| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Original Research |
1 Department of Radiology, University of Ulsan–Asan Medical Center, 388-1
Pungnap-dong, Songpa-gu, Seoul 138-736, South Korea.
2 Department of Diagnostic Radiology, Kyung Hee University Hospital, Seoul,
South Korea.
3 Department of Preventive Medicine, University of Ulsan–Asan Medical
Center, Seoul, South Korea.
Received December 22, 2004;
accepted after revision February 22, 2005.
Address correspondence to M.-G. Lee
(mglee{at}amc.seoul.kr).
Abstract
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
MATERIALS AND METHODS. Forty-one patients who were known to have or
suspected of having hepatic metastasis from colorectal carcinoma underwent
mangafodipir trisodium- or ferucarbotran-enhanced MRI in block randomization
methods. Two radiologists independently reviewed the MR images to determine
the number of hepatic lesions and to characterize the lesions as malignant or
benign. Each lesion was assessed according to its size (small, 
2 cm;
large, > 2 cm in diameter) on both mangafodipir trisodium- or
ferucarbotran-enhanced MRI. The data were correlated with the reference
diagnosis: histopathology and intraoperative sonography (n = 16);
intraoperative sonography (n = 4); and imaging and clinical diagnosis
with follow-up (> 3 months; n = 21). The detection rates and
diagnostic accuracies of hepatic lesions on both sets of MR images were
assessed using Fisher's exact test.
RESULTS. Eighty-two hepatic lesions (53 metastatic and 29 benign) were identified in 41 patients. No significant differences were seen between mangafodipir trisodium- and ferucarbotran-enhanced MRI for detecting all hepatic lesions (p = 0.183), small hepatic lesions (p = 0.299), all metastases (p = 0.695), and small metastases (p = 0.689). The diagnostic accuracies of mangafodipir trisodium- and ferucarbotran-enhanced MRI showed no significant differences in all hepatic lesions (p = 0.624) and small hepatic lesions (p = 0.641).
CONCLUSION. Mangafodipir trisodium- and ferucarbotran-enhanced MRI are similar in hepatic lesion detection and characterization in colorectal cancer patients.
Keywords: colorectal cancer • contrast media • ferucarbotran • liver • mangafodipir trisodium • metastases • MRI
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
The advent of liver-specific MRI contrast materials, which are agents targeted to enhance hepatocytes or Kupffer cells, has facilitated an increase in the accuracy of MRI in liver metastasis detection [3]. Mangafodipir trisodium (Teslascan, Amersham) is a hepatobiliary contrast agent that predominantly causes T1 shortening. After the IV administration of mangafodipir trisodium, the normal liver shows increased signal intensity on T1-weighted images while the metastases are unchanged in signal intensity. A multicenter phase 3 trial of mangafodipir trisodium reported that detection, classification, and diagnosis of focal hepatic lesions on manganese-enhanced MRI were comparable or superior to those attained on unenhanced MRI and on enhanced CT [4]. Kim et al. [5] also reported that mangafodipir trisodium-enhanced MRI was superior to helical CT in the characterization of focal hepatic lesions in colorectal cancer patients.
Ferucarbotran (Resovist, Schering) is a new class of superparamagnetic iron oxide (SPIO) agents administered IV with a bolus; therefore, both dynamic-phase and accumulation-phase MRI are now available. In a phase 3 study, ferucarbotran-enhanced T2-weighted imaging documented a significant signal intensity loss for benign lesions versus malignant lesions when compared with the results obtained with ferumoxides-enhanced MRI [6].
|
|
|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Between June 2003 and February 2004, 48 consecutive patients were enrolled in this study. Seven patients were excluded because of the following conditions: multiple (> 5) hepatic metastases on MRI (n = 3), unavailable follow-up imaging study (n = 2), and pathologically proven hepatocellular carcinoma (n = 1) or cholangiocellular carcinoma (n =1). Finally, 41 patients constituted the study population (27 men, 14 women; age range, 24–77 years; mean age, 57 years). Hepatic metastases were suspected during the follow-up period after resection of colorectal carcinoma in 21 patients and at the time of initial diagnosis of colorectal carcinoma in 20 patients.
MRI Examinations
All examinations were performed on a 1.5-T MRI scanner (Magnetom Vision,
Siemens Medical Solutions); a circularly polarized body phased-array surface
coil was used for signal reception.
