, Germany) with a semiflexible circular polarized array coil. For all patients, T1-weighted gradient-echo sequences (TR/TE, 140/4.1; flip angle, 75°; matrix, 144x256) and T2-weighted fast spin-echo sequences with spectral fat suppression (3300/138; matrix, 116 x 256) were performed before and after the administration of ferumoxides, and T2^*-weighted gradient-echo sequences (140/10; flip angle, 30°; matrix, 115 x 256) were added to the contrast-enhanced imaging protocol. Each acquisition was performed during a breath-hold of 17-20 sec. A section thickness of 8 mm was used with an interval of 2 mm in all sequences. The number of slices was 16-20 depending on the length of the liver. After unenhanced imaging, a dose of 15 µmol/kg of body weight of ferumoxides (Endorem; Laboratoire Guerbet, Rossy, France) was diluted with 100 mL of a 5% glucose solution and infused for 30 min to 1 hr. The interval between the end of the infusion and the initiation of contrast-enhanced scanning was approximately 30-60 min.

Qualitative Analysis
We evaluated four image sets defined in an additive fashion (i.e., an image set comprised the former image set plus a new imaging sequence) (Table 1). Three observers who were unaware of the results of other observers, of the findings at surgery, and of the histopathologic findings reviewed the four image sets in numeric order from set 1 to set 4. The observers recorded sizes, relative signal intensities to surrounding liver parenchyma, and sites according to Couinaud's segments for all focal liver lesions. For each image set, each lesion was characterized with the following five grades: 1, definitely benign; 2, probably benign; 3, possibly malignant; 4, probably malignant; and 5, definitely malignant. The observers assigned one of five confidence levels for the presence of malignant lesions for both liver lobes as follows: 1, definitely absent; 2, probably absent; 3, possibly present; 4, probably present; and 5, definitely present. When a solitary lesion was present in a liver lobe, the confidence level for the presence of malignant lesions in the lobe was judged to be the same number as the characteristic grade of the lesion. When two or more lesions were found in a lobe, the confidence level depended on the highest characteristic grade of the lesions. If the observers did not find any lesions in a lobe, the confidence level assigned was 1 or 2.

Receiver operating characteristic curves for each observer and each image set were calculated on a per-lobe basis using ROCKIT 0.9.1B software (Metz CE, Chicago, IL), which plotted the true-positive fraction against the likelihood of obtaining a false-positive image. The area under each receiver operating characteristic curve (A_z) indicated the overall diagnostic accuracy of observers and image sets. Sensitivity and specificity for the detection of malignant lesions were calculated for each observer and for each image set on a per-patient basis. The patients who had at least one lobe with a confidence level of 3 or greater were defined as patients with positive findings for malignant lesions. Sensitivity for detecting the lesions with histopathologic correlation was also calculated on a per-lesion basis. The mean values of A_z, sensitivity, and specificity were calculated for each image set. The differences of these mean values among the four image sets were assessed using one-way analysis of variance followed by Tukey's multiple-comparisons test. Additionally, the sensitivity of all sequences used (set 4) was also calculated on a per-patient basis except in the 10 patients who had known hepatic tumors at the time of MR imaging to avoid positive bias in accordance with the study by Krinsky et al. [17]. To assess interobserver variability in interpreting images, we measured the degree of agreement between each pair of observers in each image set using kappa statistics with binary data defined in terms of the presence (confidence level 3) or absence (confidence level = 1 or 2) of malignant hepatic lesions on a per-lobe basis. Kappa values of less than 0.40 indicated positive but poor agreement; of 0.41-0.75, good agreement; and of greater than 0.75, excellent agreement [18].

