| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Gastrointestinal Imaging |
1 From the Departments of Radiology (E.L.H., H.A., N.H., J.R., R.F.), Visceral and Transplant Surgery (J.L., P.N.), and Gastroenterology and Hepatology (M.B., R.H., B.W, S.R.), Charité Medical University Center, Campus Virchow Clinic, Humboldt University, Augustenburger Platz 1, D-13353 Berlin, Germany. Received April 17, 2001; revision requested June 4; revision received August 27; accepted October 10. Address correspondence to E.L.H. (e-mail: enrique.lopez_haenninen@charite.de).
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODS RESULTS DISCUSSIONREFERENCES |
|---|
MATERIALS AND METHODS: Sixty-six patients suspected of having pancreatic tumors underwent MR imaging (unenhanced and contrast material–enhanced MR, MRCP, and contrast-enhanced MR angiography). Two blinded readers prospectively analyzed the images by consensus, and results were correlated with surgery, biopsy, or follow-up findings. Results were tabulated in two-by-two tables.
RESULTS: MR assessment of pancreatic lesion status (differentiation of benign vs malignant) resulted in 60 correct diagnoses (accuracy, 91%), and six (10%) false diagnoses. Among histologically proved malignant tumors, MR imaging yielded correct diagnoses in 42 of 44 patients (sensitivity, 95%; 95% CI: 85%, 99%), whereas 18 of 22 patients with benign findings were classified correctly. At MR imaging, findings in four patients with chronic pancreatitis were wrongly categorized as malignant tumors (specificity, 82%; 95% CI: 60%, 95%), and in one patient, a distal common bile duct carcinoma was not detected. In no patient with pancreatic adenocarcinoma was this tumor misdiagnosed as benign. In patients with malignant tumors who underwent resection, local-regional tumor growth and vascular infiltration were accurately classified in 89% and 94%, respectively. MR imaging depicted histologically proved synchronous hepatic metastases in 82%. The positive and negative predictive values for cancer nonresectability were 90% and 83%, respectively, and the accuracy, sensitivity, and specificity were 85%, 69%, and 95%, respectively.
CONCLUSION: Unenhanced and contrast-enhanced MR imaging with MRCP and MR angiography offers potential as a noninvasive tool for assessment of patients suspected of having pancreatic tumors.
© RSNA, 2002
Index terms: Magnetic resonance (MR), cholangiopancreatography, 770.1222 • Magnetic resonance (MR), vascular studies, 770.12142 • Pancreas, MR, 770.121412, 770.121415, 770.12142 • Pancreas, neoplasms, 770.32, 770.364 • Pancreatitis, 770.291
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTS DISCUSSIONREFERENCES |
|---|
One crucial consideration in the treatment of patients suspected of having pancreatic tumors is how to proceed diagnostically. So far, ultrasonography (US) and contrast material–enhanced computed tomography (CT) have been widely used to diagnose pancreatic disease (5,6). However, in previous series, differentiating chronic pancreatitis from pancreatic cancer was considerably difficult (7). This dilemma is clinically relevant, because the detection of pancreatic cancer mandates surgical resection in patients in whom surgery can be performed, whereas in patients suspected of having chronic pancreatitis, a more observational approach, including clinical and diagnostic follow-up examinations, is justified. To overcome this dilemma, the development of sensitive and specific imaging modalities appears highly desirable.
More recently (8–10), the use of magnetic resonance (MR) imaging for detection of pancreatic tumors was demonstrated. In particular, faster sequences reduced motion artifacts substantially and facilitated successful characterization of pancreatic lesions. In addition, one major advantage of MR imaging is the possibility to examine the pancreatobiliary system noninvasively. Endoscopic retrograde cholangiopancreatography (ERCP) is still considered the reference standard for pancreatic duct evaluation and consecutive therapeutic interventions. However, since publication of the study by Wallner et al (11) concerning clinical data about the use of MR cholangiopancreatography (MRCP), investigators have evaluated its potential for use in pancreatic disease. In those series (12–14), MRCP demonstrated a high accuracy in the depiction of diverse abnormalities involving diseases of the pancreaticobiliary system, including duct dilatation, tumors, strictures, and stones.
Moreover, the introduction of contrast-enhanced three-dimensional MR angiography with fast gradient-echo (GRE) sequences facilitated the visualization of the vascular anatomy with the breath-hold technique and, thus, permitted multiphase assessment of abdominal arteries and veins with excellent quality (15,16).