Before contrast injection, a T1-weighted spoiled gradient-echo sequence (fast low-angle shot [FLASH]) was performed. Imaging parameters were TR/TE, 149/4.1; flip angle, 80°; acquisition time, 19 sec; and matrix, 132 x 256.
T2-weighted imaging was performed with HASTE and true fast imaging with steady-state free precession (FISP) sequences. The parameters for the HASTE sequence were as follows: TR/effective TE, infinite/134; echo spacing, 4.6 msec; echotrain length, 104; flip angle 150°; acquisition time, 20 sec; and matrix, 192 x 256. For the true FISP sequence, the following parameters were used: 4.8/2.3; flip angle, 70°; acquisition time, 17 sec; and matrix, 150 x 256. A T2*-weighted FLASH sequence was used only in patients who were enrolled in the ferucarbotran group; the parameters were as follows: 149/10; flip angle, 30°; and matrix, 112 x 256. One acquisition of T2*-weighted FLASH sequence obtained 5 images, and 3–5 acquisitions were required to image the entire liver. Both T1- and T2-weighted images were obtained in the axial plane. A section thickness of 8 mm, intersection gap of 2 mm, and a field of view of 350 mm were applied in all pulse sequences on both mangafodipir- and ferucarbotran-enhanced MRI.
|
|
|
|
Ferucarbotran, 0.45 mmol/kg (35–60 kg of body weight) or 0.7 mmol/kg (> 60 kg of body weight), was administered through a 5-µm filter as an IV bolus injection over 3–5 sec. Immediately after injection, the connecting tube was flushed with 10 mL of saline (0.9% NaCl). First, dynamic phase MR images were obtained at 10 sec, 50 sec, and 5 min after the ferucarbotran injection using the FLASH sequence and the same parameters as those of the unenhanced study. Next we obtained accumulation phase images, which were obtained at 10 min after the contrast injection using both T1- and T2*-weighted FLASH sequences with the same parameters as those of the unenhanced study. When needed in particular patients, thin-section axial or coronal images (5 mm for FLASH and 6 mm for T2*-weighted FLASH) were obtained; however, these images were not included in the image evaluation. For ferucarbotran-enhanced MRI, about 60 min were needed to complete the entire study.
When selecting the MR contrast agents at each examination—that is, either mangafodipir trisodium or ferucarbotran—we used a block randomization method. Therefore, the physicians who referred the MR examination, the radiologists who were involved in the MR examination during the study period, and the patients who were referred for MRI examination were not aware of the kind of contrast agent to be administered before the patient was identified as being eligible to be included in the study population and decided to participate in this study.
Image Evaluation
Two radiologists participated in the image analysis who were unaware of any
clinical, laboratory, or other imaging information except that the patients
presented with or had a history of colorectal carcinoma. Both observers had 7
years' experience in liver MRI. The combined unenhanced and contrast-enhanced
MR images were randomly assigned to each observer. Each observer independently
and separately interpreted the MR images. All MR images were evaluated on a
PACS (Radpia, Hyundai Information Technology) with the annotations masked.
Each observer recorded the number, size, and segmental location of the hepatic lesions according to the Couinaud classification. Observers also recorded their diagnostic confidence as to whether the lesions appeared to be benign or metastatic on a 5-point scale (1: definitely benign; 2: probably benign; 3: possibly malignant; 4: probably malignant; and 5: definitely malignant). Two examples each of categories 1, 2, 4, and 5 were presented to both observers. The lesions for which it was difficult to determine whether they were benign or metastatic were to be categorized as 3, but examples were not given. These examples were obtained before May 2003, so they were not included in the study population. In two patients who had multiple simple cysts, the organizer assigned the most important five lesions primarily according to their size. Observers also recorded whether contrast-enhanced MRI was helpful in characterization of the metastases and benign liver lesions. Peripheral rim enhancement with a hypointense center on mangafodipir trisodium-enhanced T1-weighted images [7] (Figs. 1A, 1B, and 1C) and peripheral rim enhancement on ferucarbotran-enhanced dynamic T1-weighted images were considered helpful in characterizing the metastases [8] (Figs. 2A, 2B, 2C, and 2D). An early increase in signal intensity with a further steady fill-in phenomenon on ferucarbotran-enhanced dynamic T1-weighted images was considered helpful in characterizing the hemangiomas [8].
