Intraductal papillary mucinous neoplasm (IPMN) is a distinct

CLINICAL GASTROENTEROLOGY AND HEPATOLOGY 2008;6:815 819 Evaluation of the Guidelines for Management of Pancreatic Branch-Duct Intraductal Papillary Mucinous Neoplasm RAYMOND S. TANG,* BENJAMIN WEINBERG, DAVID W. DAWSON, HOWARD REBER, OSCAR J. HINES, JAMES S. TOMLINSON, VINIKA CHAUDHARI, STEVEN RAMAN, and JAMES J. FARRELL *Department of Medicine, Division of Digestive Diseases, Department of Pathology and Laboratory Medicine, Department of Surgery, and Department of Radiological Sciences, UCLA David Geffen School of Medicine, Los Angeles, California See CME exam on page 719. See editorial on page 724. Background & Aims: The 2006 Sendai Consensus Guidelines recommend surgical resection for all suspected branch-duct intraductal papillary mucinous neoplasm (BD- IPMN) greater than 3 cm irrespective of symptoms, and those less than 3 cm with worrisome features. We aimed to evaluate the surgical characteristics of these guidelines retrospectively in pathologically confirmed cases of BD-IPMN. Methods: IPMNs resected at our institution (1995 2006) were classified as main-duct predominant or branch-duct (BD) predominant based on preoperative imaging and postoperative histology. Resected BD-IPMNs were classified histologically: low risk (adenoma, borderline) and high risk (carcinoma in situ or invasive cancer). Clinical data (presence of symptoms, mural nodule, dilated pancreatic duct, and cyst size) were correlated with pathology. Results: Between 1995 and 2006, there were 204 patients who underwent surgical resection of pancreatic cysts. Sixtyone patients had IPMN including 31 with BD-IPMN. A total of 74.2% (23 of 31) of BD-IPMNs would have been recommended for surgical resection including 69.2% (18 of 26) of low-risk lesions and 100% (5 of 5) of high-risk lesions. All 8 cases of BD-IPMN that would have been recommended for nonsurgical management were low-risk lesions. The positive predictive value of the guidelines is 21.7% (95% confidence interval, 9.7% 41.9%). The negative predictive value is 100% (95% confidence interval, 67.6% 100.0%). Between 2000 and 2007, 351 patients with likely BD-IPMN were evaluated but not resected. Conclusions: Implementation of the Consensus Guidelines to our singleinstitution, referral-based, surgical BD-IPMN population would have recommended resection of all histologically high-risk lesions. All lesions recommended for nonsurgical management were histologically low-risk lesions. For presumed BD-IPMNs less than 3 cm, the application of the Consensus Guidelines may reduce the resection rate for low-risk lesions. Intraductal papillary mucinous neoplasm (IPMN) is a distinct pancreatic cystic neoplasm that has received greater recognition in recent years. 1 Depending on their predominant pancreatic duct of origin, IPMNs are categorized as main duct (MD), branched duct (BD), or combined subtypes based on imaging studies and/or pathologic examination. 2 MD-IPMNs have a high risk of malignant degeneration, with prevalence of cancer ranging from 57% to 92%, and should undergo surgical resection. 3 10 On the other hand, BD-IPMNs appear to carry a much smaller risk of malignancy, with prevalence of cancer ranging from 6% to 46%. 3 10 Lévy et al 11 studied the natural history of IPMNs of 106 patients with highly probable or histologically proven IPMNs, including 53 BD-IPMNs, with a median follow-up period of 21 months, and estimated the 5-year actuarial risk for a BD-IPMN progressing to high-grade dysplasia or malignancy to be 15%. Compared with MD-IPMN, the management of BD-IPMN has been the subject of some controversy and conservative therapy with surveillance has been recommended for some of these lesions. 1 Attempts to diagnose and risk-stratify BD-IPMN with preoperative imaging is difficult. 12 Computed tomography (CT), magnetic resonance imaging (MRI), and endoscopic ultrasound (EUS) imaging features suggestive of BD-IPMN include isolated collection of pancreatic cysts in the setting of normal pancreatic duct, ductal communication, and occasionally mural nodules or masses associated with BD-IPMNs that may be suggestive of malignancy. 13 Cyst fluid analysis, especially when the carcinoembryonic antigen level is increased, also is suggestive of a mucinous process but is not useful in predicting the malignant nature or existence of BD-IPMN. 14 In 2006, the working group of the International Association of Pancreatology proposed Consensus Guidelines for the management of BD-IPMN (Sendai Consensus Guidelines). The Sendai guidelines recommend surgical resection for all BD-IPMNs greater than 3 cm irrespective of symptoms, and for all BD- IPMNs less than 3 cm with any one of several worrisome features including the presence of cyst-related symptoms, mural nodule, or a dilated main pancreatic duct greater than 6 mm. 15 The aim of this study was to evaluate the surgical characteristics of these guidelines retrospectively in radiologically and pathologically confirmed cases of BD-IPMN at our institution. Abbreviations used in this paper: BD, branch duct; CIS, carcinoma in situ; CT, computed tomography; EUS, endoscopic ultrasound; IPMN, intraductal papillary mucinous neoplasm; MD, main duct; MRCP, magnetic resonance cholangiopancreatography; MRI, magnetic resonance imaging. 2008 by the AGA Institute 1542-3565/08/$34.00 doi:10.1016/j.cgh.2008.04.005