Quantitative Analysis
A region-of-interest analysis was performed for the malignant lesions with histopathologic correlations. Signal intensities of the lesions and surrounding liver parenchyma were measured at the same anatomic level in each sequence, excluding larger vascular structures. The largest possible regions of interest were chosen according to the lesion sizes but were at least 0.4 cm². If a lesion was not visible on a sequence, a region of interest with adequate size and shape was placed in the similar anatomic location as in another sequence in which the lesion was visible. The SD of background signal intensities was measured in the phase-encoding direction, avoiding areas of motion artifacts. To evaluate the conspicuity of malignant lesions in each sequence, the contrast-to-noise ratio was calculated by dividing the difference of signal intensities between lesion and liver parenchyma by the SD of background signal intensities. To assess the signal intensity change after the administration of ferumoxides, the signal-to-noise ratios of lesions and liver parenchyma were calculated on unenhanced and ferumoxides-enhanced T1-weighted gradient-echo and T2-weighted fast spin-echo images by dividing the signal intensity of lesions or liver parenchyma by the SD of background signal intensities. The differences of the mean contrast-to-noise ratios among the five sequences were assessed using one-way analysis of variance followed by Tukey's multiple-comparisons test. The differences of the signal-to-noise ratios between unenhanced and ferumoxides-enhanced MR images were assessed for both lesions and liver parenchyma using the Student's t test for paired groups.

Histopathologic Review
In the pathologic examinations, all explanted livers were sectioned at intervals of 5-10 mm and were inspected visually and by palpation for focal nodular lesions. The presence of the lesions, their sizes, and their locations were recorded. In cases with four or fewer lesions, detailed histopathologic examinations were performed for each lesion; but in cases with five or more lesions, only the largest lesions were described in detail. The underlying liver pathology was also evaluated. Each focal lesion shown on MR imaging was correlated with the corresponding lesion documented in the histopathologic reports according to its location and size. In cases with discrepancies between the findings of MR imaging with all sequences and histopathologic reports, a radiologist and a pathologist reviewed the MR images and the coverglass preparations and discussed the causes of the discrepancies. The lesion sizes and the times between MR imaging and transplantation were compared for the false-negative and the true-positive findings using the Mann-Whitney U test to clarify the influences of the sizes and intervals for oversights of malignant lesions.

Results

Top Abstract Introduction Materials and Methods Results Discussion References

 
Correlation of Radiologic and Histopathologic Findings
In the histopathologic examinations of the 48 explanted livers, 20 patients were diagnosed as having malignant hepatic lesions, including 19 with hepatocellular carcinomas and one with a multifocal cholangiocellular carcinoma. Of the 19 patients with hepatocellular carcinomas, 12 had a solitary lesion, two had two lesions, one had four lesions, and four had five or more lesions. Retrospectively, 24 hepatocellular carcinomas in 19 patients, including all 20 lesions in patients with four or fewer lesions and all four largest lesions in patients with five or more lesions, could be correlated with the findings of MR imaging. Histopathologically, the 24 lesions comprised 15 moderately to poorly differentiated, eight well-differentiated, and one fibrolamellar hepatocellular carcinomas. The lesion sizes in the histopathologic reports ranged from 0.7 to 10.5 cm (mean, 3.2 cm) in diameter. The cholangiocellular carcinoma was a moderately differentiated adenocarcinoma and was not depicted on any sequences in spite of its multifocal nature. In addition, five benign lesions—two dysplastic nodules, two hemangiomas, and a cyst—were found in the histopathologic examinations and were correlated with the findings of MR imaging.

Qualitative Analysis
Of the 96 liver lobes analyzed in the receiver operating characteristic analysis, 23 were proven to include malignant lesions. The individual observers' and mean A_z values of each image set on a per-lobe basis are shown in Table 2. The mean A_z values were significantly greater for sets 3 and 4 than for sets 1 and 2. The individual and mean sensitivities and specificities of each image set on a per-patient basis are shown in Table 3. The mean sensitivity was also significantly greater for sets 3 and 4 than for sets 1 and 2, although no significant differences were noted between any pair of the four image sets for mean specificity. The individual and mean sensitivities of each image set on a per-lesion basis are shown in Table 4. The mean sensitivity was also significantly greater for sets 3 and 4 than for sets 1 and 2. When the patients included were limited to those with no known hepatic tumors at the time of MR imaging, the individual sensitivities for image set 4 were 70% (7/10), 70% (7/10), and 80% (8/10), with a mean value of 73%, on a per-patient basis. The kappa values between each pair of the three observers for each image set are shown in Table 5. All values indicated good or excellent degrees of agreement.

