| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Original Research |
Department of Radiology, Massachusetts General Hospital, 55 Fruit St., Boston, MA 02114-2696.
Received April 1, 2004;
accepted after revision September 29, 2004.
Address correspondence to D. V. Sahani
(dsahani{at}partners.org).
Abstract
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
MATERIALS AND METHODS. Imaging data of 34 patients (23 men, 11 women; age range, 44-78 years) with adenocarcinoma of the colon (n = 27) or adenocarcinoma of the pancreas (n = 7) who had undergone mangafodipir trisodium-enhanced liver MRI and whole-body FDG PET were retrospectively reviewed for the presence and number of liver metastases. Histopathology (n = 25) or follow-up imaging (n = 9) served as the standard of reference. Breath-hold T1-weighted gradient-recalled echo, respiratory-triggered T2-weighted fast spin-echo, and mangafodipir trisodium-enhanced axial fat-saturated high-spatial-resolution (256 x 512) T1-weighted gradient-recalled echo images were obtained on a 1.5-T scanner. FDG PET was performed after the injection of 15-20 mCi (555-740 MBq) of FDG. The sensitivity, positive predictive value, and accuracy were calculated for each technique. The performances of the two techniques were compared using the Fisher's exact test.
RESULTS. Thirty patients had hepatic metastases and four had no hepatic metastases according to the standard of reference. The total number of metastases was 79, including 33 that measured less than 1 cm. Based on a per-patient analysis, MRI and FDG PET showed sensitivities of 96.6% and 93.3%, positive predictive values of 100% and 90.3%, and accuracies of 97.1% and 85.3%, respectively. According to a per-lesion analysis, MRI and FDG PET showed sensitivities of 81.4% and 67.0%, positive predictive values of 89.8% and 81.3%, and accuracies of 75.5% and 64.1%, respectively. MRI detected more hepatic metastases than FDG PET (p = 0.016). Of the 33 subcentimeter lesions confirmed on the standard of reference, all were identified on MRI, whereas only 12 were detected on FDG PET (p = 0.0001).
CONCLUSION. In patients with colon and pancreatic adenocarcinoma, high-spatial-resolution mangafodipir trisodium-enhanced liver MRI and whole-body FDG PET were comparable in the detection of patients with liver metastases. FDG PET provided additional information about extrahepatic disease and was useful in initial staging. However, significantly more and smaller liver metastases were detected on MRI than on FDG PET.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Imaging plays a significant role in establishing metastatic disease in the liver, which influences the treatment strategy. For example, patients with resectable lesions may benefit from curative resection, whereas the remaining patients may benefit from other therapies such as hepatic artery chemoembolization, radiofrequency ablation, or systemic chemotherapy [4]. CT arterioportography has been reported as the most sensitive test for the detection of hepatic metastases [6]. MRI has been shown to be equal in sensitivity and specificity to CT arterioportography in the detection of focal liver lesions [7]. PET with fluorine-18 FDG is established as an important method for staging gastrointestinal malignancies [8].
A limited number of publications have directly compared MRI and FDG PET for the detection of liver metastases [9]. We retrospectively reviewed imaging, pathology, and surgical data in patients with colonic or pancreatic malignancy who underwent both mangafodipir trisodium-enhanced liver MRI and whole-body FDG PET for the detection of metastatic disease to assess the relative performance of each technique in liver metastasis detection.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Patients
Our study cohort comprised 23 men and 11 women who ranged in age from 44 to
78 years, with an average age of 60 years. The histopathologic distribution of
the primary tumor included adenocarcinoma of the colon in 27 patients and
adenocarcinoma of the pancreas in seven patients. Three patients had
previously undergone partial hepatectomy for metastases from colon cancer and
were being evaluated for recurrent disease in the liver; the remaining 31
patients had no prior liver surgery. Mangafodipir trisodium-enhanced MRI of
the liver and whole-body FDG PET were performed in this group as part of the
metastatic workup.
MRI Technique
MRI of the liver was performed on a 1.5-T system (Signa, GE Healthcare).
