A Single-Center Experience of Endoscopic Ultrasonography for Enlarged Pancreas on Computed Tomography

Transcription

1 CLINICAL GASTROENTEROLOGY AND HEPATOLOGY 2006;4: A Single-Center Experience of Endoscopic Ultrasonography for Enlarged Pancreas on Computed Tomography SAMMY HO,* ROBERT J. BONASERA, BONNIE J. POLLACK, JAMES GRENDELL, MARTIN FEUERMAN, and FRANK GRESS* *Department of Gastroenterology, Duke University Medical Center, Durham, North Carolina; and Winthrop University Hospital, Mineola, New York Background & Aims: The clinical significance of fullness or enlargement of the pancreas (FP/EP) is not well established. The objective of this study was to report our experience with endoscopic ultrasonography (EUS) in evaluating patients referred for FP/EP found on computed tomography (CT). Methods: Patients referred to our center for EUS evaluation of FP/EP between January 1998 and December 2003 were studied. Patient demographics, clinical history, endoscopic findings, and follow-up were recorded. Multivariate analysis was used to identify predictors of pancreatic malignancy. Results: A total of 50 patients: 46% (23/50) male, mean age 59 years (range, 18 90) made up our studied population. EUS demonstrated normal findings in 42% (21/50), prominent ventral anlage (embryologic variant) in 14% (7/50), and chronic pancreatitis in 22% (11/50). In 22% (11/50), a suspicious mass was noted and fine-needle aspiration (FNA) was performed. Cytology revealed chronic inflammation in 7 patients, while adenocarcinoma was found in the remaining 4. Median follow-up was 27 months, and the diagnosis did not change in any of the 50 patients. There were no procedure-related complications. After multivariant regression analysis, the factors that were statistically associated with malignancy were a CA19-9 level >300 (P.0002) and weight loss (P <.006). Conclusions: The majority of patients presenting with FP/EP had benign disease, but 8% had pancreatic cancer. Elevated CA19-9 and weight loss were predictive of pancreatic malignancy. EUS and EUS-FNA are safe and accurate diagnostic tests and can play an important role in evaluating patients with FP/EP. Pancreatic cancer is the fourth leading cause of cancerrelated deaths in the United States. It is estimated by the American Cancer Society that 30,300 patients developed pancreatic cancer in 2002, and almost all these patients will die of the disease. 1 Unfortunately, most patients with pancreatic cancer present late in the course of their illness with either locally advanced or metastatic disease. 2,3 The late presentation, aggressive nature of the tumor, and lack of effective therapies all result in a poor prognosis. Only 10% 20% of patients are considered candidates for curative resection at the time of diagnosis. 3,4 Even though surgery offers the only chance for cure, the prognosis remains grim even for patients with potentially resectable cancers. This is evident by a 5-year survival after surgical resection of about 20%. 5,6 Until recently, computed tomography (CT) has been considered to be the imaging modality of choice for the detection of pancreatic cancer. This stems from its easy application, universal experience, and cost-effectiveness. In the published literature, the overall accuracy of contrast-enhanced CT for the detection of pancreatic tumors is approximately 80% 91%. 7,8 Contrast CT enhances most pancreatic adenocarcinomas less than the adjacent normal pancreatic parenchyma and will therefore appear relatively lower in attenuation. Large tumors will distort the normal contour of the pancreas and might infiltrate the adjacent fat, encase nearby vessels, or obstruct the common or pancreatic duct. On the other hand, smaller tumors might not distort the contour of the gland and for this reason might not be detected. Another role for CT is to obtain a tissue diagnosis in patients with suspected pancreatic cancer, especially in nonsurgical cases when chemotherapy is planned. Although CT-guided fine-needle aspiration (FNA) can be used for this purpose, reports of increased risk of peritoneal dissemination of cancer cells and a false-negative rate of nearly 20% make this modality a poor choice for obtaining diagnostic tissue sampling. 