| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
1 Department of Radiology, University of Vienna, Waehringer Guertel 18-20,
A-1090 Vienna, Austria.
2 Department of Surgery, University of Vienna, A-1090 Vienna, Austria.
3 Department of Vascular/Interventional Radiology, Duke University Medical
Center, DUMC 3808, Durham, NC 27710.
4 Department of Internal Medicine 4, Division of Gastroenterology, University of
Vienna, A-1090 Vienna, Austria.
Received November 2, 2001;
accepted after revision February 19, 2002.
Supported by the Ludwig Boltzmann Institute for Clinical and Experimental
Radiologic Research.
Abstract
 Top Abstract IntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
SUBJECTS AND METHODS. Forty-two patients with suspected pancreatic masses underwent contrast-enhanced helical CT and mangafodipir trisodium—enhanced MR imaging at 1.5 T. The images were assessed for the presence or absence of tumors; characterization of masses; and presence of vascular invasion, lymph node metastases, or liver metastases. Imaging findings were correlated with findings at laparotomy, laparoscopy, biopsy, or follow-up.
RESULTS. Focal masses were present in 36 patients (cancer, n = 26; focal pancreatitis, n = 7; other, n = 3). The sensitivity for lesion detection of MR imaging was 100% and of CT, 94%. Two small malignant lesions were missed on CT. For the diagnosis of tumor nonresectability, the sensitivity of MR imaging and CT was 90% and 80%, respectively. Liver metastases were missed on MR imaging in one of the eight patients and on CT in four. For differentiation between adenocarcinoma and nonadenocarcinoma, the sensitivity of MR imaging was 100% (positive predictive value, 90%; negative predictive value, 100%), and the sensitivity of CT was 92% (positive predictive value, 80%; negative predictive value, 67%). Receiver operating characteristic analysis revealed that the mean area under the curve for MR imaging was 0.920 and for CT, 0.832 (not significant).
CONCLUSION. Mangafodipir trisodium—enhanced MR imaging is as accurate as contrast-enhanced helical CT for the detection and staging of pancreatic cancer but offers improved detection of small pancreatic metastases and of liver metastases compared with CT.
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
2 cm in diameter) have the best prognosis
[2]. Therefore, early diagnosis
and assessment of tumor resectability are the important tasks for diagnostic
imaging. Although endosonography
[3] has recently been reported
to offer an alternative imaging method for diagnosis, helical CT has been the
standard technique for the diagnosis and staging of pancreatic carcinoma.
Helical CT is a robust modality and provides excellent anatomic resolution if
thin-slice techniques are used. The technical aspects of single-slice helical CT have been improved by the development of dual-phase CT and hydro-CT [4]. In dual-phase CT, the pancreas is scanned twice after a single injection of contrast material: once during the pancreatic parenchymal phase, which provides the best tumor contrast and good vessel opacification, and a second time during the portal venous phase to scan for liver metastases [4, 5]. In helical hydro-CT, the oral administration of a positive contrast agent before scanning has been replaced by the oral administration of approximately 1000 mL of water and IV administration of hyoscine-N-butyl bromide (Buscopan; Boehringer, Ingelheim, Germany) to reduce peristalsis. This technique distends the stomach and duodenum more and better delineates the pancreas against the water-filled duodenum [6].
Gadolinium-enhanced MR imaging has been shown to be equivalent or even superior to CT for imaging the pancreas [7,8,9]. Recently, technical advances, such as high-field-strength MR imaging with a phased array coil [10] and ultrafast imaging [8], have provided excellent results in terms of the detection and staging of pancreatic cancer. Mangafodipir trisodium (formerly known as manganese dipyridoxal diphosphate [Mn-DPDP]; Amersham Health, Oslo, Norway) is a new organ-specific contrast agent that was originally developed for MR imaging of the liver. However, Gehl et al. [11, 12] have shown that uptake of mangafodipir trisodium also occurs in the pancreatic parenchyma but not in pancreatic tumors, leading to improved conspicuity of pancreatic cancers. With its propensity to be taken up by the liver and pancreas but not metastases, mangafodipir trisodium could be an efficient contrast agent for the MR imaging detection and staging of pancreatic cancer. In a study comparing gadolinium- and mangafodipir trisodium—enhanced MR images of patients with suspected pancreatic cancer, gradient-recalled echo images that were enhanced with mangafodipir trisodium were significantly better at delineating pancreatic tumors than those enhanced with gadolinium chelates [13].