Results of the interpretations by the two observers were collected and compared. A lesion was considered to be present when both observers detected it. For lesions about which the two observers agreed, their agreement was accepted as the final interpretation: on a 5-point scale: 1 or 2 was considered an agreement for benign, and 3, 4, or 5, for metastasis. In cases of disagreement, a consensus panel consisting of the two original observers plus a third party (with 20 years' experience) who was also blinded made the final decision as to whether the lesion was benign or metastatic. Consensus was also reached as to whether enhanced MRI was helpful in characterizing metastasis. These final interpretations were used to determine the detection rate, diagnostic accuracy, and rate of characteristic enhancement patterns of the hepatic lesions with each contrast agent.
Statistical Analysis
All statistical analyses were performed using standard statistical software
(SPSS version 11 for Windows, SPSS). For analysis, hepatic lesions were
classified as follows: all hepatic lesions (all benign lesions and
metastases); all metastases; small hepatic lesions (all benign lesions and
metastases 2 cm in maximum diameter); and small metastases ( 2 cm in
maximum diameter).
The two observers' interpretations were used for the following analysis: Interobserver agreement on the diagnostic confidence level (the 5-point scale described earlier) was assessed to establish the reliability of the image interpretation. The degree of interobserver agreement in distinguishing metastases from benign lesions was calculated using weighted kappa statistics. In general, a kappa value greater than 0.75 was considered to indicate excellent agreement beyond chance.
Consensus interpretations were used for the following analyses: The detection rate, diagnostic accuracy, and rate of characteristic enhancement patterns on mangafodipir trisodium- and ferucarbotran-enhanced MRI were analyzed using Fisher's exact test.
Standard of Reference
One of the authors who did not participate in the MRI interpretation formed
the standard of reference. In 20 patients, intraoperative sonography by a
radiologist, after careful inspection and palpation by a surgeon, was the
standard of reference. Among these 20 patients, surgical removal of metastases
was possible in 16 patients. The average interval between MRI and surgery was
10 days (range, 1–30 days). In the remaining 21 patients, intraoperative
sonography was not performed because the lesion was interpreted as a benign
lesion on MRI (n = 14), radiofrequency ablation was performed
(n = 5), or peritoneal seeding was found during laparotomy
(n = 2). In these patients, the reference diagnosis was formed by
means of interpretation of all available data for each patient. These data
included all imaging data (findings of sonography, CT, MRI, and other imaging
if available), clinical data, laboratory data, and data from follow-up CT or
MRI. In all 41 patients, follow-up CT or MRI was performed within 3–6
months.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
|
Histopathologic confirmation or intraoperative sonography was not available in 14 metastases for the following reasons: radiofrequency ablation (n = 7), misinterpretation as a benign lesion (n = 3), not detected on MRI (n = 2), or peritoneal seeding at the time of surgery (n = 2). Seven metastases treated by radiofrequency ablation were interpreted as metastases on sonography, CT, and MRI. Except for the seven metastases treated by radiofrequency ablation, an additional seven metastases that were untreated showed an increase in size and level of carcinoembryonic antigen (n = 6) or a decrease in size by chemotherapy (n = 1) during the follow-up period.
The benign lesions were simple hepatic cysts (n = 20), a hemorrhagic cyst (n =1), hemangiomas (n = 4), focal eosinophilic necrosis (n = 1), a previous biopsy site (n =1), a previous surgery site (n = 1), and a hepatic calcification (n = 1). Four hepatic cysts were confirmed by histopathology and intraoperative sonography. Four other hepatic cysts were confirmed by intraoperative sonography. The remaining 12 hepatic cysts, four hemangiomas, one hemorrhagic cyst, one focal eosinophilic necrosis, one previous biopsy site, one previous surgery site, and one hepatic calcification were confirmed by all available data and follow-up imaging. Histopathologic confirmation or intraoperative sonography was limited in benign hepatic lesions (21/29, 72%) because biopsy or intraoperative sonography of definitely benign lesions on MRI was considered unethical.
The results of the two observers' analysis were as follows: One observer
detected 84 hepatic lesions and the other observer, 83. Three lesions were
detected by only one of two observers. Only 82 focal hepatic lesions were
detected by both observers. The weighted kappa value in analyzing the
interobserver agreement for differentiating metastases from benign lesions
showed excellent agreement (
, 0.833; 95% confidence interval,
0.759–0.908). An incorrect diagnosis was made in six lesions by one
observer and in four lesions by the other observer; even after consensus
interpretation, an incorrect diagnosis was made in four lesions.