Unenhanced T1-weighted images were obtained using a breath-hold 2D
gradient-echo sequence during in phase and out of phase (TR range/first-echo
TE, second-echo TE, 150-200/1.8, 4.2; slice thickness, 6-9 mm; interslice gap,
0-2 mm; and acquisition matrix, 256 x 128-192). A slow IV infusion of
mangafodipir trisodium (Teslascan [0.5 µmol/kg], Nycomed) was
given over 1-2 min. This was followed by the acquisition of
respiratory-triggered fast spin-dual-echo T2-weighted images (4,000-6,000/102,
135; slice thickness, 6-9 mm; interslice gap, 0-2 mm; and acquisition matrix,
256 x 192). Breath-hold high-spatial-resolution T1-weighted
gradient-recalled echo images (TR/TE, 200/2.1; slice thickness, 5-6 mm; no
interslice gap; and 256 x 512 matrix) were subsequently obtained. With
this sequence, the entire liver was covered in two to three overlapping
acquisitions, each of a 30-sec breath-hold duration.
FDG PET Technique
FDG PET was performed on an ECAT HR+ PET scanner (Siemens Medical
Solutions). The patients were advised to fast for at least 6 hr before the
study, and any IV glucose administration was stopped. Blood glucose levels
were checked before starting the procedure, and imaging was performed if blood
glucose levels were less than 200 mg/dL (range, 41-174 mg/dL; mean, 93.1
mg/dL; median, 93.0 mg/dL). Four patients had levels more than 100 mg/dL.
FDG (15-20 mCi [555-740 MBq]) was injected into a peripheral vein, and imaging was performed after 1 hr. Patients were positioned supine on the scanner, and emission images (7 min) were acquired in 6-7 bed positions from mandible to mid thigh level. Transmission images, acquired with rotating sources containing germanium-68 (3 min per bed position), were used for attenuation correction. Images were reconstructed using the ordered subset expectation maximization (OSEM) algorithm (2 iterations, 8 subsets). A gaussian filter with the cutoff based on patient weight (< 180 lb [< 82 kg], 7; 180-200 lb [82-91 kg], 8; 200-250 lb [91-113 kg], 9; > 250 lb [113 kg], 10) was used for reconstruction. The images were displayed in rotating maximum intensity projections and in axial, coronal, and sagittal planes on a workstation (ECAT workstation [software version 7.2.2], Siemens Medical Solutions) and interpreted by a board-certified nuclear medicine physician. Focal areas of increased uptake that were greater than normal hepatic parenchyma were considered to represent tumor. The standard uptake values (SUVs) were not calculated because this calculation was not a part of our routine practice.
Image Analysis
The liver MR images and whole-body FDG PET images were analyzed
independently by two reviewers experienced in interpreting liver MR images
(reviewer 1) and whole-body FDG PET images (reviewer 2) without prior
knowledge of the histopathologic or follow-up imaging findings. The FDG PET
and MRI reviewers were blinded to the results of the other technique. The
reviewers had interpreted a few of these studies initially as a part of
routine clinical practice (reviewer 1, n = 7 MRI studies; reviewer 2,
n = 5 PET studies). The time interval between the initial
interpretation and the second interpretation for this study was 13-18
months.
Each study was evaluated for the presence and number, characterization (benign vs malignant), and location of liver lesions. The size of the lesion was measured on contrast-enhanced T1-weighted fat-saturated MR images. Mangafodipir trisodium-enhanced images were used primarily for lesion detection, and lesion characterization was based primarily on the unenhanced T1 and T2 characteristics of the lesion.
Standard of Reference
Histopathology obtained after surgical resection and intraoperative
sonography constituted the standard of reference in 25 patients. In these
patients, lesion registration was based on pathologist's description of the
lesions, the surgeon's description of the bimanual palpation, and findings on
intraoperative sonography.
In nine patients who were not considered suitable for surgery, follow-up imaging performed after 4-10 months (average time, 8.2 months) on CT (n = 1), MRI (n = 7), or FDG PET (n = 1) served as the standard of reference. Among the patients who had follow-up imaging as the standard of reference (n = 9), seven had pancreatic adenocarcinoma and two had adenocarcinoma of the colon. In these patients, surgery was not performed because of the presence of distant metastases or an inoperable primary tumor. All these patients had extrahepatic disease identified on FDG PET. A lesion was considered stable or benign if there was no change in the size of the lesion during the next follow-up anatomic imaging study. A lesion was considered malignant if it had changed more than 25% in size or if it showed persistent focal abnormal activity on the FDG PET scan.
