ORIGINAL ARTICLE: Clinical Endoscopy

Combined EUS with FNA and ERCP for the evaluation of patients with obstructive jaundice from presumed pancreatic malignancy
William A. Ross, MD, Sanjeev M. Wasan, MD, Douglas B. Evans, MD, Robert A. Wolff, MD, Leonard V. Trapani, MD, Gregg A. Staerkel, MD, Thomas Prindiville, MD, Jeffrey H. Lee, MD Houston, Texas, USA

Background: An EUS-guided FNA (EUS-FNA) and a therapeutic ERCP are frequently required for the evaluation of patients who were seen for an obstructing periampullary lesion. Objective: To determine the feasibility and outcomes of combining an EUS-FNA and a therapeutic ERCP into a single session. Design: Retrospective single-center study. Setting: Tertiary-referral cancer center. Patients: A total of 114 patients with a suspected malignant obstructing lesion in the pancreatic head. Interventions: An EUS with or without FNA plus an ERCP. Main Outcome Measurements: Duration, diagnostic yield, and complication rate of the combined procedures. Results: The mean (SD) total procedure time (EUS, with or without FNA plus ERCP) was 73.6 30 minutes, with a median of 66 minutes (range 25-148 minutes). In many cases, cytologic diagnosis from FNA became available during an ERCP, which obviated the need for further sampling. EUS-FNA had a sensitivity, specicity, positive predictive value, negative predictive value, and overall accuracy of 84.6%, 100%, 100%, 62.9%, and 87.8%, respectively. During an ERCP, endoscopic sphincterotomies were performed in 51 patients, and biliary stents were placed in 96 patients. Twelve patients (10.5%) had a complication, with 6 having postprocedural pancreatitis. Limitations: Retrospective single-center experience. Conclusions: Combined EUS-FNA and therapeutic ERCP is technically feasible, with a complication rate no higher than the component procedures, while efciently providing tissue diagnosis and biliary drainage. (Gastrointest Endosc 2008;68:461-6.)

EUS has a complementary role to helical CT for the evaluation of patients with pancreatic adenocarcinoma.1 Despite advances in CT technology, EUS continues to have better sensitivity for the detection of pancreatic cancers smaller than 3 cm in size.2,3 A frequent presentation of pancreatic cancer is painless jaundice, with a CTor a transabdominal US suggestive of an obstruction in the area of the pancreatic head. The next goals in the evaluation involve establishing a diagnosis and relieving the biliary obstruction.

Abbreviations: EST, endoscopic sphincterotomy; EUS-FNA, EUS-guided FNA; NPV, negative predictive value; PPV, positive predictive value. Copyright 2008 by the American Society for Gastrointestinal Endoscopy 0016-5107/$32.00 doi:10.1016/j.gie.2007.11.033

An ERCP as the next step will relieve the obstruction and may establish a diagnosis. However, brushings obtained during an ERCP have a low sensitivity for pancreatic cancer, and an ERCP produces little additional information regarding tumor stage.4,5 In addition, an ERCP carries a risk of pancreatitis, which can make interpretation of subsequent imaging studies difcult, and increases the risk of complications during a future FNA.6 Also, the presence of a biliary stent placed during an ERCP may adversely affect the accuracy of a subsequent EUS.7 However, performing an EUS rst enables staging and pathologic diagnosis but leaves the obstruction unaddressed until another day. Our institutional approach to patients with potentially resectable locally advanced pancreatic cancer is to provide preoperative chemoradiation. As a result, tissue
Volume 68, No. 3 : 2008 GASTROINTESTINAL ENDOSCOPY 461

www.giejournal.org

Obstructive jaundice from presumed pancreatic malignancy

Ross et al

conrmation of the suspected diagnosis and a normal serum bilirubin are required before the initiation of chemotherapy. To facilitate the evaluation of patients seen with obstructive jaundice and suspected pancreatic malignancy, we decided to combine EUS-FNA and ERCP into a single session to provide both prompt tissue diagnosis and relief of obstructive jaundice. This approach enables us to take advantage of the strengths of both modalities and reduces the delay for the initiation of chemoradiation. To demonstrate the feasibility and results of this approach, we analyzed our experience in 114 patients who had a combined ERCP and EUS for evaluation of obstructive jaundice and/or a periampullary mass.