Quantitative Analysis
The 24 hepatocellular carcinomas with histopathologic correlations were evaluated. The mean contrast-to-noise ratios for the five sequences are shown in Table 6. No significant difference was seen between the unenhanced sequences and the ferumoxides-enhanced T2-weighted fast spin-echo sequences. The ferumoxides-enhanced T1-weighted gradient-echo sequences produced a significantly greater mean contrast-to-noise ratio than the unenhanced sequences and the ferumoxides-enhanced T2-weighted fast spin-echo sequences. The mean contrast-to-noise ratio of the ferum-oxides-enhanced T2^*-weighted gradient-echo sequences was significantly greater than that of any other sequences. The mean signal-to-noise ratio of the liver parenchyma decreased significantly with ferumoxides, from 48.9 (95% confidence interval, 40.6-57.2) to 40.7 (34.1-47.3) on T1-weighted gradient-echo images (p = 0.008) and from 14.7 (11.8-17.6) to 10.1 (7.7-12.5) on T2-weighted fast spin-echo images (p = 0.004). On T2-weighted fast spin-echo images, the signal-to-noise ratios of the lesions before and after the administration of ferumoxides were 19.2 (15.7-22.7) and 16.0 (12.6-19.4), respectively, and no significant change was observed (p = 0.074). On T1-weighted gradient-echo images, however, the signal-to-noise ratios of the lesions increased significantly with ferumoxides, from 50.9 (42.9-58.9) to 57.5 (48.1-66.9) (p = 0.040).

False-Negative and False-Positive Lesions
As a result of the individual observations of all used sequences (set 4), five hepatocellular carcinomas in five patients and the occult cholangiocellular carcinoma in one patient were overlooked by at least one observer (Fig. 1A,1B,1C,1D,1E,1F). The lesion sizes ranged from 1.2 to 5.5 cm (mean, 2.9 cm) in diameter, and the time between MR imaging and orthotopic liver transplantation ranged from 11 to 184 days (mean, 111 days). In the other 14 patients whose lesions were diagnosed as true-positive by all three observers, the lesion sizes and the times ranged from 0.7 to 10.5 cm (mean, 3.2 cm) and from 19 to 183 days (mean, 100 days), respectively. The mean lesion sizes and the mean times were not significantly different between the false-negative and the true-positive patients. On the other hand, 22 lesions in 13 tumor-free patients were misdiagnosed as possible malignant lesions with a characteristic grade of 3 or greater by at least one observer (Fig. 2A,2B,2C,2D,2E,2F). The sizes of those lesions ranged from 0.5 to 5 cm (mean, 2.2 cm). According to the review of MR imaging and the histopathologic coverglass preparations, the false-positive findings were attributed to areas of dense fibrous stroma and small regenerative nodules (n = 7), dense fibrous stroma (n = 5), focal fatty deposition (n = 4), partial volume of vessel (n = 4), and focal necrotic change caused by cirrhosis (n = 1). The other one lesion was misdiagnosed as a hepatocellular carcinoma by all three observers and suspected, also retrospectively, to be a neoplastic lesion.

View larger version (148K): [in this window] [in a new window] [as a PowerPoint slide]   Fig. 1A. —47-year-old man with solitary hepatocellular carcinoma related to viral hepatitis type C—induced cirrhosis. Two of three observers missed lesion on image set 4. Transverse unenhanced T1-weighted gradient-echo MR image (TR/TE, 140/4.1) shows no focal lesion.