Statistical Analysis
The sensitivity, positive predictive value, and accuracy were calculated
for each technique. The performance of the two techniques was compared using
the Fisher's exact test. A p value of less than 0.05 was considered
significant.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Of the 25 patients who underwent surgical resection, 23 had liver metastases. There were 79 lesions in these 23 patients. In two patients who underwent surgical resection, one had granulation tissue on pathology at the site of suspected tumor recurrence and the other had no identifiable tumor at surgery. In nine patients who did not undergo surgery, follow-up imaging studies showed disease progression or persistent focal hepatic FDG uptake considered to be consistent with liver metastases. Eighteen lesions in these nine patients were identified on follow-up imaging.
MRI
MRI correctly identified metastases in 29 patients. In five patients, MRI
was negative for metastatic disease. However, one metastasis was seen in one
of these five patients on FDG PET and on the follow-up CT performed after 6
months. MRI showed a total of 88 lesions, of which 79 were confirmed
metastases on the standard of reference. Of the 88 lesions detected on MRI, 39
measured less than 1 cm and 33 of these 39 lesions were metastases on the
standard of reference. MRI failed to depict 18 metastases.
Among the patients who had imaging follow-up as the standard of reference (n =9), MRI revealed 15 lesions, of which 10 lesions showed progression on follow-up and were considered true-positives.
FDG PET
The FDG PET findings were positive for malignant lesions in the liver in 31
of the 34 patients. Of these, 28 had metastases on the standard of reference
and the remaining three patients did not and were considered false-positive
cases. FDG PET findings were falsely negative in two patients. In one of these
patients, MRI showed eight metastatic foci, all of which measured less than 1
cm in the liver, that were confirmed on the standard of reference. In the
other patient, MRI depicted a solitary liver metastasis that was resected and
confirmed at pathology.
FDG PET detected a total of 65 lesions, 50 of which were confirmed metastases on the standard of reference. Of the 50 metastases, 12 lesions measured less than 1 cm. Thirty-two metastases were not identified on FDG PET (Figs. 1A, 1B, 1C, 1D, 2A, 2B, 2C, 2D, 2E, and 2F), and 21 of these measured less than 1 cm.
|
|
|
|
|
|
|
|
|
|
Among the patients who had imaging follow-up as the standard of reference (n =9), FDG PET depicted 17 lesions, of which 11 showed progression and were considered true-positives.
FDG PET identified extrahepatic disease in nine patients. These included abdominal lymph node metastases in four patients, mediastinal lymph node metastases in three patients, abdominal wall metastasis in one patient, and a metastasis to the lumbar spine in one patient.
MRI Versus FDG PET
MRI and FDG PET interpretations were concordant in 28 of the 34 patients.
In 27 patients, both FDG PET and MRI were positive for liver metastases and in
one patient, both were negative for metastatic disease. However, in six
patients, MRI and FDG PET findings were not in agreement. Three of these six
patients had no metastatic disease on MRI and the standard of reference but
had a false-positive FDG PET study. In another patient, FDG PET correctly
identified a solitary metastasis in the left lobe of liver that was not
evident on MRI. In two patients, MRI correctly identified metastases, whereas
FDG PET was falsely negative (Table
1).
|
Based on the number of lesions documented by the standard of reference, MRI identified 21 lesions that were not evident on FDG PET (Figs. 1A, 1B, 1C, 1D, 2A, 2B, 2C, 2D, 2E, and 2F) and the latter detected seven lesions that were not evident on MRI (Table 2).
|
Statistical Analysis
According to a per-patient analysis, MRI showed a sensitivity of 96.6%,
positive predictive value of 100%, and accuracy of 97.1%. FDG PET showed a
sensitivity of 93.3%, positive predictive value of 90.3%, and accuracy of
85.3% (Table 3). There was no
statistically significant difference between MRI and FDG PET for the detection
of patients with metastatic disease (p = 0.61). FDG PET and MRI were
in agreement in 27 (90%) of the 30 patients.
|
According to a per-lesion analysis, MRI and FDG PET showed sensitivities of 81.4% and 67.0%, positive predictive values of 89.8% and 81.3%, and accuracies of 75.5% and 64.1%, respectively. MRI detected more hepatic metastases than FDG PET (p = 0.016), and this observation was statistically significant. Among metastases that measured less than 1 cm, lesion detection was better with MRI than with FDG PET. Of 33 subcentimeter metastases confirmed on the standard of reference, all were identified on MRI, whereas only 12 were detected on FDG PET. MRI detected significantly more metastases among lesions measuring less than 1 cm (p = 0.0001).