9 There are also circumstances in which inconclusive imaging of the pancreas is obtained after a contrastenhanced CT scan is performed. The radiologic terms often used to describe these findings are fullness, enlargement, or prominence of the pancreas (Figure 1). The clinical significance of these indeterminate CT findings is not established, especially in a patient cohort with a relatively low incidence of pancreatic cancer. However, these findings in the proper clinical setting are a cause for Abbreviations used in this paper: CT, computed tomography; EUS, endoscopic ultrasonography; FNA, fine-needle aspiration by the American Gastroenterological Association /06/$32.00 PII: /S (05)

2 January 2006 EUS FOR EVALUATING ABNORMAL PANCREAS ON CT 99 EUS-FNA has the advantage of a shorter needle tract and has been shown to have a lower likelihood of tract seeding compared with percutaneous methods. 22 The safety profile of pancreatic EUS-FNA has been welldemonstrated, with complications occurring in less than 1% of patients. Infection, bleeding, and pancreatitis are the most common complications. 23 There is abundant literature on the clinical utility of EUS and EUS-FNA in patients with suspected pancreatic cancer. However, the clinical significance of indeterminate CT findings, such as fullness, enlargement, or prominence of the pancreas, has not been described. The aim of this study was to report our experience with EUS and EUS-FNA for evaluating patients with fullness, enlargement, or prominence of the pancreas on CT scan. Figure 1. CT scan showing fullness of the pancreas. Photo courtesy of Eric K. Paulson, MD. concern. There is also no consensus on the management of patients presenting with a fullness, enlargement, or prominence of the pancreas. Because of superior imaging of the pancreas, endoscopic ultrasonography (EUS) has become the diagnostic test of choice in many institutions for evaluating pancreatic masses and particularly in these inconclusive cases. 10,11 The high resolution of these images allows for identification of lesions as small as 2 3 mm and can delineate their relationship to adjacent blood vessels such as the portal vein and mesenteric vasculature. In comparison, EUS has the highest detection rate for a pancreatic mass compared with other pancreatic imaging modalities, such as abdominal ultrasonography, CT, and endoscopic retrograde cholangiopancreatography. 12,13 Although EUS is very sensitive for detecting pancreatic masses, endosonography also has its limitations. Similar to CT scan, EUS cannot distinguish malignancy from an inflammatory process such as pancreatitis. In one large series, EUS was only 76% accurate in the diagnosis of malignancy and 46% accurate for focal inflammation. 14 The development of real-time EUS-guided FNA has revolutionized our ability to differentiate benign from malignant pancreatic masses largely by its ability to obtain cytologic tissue from suspicious pancreatic lesions. This is of clinical importance because the majority of pancreatic mass lesions are aggressive neoplasms and thus would benefit from early detection, diagnosis, and surgical intervention. 15,16 Compared with other imaging modalities, the results of EUS-FNA of pancreatic masses are excellent, with a sensitivity reported to be greater than 80% and a specificity of virtually 100% In addition, the safety of CT-FNA is uncertain because of concern about peritoneal seeding via the needle tract. 