The purpose of our study was to compare prospectively the diagnostic yield of mangafodipir trisodium—enhanced MR imaging performed with a phased array coil and contrast-enhanced helical CT in the detection of pancreatic masses and staging of pancreatic cancer.
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
Final diagnoses were cancer in 26 patients, focal pancreatitis in seven patients, a cyst in one patient, a teratoma in one patient, metastases from hypernephroma in one patient, and no tumor in six patients. The mean lesion size was 3.7 cm (range, 1-8 cm). Twelve lesions were smaller than 3 cm in diameter, and seven tumors had a diameter of 2 cm or less. The final diagnoses were based on laparotomy or laparoscopy with intraoperative sonography and biopsy in 21 patients and on CT-guided biopsy of suspected liver metastases in three patients, of the primary tumor in three patients, and of a pseudocyst in one patient. A biopsy of tumor using endoscopic retrograde cholangiopancreatographic guidance and subsequent follow-up were performed in one patient. In the other 13 patients, follow-up results over 12-24 months were available to corroborate the diagnosis of benignity.
MR Imaging
In all patients, MR imaging was performed on a 1.5-T unit (Vision
[n = 30 patients], Siemens, Erlangen, Germany; or ACS-NT [n
= 12 patients], Philips, Best, The Netherlands) using a phased array torso
coil. For both MR units, almost identical imaging parameters were used in all
patients.
When the Vision MR unit was used, T1-weighted breath-hold fast low-angle shot (FLASH) gradient-recalled echo images with and without fat suppression were obtained before and after the administration of contrast material. Imaging parameters of the FLASH pulse sequences were as follows: TR/TE, 174.9/4.1; flip angle, 80°; number of acquisitions, 1; matrix size, 129 x 256; 1 breath-hold; and acquisition time, 22 sec. The slice thickness was 8 mm and the slice gap was 0.8 mm for evaluation of the liver and 5 and 0.5 mm, respectively, for evaluation of the pancreas. The imaging parameters for the fat-suppressed FLASH sequences were as follows: 86.5/4.1; number of acquisitions, 1; matrix size, 160 x 256; 2 breath-holds; acquisition time, 2 x 19 sec; slice thickness, 5 mm; and slice gap, 0 mm.
In all patients, mangafodipir trisodium (Teslascan, Amersham Health) was administered as an IV infusion at a dosage of 10 µmol/kg of body weight over 10-15 min, depending on the volume of contrast material and venous access. After starting the infusion, we obtained T2-weighted turbo spin-echo images with and without fat suppression. To evaluate the liver, we obtained T2-weighted fat-suppressed turbo spin-echo images using the following imaging parameters: 3500/88; number of acquisitions, 1; matrix size, 132 x 256; echo-train length, 33; 2 breath-holds; acquisition time, 2 x 17 sec; and slice thickness, 8 mm/0 mm. To evaluate the pancreas, we obtained T2-weighted turbo spin-echo images using the following imaging parameters: 3300/138; number of acquisitions, 1; matrix size, 116 x 256; echo-train length, 29; 2 breath-holds; acquisition time, 2 x 17 sec; and slice thickness, 5 mm/0 mm. Optionally, additional multidetector MR cholangiopancreatographic pulse sequences (half-Fourier acquisition single-shot turbo spin echo [HASTE]) were performed if tumors were present in the head of the pancreas. Twenty minutes after the start of contrast infusion, the T1-weighted pulse sequences were repeated.