Capsule Summary
What is already known on this topic
d

An ERCP may relieve biliary obstruction, but specimens obtained during the procedure have a low sensitivity for malignancy. Complications of ERCP, such as pancreatitis and bile duct stents, can adversely affect the accuracy of a subsequent EUS. Performing an EUS before an ERCP facilitates staging and pathologic diagnosis but does not treat the obstruction.

What this study adds to our knowledge

d

PATIENTS AND METHODS

Data were collected retrospectively from a review of the electronic medical record and endoscopy database on all patients who underwent an EUS and ERCP in a single session at our center from July 1, 2001, to October 31, 2004. All patients in the study were seen with obstructive jaundice and/or abnormal periampullary imaging that raised a concern for periampullary (including pancreatic) malignancy. The decision to arrange a combined EUSERCP was made jointly by the referring physician and the endoscopist. Patients had a thin-slice multidetector dual-phase helical CT study of the abdomen and/or abdominal magnetic resonance imaging at our institution before the EUS-ERCP.8 Combined EUS-ERCP was done in 1 of 2 dedicated uoroscopy suites, with total intravenous anesthesia of the patient by using propofol with supplemental midazolam and/or fentanyl administered by an anesthesiologist and nurse anesthetist team. By using a linear-array endoscope with or without a radial echoendoscope (Pentax EG-3830U and EG-3630UR, Pentax Corp, Tokyo, Japan, or Olympus/ Aloka GFUC-130, Aloka Medical Device Co, Tokyo, Japan), an EUS evaluation of the ampulla, pancreas, bile duct, and duodenum was done. When a pancreatic mass or a thickened bile-duct wall was present, 2 to 4 aspirates were obtained with a ne needle (Echo-1-22 needle; Cook Endoscopy, Winston-Salem, NC). The specimens were immediately assessed for adequacy by a cytotechnician and then were stained with Papanicolaous stain for examination by a staff cytologist to establish a tissue diagnosis. Tissue processing and interpretation took 5 to 30 minutes, with results being called to the endoscopy suite by the cytopathologist. While the FNA specimens were being evaluated, an ERCP was performed. When the FNA specimens were adequate to establish a diagnosis, no further tissue sampling was attempted during the ERCP. However, if the initial specimens were not positive for cancer or if the reporting was delayed, then additional tissue samples were obtained by brushing and/or a biopsy. If the FNA specimens were not
462 GASTROINTESTINAL ENDOSCOPY Volume 68, No. 3 : 2008
d

In a retrospective single-center study of 114 patients with suspected malignant obstruction in the pancreatic head, an EUS-guided FNA and therapeutic ERCP performed in a single session had a mean total procedure time of 73.6 minutes and a complication rate of 10.5%. Specimens from EUS-guided FNA had a sensitivity, specificity, positive predictive value, negative predictive value, and overall accuracy of 83.3%, 100%, 100%, 60.7%, and 86.7%, respectively.

thought to be adequate, then another FNA with the reintroduction of the linear EUS endoscope after completion of the ERCP was carried out. Antibiotics were given prophylactically at the discretion of the endoscopist. Procedural time was calculated from the insertion of the rst instrument to the withdrawal of the last instrument. For the purpose of analysis, atypia on cytology was considered negative for cancer, and a suspicious result on cytology was considered separately from positive or negative results. In the absence of a tissue diagnosis of cancer, malignancy was thought to be present if the primary tumor enlarged and if evidence for metastatic disease was present during follow-up. The severity of complications was graded by criteria dened by Cotton et al.9 The Fisher exact test was used to evaluate the association between procedural features and complications.

RESULTS Demographic data

A total of 114 patients had an EUS combined with an ERCP during the study period. There were 66 men and the overall mean (SD) age was 62.6 11.8 years. The most common indications for an EUS were a periampullary mass (67%), ductal abnormalities (18%), and staging (6%). The most common indications for an ERCP were jaundice (61%), stent revision (18%), and a ductal stricture (8%). Fifty-nine patients had a prior attempt at tissue diagnosis that was nondiagnostic. At the time of their combined procedures, 43% of all the patients had prior biliary stenting. Of the 59 patients with prior nondiagnostic biopsies, 44 patients had a single attempt for tissue diagnosis at other
www.giejournal.org