21 Materials and Methods All patients referred to our center for EUS evaluation of the pancreas for indeterminate CT abnormalities between January 1998 and December 2003 were reviewed. Patients with radiologic reports that included terms such as enlargement, fullness, or prominence of the pancreas were included. Patients with recent acute pancreatitis ( 6 weeks) and those with CT findings showing chronic pancreatitis, dilated pancreatic duct, and discrete mass lesions were excluded. Subjects were identified through a query of our computerized EUS database. This database maintains all cases of EUS and EUS- FNA performed at our facility since Medical records were reviewed in a retrospective fashion. Patient demographics, clinical history including follow-up, laboratory results, endosonographic findings, and FNA cytology results when performed were recorded. Follow-up was achieved by medical record review. If this was insufficient, then referring physicians and/or patients were contacted to obtain clinical follow-up. A normal pancreatic EUS was defined as no evidence of mass/ tumor/cyst and no changes of chronic pancreatitis. The endosonographic diagnosis of chronic pancreatitis was defined as the presence of 3 or more of the following EUS criteria: hyperechoic foci, stranding, increased parenchymal lobularity, cystic spaces, calcifications, and ductal abnormalities (main duct dilation 4 mm, irregularity, visible side branches, or ductal calculi). EUS-FNA was performed in cases in which a suspicious mass was noted. This study was approved by our institutional review board as required by New York State law for review of medical records. Study Site All EUS and EUS-FNA examinations were performed at Winthrop University Hospital, a major teaching hospital for the State University of New York at Stony Brook, by 2 experienced endosonographers (F.G., B.P.) in conjunction with an advanced EUS training fellow. Endoscopic Ultrasonography Technique EUS and EUS-FNA techniques have been described previously. 24,25 In brief, target lesions were initially identified

3 100 HO ET AL CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 4, No. 1 with either a radial scanning echoendoscope (GF-UM-20 or GF-UM-30; Olympus America Inc, Melville, NY) or linear echoendoscope (GFUCT-140; Aloka/Olympus America, Melville, NY or FG-32UA or FG-36UX; Pentax Precision Co, Orangeburg, NY). If the EUS findings detected a mass or raised the suspicion that a focal lesion was present, EUSguided FNA was performed by using a curved linear array echoendoscope (GFUCT-140 or FG-32UA or FG-36UX). All patients were sedated with either conscious sedation by using meperidine/midazolam or propofol titrated to patient comfort to achieve optimal sedation. Propofol was administered by an anesthesiologist in all cases. Prophylactic antibiotics were not given routinely. All biopsies were performed by using a disposable 22- or 25-gauge EUS-FNA needle system (Wilson-Cook Inc, Winston-Salem, NC). Lesions located in or adjacent to the head of the pancreas were biopsied with the transducer in the duodenal bulb or second portion of the duodenum, whereas lesions in the body and tail were sampled via a transgastric approach. Because the scanning plane is in the long axis of the instrument, real-time visualization of the biopsy needle is permitted. In some cases, color Doppler was used to exclude the presence of vessels lying in the path of the needle before the lesion was punctured under EUS guidance. After localization of the mass, the EUS-FNA needle was slowly advanced into the lesion under real-time EUS imaging. At our center, largely as a result of economic factors, an on-site cytopathologist was not present during EUS-FNA sampling. An experienced EUS assistant was present at all procedures and was trained to assist with specimen acquisition and preparation of sample slides for review by the cytopathologist at a later time. In general, our institution abides by the practice of obtaining a minimum of 7 passes into a pancreatic mass and 5 passes into suspicious lymph nodes. 