The following are the results of the consensus analysis: The detection rates for hepatic lesions and metastases showed no statistically significant differences between ferucarbotran- and mangafodipir trisodium-enhanced MRI (Table 2). The rates of the characteristic enhancement pattern for distinguishing metastases from benign lesions after contrast injection also showed no statistical significance (Table 3).
|
|
On mangafodipir trisodium-enhanced MRI, a correct diagnosis was made in 37 (93%) of the 40 hepatic lesions detected on MRI (Figs. 1A, 1B, and 1C). A false-negative diagnosis was made in one small metastasis (0.6 cm in maximum diameter), which was interpreted as a benign lesion on MRI. A false-positive diagnosis was made in two benign lesions. One was a previous biopsy site (0.6 cm in maximum diameter) and the other was a previous surgery site (1.5 cm in maximum diameter) (Figs. 3A, 3B, and 3C). These two lesions were interpreted as metastasis on MRI.
|
|
|
|
|
|
|
2
cm), the diagnostic accuracies of mangafodipir trisodium- (27/30, 90%) and
ferucarbotran-enhanced (19/20, 95%) MRI showed no significant differences
(p = 0.641). |
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Some reports have indicated that mangafodipir trisodium-enhanced MRI showed higher detectability of hepatic lesions than either dynamic CT or unenhanced MRI [4, 7, 9]; however, other reports have described similar detection sensitivity using mangafodipir trisodium [5, 10]. Ferumoxides- or ferucarbotran-enhanced MRI has also shown comparably high detectability with CT during arterial portography and intraoperative sonography [11–14].
Our results showed that the detection rates of mangafodipir trisodium-enhanced MRI for all metastases (82%) were similar [10] or slightly lower than those in previous reports [5] and that the detection rates for ferucarbotran-enhanced MRI for all metastases (90%) were similar to those of previous reports using ferucarbotran [11, 15].
Our results also showed that the detection rates of hepatic lesions and metastases were slightly higher on ferucarbotran-enhanced MRI than on mangafodipir trisodium-enhanced MRI; however, no statistical significance was seen between the two contrast agents. Also, no statistically significant differences were seen between the diagnostic accuracies of each contrast agent concerning differentiation between metastases and benign lesions. These results do not necessarily mean that the diagnostic performance of these two contrast agents is the same because we did not perform a separate analysis for each unenhanced and contrast-enhanced image. In practice, unenhanced T1- and T2-weighted images are necessary to decrease the number of false-positive findings attributed to small vessels and to characterize hepatic lesions [16]. Previous reports have shown that combined analysis of unenhanced and contrast-enhanced MR images showed better results than those of contrast-enhanced MR images alone [15, 17].
Several recent studies have compared mangafodipir trisodium and ferumoxides with other contrast agents [3, 18, 19]. In a comparison between ferumoxides- and mangafodipir trisodium-enhanced MRI for the detection of hepatocellular carcinoma, ferumoxides-enhanced MRI had superior diagnostic accuracy in lesions smaller than 10 mm and superior lesion conspicuity compared with mangafodipir trisodium-enhanced MRI [18]. In a comparison between ferumoxides and gadobenate dimeglumine for the detection of hepatic metastases, ferumoxides-enhanced MRI had superior sensitivity to gadobenate dimeglumine [3]. In another study comparing gadobenate dimeglumine and mangafodipir trisodium for various focal hepatic lesions, both provided equal liver enhancement and lesion conspicuity on contrast-enhanced imaging [19]. Considering these two reports [3, 19], it might be assumed that ferumoxides have superior sensitivity to mangafodipir trisodium for hepatic lesion detection. However, in an indirect comparison of different reports [3, 19] with different patient populations, different MR sequences, and different methods of analysis, there is always the chance of making an error in judgment. Although ferucarbotran was used instead of ferumoxides, our study showed similar results between ferucarbotran- and mangafodipir trisodium-enhanced MRI in the detection and characterization of metastases and benign lesions of the liver in colorectal cancer patients.
Our choice of pulse sequences and delay time after contrast injection requires an explanation because the application of an inadequate pulse sequence and delay time may cause a serious bias in the comparison of each contrast agent.