Ross et al

Obstructive jaundice from presumed pancreatic malignancy

institutions: biliary brushings alone (23), biopsies (11), an EUS-FNA (5), CT-guided FNA (4), and peritoneal washing (1). Ten of the 59 patients had multiple nondiagnostic attempts at tissue diagnosis at other institutions, including biliary brushings (10), an EUS-FNA (7), endoscopic biopsy (3), and surgery (1). The remaining 5 of the 59 patients had 1 or more nondiagnostic studies at M.D. Anderson before the combined procedures: biliary brushings (3), endoscopic biopsies (2), an EUS-FNA (2), or a CT-guide FNA (1). After the combined procedures, a malignant diagnosis was established for 80 patients by various means of tissue acquisition (78) or clinical course (2). Benign processes were diagnosed in the remaining 34 patients (Table 1).

TABLE 1. Final diagnosis Malignant (n Z 80) Pancreatic Ampullary Cholangiocarcinoma Neuroendocrine Hepatoma Pancreatic adenosquamous Pancreatic lymphoma Benign (n Z 34) Stricture Chronic pancreatitis Papillary stenosis Ampullary adenoma Autoimmune pancreatitis Common duct stone Intraductal papillary mucinous neoplasm Pancreatic divisum Pancreatic pseudocyst Serous cystadenoma Sclerosing cholangitis Adenomyosis Choledochocele 11 6 3 3 2 2 1 1 1 1 1 1 1 68 5 2 2 1 1 1

EUS-FNA
A total of 87 patients had an EUS-FNA, and these specimens were positive for malignancy in 55 (53 adenocarcinomas, 1 adenosquamous carcinoma, and 1 neuroendocrine tumor), suspicious in 4, and negative in 28. Of the 4 patients with suspicious FNA specimens, 1 was diagnosed with lymphoma by another EUS-FNA, and the other 3 were diagnosed with pancreatic cancer at surgery. Of the 28 patients with negative FNA specimens, 11 were found to have cancer by other means: 5 at surgery, 2 by CT-guided FNA, 1 by a later endoscopic biopsy of an ampullary mass, 1 by another EUS-FNA, and 2 by clinical means only. Of the 55 positive FNAs, 15 were conrmed with a surgical course and the remainder by a clinical course. Excluding the 4 suspicious cases, EUS-FNA had a sensitivity of 83.3%, a specicity of 100%, a positive predictive value (PPV) of 100%, a negative predictive value (NPV) of 60.7%, and an overall accuracy of 86.7% in differentiating cancer from nonmalignant conditions (Table 2).

Brushings of biliary strictures

Biliary brushings were obtained in 54 patients: the brushings were positive for malignancy in 4, suspicious in 4, and negative or inconclusive in 46. The 4 positive and 4 suspicious results were veried at surgery as cancer. In 26 of the 46 patients with negative brushings, a diagnosis of malignancy was established by other means: 13 by a positive FNA at the time of the EUS-ERCP, 1 by a positive biopsy specimen after a negative FNA, 6 at surgery, 2 by a second session of EUS-FNA, 3 by CT-guided FNA, and 1 by transabdominal USguided FNA. With suspicious results excluded, biliary brushings had a sensitivity of 13.3%, a specicity of 100%, a PPV of 100%, a NPV of 43.4%, and an accuracy of 48% for differentiating malignant from benign conditions.

Biopsy of biliary strictures

Biopsies of bile-duct strictures were performed during an ERCP in 20 patients. Results were positive for malignancy in 3 cases, suspicious in 1, and negative in 16. Of the 16 patients with negative biopsy specimens, 7 were
www.giejournal.org

found to have cancer by other means: 5 at surgery, 1 by a later endoscopic biopsy, and the other by a positive EUS-FNA. The patient with a suspicious biopsy had a negative FNA and brushing results but was found to have pancreatic adenocarcinoma at surgery. Overall, 73 patients with a nal cancer diagnosis had a biopsy, a brushing, and/or an FNA during the combined procedures. Without considering suspicious results, combined endoscopic techniques demonstrated a sensitivity of 82.6% (57/69). Seventy-one patients with cancer died at a mean of 388 days. The other 9 patients were observed for a median of 1248 days (range 609-1818 days). Patients with a benign diagnosis were observed for a median of 504 days (range 7-2022 days) with only 1 new cancer diagnosis. The new cancer diagnosis was a pancreatic adenocarcinoma in a patient 29 months after combined procedures for a choledochocyst. The cancer was located in the neck of the pancreas and was distinct from the choledocalcyst on CT imaging and, therefore, was thought to be unrelated to the ndings at the earlier EUS-ERCP.
Volume 68, No. 3 : 2008 GASTROINTESTINAL ENDOSCOPY 463