26 Cytologic Interpretation of Endoscopic Ultrasonography Fine-Needle Aspiration Specimens A smear was interpreted as positive for malignancy if malignant cells were clearly identified. A diagnosis of negative for malignancy was rendered if only benign ductal epithelial cells were present with or without accompanying necrosis/inflammation. Statistical Methods For descriptive analyses, continuous variables are reported as mean standard deviation. Proportions such as sensitivity and specificity are reported with their corresponding exact 95% confidence intervals. The 2 test or Fisher exact test was used to evaluate possible categorical predictors of malignancy. Multivariate analysis was not done because of the small number of malignant cases. The kappa statistic and confidence interval are provided to measure the agreement between the CA 19-9 marker and biopsy findings. All calculations were performed by using SAS 8.2 (SAS Institute Inc, Cary, NC) for Windows. Results were considered statistically significant when P value was.05. Results A total of 50 patients underwent EUS evaluation for fullness, enlargement, or prominence of the pancreas found on CT scan during the study period. Table 1 summarizes the main demographic and clinical characteristics of our patient population. There were 23 (46%) men and 27 (54%) women in the study. The mean age was years (range, years). Presenting symptoms included abdominal pain (86%), prior pancreatitis (8%), and bloating (6%). The enlargement, fullness, or prominence was in the head of the pancreas in 88% (44/50), body in 2% (1/50), and in the tail in 10% (5/50) of patients. EUS showed normal findings in 42% (21/50) of patients (Figure 2). In 14% (7/50), a prominent ventral anlage was evident. This endosonographic finding was considered an embryologic variant with minimal clinical significance. In 22% (11/50) of patients, EUS findings were consistent with chronic pancreatitis. Specifically, 3 patients were found to have a dilated pancreatic duct ( 4 mm) on EUS. In the remaining 22% (11/50) of patients, a suspicious mass was noted, and FNA was performed (Figure 3). Cytology showed chronic inflammation in 7 patients, whereas adenocarcinoma was found in the remaining 4. Chi-square analysis did not show diabetes, alcohol use, abnormal liver function test results, dilated common bile duct ( 1 cm), and retroperitoneal lymphadenopathy as predictors of malignancy (Table 2). Significantly (P.05) associated with malignancy were weight loss (P.006) and CA19-9 level 300 (P.0002), sensitivity 75%, specificity 100% (Table 3). Table 1. Patient Demographics (n 50) n Frequency (%) Gender Female Male Location Body Head Tail Analysis with characteristics Diabetes mellitus 6/50 12 Ethyl alcohol 11/50 22 Weight loss 15/50 30 Jaundice 2/50 4 Elevated CA /50 6 Dilated common bile duct 3/50 6 Lymphadenopathy 2/50 4 Mean age (y) standard deviation ; range, Median follow-up (mo) 27; range, 6 63 Mean follow-up (mo) standard deviation

4 January 2006 EUS FOR EVALUATING ABNORMAL PANCREAS ON CT 101 Figure 2. EUS findings. There was no documentation of EUS-related complications in the study group. Median follow-up was 27 months, and the diagnosis did not change in any of the 50 patients. No patient with a normal EUS of the pancreas developed pancreatic cancer during the follow-up period. Of the 4 patients found to have cancer, 3 underwent chemoradiation therapy, and none went on to have surgical resection caused by comorbidities and/or the unresectability of the tumor. Three of the 4 patients are deceased, 1 by cause unrelated to underlying pancreatic cancer. On the basis of our survival follow-up data, the negative predictive value of a normal pancreatic EUS for the diagnosis of pancreatic cancer was 100% in this study cohort of patients. Discussion The overall survival of patients with pancreatic cancer has not changed much during the past several decades. This might in part be due to the difficulty in reliably detecting early stage pancreatic tumors. Small pancreatic tumors are asymptomatic and are frequently missed by conventional imaging techniques including CT, abdominal US, and magnetic resonance imaging. Prior studies have shown that EUS can detect pancreatic tumors undetected by other modalities, and EUS-guided FNA permits accurate histologic confirmation of malignancy. 