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
In the past, contrast-enhanced helical CT arterioportography was considered the most sensitive preoperative imaging tool for the detection of hepatic metastases [6]. However, advancements in liver MRI technology, especially with contrast agent developments, have improved the sensitivity and specificity of contrast-enhanced MRI for liver tumor detection [7, 11-18]. Several studies performed with extracellular gadolinium chelates or with the reticuloendothelial cell-specific (RES) contrast agents (ferumoxides) have shown that MRI performs better than helical CT and has a sensitivity almost equal to that of CT arterioportography with a more favorable specificity [2, 7, 13-17]. Results from a multicenter phase III clinical trial have shown lesion detection is improved with mangafodipir trisodium-enhanced liver MRI in comparison with dual-phase helical CT [18].
FDG PET has been shown to be an excellent tool for gastrointestinal cancer staging [19-22]. However, its role in the detection of liver metastases has not been established. A number of studies have reported superior sensitivity and specificity with FDG PET as compared with multiphasic helical CT in the detection of liver metastases [19-24]. One study has shown a sensitivity of 89% for FDG PET as compared with 71% for CT in detecting liver metastases from colon cancer [19]. However, in this and several other articles comparing PET and CT, there have been potential sources of bias that could have favored PET over CT including the interval between CT and PET, unequal skill in test performance, variations in CT technology, and bias in test interpretation.
Yang et al. [9] found no significant difference in the detection of liver metastases with gadolinium chelate-enhanced liver MRI and FDG PET. In another study, comparing FDG PET with other cross-sectional imaging techniques for the detection of primary and secondary liver malignancies, FDG PET performed better than sonography and CT; however, gadolinium chelate-enhanced MRI had the highest sensitivity and specificity [23]. In that study, the sensitivity, specificity, and positive predictive value of PET for all malignant hepatic lesions were 82%, 25%, and 96%, respectively, compared with 63%, 50%, and 96% for abdominal sonography; 71%, 50%, and 97% for CT; and 83%, 57%, and 97% for MRI. To our knowledge, there are no published studies available comparing the performance of liver MRI using a liver-specific contrast agent such as mangafodipir trisodium with FDG PET.
In our study, mangafodipir trisodium-enhanced liver MRI and whole-body FDG PET were comparable in the detection of patients with metastatic disease. There was no significant difference between MRI and FDG PET for the detection of the liver metastases on a patient-by-patient basis. However, for liver lesion detection, MRI detected significantly more and smaller metastases than FDG PET. This observation was most significant for lesions that measured less than 1 cm, because only 12 lesions that measured less than 1 cm were detected on FDG PET in comparison with 33 lesions detected on MRI. This low lesion detection rate with FDG PET may be due to the relatively poorer spatial resolution of PET in comparison with MRI. In addition, the use of the liver-specific contrast agent, mangafodipir trisodium, may have further aided in improving liver lesion detection. This paramagnetic contrast agent is taken up by functioning hepatocytes, causing T1 shortening of the liver parenchyma [25]. Tumors lacking hepatocytes maintain their native signal, improving the lesion-to-liver contrast. In addition, a long imaging window of several hours after the administration of this contrast agent allows acquisition of both 2D and 3D high-spatial-resolution thin-slice images through even the largest liver [26]. The other advantages of MRI include specific anatomic localization and characterization of lesions detected [8].
These findings may have significant implications in the management of patients with colon and pancreatic cancer. Because the sensitivity for the detection of metastatic disease in the liver is equal on FDG PET and mangafodipir trisodium-enhanced MRI, patients with suspected metastatic disease may benefit from initial FDG PET given that it allows the detection of extrahepatic disease. Indeed, in our study, FDG PET identified extrahepatic disease in nine of the 34 patients. However, in patients with no extrahepatic metastatic disease, MRI would be beneficial in identifying all the hepatic lesions and determining the resectability of those lesions.
FDG PET is based on the principle that increased metabolic activity, such as that seen in malignant tissue, is accompanied by increased glucose uptake relative to that of the surrounding normal tissue and that this increased metabolic focus is seen as more intense uptake on FDG PET images [21]. However, the increased FDG uptake is nonspecific and may occur in any condition associated with increased tissue metabolism, such as acute or chronic inflammatory processes [21]. Because tumors are metabolically active, the functional information provided by FDG PET is expected to aid in the detection of early metastatic or recurrent disease before it is evident on other imaging studies [21].