The purpose of our study, therefore, was to prospectively assess the accuracy of MR imaging, MRCP, and MR angiography in the evaluation of patients suspected of having pancreatic tumors.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTS DISCUSSIONREFERENCES |
|---|
Inclusion criteria were that the patient was suspected of having a pancreatic mass and that the patient was 18 years old or older. Patients were excluded if they had a known hypersensitivity to gadopentetate dimeglumine, they had liver metastases, they were unable to complete the examination for various reasons, they had mental retardation, or their medical history was not available. The study received institutional review board approval, and written informed consent was obtained from all patients before entry into this study. At admission, all patients already had undergone US, which generally was performed by the referring colleague; in addition, a minor proportion (n = 24 [36%]) of patients underwent CT that was performed at different institutions and with different techniques prior to admission to our referral center.
Imaging Techniques
Patients underwent MR imaging with a 1.5-T unit (Gyroscan ACS-NT; Philips, Best, the Netherlands) with a gradient strength of 25 mT/m. In this study, the sequences used with a surface coil were transverse T2-weighted turbo spin-echo (2,000/90 [repetition time msec/echo time msec]) followed by transverse unenhanced T1-weighted GRE with (126/1.7; flip angle, 80°) and without (30/15; flip angle, 30°) fat suppression. Section thickness was 6 mm; field of view, 375 mm; and matrix, 256 x 256.
MRCP was performed with unenhanced T2-weighted fast turbo spin-echo sequences (8,000/1,000; turbo factor, 255; matrix, 512 x 512; section volume, 50 mm) in nine coronal oblique orientations (-30° through 30°) with a single-section thick-slab technique. Acquisition time was 9 seconds per section (total, 1.12 minutes).
Unenhanced and contrast-enhanced MR angiography (5.2/1.4; flip angle, 40°; matrix, 163 x 512) was performed with coronal oblique sections, and gadopentetate dimeglumine (Magnevist; Schering, Berlin, Germany) was administered at a dose of 0.2 mmol/kg of body weight through an 18-gauge catheter in an antecubital vein. For contrast material injection, an MR imaging–compatible power injector (Spectris; Medrad, Pittsburgh, Pa) and an injection rate of 2 mL/sec were used. Adequate timing was determined with a test bolus of 2 mL of gadopentetate dimeglumine, followed by a 20-mL saline flush and repeated measurements over a period of 50 seconds. On the basis of these data, perfusion during the arterial phase, the portal venous phase (fixed timing, 7 seconds after the end of the arterial phase sequence), and the venous phase (fixed timing, 10 seconds after the end of the portal venous phase sequence) acquisitions was documented. The acquisition time for each MR angiographic sequence was 16 seconds.
Finally, transverse contrast-enhanced GRE sequences (30/15; flip angle, 30°) were performed. The overall time of the examinations used in the "one-stop shop" approach described herein was approximately 50–55 minutes.
Image Analysis
MR images were prospectively evaluated with the consensus of two radiologists (E.L.H., J.R.) who were blinded to clinical and serologic factors. All components (MR imaging, MRCP, and MR angiography) of the examination were evaluated as a whole. The size and localization of the pancreatic lesions were documented for unenhanced and contrast-enhanced images. In addition, the signal intensity pattern of the lesions (ie, hyperintense, hypointense, or isointense as compared with adjacent pancreatic parenchyma) for the different sequences was determined.
Moreover, the evaluation included the assessment of local-regional tumor extent. Peripancreatic tumor growth was diagnosed with irregular signal of the peripancreatic fat or infiltration of adjacent structures (eg, duodenum, colon, stomach). Lymph nodes with short-axis diameters greater than 1 cm were considered metastatic. Presence of peritoneal carcinomatosis and mesenteric implants was assessed. In addition, focal liver lesions were documented and categorized as benign or metastatic. Lesions were considered metastatic if they were mildly hyperintense to the liver on T2-weighted images or they were hypointense to the liver on unenhanced T1-weighted images and demonstrated contrast enhancement on T1-weighted images after administration of gadopentetate dimeglumine.
For assessment of the pancreaticobiliary system, MRCP images were interpreted and assessed for the presence of pathologic morphologic features. This included measurements of the maximum ductal diameters of the common bile duct and pancreatic duct (three measurements: head, body, tail); furthermore, the presence of dilatations, stenoses, or obstructions was recorded. MRCP images were ranked according to a four-category scale as follows: nondiagnostic, anatomic structures not visible; poor diagnostic quality, anatomic structures partially visible; adequate diagnostic quality, anatomic structures mainly visible; and excellent diagnostic quality, excellent visualization of anatomic structures.