27 There are occasions when CT imaging reveals subtle abnormalities of the pancreas in the absence of a discrete mass lesion. These radiographic findings are often seen as secondary changes in patients with chronic pancreatitis 28 and are sometimes difficult to distinguish from carcinoma or other tumors. EUS often is performed because of a clinical suspicion of pancreatic cancer when the results of other diagnostic imaging studies are indeterminate. In the present study, 50 patients underwent EUS evaluation for fullness or enlargement of the pancreas found on CT scan. Although the majority of patients had benign disease, 8% were found to have pancreatic cancer. In addition, most of the potential predictors for malignancy were not reliable in predicting underlying pancreatic cancer in this group. An elevated CA 19-9 level ( 300) was predictive of pancreatic malignancy with high specificity. However, CA 19-9 can be falsely elevated in certain situations, such as benign biliary obstruction, limiting its value in patients with jaundice. Therefore, EUS and EUS-FNA are valuable means to document the absence of malignancy in patients with inconclusive CT imaging of the pancreas. On the basis of the results of this study, we believe that it is important to categorize our studied population into 2 groups depending on the pre-test probability of having underlying malignancy. In our study of 50 patients, there were 3 patients with elevated CA 19-9 and 15 with weight loss (as indicated in Table 1). These 15 patients included the 3 with elevated CA 19-9 levels. Therefore, this subgroup of 15 patients might be considered as having a higher pre-test probability of pancreatic cancer, whereas the other 35 patients would be viewed as having a lower pre-test likelihood. For this group of 35 patients, all had negative EUS, with no patient developing cancer during follow-up. For these patients then, after EUS, we obtained a lower post-test likelihood. For the group of 15 patients, however, 4 were found to have cancer by EUS, and the diagnosis did not change after the follow-up period. For this subgroup of patients with higher pre-test likelihood, the prevalence was increased from 8% (4/50) to 27% (4/15), with resulting negative predictive value still 100%. Future larger studies would be needed to confirm that patients with such higher clinical suspicion would yield similarly high negative value. In our study, all patients who had a clinical history and/or radiologic findings of chronic pancreatitis were excluded. Interestingly, 11 patients were found to have chronic pancreatitis as defined by the presence of 3 or Figure 3. (A) A hypoechoic mass found in the head of pancreas on EUS. (B) FNA of the pancreatic mass with a 25-gauge FNA needle.

5 102 HO ET AL CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 4, No. 1 Table 2. Predictors of Malignancy Variable Negative for malignancy (n 46) Positive for malignancy (n 4) P value Age (y) Mean time followed (mo) With characteristics Dilated common bile duct 2 (4.3%) 1 (25%).23 Lymphadenopathy 1 (2.2%) 1 (25%).15 Diabetes 4 (8.7%) 2 (50%).066 Ethyl alcohol 9 (19.6%) 2 (50%).21 Jaundice 1 (2.2%) 1 (25%).155 Weight loss 11 (23.9%) 4 (100%).006 CA (0%) 3 (75%).0002 Female 25 (54.3%) 2 (50%).99 NOTE. P values for Fisher exact test for categorical variable; P values for rank sum test for continuous variables. more EUS criteria. The ability of EUS to identify subtle changes on the pancreas makes this imaging modality more sensitive than CT in detecting chronic pancreatitis. 29 There are 2 notable limitations in our study. Helical CT with thin-section imaging techniques through the pancreas were not standardized for all the patients included in this study. In recent years, advances in CT technology have resulted in vast improvements in pancreatic imaging. Multiphase helical scanners with rapid infusion of intravenous contrast medium allow for finer delineation of the pancreatic parenchyma. 