View larger version (141K): [in this window] [in a new window] [as a PowerPoint slide]   Fig. 1B. —47-year-old man with solitary hepatocellular carcinoma related to viral hepatitis type C—induced cirrhosis. Two of three observers missed lesion on image set 4. Transverse unenhanced T2-weighted fast spin-echo MR image (3300/138) (B) and transverse ferumoxides-enhanced T2-weighted fast spin-echo MR image (3300/138) (C) show no focal lesion. Note severe motion-related artifacts (arrows).

View larger version (157K): [in this window] [in a new window] [as a PowerPoint slide]   Fig. 1C. —47-year-old man with solitary hepatocellular carcinoma related to viral hepatitis type C—induced cirrhosis. Two of three observers missed lesion on image set 4. Transverse unenhanced T2-weighted fast spin-echo MR image (3300/138) (B) and transverse ferumoxides-enhanced T2-weighted fast spin-echo MR image (3300/138) (C) show no focal lesion. Note severe motion-related artifacts (arrows).

View larger version (153K): [in this window] [in a new window] [as a PowerPoint slide]   Fig. 1D. —47-year-old man with solitary hepatocellular carcinoma related to viral hepatitis type C—induced cirrhosis. Two of three observers missed lesion on image set 4. Transverse ferumoxides-enhanced T1-weighted gradient-echo MR image (140/4.1) shows lesion with decreased signal loss. Lesion (arrow) is 1.5 cm in diameter in segment VII.

View larger version (146K): [in this window] [in a new window] [as a PowerPoint slide]   Fig. 1E. —47-year-old man with solitary hepatocellular carcinoma related to viral hepatitis type C—induced cirrhosis. Two of three observers missed lesion on image set 4. Transverse ferumoxides-enhanced T2*-weighted gradient-echo MR image (140/10) shows best contrast between lesion and surrounding liver (arrow). Note fibrotic changes of decreased signal loss (arrowheads) caused by severe cirrhosis.

View larger version (188K): [in this window] [in a new window] [as a PowerPoint slide]   Fig. 1F. —47-year-old man with solitary hepatocellular carcinoma related to viral hepatitis type C—induced cirrhosis. Two of three observers missed lesion on image set 4. Photomicrograph of histologic specimen from nodule shows moderately differentiated hepatocellular carcinoma with tubercular sinusoidal structure (H and E, x50). Insert (lower right) shows foci of clear cell differentiation (H and E, x200).

View larger version (145K): [in this window] [in a new window] [as a PowerPoint slide]   Fig. 2A. —40-year-old man with fibrotic lesion in alcohol-induced cirrhosis. Two of three observers recognized lesion as possibly malignant on image set 4. Transverse unenhanced T1-weighted gradient-echo MR image (TR/TE, 140/4.1) (A), transverse unenhanced T2-weighted fast spin-echo MR image (3300/138) (B), and transverse ferumoxides-enhanced T2-weighted fast spin-echo MR image (3300/138) (C) show no focal lesion.

View larger version (160K): [in this window] [in a new window] [as a PowerPoint slide]   Fig. 2B. —40-year-old man with fibrotic lesion in alcohol-induced cirrhosis. Two of three observers recognized lesion as possibly malignant on image set 4. Transverse unenhanced T1-weighted gradient-echo MR image (TR/TE, 140/4.1) (A), transverse unenhanced T2-weighted fast spin-echo MR image (3300/138) (B), and transverse ferumoxides-enhanced T2-weighted fast spin-echo MR image (3300/138) (C) show no focal lesion.

View larger version (154K): [in this window] [in a new window] [as a PowerPoint slide]   Fig. 2C. —40-year-old man with fibrotic lesion in alcohol-induced cirrhosis. Two of three observers recognized lesion as possibly malignant on image set 4. Transverse unenhanced T1-weighted gradient-echo MR image (TR/TE, 140/4.1) (A), transverse unenhanced T2-weighted fast spin-echo MR image (3300/138) (B), and transverse ferumoxides-enhanced T2-weighted fast spin-echo MR image (3300/138) (C) show no focal lesion.