When the ACS-NT MR unit was used, T1-weighted turbo field-echo, spin-echo, and fat-suppressed spin-echo sequences were performed. The following parameters were used to obtain the turbo field-echo images: 15/4.8; number of acquisitions, 1; matrix size, 179 x 256; acquisition time, 3 min 39 sec; slice thickness, 5 mm; and slice gap, 2.5 mm over contiguous slices. For the spin-echo imaging, the following parameters were used: 600/14; number of acquisitions, 4; matrix size, 230 x 256; acquisition time, 4 min 45 sec; slice thickness, 5 mm; and slice gap, 0.5 mm. The fat-suppressed spin-echo images were obtained with the following parameters: 700/14; number of acquisitions, 4; matrix size, 230 x 256; acquisition time, 5 min 31 sec; slice thickness, 8 mm; and slice gap, 0.8 mm. After starting the infusion of mangafodipir trisodium, we obtained T2-weighted fat-suppressed turbo spin-echo images to evaluate the liver (3000/100; number of acquisitions, 4; matrix size, 179 x 256; echo-train length, 21; acquisition time, 2 min 31 sec; slice thickness, 8 mm; and slice gap, 0.8 mm). Non—fat-suppressed turbo spin-echo images were obtained to evaluate the pancreas (3000/100; number of acquisitions, 4; matrix size, 169 x 256; echo-train length, 18; acquisition time, 3 min 12 sec; slice thickness, 5 mm; and slice gap, 0.5 mm). Optionally, a two- or three-dimensional turbo spin-echo MR cholangiopancreatographic pulse sequence was performed if masses were present in the pancreatic head. Twenty minutes after starting the infusion, we repeated the T1-weighted pulse sequences.
CT
CT scans were obtained using either of two helical scanners (AVE1, Philips;
or Somatom Plus 4, Siemens). After unenhanced scans of the liver and pancreas
were obtained, dual-phase contrast-enhanced scans of the liver and pancreas
were obtained. Before scanning was performed, 750-1000 mL of water was
administered orally as a contrast agent and 20 mg of hyoscine-N-butylbromide
(Buscopan) was IV administered to reduce peristaltic artifact. Two helical
sequences were performed after the IV infusion of 120-140 mL (300 mg/mL) of
nonionic contrast material at a rate of 4 mL/sec. The first sequence was
performed 40 sec after the initiation of infusion, consisted of thin-slice
images (slice thickness, 3 mm; table feed, 5 mm; reconstruction interval, 3
mm), and covered the pancreas in the pancreatic phase. The second sequence
(slice thickness, 5 mm; table feed, 8 mm; reconstruction interval, 4 mm) began
after a minimal interscan delay (68-75 sec after the initiation of infusion)
during the portal venous phase and covered the liver and pancreas.
Image Analysis
CT and MR images of all patients were separately analyzed in a prospective
fashion. Two radiologists who were unaware of the results of other imaging
studies and the findings at surgery and histopathologic examination evaluated
all the images in consensus. To minimize any learning bias, we scheduled a
3-month interval between viewing the CT scans and viewing the MR images of the
same patient. Pancreatic adenocarcinoma was characterized on helical CT and on
mangafodipir trisodium—enhanced MR imaging as a hypodense mass or on
T1-weighted as a hypointense mass that showed less or no enhancement,
respectively, compared with the surrounding pancreatic parenchyma
[12].
Tumor Detection
Tumor size was determined by measuring the hypodense or hypointense area on
CT or MR images. When pancreatitis was severe and the tumor margins could not
be delineated completely, the tumor was measured in the longest diameter
assessable. For all MR images and CT studies, the observers classified the
presence or absence of lesions in the pancreas on a 5-point confidence scale
(1 = definitely present, 2 = probably present, 3 = indeterminate, 4 = probably
absent, 5 = definitely absent). If a lesion was present, the observers
recorded its size and location.