For mangafodipir trisodium-enhanced MRI, we used T1-weighted breath-hold FLASH sequences obtained 30 min and 2 hr after contrast injection. A previous study reported that 24-hr delayed imaging with mangafodipir trisodium provided additional information in 49 (38%) of the 129 cases; in five cases this additional information led to a change in patient management [20]. However, 24-hr delayed imaging is not practical in a clinical setting [18]. The useful imaging window in the liver is about 2 hr from the end of the mangafodipir trisodium injection [21], and in recent reports, 24-hr delayed imaging is no longer used [5, 7, 10, 18]. We also used gradient-echo sequences instead of spin-echo sequences. Gradient-echo sequences are known to be superior to spin-echo sequences for mangafodipir trisodium-enhanced T1-weighted imaging [21].
For ferucarbotran-enhanced MRI, we used T2*-weighted FLASH sequences. In comparative studies of static imaging after SPIO administration on a 1.5-T MR imager, T2-weighted fast spin-echo and T2*-weighted gradient-echo sequences were considered to be optimal for the detection of primary and secondary hepatic malignancies [22–24]. Of these previous reports [22–24], one study [24] using ferucarbotran showed that high tumor detectability and contrast-to-noise ratio were noted on T2*-weighted gradient-echo sequences. For the delay time for static imaging after ferucarbotran administration, previous studies have documented that an accumulation phase could begin as early as 10 min after the injection of ferucarbotran [25, 26].
Our study had several limitations. First, histopathologic proof was not available for nonresected segments. Our standard of reference included all available data not confirmed by intraoperative sonography or histopathology. A suboptimal standard of reference may result in a too-small fraction of small lesions and a subsequent overestimation of the detection rate [27]. We assume that this bias in standard of reference might influence both contrast agents similarly and so would have little effect on the results of our study because the purpose of our study was to compare two contrast agents for the detection and characterization of hepatic lesions.
Second, the section thickness of the MR images was 8 mm, which was relatively thick. On the mangafodipir trisodium-enhanced T1-weighted FLASH sequence, the section thickness may be lowered to 5 mm; and on the ferucarbotran-enhanced T2*-weighted sequence, the section thickness may be lowered to 6 mm on our 1.5-T MR imager. In practice, we also use these thin sections on contrast-enhanced MRI if there is uncertainty on the 8 mm-section thickness images. Controversy may exist that mangafodipir trisodium-enhanced MRI, which can use thinner sections than ferucarbotran-enhanced MRI, may be in a disadvantageous position for comparison when the same section thickness is used.
Third, ferucarbotran-enhanced MRI had more images than mangafodipir trisodium-enhanced MRI because of dynamic phase MRI. A different number of MR images might influence the diagnostic performance of both contrast agents. However, dynamic phase MRI could not be omitted because it was an important sequence for the characterization of hepatic lesions on ferucarbotran-enhanced imaging.
Fourth, we used 2-hr delayed imaging on mangafodipir trisodium-enhanced MRI. In recently published articles, a 2-hr delayed imaging is no longer used [7, 10, 18] with the exception of one study [5]. If we deleted the 2-hr delayed imaging in our study protocol, the examination time might be similar using both contrast agents. The usefulness of 2-hr delayed imaging on mangafodipir trisodium-enhanced MRI is not a part of this study. Fifth, in our study, mangafodipir trisodium- and ferucarbotran-enhanced MRI were not performed in the same patients. This may lower the value of our study. However, injection of two different contrast agents in the same patients might present an ethical problem. Sixth, the standard of reference consisted of a relatively small number of patients (n = 41) and hepatic lesions (n = 82). Last, our study lacked comparison with gadolinium-enhanced MRI, a commonly used nonspecific MRI contrast agent.
In conclusion, our results show similar hepatic lesion detection and characterization in colorectal cancer patients when two liver-specific contrast agents are used. Ferucarbotran-enhanced MRI was slightly superior to mangafodipir trisodium-enhanced MRI, but this was not of statistical significance.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
2cm) hepatic lesions in
colorectal cancer patients: detection and characterization on mangafodipir
trisodium-enhanced MRI. AJR 2004;182
: 1233-1240
CiteULike 
Complore 
Connotea 
Del.icio.us 
Digg 
Reddit 
Technorati What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