View larger version (140K): [in this window] [in a new window] [as a PowerPoint slide]   Fig. 2D. —40-year-old man with fibrotic lesion in alcohol-induced cirrhosis. Two of three observers recognized lesion as possibly malignant on image set 4. Transverse ferumoxides-enhanced T1-weighted gradient-echo MR image (140/4.1) (D) and transverse ferumoxides-enhanced T2*-weighted gradient-echo MR image (140/10) (E) show masslike lesion with decreased signal loss. Lesion (arrows) is 4.5 cm in diameter in segment VIII.

View larger version (143K): [in this window] [in a new window] [as a PowerPoint slide]   Fig. 2E. —40-year-old man with fibrotic lesion in alcohol-induced cirrhosis. Two of three observers recognized lesion as possibly malignant on image set 4. Transverse ferumoxides-enhanced T1-weighted gradient-echo MR image (140/4.1) (D) and transverse ferumoxides-enhanced T2*-weighted gradient-echo MR image (140/10) (E) show masslike lesion with decreased signal loss. Lesion (arrows) is 4.5 cm in diameter in segment VIII.

View larger version (157K): [in this window] [in a new window] [as a PowerPoint slide]   Fig. 2F. —40-year-old man with fibrotic lesion in alcohol-induced cirrhosis. Two of three observers recognized lesion as possibly malignant on image set 4. Photomicrograph of histologic specimen from lesion shows cirrhotic liver with dense fibrous stroma, proliferation of bile ducts, and mild chronic inflammation. (H and E, x50)

Discussion

Top Abstract Introduction Materials and Methods Results Discussion References

 
Our results showed that the mean sensitivities of all sequences used (set 4) were 85% and 88% on a per-patient and a per-lesion basis, respectively. These values were significantly greater than those of the unenhanced MR imaging (set 1). In comparison with the sensitivity of 54% with gadolinium-enhanced MR imaging reported by Krinsky et al. [17], the mean sensitivity in our study seems greater on a per-patient basis. However, when only the patients who did not have any known hepatic tumors at the time of MR imaging were included in our study as well as in theirs, the sensitivity decreased to 73%. On the other hand, the specificity was 85% on a per-patient basis in their study, whereas the mean specificity of all used sequences was only 74% in ours. Moreover, the mean specificity was not different from that of the unenhanced MR imaging in our study. Thus, this low value of specificity seems to be a substantial flaw of ferumoxides-enhanced MR imaging for detecting malignant hepatic lesions in patients with end-stage cirrhosis. In most of our patients, the signal intensities of the liver parenchyma were inhomogeneous after the administration of ferumoxides because fibrotic changes caused by liver cirrhosis appeared as areas of decreased signal loss. Most fibrotic changes could be distinguished from malignant lesions because of their linear or stellate configurations. However, when fibrotic changes had masslike shapes, they tended to be misdiagnosed as malignant lesions. According to our radiologic and histopathologic review of false-positive lesions, 12 (55%) of the 22 lesions were attributed to fibrotic changes caused by cirrhosis. The observers' responses for this kind of signal decay were relatively variable because we had not set any diagnostic criteria for such lesions. The specificities for the three observers were therefore relatively different from each other, although the sensitivities were similar. Recently, the combination of ferumoxides- and dynamic gadolinium-enhanced MR imaging was reported to be a more accurate method for detecting hepatocellular carcinoma than ferumoxides-enhanced MR imaging alone [19]. Further studies are needed for confirmation of these results.