Tumor Extension and Resectability
To assess for vascular invasion, the observers focused their attention on
six vascular structures: the main portal vein, portal venous confluence,
superior mesenteric artery and vein, celiac trunk, and hepatic artery.
Invasion of the splenic artery, splenic vein, and spleen was not considered
critical because they can be resected in pancreatic tail tumors. CT or MR
imaging evidence of vascular compromise was defined as vessel occlusion or
invasion. We used the criteria established by Lu et al.
[14] and O'Malley et al.
[15] with some modifications
to define the degree of vascular invasion. Tumors with no contiguity (defined
as grade 0 by Lu et al.) and tumors with less than one-quarter circumference
of arteries (defined as grade 1 by Lu et al.) were deemed resectable. Limited
invasion of the portal vein, venous confluence, or the superior mesenteric
vein was not considered an absolute contraindication to surgery because a
radical tumor resection with venous reconstruction can be performed by our
pancreas surgeons. Therefore, between one quarter and one half of the
circumference of the vein contiguous with tumor (defined as grade 2 by Lu et
al.) was still considered resectable
[15].
The short-axis diameter of enlarged lymph nodes and the presence or absence of liver metastases and of peritoneal implants were recorded. Tumors were deemed unresectable when enlarged distant (nonperipancreatic) lymph nodes, liver metastases, or peritoneal implants were present.
On the basis of the morphology and signal intensity of the lesion and on the aforementioned criteria, the observers then characterized each lesion as benign or malignant using a 5-point scale. In patients with suspected malignant masses, resectability or nonresectability of the tumor was rated on a 5-point scale (1 = definitely resectable, 2 = probably resectable, 3 = indeterminate, 4 = probably nonresectable, 5 = definitely nonresectable).
Statistical Analysis
To evaluate the diagnostic yield of MR imaging and CT studies, we
determined the sensitivity and specificity of each modality for the detection
of pancreatic tumors, for the differentiation between benign and malignant
masses, and for the assessment of tumor resectability, and we calculated the
95% confidence intervals (Cls). The chi-square test was used to compare
sensitivities and specificities. A p value of less than 0.05 was
considered significant. The performance of CT and MR imaging in
differentiating between benign and malignant masses was also compared using
receiver operating characteristic analysis
[16]. Alternative—free
response receiver operating characteristic curves were calculated for the
observers' CT and the MR imaging interpretations by plotting the true-positive
fraction against the likelihood of obtaining a false-positive diagnosis. The
diagnostic accuracy of each modality was estimated by calculating the mean
area under the receiver operating characteristic curve (i.e., the mean
Az index).
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
Tumor Resectability
Of the 26 patients with proven pancreatic cancer, one was not evaluated
further because of his poor general medical condition. In the 25 remaining
patients, 10 pancreatic tumors were found to be resectable and 15,
nonresectable. Mangafodipir trisodium—enhanced MR imaging enabled the
correct diagnosis of resectability in nine patients and of nonresectability in
14 patients; one false-negative diagnosis (missed liver metastases) and one
false-positive diagnosis (invasion of the portal vein, which could be treated
by radical resection with venous interposition) were recorded. The sensitivity
for diagnosing nonresectability was 90% (95% CI, 0.84-0.96) and the
specificity, 93% (95% CI, 0.90-0.96). Contrast-enhanced helical CT enabled the
correct diagnosis of resectability in nine patients and of nonresectability in
12 patients; three false-negative diagnoses (missed liver metastases) (Fig.
4A,4B)
and one false-positive diagnosis (invasion of the portal vein, which could be
treated by radical resection with venous interposition) were recorded. The
sensitivity was 80% (95% CI, 0.75-0.85) and specificity, 60% (95% CI,
0.50-0.70).