For MR angiography, visualization of vascular anatomy and assessment of vascular infiltration included separate identification of the celiac axis, the common hepatic artery, the gastroduodenal artery, the left gastric artery, the splenic artery, the superior mesenteric artery, the portal vein, the superior mesenteric vein, and the splenic vein. For assessment of vascular anatomy and evaluation of potential tumor infiltration, both original paracoronal source data (OD) and results from the maximum intensity projection (MIP) image were viewed.
Image quality was categorized with a five-point scale as follows: insufficient (score of 1, vessel not visible on OD and MIP image), poor (score of 2, vessel not visible on MIP image and partially visible on OD with low contrast), fair (score of 3, vessel fully visible on OD and MIP image with low vessel contrast), good (score of 4, good visualization on OD and MIP image with good vascular contrast), and very good (score of 5, excellent vessel visualization and contrast on OD and MIP image). Vascular tumor infiltration was assumed on the basis of the presence of either vessel caliber reduction, vessel encasement (>90% circumferential involvement was considered positive for this presence in the portal venous system), or vessel occlusion.
Final diagnoses of tumors were categorized as follows: (a) Definitely benign. No delineated solid tumor was present or the pancreatic cavity was considered benign (eg, cyst, pseudocyst, or abscess), and no pancreatobiliary duct stenosis or obstruction was present or pancreatobiliary duct stenosis or obstruction with or without upstream dilatation was considered benign. (b) Probably benign. No delineated solid tumor was present, or the pancreatic cavity was considered benign, or no pancreatobiliary duct stenosis or obstruction was present or pancreatobiliary duct stenosis or obstruction with or without upstream dilatation was considered benign. (c) Inconclusive. Findings could not be classified as benign or malignant. (d) Probably malignant. A delineated solid lesion was present or pancreatobiliary duct stenosis or obstruction with or without upstream dilatation was considered malignant. (e) Definitely malignant. A delineated solid lesion was present, and consistent pancreatobiliary duct stenosis or obstruction was considered malignant (17).
Resectability of tumors with or without vascular reconstruction was assumed when there were no signs of infiltration of the superior mesenteric artery. Tumors that infiltrated the portal venous system were considered resectable by means of reconstructive surgical measures. The criterion for local-regional nonresectability was infiltration of the superior mesenteric artery, of the mesentery root, or of the mesocolon. On the basis of these findings, tumors were classified as definitely resectable or definitely nonresectable.
Verification of Diagnosis and Statistical Analysis
MR results were subsequently correlated with the results from surgery, histopathologic analysis, or follow-up (ie, MR reevaluation including assessment of clinical symptoms and of biochemical tumor markers CA 19-9, chromogranin A, and 24-hour 5-hydroxyindoleacetic acid) after 6, 12, and 18 months; with unchanged or regressive findings, the diagnosis of pancreatic carcinoma was considered highly unlikely. All abnormal findings at surgery were documented, and locations and diameters of tumors were noted and assessed histopathologically (open surgery, n = 47; laparoscopy, n = 1). Thirteen patients underwent percutaneous biopsy. Five patients were examined at follow-up. Tumor classification was performed according to Union Internationale Contre le Cancer (UICC) guidelines. The sensitivity, specificity, positive predictive value, and negative predictive value were calculated for both malignant and benign findings and tabulated in two-by-two tables; for diameter assessment of tumors depicted at MR imaging and examined at histopathologic analysis, the correlation coefficient was calculated; 95% CIs were determined by using the Geigy tables (18,19).
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTS DISCUSSIONREFERENCES |
|---|
Overall, MR assessment of pancreatic disease resulted in 60 correct diagnoses (accuracy, 91%) and six (10%) incorrect diagnoses (including the inconclusive MR diagnosis mentioned later). Among patients with a final diagnosis of malignancy, MR imaging yielded correct diagnoses in 42 of 44 patients (sensitivity, 95%; 95% CI: 85%, 99%); MR imaging yielded a correct diagnosis in 18 of 22 patients with benign pancreatic findings (specificity, 82%; 95% CI: 60%, 95%). Among patients with benign final diagnoses, MR findings in four patients with confirmed chronic pancreatitis were wrongly considered malignant. The positive predictive value was 91% (42 of 46), and the negative predictive value was 90% (18 of 20).