In our study, both qualitative and quantitative analyses were performed to evaluate the diagnostic accuracy of each image set and the lesion conspicuity of each sequence. The ferumoxides-enhanced T2-weighted fast spin-echo sequences alone improved neither mean A_z value nor mean sensitivity compared with the unenhanced imaging. Ward et al. [20] considered the following three reasons for the poor diagnostic accuracy of ferumoxides-enhanced T2-weighted fast spin-echo sequences: low contrast-to-noise ratio, signal loss of both liver parenchyma and focal lesions due to magnetization transfer effects, and low sensitivity to the susceptibility effects of ferumoxides. In accordance with their consideration, the mean contrast-to-noise ratio of the ferumoxides-enhanced T2-weighted fast spin-echo sequence was significantly lower than that of the gradient-echo sequences in our study. Moreover, even after the administration of ferumoxides, no significant improvement in the mean contrast-to-noise ratio was observed on ferumoxides-enhanced T2-weighted fast spin-echo sequences in spite of the significant decrease of the parenchymal signal-to-noise ratio. The magnetization transfer effects might explain this phenomenon because the mean signal-to-noise ratio of the lesions tended to decrease after the administration of ferumoxides. However, the difference did not reach the level of significance in our series. In addition, motion-related artifacts were sometimes observed and resulted in degradation of image quality even though a breath-hold and fat-suppressed sequence was performed using a semiflexible circular polarized array coil with an up-to-date 1.5-T MR unit.

Obvious improvements of the mean A_z value and mean sensitivity were observed with the addition of the ferumoxides-enhanced T1-weighted gradient-echo sequences. In the quantitative analysis, the mean contrast-to-noise ratio was also significantly greater for these sequences than for unenhanced sequences and ferumoxides-enhanced T2-weighted fast spin-echo sequences. On the T1-weighted gradient-echo sequences, the mean signal-to-noise ratio of the liver parenchyma decreased significantly after the administration of ferumoxides. Conversely, that of the lesions increased significantly with ferumoxides. Ferumoxides are already known to have a T1 relaxivity effect at a lower concentration in plasma; they act as positive contrast enhancers in addition to the main effect of T2 shortening [21]. Two groups of investigators have reported the usefulness of the signal enhancement caused by the T1 effect for the diagnosis of hemangiomas on ferumoxides-enhanced T1-weighted images [22, 23]. However, the usefulness of the T1 effect for the diagnosis of hepatocellular carcinoma was controversial. Grangier et al. [22] performed a region-of-interest analysis in 10 lesions and reported that no significant signal increase was observed after the administration of ferumoxides. However, in a study by Mergo et al. [24], the signal-to-noise ratios of seven lesions increased significantly after the administration of ferumoxides. This observation is in accordance with our results with 24 lesions. Smaller fragments of ferumoxides should be retained in plasma at the time of acquisition of the contrast-enhanced images, because vessels were obviously enhanced on ferumoxides-enhanced T1-weighted gradient-echo sequences, as described by van Gansbeke et al. [23]. Hence, the ferumoxides-enhanced T1-weighted gradient-echo sequences were useful not only for detecting but also for characterizing hypervascularity of hepatocellular carcinomas, although the mean contrast-to-noise ratio of those sequences was significantly inferior to that of ferumoxides-enhanced T2^*-weighted gradient-echo imaging.

Our study has several limitations, mainly because of the retrospective fashion of histopathologic correlation. First, we could not perform a complete lesion-by-lesion analysis because only the largest lesions were documented in detail in the histopathologic reports in patients with five or more lesions. In such cases, we could not confirm the histopathology of the other small lesions detected on MR imaging; therefore, these small nodules were not included on a per-lesion basis. Thus the analyzed lesions were relatively larger (mean, 3.2 cm) than those in previous studies with reported lesion sizes of 1.9 and 1.8 cm [6, 17]. Second, we selected 6 months as the longest time between MR imaging and transplantation because screening radiologic tests, including CT and sonography, are performed every 6 months at many centers [3]. With this relatively long interval, the sensitivity could be estimated lower as a result of incorrect false-negative diagnoses of malignant lesions because the shortest volume doubling time of hepatocellular carcinoma is estimated to be 27-41 days [25,26,27,28]. In our study, however, the hepatocellular carcinomas in all 19 patients were correctly diagnosed as true-positive by at least one observer on a per-patient basis. Thus, in only one of the 48 patients—the one who had an occult, multifocal cholangiocellular carcinoma—could findings be incorrectly diagnosed as false-negative, although to our knowledge the doubling time of this kind of neoplasm is not known. In addition, no significant difference was seen between the intervals of the false-negative findings and the true-positive findings.