Obstructive jaundice from presumed pancreatic malignancy

Ross et al

TABLE 2. Comparison of diagnostic methods in 114 patients with suspected pancreatic malignancies, excluding suspicious results Method EUS-FNA (n Z 82) Brushings (n Z 50) Sensitivity, % 83.3 (55/66) 13.3 (4/30) Specificity, % 100 (17/17) 100 (20/20) PPV, % 100 (55/55) 100 (4/4) NPV, % 60.7 (17/28) 43.4 (20/46) Accuracy, % 86.7 (72/83) 48 (24/50)

Duration of combined procedures

The mean (SD) total procedure time (an EUS with or without an FNA plus ERCP) was 73.6 30 minutes and the median was 66 minutes. The range was wide, extending from 25 to 148 minutes. The median procedural time with an FNA was 69 minutes and without an FNA was 65 minutes. In many cases, the nal FNA results became available during an ERCP, which obviated the need for further sampling. Fifty-one endoscopic sphincterotomies (EST), including 15 pre-cut papillotomies with a needle knife, were performed during ERCP. In 96 cases, biliary stents, including 26 metal stents, were placed.
TABLE 3. Complications of EUS-ERCP No. patients 5 3 1 1 1 1

Complication Acute pancreatitis Early cholangitis Combined cholangitis/pancreatitis Nausea/vomiting with pain but with normal amylase Hemobilia Biloma

Complications
Complications were seen in 12 patients (10.5%), with pancreatitis being the most common (Table 3). Of the 12 patients with complications, 10 had pancreatic masses that averaged 2.81 cm when seen on an EUS examination compared with a mean mass size of 3.1 cm in those with no complications. The difference in mass size did not reach statistical signicance. Eight of the 10 masses proved to be malignant. The diagnoses in the other 4 patients with complications were chronic pancreatitis (1), choledochocele (1), papillary stenosis (1), and benign common bile duct stricture (1). All 12 patients were outpatients at the time of the combined procedures and required hospitalization for the complications. Four of the complications were mild, 7 were moderate, and 1 was severe. No perforations or procedure-related deaths occurred. No surgical intervention was required to deal with the complications. Other than delaying onset of therapy, oncologic management was not affected. Various procedural features, such as a biliary brushing, an FNA, an EST, stent placement, or metal-stent usage, were evaluated for their relationship to overall complications and pancreatitis. No features were found to be associated with overall complications.

DISCUSSION
In a single session, we were able to establish a tissue diagnosis, complete local staging, and relieve biliary obstruction. Our institutional preference for preoperative chemoradiation in patients with potentially resectable pancreatic cancer requires that the diagnosis be established and the bilirubin normalize before initiation of therapy. Therefore, the combined procedures enable com464 GASTROINTESTINAL ENDOSCOPY Volume 68, No. 3 : 2008

pletion of planning for subsequent therapy and rapid initiation of appropriate care. All of this was accomplished with a complication rate no greater than that for the component procedures. The reported rate of signicant complications with pancreatic FNA is 2.5% to 5%, and the risk of pancreatitis is 0.5% to 2.0%.10-13 The risk of pancreatitis with a diagnostic ERCP is reported to be 5%.14 In 3 recent ERCP series where the majority of cases were therapeutic, the rate of signicant complications was 10% to 13%15-17 In particular, biliary-stent placement was associated with increased risks of both pancreatitis and cholangitis.17 In our series, the overall complication rate was 10.5%, and the pancreatitis rate was 4.3%. These rates are within the range of reported complication rates for the individual procedures. Numerous procedural details were studied to determine if there was association with overall complications or pancreatitis. However, no statistically signicant association was identied. The performance of an EUS-FNA for detecting malignancy in the combined procedures was comparable with earlier reports from our institution and those of other centers when an FNA alone was done.3,7,8,11,18,19 Although the sensitivity of brushings and biopsies was much lower than that of an EUS-FNA in our study, a biliary biopsy did detect 2 cancers missed by an FNA, and another cancer was suggested by results of brushings. Our yield from brushings was adversely affected by several factors: (1) selection bias, in that 33 of our patients had already had nondiagnostic brushings before the combined procedures, (2) interpretation of biliary brushing may be complicated by stent-induced inammation, which was a factor in 43%
www.giejournal.org