30 Therefore, one could argue that our results could be different if the newer multidetector row CT scanners were used in our study. However, a review 31 that pooled data from 4 studies comparing the accuracy of EUS with helical CT in the evaluation of confirmed pancreatic cancer found that EUS detected more tumors (97% vs 73%), was more accurate for determining tumor resectability (91% vs 83%), and was more sensitive for detecting vascular Table 3. CA19-9 Statistical Significance Biopsy result Total a CA a Total Operating characteristics Value 95% Confidence interval Sensitivity 3/4 (75%) Specificity 46/46 (100%) Positive predictive value 3/3 (100%) Negative predictive value 46/47 (97.9%) Raw agreement 49/50 (98%) NOTE. Kappa ; P a A negative finding means no mass was discovered, or the mass was interpreted as inflammatory pathologically, and thus negative. invasion (91% vs 64%). A more recent published study by Agarwal et al 32 confirmed that EUS was more accurate than spiral CT for diagnosing pancreatic cancer (94% vs 74%). Although our study did not uniformly use the newer multidetector row CT scanners, we believe that our results and conclusions are applicable to the current standard of practice. The second potential limitation to our study was that our patient population was referred to our center for EUS evaluation from various institutions and facilities in the community. One has to acknowledge that CT interpretation and technique are operator and instrument dependent, and results might vary widely between institutions. Subtle changes such as enlargement or fullness of the pancreas might be interpreted subjectively as a mass by one radiologist and as a normal variant by another. This could suggest that our results might not be generally applicable because we did not have a single radiologist review all CT studies. However, we believe that study design provides a more realistic and practical representation of the current practice in the community, in which patients are often referred to tertiary centers from local community hospitals for evaluation. We believe that our results would not have been as representative of real-life practice if all patients had their CT scans reviewed by the same academic university based radiologist. In conclusion, EUS is an accurate and safe modality for evaluating patients with inconclusive CT imaging of the pancreas. In addition, EUS-FNA is an effective modality for establishing a tissue diagnosis in patients with pancreatic lesions detected by EUS. As demonstrated in our present study, a normal EUS of the pancreas rules out a diagnosis of pancreatic cancer with a high degree of certainty when there is a clinical suspicion of pancreatic malignancy based on equivocal CT findings. Therefore, if there is high confidence that the pancreas has been examined thoroughly at EUS and found to be normal,

6 January 2006 EUS FOR EVALUATING ABNORMAL PANCREAS ON CT 103 then follow-up imaging studies might not be necessary. This recommendation has been suggested by Catanzaro et al. 33 However, we believe that in patients who have a high pre-test probability of pancreatic cancer (elevated CA 19-9 and weight loss) and a normal EUS examination, the decision to obtain further imaging and the timing of such imaging should be decided on an individual basis. We believe that EUS should be strongly considered as the next step in the evaluation of patients with focal enlargement or fullness of the pancreas, especially when the clinical index of suspicion is high for a malignancy. Although continued advancements in CT technology have improved the accuracy of this modality for detecting pancreatic tumors, the impact of new multi-slice CT scanners on the diagnostic accuracy of pancreatic mass detection will need to be evaluated further. Future prospective studies comparing EUS and spiral CT for evaluating the pancreas in patients with abdominal symptoms or suspected pancreatic disease are warranted. References 1. Cancer facts and figures Atlanta: American Cancer Society, Moossa AR, Gamagami RA. Diagnosis and staging of pancreatic neoplasms. Surg Clin North Am 1995:75: Palazzo L. Staging of pancreatic carcinoma by endoscopic ultrasound. Endoscopy 1998;30A Ahmad NA, Lewis JD, Ginsberg GG, et al. Long-term survival after pancreatic resection for pancreatic adenocarcinoma. Am J Gastroenterol 2001;96: Sohn TA, Yeo CJ, Cameron JL, et al. Resected adenocarcinoma of the pancreas-616 patients: results, outcomes, and prognostic indicators. J Gastrointest Surg 2000;4: Cameron JL, Crist DW, Sitzmann JV, et al. Factors influencing survival after pancreaticoduodenectomy for pancreatic cancer. Am J Surg 1991;160: Bluemke