In conclusion, although the mean A_z value and mean sensitivity were significantly improved after the administration of ferumoxides, specificity was not improved in patients with end-stage cirrhosis who were awaiting orthotopic liver transplantation because of the false-positive lesions mainly caused by fibrotic changes. Among the three ferumoxides-enhanced sequences evaluated in our study, the T2^*-weighted gradient-echo sequences are the most sensitive for detecting malignant lesions. T1-weighted gradient-echo sequences are much more useful for detecting and characterizing hypervascular hepatocellular carcinomas than T2-weighted fast spin-echo sequences because of the greater mean contrast-to-noise ratio produced by the T1 relaxivity of hepatocellular carcinomas and the T2 shortening of liver parenchyma. No diagnostic parameters were improved with T2-weighted fast spin-echo sequences alone.

References

Top Abstract Introduction Materials and Methods Results Discussion References

 

1. 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]
2. Chalasani N, Horlander JC, Said AS, et al. Screening for hepatocellular carcinoma in patients with advanced cirrhosis. Am J Gastroenterol 1999;94:2988 -2993[Medline]
3. Gambarin-Gelwan M, Wolf DC, Shapiro R, Schwarz ME, Min AD. Sensitivity of commonly available screening tests in detecting hepatocellular carcinoma in cirrhotic patients undergoing liver transplantation. Am J Gastroenterol 2000;95:1535 -1538[Medline]
4. Shapiro RS, Katz R, Mendelson DS, Halton KP, Schwartz ME, Miller CM. Detection of hepatocellular carcinoma in cirrhotic patients: sensitivity of CT and ultrasonography. J Ultrasound Med 1996;15:497 -502[Abstract]
5. Miller WJ, Baron RL, Dodd GD III, 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]
6. Lim JH, Kim CK, Lee WJ, et al. Detection of hepatocellular carcinomas and dysplastic nodules in cirrhotic livers: accuracy of helical CT in transplant patients. AJR 2000;175:693 -698[Abstract/Free Full Text]
7. Peterson MS, Baron RL, Marsh JW Jr, Oliver JH III, Confer SR, Hunt LE. Pretransplantation surveillance for possible hepatocellular carcinoma in patients with cirrhosis: epidemiology and CT-based tumor detection rate in 430 cases with surgical pathologic correlation. Radiology 2000;217:743 -749[Abstract/Free Full Text]
8. Taourel PG, Pageaux GP, Coste V, et al. Small hepatocellular carcinoma in patients undergoing liver transplantation: detection with CT after injection of iodized oil. Radiology 1995;197:377 -380[Abstract/Free Full Text]
9. Bizollon T, Rode A, Bancel B, et al. Diagnostic value and tolerance of Lipiodol-computed tomography for the detection of small hepatocellular carcinoma: correlation with pathologic examination of explanted livers. J Hepatol 1998;28:491 -496[Medline]
10. Spreafico C, Marchiano A, Mazzaferro V, et al. Hepatocellular carcinoma in patients who undergo liver transplantation: sensitivity of CT with iodized oil. Radiology 1997;203:457 -460[Abstract/Free Full Text]
11. Ba-Ssalamah A, Heinz-Peer G, Schima W, et al. Detection of focal hepatic lesions: comparison of unenhanced and SHU 555 A-enhanced MR imaging versus biphasic helical CTAP. J Magn Reson Imaging 2000;11:665 -672[Medline]
12. Choi D, Kim SH, Lim JH, et al. Preoperative detection of hepatocellular carcinoma: ferumoxides-enhanced MR imaging versus combined helical CT during arterial portography and CT hepatic arteriography. AJR 2001;176:475 -482[Abstract/Free Full Text]
13. Ward J, Naik KS, Guthrie JA, Wilson D, Robinson PJ. Hepatic lesion detection: comparison of MR imaging after the administration of superparamagnetic iron oxide with dual-phase CT by using alternative-free response receiver operating characteristic analysis. Radiology 1999;210:459 -466[Abstract/Free Full Text]