Table 1 gives the blinded readers’ final diagnoses for groups with both malignant and benign pancreatic tumors. The blinded readers’ final diagnoses for the 37 pancreatic adenocarcinomas were definitely malignant in 86% (n = 32), probably malignant in 11% (n = 4), and inconclusive in 3% (n = 1). In one patient, a confirmed NET was correctly classified as definitely malignant. In five patients, papillary carcinoma or distal common bile duct carcinoma and one non-Hodgkin lymphoma were correctly categorized as probably (n = 1) and definitely (n = 4) malignant. One distal common bile duct carcinoma was incorrectly categorized as probably benign. In no patient with confirmed pancreatic malignancy was the tumor wrongly categorized as benign.
|
Assessment of Pancreatic Lesions and Pancreatic Duct Characteristics
Qualitative characteristics of malignant pancreatic lesions detected with MR imaging demonstrated that 16 (43%) lesions were mildly hyperintense on T2-weighted images (all adenocarcinomas), whereas 18 and three tumors were isointense and hypointense, respectively, on T2-weighted images (adenocarcinomas, n = 16; NET, n = 1). Thirty-one (84%) malignant pancreatic lesions were hypointense and six (adenocarcinomas, n = 5, NET, n = 1) were isointense on unenhanced T1-weighted GRE images (Fig 1). With contrast-enhanced GRE sequences, 25 lesions were hypointense (adenocarcinomas, n = 24, NET, n = 1), seven were isointense, and five were hyperintense. Among 22 patients with benign final diagnoses, no difference in signal intensity pattern of pancreatic lesions could be detected in 16 patients. Of 17 patients with chronic pancreatitis, four had variable heterogeneous parenchymal signal intensity differences that were wrongly determined as malignant with MR imaging (Fig 2).
|
|
|
|
|
|
|
|
|
Assessment of Vascular Infiltration
The qualitative evaluation of the vascular system showed good and excellent visualization (scores of 4 and 5, respectively) in the majority of patients. No nondiagnostic ratings were observed for the arterial and portal venous systems. Overall scores were 4.96 for the celiac axis, 4.95 for the splenic artery, 4.94 for the common hepatic artery, 4.65 for the gastroduodenal artery, 4.55 for the left gastric artery, and 4.99 for the proximal superior mesenteric artery. Overall scores for the portal venous system were 4.95 for the portal vein, 4.85 for the splenic vein, and 4.90 for the superior mesenteric vein.
Among patients with a final diagnosis of malignancy who underwent surgery or exploratory surgery (n = 34), six had surgically proved vascular infiltration (portal vein or confluens), and five (sensitivity, 83%) of these were correctly depicted with MR imaging; in contrast, vascular infiltration of the celiac axis was overestimated in one patient, and in one patient, infiltration of the hepatic artery was missed. Accuracy in depiction of vascular infiltration with MR imaging was 94% (32 of 34; Table 2). Figure 1c shows a coronal MR angiogram obtained in a patient with pancreatic adenocarcinoma and surgically confirmed portal vein infiltration in the region of the pancreatic head, with consecutive collateral formation.
Assessment of Tumor Resectability and Hepatic Metastases
Of all patients with final diagnoses of malignancy, 11 had histologically proved liver metastases, and MR imaging depicted true-positive lesions in nine patients (sensitivity, 82%). In one patient, 5-mm small hepatic metastases were missed at MR imaging, and in one patient with multiple cysts, concurrent metastases were not recognized (Table 2).
Overall accuracy for determining tumor nonresectability was 85% (29 of 34), sensitivity was 69% (nine of 13), and specificity was 95% (20 of 21). Two false-negative readings were caused by tumor presence in the mesentery and the mesocolon. In addition, two patients had unrecognized hepatic metastases. In one patient, nonresectability was assumed to be caused by multiple enlarged but not metastatic lymph nodes at the celiac axis. The positive predictive value for nonresectability was 90% (nine of 10), whereas the negative predictive value was 83% (20 of 24).
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTS DISCUSSIONREFERENCES |
|---|
The current study, therefore, was initiated to prospectively determine the accuracy of a single noninvasive modality in the clinical work-up of patients suspected of having pancreatic tumors who were referred for evaluation with regard to lesion detectability, assessment of lesion status (malignant vs benign), and resectability of the lesion.