Overall, liver metastases were present in eight patients. In four patients (multiple metastases, n = 3; solitary, n = 1), liver metastases were shown on both CT and MR imaging. The observers missed small liver metastases (0.5-1 cm in diameter) on MR imaging in one patient and on CT in four patients (Fig. 6A,6B,6C,6D). However, in one of these patients, unresectability was correctly diagnosed at CT because vascular encasement was present.
|
|
|
|
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
The purpose of our study was to compare the diagnostic accuracy of contrast-enhanced single-detector helical CT and MR imaging in the detection of focal pancreatic lesions and the staging of pancreatic cancer. Some studies have been performed to compare MR imaging and CT for the detection of pancreatic cancer [7, 9, 18]. The results of the study conducted by the Radiology Diagnostic Oncology Group indicated that dynamic CT is superior to unenhanced non—breath-hold MR imaging for the assessment of resectability [7]. Gadolinium-enhanced dynamic MR imaging performed during breath-holding proved to be equal to or slightly better than single-slice helical CT in lesion detection [9, 18]. However, these studies have shown mixed results for the determination of local tumor extension and vascular involvement on CT and MR imaging [9, 18]. In both studies, the question about the presence of liver metastases was not addressed.
Recently, mangafodipir trisodium—enhanced MR imaging using a whole-body coil has been shown to be effective in the detection and staging of pancreatic cancer [19, 20]. In the study of Rieber et al. [19], helical CT enabled detection of eight of eight pancreatic tumors and MR imaging, seven of eight; however, mangafodipir trisodium—enhanced MR imaging improved tumor delineation and diagnostic confidence compared with helical CT. In the study performed by Romijn et al. [20], the detection rate for cancer was slightly improved using mangafodipir trisodium—enhanced MR imaging than using single-detector helical CT. However, the accuracy for tumor staging was not improved [20]. MR imaging was equal to CT for the assessment of vascular invasion, and both modalities missed omental metastases [20].
One of the most difficult issues in imaging patients with pancreatic cancer is the assessment of vascular invasion [14, 15]. Our findings confirm the results of Romijn et al. [20]: Mangafodipir trisodium—enhanced MR imaging does not improve the staging of pancreatic cancer by revealing vascular encasement. The presence of duodenal invasion is displayed even better on helical CT because the duodenum can be distended after peroral administration of a negative contrast agent, water. Unless a costly negative oral contrast agent is used at MR imaging, delineation of the duodenal wall is difficult. Extensive vascular invasion is also better displayed by helical CT because it may show dilatation of the peripancreatic veins, a finding that is suggestive of peripancreatic tumor invasion [21]. Even gadolinium-enhanced MR imaging is limited in its ability to exactly delineate vascular involvement, especially of the portal vein as a result of its oblique orientation [10, 18]. Helical CT may take another big step forward with the use of multidetector CT. Preliminary experience with multidetector array CT has shown promising results regarding the three-dimensional display of vascular encasement [22]. Mangafodipir trisodium is an organ-specific MR contrast agent administered slowly by infusion. This agent does not lead to a T1-shortening of the blood sufficient to be exploited for MR angiography. However, the problem of black blood imaging can be overcome by new breath-hold gradient-recalled echo pulse sequences (e.g., fast imaging with steady-state free precession [Siemens], balanced fast field echo [Philips]), which render the vasculature bright even without a contrast agent. These pulse sequences may be used to supplement contrast-enhanced MR images obtained during the pancreatic phase [23].
We found that MR imaging was superior to CT for the detection of liver metastases, although the difference did not reach statistical significance because too few patients presented with metastases. This finding is of no clinical relevance because if contrast-enhanced MR imaging reveals more liver metastases than CT, nonresectability is indicated by the presence of at least one metastasis [24]. In three of the eight patients with liver metastases in our study, CT findings were negative, but MR imaging revealed small metastases. Although one may speculate that mangafodipir trisodium—enhanced MR imaging may be superior to helical CT for the depiction of liver metastases, the number of patients with liver metastases in this series was too small to prove this hypothesis. In these patients, the surgical strategy may be changed. Before laparotomy is performed, laparoscopy, which is associated with only minimal morbidity, can be used for the detection and biopsy of small metastases [24].