14. Reimer P, Jaehnke N, Fiebich M, et al. Hepatic lesion detection and characterization: value of un-enhanced MR imaging, superparamagnetic iron oxide-enhanced MR imaging, and spiral CT—ROC analysis. Radiology 2000;217:152 -158[Abstract/Free Full Text]
15. Imai Y, Murakami T, Yoshida S, et al. Superparamagnetic iron oxide-enhanced magnetic resonance images of hepatocellular carcinoma: correlation with histological grading. Hepatology 2000;32:205 -212[Medline]
16. 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 1999;172:1547 -1554[Abstract/Free Full Text]
17. Krinsky GA, Lee VS, Theise ND, et al. Hepatocellular carcinoma and dysplastic nodules in patients with cirrhosis: prospective diagnosis with MR imaging and explantation correlation. Radiology 2001;219:445 -454[Abstract/Free Full Text]
18. Landis JR, Koch GG. The measurement of observer agreement for categorical data. Biometrics 1977;203:759 -765
19. Ward J, Guthrie JA, Scott DJ, et al. Hepatocellular carcinoma in the cirrhotic liver: double-contrast MR imaging for diagnosis. Radiology 2000;216:154 -162[Abstract/Free Full Text]
20. 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]
21. Chambon C, Clement O, Le Blanche A, Schouman-Claeys E, Frija G. Superparamagnetic iron oxide as positive MR contrast agents: in vitro and in vivo evidence. Magn Reson Imaging 1993;11:509 -519[Medline]
22. Grangier C, Tourniaire J, Mentha G, et al. Enhancement of liver hemangiomas on T1-weighted MR SE images by superparamagnetic iron oxide particles. J Comput Assist Tomogr 1994;18:888 -896[Medline]
23. van Gansbeke D, Metens TM, Matos C, et al. Effects of AMI-25 on liver vessels and tumors on T1-weighted turbo-field-echo images: implications for tumor characterization. J Magn Reson Imaging 1997;7:482 -489[Medline]
24. Mergo PJ, Engelken JD, Helmberger T, Ros PR. MRI in focal liver disease: a comparison of small and ultra-small superparamagnetic iron oxide as hepatic contrast agents. J Magn Reson Imaging 1998;8:1073 -1078[Medline]
25. Sheu JC, Sung JL, Chen DS, et al. Growth rate of asymptomatic hepatocellular carcinoma and its clinical implications. Gastroenterology 1985;89:259 -266[Medline]
26. Ebara M, Ohto M, Shinagawa T, et al. Natural history of minute hepatocellular carcinoma smaller than three centimeters complicating cirrhosis: a study in 22 patients. Gastroenterology 1986;90:289 -298[Medline]
27. Okazaki N, Yoshino M, Yoshida T, et al. Evaluation of the prognosis for small hepatocellular carcinoma based on tumor volume doubling time: a preliminary report. Cancer 1989;63:2207 -2210[Medline]
28. Barbara L, Benzi G, Gaiani S, et al. Natural history of small untreated hepatocellular carcinoma in cirrhosis: a multivariate analysis of prognostic factors of tumor growth rate and patient survival. Hepatology 1992;16:132 -137[Medline]

CiteULike

Complore

Connotea

Del.icio.us

Digg

Reddit

Technorati

This article has been cited by other articles:

				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]

				D. V. Sahani and S. P. Kalva Imaging the Liver Oncologist, July 1, 2004; 9(4): 385 - 397. [Abstract] [Full Text] [PDF]

This Article Abstract Figures Only Full Text (PDF) Alert me when this article is cited 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 Mori, K. Articles by Reiser, M. Search for Related Content PubMed PubMed Citation Articles by Mori, K. Articles by Reiser, M. Social Bookmarking                      What's this? Hotlight (NEW!) What's Hotlight?