In our study, overall accuracy was calculated at 90% and, thus, was superior or similar to accuracy in previous studies in which MR imaging was used for pancreatic cancer evaluation (22,23). Nevertheless, among the 37 confirmed pancreatic adenocarcinomas described in this series, surgical and histopathologic assessment demonstrated no tumor that was locally confined to the pancreas. Compared with findings in previous studies that have been published so far in which CT, US, and PET were used for pancreatic tumor detection, the sensitivity of MR imaging appeared to be superior to that of CT and similar to the best values for sensitivity with US and PET. However, to our knowledge, so far no article has been published in which results were assessed with multisection CT, in which the speed of scanning and reconstruction of data may provide valuable contributions to the staging of pancreatic tumors (5,6,19,23).
Consistent with findings in previous MR imaging studies with unenhanced and contrast-enhanced GRE sequences in addition to T2-weighted MR imaging, the signal intensity characteristics of most pancreatic adenocarcinomas were hypointense to the pancreas with unenhanced T1-weighted GRE sequences and hypointense or isointense with T2-weighted imaging. However, a proportion of 44% of patients with pancreatic adenocarcinomas exhibited a mildly hyperintense signal intensity on T2-weighted images. Whereas NETs tend to show low signal intensity on T1-weighted images and high signal intensity on T2-weighted images (8–10,24–26), in one patient with an NET in this study low signal intensity was exhibited on T1-weighted GRE images and isointense signal intensity was exhibited on T2-weighted images.
As previously described, the presence of chronic pancreatitis represents an extremely difficult challenge in the differential diagnosis of pancreatic cancer. Whereas in this study no patients with malignant final diagnoses had tumors that were categorized as probably or definitely benign, in four patients with confirmed chronic pancreatitis, the lesions were wrongly categorized as malignant at MR imaging. In addition, one patient with an inconclusive MR imaging result (atrophic pancreas with no delineated parenchymal lesion) had a 15-mm carcinoma of the pancreatic head (UICC, T3 N0) at surgery. As described herein, we obtained contrast-enhanced T1-weighted GRE images after acquisition of MR angiographic sequences. Thus, it remains unanswered whether or not different MR imaging approaches (eg, GRE sequences performed immediately after or <1 minute after administration of gadopentetate dimeglumine, dynamic studies) would have substantially contributed to a correct diagnosis in the minor proportion of patients with wrong or inconclusive diagnoses determined with MR imaging.
In our study, MRCP yielded excellent visualization of the common bile duct and the pancreatic duct in most patients. This was consistent with findings in a previous study regarding comparison of MRCP versus ERCP, which demonstrated not only excellent ductal visualization but also high sensitivity and specificity for pancreatic disease (27). As described herein, the majority of patients with pancreatic carcinomas had ductal stenosis or obstruction with consecutive dilatation of the pancreatic duct in the body or tail, with and without dilatation of the common bile duct.
In contrast, findings in chronic pancreatitis tended to be restricted to the main pancreatic duct and side branch dilatations. Whereas T1-weighted MR image cross sections with signal intensity loss and peripancreatic infiltration are most suggestive of pancreatic cancer, in the proportion of patients with poorly delineated parenchymal lesions or diffuse pancreatic signal intensity alterations, the comprehensive approach with MR imaging cross sections and MRCP appeared highly advantageous for the distinction between pancreatic cancer and chronic pancreatitis. Thus, MRCP was an extremely helpful tool for detailed ductal visualization and characterization of pancreatic disease.
In pancreatic carcinoma, direct invasion of adjacent organs is frequently observed. In addition, with complex lymphatic vasculature and no capsule, pancreatic carcinomas tend to metastasize to regional lymph nodes early. In this study, local-regional tumor infiltration and lymphatic involvement could be accurately determined in 89% and 76% of patients, respectively. Limitations were (a) unreliable detection of small metastatic lymph nodes less than 10 mm in diameter and (b) detection of invasion of the mesentery. Moreover, MR imaging assessment of vascular infiltration had a sensitivity and specificity of 83% and 96%, respectively. Within the broad range of results reported in previous studies regarding use of CT and endoscopic sonography for evaluation of vascular infiltration, our results, comparatively, were similar to the highest sensitivity and specificity (90%–98%) reported so far (23,28).