DA, Cameron JL, Hruban RH, et al. Potentially resectable pancreatic adenocarcinoma: spiral CT assessment with surgical and pathologic correlation. Radiology 1995;197: Freeny PC, Traverso LW, Ryan JA. Diagnosis and staging of pancreatic adenocarcinoma with dynamic computed tomography. Am J Surg 1993,165: Bret PM, Nicolet V, Labadie M. Percutaneous fine-needle aspiration biopsy of the pancreas. Diagn Cytopathol 1986:2: Catanzaro A, Richardson S, Veloso H, et al. Long-term follow-up of patients with clinically indeterminate suspicion of pancreatic cancer and normal EUS. Gastrointest Endosc 2003;58: Legmann P, Vignaux O, Dousset B. Pancreatic tumors: comparison of dual-phase helical CT and endoscopic ultrasonography. Am J Roentgenol 1998;170: Protiva P, Sahai AV, Agarwal B. Endoscopic ultrasound in the diagnosis and staging of pancreatic neoplasms. Int J Gastrointest Cancer 2001;30: Bhutani MS. Endoscopic ultrasound in pancreatic disease. Semin Gastrointest Dis 1998;9: Rosch T, Lorenz R, Braig C, et al. Endoscopic ultrasound in pancreatic tumor diagnosis. Gastrointest Endosc 1991;37: Wiersema MJ. Accuracy of endoscopic ultrasound in diagnosing and staging pancreatic carcinoma. Pancreatology 2001;1: Gress F, Gottlieb K, Sherman S, et al. Endoscopic ultrasonography-guided fine-needle aspiration biopsy of suspected pancreatic cancer. Ann Intern Med 2001;134: Varadarajulu S, Wallace MB. Applications of endoscopic ultrasonography in pancreatic cancer. Cancer Control 2004:11: Levy MJ, Wiersema MJ. Endoscopic ultrasound in the diagnosis and staging of pancreatic cancer. Oncology 2002;16: Suits J, Frazee R, Erickson RA. Endoscopic ultrasound and fine needle aspiration for the evaluation of pancreatic masses. Arch Surg 1999;134: Bhutani MS. Endoscopic ultrasound in pancreatic disease. Semin Gastrointest Dis 1998;9: Ferrucci JT, Wittenberg J, Margolies MN, et al. Malignant seeding of the tract after thin-needle aspiration biopsy. Radiology 1979: 130: Micames C, Jowell PS, White R, et al. Lower frequency of peritoneal carcinomatosis in patients with pancreatic cancer diagnosed by EUS-guided FNA vs percutaneous FNA. Gastrointest Endosc 2003:58: O Toole D, Palazzo L, Arotcarena R. Assessment of complications of EUS-guided fine-needle aspiration. Gastrointest Endosc 2001; 53: Wiersema MJ, Vilmann P, Giovannini M, et al. Endosonographyguided fine-needle aspiration biopsy: diagnostic accuracy and complication assessment. Gastroenterology 1997;112: Harewood GC, Wiersema MJ. Endosonography-guided fine needle aspiration biopsy in the evaluation of pancreatic masses. Am J Gastroenterol 2002;97: LeBlanc JK, Ciaccia D, Al-Assi MT, et al. Optimal number of EUS-guided fine needle passes needed to obtain a correct diagnosis. Gastrointest Endosc 2004;59: Gress FG, Hawes RH, Savides TJ. Role of EUS in the preoperative staging of pancreatic cancer: a large single-center experience. Gastrointest Endosc 1999;50: Shams J, Stein A, Cooperman AM. Computed tomography for pancreatic diseases. Surg Clin North Am 2001;81: Buscail L, Escourrou J, Moreau J, et al. Endoscopic ultrasonography in chronic pancreatitis: a comparative prospective study with conventional ultrasonography, computed tomography, and ERCP. Pancreas 1995;10: Takeshita K, Furui S, Takada K. Multidetector row helical CT of the pancreas: value of three-dimensional images, two-dimensional reformations, and contrast-enhanced multiphasic imaging. J Hepatobiliary Pancreat Surg 2002;9: Hunt GC, Faigel DO. Assessment of EUS for diagnosing, staging, and determining resectability of pancreatic cancer: a review. Gastrointest Endosc 2002;55: Agarwal B, Abu-Hamda E, Molke KL. Endoscopic ultrasoundguided fine needle aspiration and multidetector spiral CT in the diagnosis of pancreatic cancer. Am J Gastroenterol 2004;99: Catanzaro A, Richardson S, Veloso H, et al. Long-term follow-up of patients with clinically indeterminate suspicion of pancreatic cancer and normal EUS. Gastrointest Endosc 2003;58: Address requests for reprints to: Sammy Ho, MD, Duke University Medical Center, Gastroenterology, Box 3189, Durham, North Carolina sammyho01@yahoo.com; fax: (919)