In a study of patients with small pancreatic metastases (2 cm) by Irie
et al. [25],
gadolinium-enhanced MR imaging revealed seven of eight and helical CT only
five of eight small pancreatic metastases. In our study, mangafodipir
trisodium—enhanced MR imaging depicted seven of seven small pancreatic
metastases, whereas CT failed to show two. Although the number of small
malignancies in each of the two series was too small to reach statistical
significance, these findings suggest that contrast-enhanced MR imaging is
superior to CT in delineating small malignancies.
Differentiation between pancreatic adenocarcinoma and focal chronic pancreatitis remains a challenge. The lack of mangafodipir trisodium uptake into the pancreatic parenchyma is not 100% specific for the presence of cancer. Neither masses associated with focal chronic pancreatitis nor atrophic parenchyma distal to an obstructed duct takes up a large amount of mangafodipir trisodium. In these areas, the presence of extensive fibrosis precludes considerable enhancement. Because pancreatic tumors with equivocal findings on CT or MR imaging are potentially resectable, an important clinical concern is that these patients not be denied surgery. Therefore, the observers in our study were more likely to err on the side of making the diagnosis of cancer in the patients with focal pancreatitis. This tendency resulted in false-positive diagnoses of cancer on both CT and MR imaging. One patient with a presumed diagnosis of cancer on CT and MR imaging underwent Whipple's procedure. Histologic evaluation of the resected specimen revealed only focal pancreatitis.
A limitation of this study lies in the small number of patients with resectable tumors. However, in general, the percentage of patients presenting with resectable pancreatic cancers is low. Moreover, almost all patients with cancer underwent either laparoscopy with intraoperative sonography or laparotomy to confirm the imaging findings. A second limitation is the fact that all the images were assessed by viewing conventional hard-copy films. Managing large numbers of images generated by MR imaging, helical CT, and, even more so, by multidetector CT is easier using cine display. Although cine display of the pancreas is preferred over film-based viewing for delineation of vascular structures, it is not superior in terms of lesion detection [26]. Even if cine viewing were advantageous, the results of this study would not have been biased because all CT and MR imaging examinations were reviewed on films.
In conclusion, our study results show that mangafodipir trisodium—enhanced MR imaging and helical CT are equivocal for local staging of pancreatic cancer but that mangafodipir trisodium—enhanced MR imaging offers advantages in the detection of small pancreatic malignancies and of liver metastases.
|
TopAbstractIntroductionSubjects and MethodsResultsDiscussionReferences |
|---|
CiteULike 
Complore 
Connotea 
Del.icio.us 
Digg 
Reddit 
Technorati What's this?
This article has been cited by other articles:
|
|
 |
|
  A. Hayashibe, M. Kameyama, M. Shinbo, and S. Makimoto Clinical Results on Intra-arterial Adjuvant Chemotherapy for Prevention of Liver Metastasis Following Curative Resection of Pancreatic Cancer Ann. Surg. Oncol., January 1, 2007; 14(1): 190 - 194. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 A.-J. Lemke, S. M. Niehues, N. Hosten, H. Amthauer, M. Boehmig, C. Stroszczynski, T. Rohlfing, S. Rosewicz, and R. Felix Retrospective Digital Image Fusion of Multidetector CT and 18F-FDG PET: Clinical Value in Pancreatic Lesions--A Prospective Study with 104 Patients J. Nucl. Med., August 1, 2004; 45(8): 1279 - 1286. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 M K Kalra, M M Maher, P R Mueller, and S Saini State-of-the-art imaging of pancreatic neoplasms Br. J. Radiol., December 1, 2003; 76(912): 857 - 865. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