Ross et al

Obstructive jaundice from presumed pancreatic malignancy

of patients, and (3) the preponderance of pancreatic cancer versus cholangiocarcinoma in this series. ERCP-based methods of tissue sampling have a role if an EUS-FNA fails to make a diagnosis. The mean procedure time for the combined procedures at our teaching hospital of 73.6 minutes compares favorably with reports in the literature regarding the time required for the component procedures. Wiersema et al20 reported that an EUS-FNA alone took 30 to 120 minutes, with the pancreatic cases being in the upper limit of this range.20 Mertz and Gautam21 reported that performing a pancreatic EUS-FNA averaged 65 minutes. A therapeutic ERCP takes on average 25 to 50 minutes.22,23 However, a recent report by Fisher et al23 raises concern that an ERCP longer than 30 minutes in duration may increase the risk of myocardial injury in patients 65 years or older. This 30-minute threshold would be exceeded by many, if not most, therapeutic ERCP procedures. It is unclear whether EUS procedures longer than 30 minutes carry the same risk. There is little in the literature that evaluates procedural duration as a risk for complications. We detected no increased risk of cardiopulmonary complications in our 51 patients who were 65 years or older. Further studies will be required to verify the ndings of Fisher et al23 and to determine how procedural duration is correlated with the risk of myocardial injury. Our report has limitations inherent in a retrospective analysis from a single center. In addition, it covers an experience over 40 months, with a heterogeneous group of patients under the care of 4 endoscopists. All 3 sampling techniques were not performed on each patient, nor was the use of EST or stent-type uniform. The literature on combined EUS-ERCP is limited, but case reports exist from other centers that detail complications.24,25 Nevertheless, our results show that the approach is feasible in patients with benign as well as malignant disease and carries no excess risk in experienced hands with appropriate anesthesia and cytopathologic support. With future advances in echoendoscope technology, we may see a linear EUS endoscope with a wide therapeutic channel that will enable both FNA sampling and biliary-pancreatic endotherapies, further simplifying the approach to these patients.26 In conclusion, we believe that an EUS with FNA and ERCP for biliary stent placement can be performed during a single session without compromising diagnostic accuracy or patient safety. Combining the 2 procedures expedited the patient evaluation, eliminated the need for a second endoscopy session, and reduced demand on endoscopic as well as anesthesia resources. DISCLOSURE The authors report that there are no disclosures relevant to this publication.