Currently, complete resection provides the only potential cure for pancreatic adenocarcinomas. However, in previous series, up to 80% of patients did not have resectable tumors owing to locally advanced or metastatic disease. Whereas classic contraindications for resection include major vessel encasement or organ invasion other than that of the duodenal wall, with refinements in surgical technique, the safety of surgical resections with, for example, portal vein involvement at a low morbidity and perioperative mortality, has been demonstrated (2,29). Thus, tumor resectability depends predominantly on the degree of vascular infiltration, mesenteric invasion, and systemic spread. We observed positive and negative predictive values for nonresectability of 90% and 83%, respectively, findings that were consistent with those in previous studies. However, in the prediction of tumor nonresectability, the sensitivity was only 69% (nine of 13), which indicates the limits on diagnosing diffuse lymphatic lesions or small metastases in the liver (23,30). However, in our study, the number of patients with these limitations was low.
It must be emphasized that the study population included five patients for whom a final diagnosis of a benign lesion was based on unchanged findings at follow-up reevaluations. We are aware that findings at follow-up evaluations do not offer absolute proof of benign lesion status and that the very rare group of NETs (<1% of all pancreatic tumors) may have a slow growth fraction (31). However, with unchanged findings at reevaluation at 6, 12, and 18 months follow-up, we consider this approach appropriate to exclude malignant tumor growth in the setting of this study. In addition, it should be noted that we performed MR imaging assessments as consensus rather than individual readings.
In summary, the results from this MR study indicate that a comprehensive combination of contrast-enhanced MR imaging sequences with MRCP and MR angiography offers excellent potential to depict and characterize pancreatic lesions. In addition, when a lesion is deemed resectable, then the patient might be spared an unnecessary ERCP and stent placement along with attendant complications, including pancreatitis, perforation, or hemorrhage. Whereas the findings in a minor proportion of patients with chronic pancreatitis provided diagnostic difficulties, local-regional extension, metastatic dissemination, and resectability of most pancreatic tumors were correctly classified.
Although combinations of different diagnostic modalities may yield the highest overall value for the evaluation of pathologic conditions of the pancreas, in consideration of increasing health care expenses, MR imaging with MRCP and MR angiography is a highly suitable noninvasive approach for the assessment of pancreatic disease, with the potential to be substituted for a variety of more conventional diagnostic work-up procedures used in the past.
|
FOOTNOTES |
|---|
Author contributions: Guarantors of integrity of entire study, E.L.H., H.A., N.H., S.R., R.F.; study concepts, E.L.H., N.H., S.R., R.F.; study design, E.L.H., S.R., R.F.; literature research, H.A., S.R.; clinical studies, H.A., S.R., M.B., J.L., R.H.; data acquisition, H.A., S.R.; data analysis/interpretation, E.L.H., H.A., N.H., J.R.; statistical analysis, E.L.H., H.A., S.R.; manuscript preparation, E.L.H., S.R.; manuscript definition of intellectual content, E.L.H., N.H., S.R., R.F.; manuscript editing, E.L.H., H.A.; manuscript revision/review, E.L.H., H.A., N.H., S.R., R.F., B.W.; manuscript final version approval, E.L.H., H.A., N.H., J.R., J.L., S.R., P.N., R.F.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTS DISCUSSIONREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  P. Ghaneh, E. Costello, and J. P Neoptolemos Biology and management of pancreatic cancer Gut, August 1, 2007; 56(8): 1134 - 1152. [Full Text] [PDF]
|
|
|
|
 |
|
 J. Ahualli The Double Duct Sign Radiology, July 1, 2007; 244(1): 314 - 315. [Full Text] [PDF]
|
|
|
|
 |
|
 H. R. Cardenes, E. G. Chiorean, J. DeWitt, M. Schmidt, and P. Loehrer Locally advanced pancreatic cancer: current therapeutic approach. Oncologist, June 1, 2006; 11(6): 612 - 623. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 Y. Mataki, S. Takao, K. Maemura, S. Mori, H. Shinchi, S. Natsugoe, and T. Aikou Carcinoembryonic Antigen Messenger RNA Expression Using Nested Reverse Transcription-PCR in the Peripheral Blood During Follow-up Period of Patients Who Underwent Curative Surgery for Biliary-Pancreatic Cancer: Longitudinal Analyses Clin. Cancer Res., June 1, 2004; 10(11): 3807 - 3814. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
R. Sugita, A. Furuta, K. Ito, N. Fujita, R. Ichinohasama, and S. Takahashi Periampullary Tumors: High-Spatial-Resolution MR Imaging and Histopathologic Findings in Ampullary Region Specimens Radiology, June 1, 2004; 231(3): 767 - 774. [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 |
| RADIOLOGY | RADIOGRAPHICS | RSNA JOURNALS ONLINE |