The other advantage of FDG PET is the ability to survey the whole body for metastatic disease. Lai et al. [27] found previously unsuspected extrahepatic disease, predominantly involving the celiac lymph nodes, in 32% patients scheduled for hepatic metastasectomy. However, FDG PET can be falsely negative in specific tumor types such as mucinous tumors of the colon, esophagus, breast, and lung and renal cell carcinomas [28, 29]. Likewise, the sensitivity of FDG PET for metastatic liver lesions varies significantly with lesion size [30]. Frohlich et al. [30] found that in patients with colorectal cancer and metastases to liver the sensitivity of FDG PET for liver metastases was 14% for lesions 1.5 cm or smaller, 84% for those between 1.5 and 3 cm, and 100% for those larger than 3 cm.
It is important to note that fatty infiltration of the liver also may result in heterogeneous uptake of FDG in the liver, and this may result in a false-positive interpretation (personal observation). Likewise, during the postoperative period, relatively higher FDG activity may be observed for several months at the site of surgery. This is typically due to postsurgical inflammation or granulation tissue that can masquerade as tumor recurrence on FDG PET [31]. Other drawbacks of FDG PET include its high cost, inherent radiation dose exposure, limited availability, and poor spatial resolution.
We also acknowledge that the MRI contrast agent mangafodipir trisodium used in our study was recently taken off the market temporarily in the United States after more than 5 years of commercial availability. This is due to some problems with the stability of compound, and we are uncertain about when this agent will become available again in the future. However, two other hepatobiliary-specific contrast agents such as gadobenate dimeglumine (MultiHance, Bracco Diagnostic) and gadolinium EOB-DTPA (Eovist, Berlex Laboratory) have successfully completed the phase III clinical trials in Europe and the United States and may soon be available for liver imaging [32, 33]. These contrast agents have a dual mechanism of action with earlier circulation in the extracellular space, as seen with gadolinium chelates, and then delayed hepatobiliary uptake, as seen with mangafodipir trisodium. These agents have also been shown to improve lesion-to-liver contrast during the delayed hepatobiliary phase and thus may potentially improve liver metastasis detection [32-34].
There are some limitations in our study. First, the retrospective nature of the data collection and analysis may introduce an inadvertent bias to one or the other imaging test. Second, surgical and histopathologic confirmation was not available for all the lesions in our study. Indeed, in nine of 34 patients, follow-up imaging was used as an alternate standard of reference. In addition, we included only patients who underwent liver MRI, whole-body FDG PET, and surgery or follow-up imaging within 6 weeks; therefore, absolutely reliable calculations of the sensitivity and specificity of these techniques remain uncertain.
With the advent of PET/CT, liver lesion detection and localization have significantly improved, and this procedure is now considered a state-of-the-art test for oncologic imaging. However, further studies are required to directly compare PET/CT and MRI in the detection of metastatic disease in the liver. Finally, we were not able to assess the impact on the treatment strategy for or survival benefit of patients in whom additional lesions were detected or missed with either technique.
In conclusion, in patients with colon and pancreatic adenocarcinoma, high-spatial-resolution contrast-enhanced liver MRI using the hepatobiliary-specific contrast agent mangafodipir trisodium and FDG PET were comparable in the detection of liver metastases. In addition, FDG PET detected extrahepatic disease; thus, it is useful for the initial staging of the malignancy. However, significantly more lesions, especially small lesions, were detected on MRI than on FDG PET. Therefore, in patients being considered for liver resection after initial staging for metastatic disease, the treatment decision should be based primarily on liver MRI because additional lesions detected on MRI may alter or preclude such surgery.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
CiteULike 
Complore 
Connotea 
Del.icio.us 
Digg 
Reddit 
Technorati What's this?
This article has been cited by other articles:
|
|
 |
|
  M. Tatsumi, J. H. Miller, and R. L. Wahl 18F-FDG PET/CT in Evaluating Non-CNS Pediatric Malignancies J. Nucl. Med., December 1, 2007; 48(12): 1923 - 1931. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 F. H. Miller, A. L. Keppke, D. Reddy, J. Huang, J. Jin, M. F. Mulcahy, and R. Salem Response of Liver Metastases After Treatment with Yttrium-90 Microspheres: Role of Size, Necrosis, and PET Am. J. Roentgenol., March 1, 2007; 188(3): 776 - 783. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
D.-M. Koh, G. Brown, Z. Meer, A. R. Norman, and J. E. Husband Diagnostic Accuracy of Rim and Segmental MRI Enhancement of Colorectal Hepatic Metastasis After Administration of Mangafodipir Trisodium Am. J. Roentgenol., February 1, 2007; 188(2): W154 - W161. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