REFERENCES
1. Michl P, Pauls S, Gress TM. Evidence-based diagnosis and staging of pancreatic cancer. Best Pract Res Gastroenterol 2006;20:227-51. 2. Volmar KE, Vollmer RT, Jowell PS, et al. Pancreatic FNA in 1000 cases: a comparison of imaging modalities. Gastrointest Endosc 2005;61: 854-61. 3. DeWitt J, Deveraeaux B, Chriswell M, et al. Comparison of endoscopic ultrasonography and multidetector computed tomography for detecting and staging pancreatic cancer. Ann Intern Med 2004;141: 753-63. 4. Mansfield JC, Griffin SM, Wadchra V, et al. A prospective evaluation of cytology from biliary strictures. Gut 1997;40:671-7. 5. Jailwala J, Fogel EL, Sherman S, et al. Triple-tissue sampling at ERCP in malignant biliary obstruction. Gastrointest Endosc 2000;51:383-90. 6. Gress F, Michael H, Gelrud D, et al. EUS-guided fine-needle aspiration of the pancreas: evaluation of pancreatitis as a complication. Gastrointest Endosc 2002;56:864-7. 7. Agrawal B, Abu-Hamda E, Molke KL, et al. Endoscopic ultrasoundguided fine needle aspiration and multidetector spiral CT in the diagnosis of pancreatic cancer. Am J Gastroenterol 2004;99:844-50. 8. Tamm EP, Silverman PM, Charnsangavej C, et al. Diagnosis, staging, and surveillance of pancreatic cancer. AJR Am J Roentgenol 2003;180:1311-23. 9. Cotton PB, Lehman G, Vennes J, et al. Endoscopic sphincterotomy complications and their management: an attempt at consensus. Gastrointest Endosc 1991;37:383-93. 10. Eloubeidi MA, Gress FG, Savides TJ, et al. Acute pancreatitis after EUSguided FNA of solid pancreatic masses: a pooled analysis from EUS centers in the United States. Gastrointest Endosc 2004;60:385-9. 11. Gress FG, Gottlieb K, Sherman S, et al. Endoscopic ultrasonographyguided fine-needle aspiration biopsy of suspected pancreatic cancer. Ann Intern Med 2001;134:459-64. 12. American Society for Gastrointestinal Endoscopy, Adler DG, Jacobson BC, et al. ASGE guidelines: complications of EUS. Gastrointest Endosc 2005;61:8-12. 13. Eloubeidi MA, Tamhane A, Varadarajulu S, et al. Frequency of major complications after EUS-guided FNA of solid pancreatic masses: a prospective evaluation. Gastrointest Endosc 2006;63:622-9. 14. Freeman ML. Post-ERCP pancreatitis: patient and technique-related risk factors. JOP 2002;3:169-76. 15. Cheng C-L, Sherman S, Watkins J, et al. Risk factors for post-ERCP pancreatitis: a prospective multicenter study. Am J Gastroenterol 2006;101:139-47. 16. Vandervoort J, Soetikno RM, Tham TCK, et al. Risk factors for complications after performance of ERCP. Gastrointest Endosc 2002;56: 652-6. 17. Christensen M, Matzen P, Schulze S, et al. Complications of ERCP: a prospective study. Gastrointest Endosc 2004;60:721-31. 18. Raut CP, Grau AA, Staerkel GA, et al. Diagnostic accuracy of endoscopic ultrasound-guided fine-needle aspiration in patients with presumed pancreatic cancer. J Gastrointest Surg 2003;7:118-26. 19. Eloubeidi MA, Chen VK, Eltoum IA, et al. Endoscopic ultrasound guided fine needle aspiration biopsy of patients with suspected pancreatic cancer: diagnostic accuracy and acute and 30-day complications. Am J Gastroenterol 2003;98:2663-8. 20. Wiersema MJ, Vilmann P, Giovanni M, et al. Endosonography-guided fine-needle aspiration biopsy: diagnostic accuracy and complication assessment. Gastroenterology 1997;112:1087-95. 21. Mertz H, Gautam S. The learning curve for EUS-guided FNA of pancreatic cancer. Gastrointest Endosc 2004;59:33-7. 22. Fanti L, Agostoni M, Casati A, et al. Target-controlled propofol infusion during monitored anesthesia in patients undergoing ERCP. Gastrointest Endosc 2004;60:361-6. 23. Fisher L, Fisher A, Thomson A. Cardiopulmonary complications of ERCP in older patients. Gastrointest Endosc 2006;63:948-55.

www.giejournal.org

Volume 68, No. 3 : 2008 GASTROINTESTINAL ENDOSCOPY 465

Obstructive jaundice from presumed pancreatic malignancy 24. Mergener K, Jowell PS, Branch MS, et al. Pneumoperitoneum complication of ERCP performed immediately after EUS-guided fine needle aspiration. Gastrointest Endosc 1998;47:541-2. 25. Di Matteo F, Shimpi L, Gabbrielli A, et al. Same-day endoscopic retrograde cholangiopancreatography after transduodenal endoscopic ultrasound guided needle aspiration: do we need to be cautious? Endoscopy 2006;38:1149-51. 26. Rocca R, De Angelis C, Castellino F, et al. EUS diagnosis and simultaneous endoscopic retrograde cholangiography treatment of common bile duct stones by using an oblique-viewing echoendoscope. Gastrointest Endosc 2006;63:479-84.

Ross et al

Received October 30, 2006. Accepted November 12, 2007. Current affiliations: Departments of Gastroenterology, Hepatology and Nutrition (W.A.R., S.M.W., T.P., J.H.L.), Gastrointestinal Medical Oncology (R.A.W.), Surgical Oncology (D.B.E.), Anesthesiology (L.V.T.), and Pathology (G.A.S.), The University of Texas M.D. Anderson Cancer Center, Houston, Texas, USA. Reprint requests: William A. Ross, MD, Department of Gastroenterology, Hepatology and Nutrition, MD Anderson Cancer Center, 1515 Holcombe Blvd, Unit 436, Houston, TX 77030.

466 GASTROINTESTINAL ENDOSCOPY Volume 68, No. 3 : 2008

www.giejournal.org