816 TANG ET AL CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 6, No. 7 Table 1. Demographics and Clinical Features of the IPMN Patients MD-IPMN (n 30) BD-IPMN (n 31) Demographics Mean age, y ( SD) 66.6 9.5 66.5 10.1 Sex, male:female 19:11 10:21 Clinical feature Symptoms from cyst 20/30 (66.7%) 18/31 (58.1%) Patients and Methods Cases of surgically resected pancreatic cysts from the surgical pathology database and the radiologic database at UCLA Medical Center from January 1995 to October 2006 were evaluated retrospectively. Patients clinical, radiologic, and pathologic data were collected by reviewing the computerized medical records and clinic charts. Assessment of the patients preoperative pancreatic cyst-related symptoms was made based on review of the preoperative surgical assessment. A single pathologist (D.W.D.) reviewed all pancreatic cyst surgical resections. The pathologic diagnosis of IPMN was made according to established criteria by the World Health Organization. 16 Histology was used as the gold standard for the final diagnosis of IPMN in these patients. IPMNs were identified and classified as main-branch predominant or BD-predominant based on preoperative imaging and postoperative histology. Main-branch IPMN encompasses both MD involvement and combined-duct involvement, whereas the BD-predominant subtype excludes MD involvement. Resected BD-IPMNs were graded pathologically as adenoma, borderline, carcinoma in situ (CIS), or invasive cancer. For the purpose of this study, BD-IPMN was considered low risk if it showed adenoma or borderline histology, whereas the presence of CIS or invasive cancer was considered a high-risk category. Mucinous cystic neoplasms were excluded on the basis of histology (the presence of ovarian-type stroma). Pathology of the resected pancreatic cysts was correlated with the patient s clinical presentation and preoperative imaging such as CT, MRI, magnetic resonance cholangiopancreatography (MRCP), endoscopic retrograde cholangiopancreatography, or EUS. We then retrospectively applied the Sendai Consensus Guidelines published in 2006 to our histologically confirmed cases of BD-IPMN. 15 All BD-IPMNs greater than 3 cm and those BD-IPMNs less than 3 cm that showed any worrisome clinical features (cyst-related symptoms based on preoperative clinical assessment, presence of mural nodule, or dilated main pancreatic duct 6 mm by MRI, CT, or EUS) were considered lesions indicating surgical resection. 15 BD-IPMNs not fulfilling the earlier-described criteria were considered lesions suitable for nonsurgical management. 15 The positive predictive value of the Consensus Guidelines, defined as the number of histologically high-risk lesions among lesions indicating resection, and the negative predictive value, defined as the number of histologically low-risk lesions suitable for nonsurgical management, then were calculated. To identify patients with possible BD-IPMN who did not undergo surgical resection during the study period, we initially selected all patients who were evaluated for a pancreatic cyst (based on International Classification of Diseases-9th revision billing codes) at UCLA Medical Center. After removing all patients who underwent surgery within 1 year of their initial presentation, we then made a presumptive diagnosis of BD- IPMN based on a consensus review of medical, radiology, endoscopic, and pathology records. Criteria favoring a diagnosis of likely BD-IPMN included the following: absence of a history of pancreatitis, isolated or multifocal pancreatic cysts, nondilated pancreatic duct, absence of imaging features compatible with pancreatic pseudocyst or serous cystadenoma, and pancreatic cyst fluid analysis compatible with BD-IPMN. We then sought follow-up evaluation on the nonsurgical group. This study was performed with approval by the UCLA Institutional Review Board. Results Between January 1995 and October 2006, there were 204 pancreatic cyst resections performed at our institution. Of the 204 pancreatic cyst patients undergoing resection, 61 patients had IPMN, including 31 patients with BD-IPMN, the majority since 2001. The demographics and clinical features of the IPMN patients are summarized in Table 1. The mean age of all patients with MD-IPMN was 66.6 9.5 years. Sixty-three percent of MD- IPMN patients were men. Cyst-related symptoms were present in 66.7% of the MD-IPMN patients. The BD-IPMN patients had a mean age of 66.5 10.1 years. Thirty-two percent of BD- IPMN patients were men. A total of 58.1% of the BD-IPMN patients had cyst-related symptoms. The mean age for patients with BD-IPMN adenoma, BD-IPMN borderline, BD-IPMN CIS, and BD-IPMN invasive cancer were 65, 66.7, 65.7, and 79.5 years, respectively. Of note, 20 BD-IPMN patients underwent EUS as part of the preoperative evaluations. Among the 31 cases of resected BD-IPMN, there were 17 adenomas, 9 borderlines, 3 CIS, and 2 invasive cancers. The pathology and clinical-pathologic and radiographic features of the BD-IPMN patients are summarized in Table 2. Twelve BD-IPMNs were greater than 3 cm in size, and 19 were less than Table 2. BD-IPMN: Pathology and Clinical-Radiologic Features Actual number of resected BD-IPMNs by histology Adenoma (n 17) Borderline (n 9) CIS (n 3) Invasive cancer (n 2) Cyst 3 cm(n 12) 4 4 2 2 Cyst 3 cm(n 19) 13 5 1 0 Cyst 3 cm with 1 of the 3 worrisome features a (n 11) 6 4 1 NA Per guidelines Proportion of cysts 3 cm that would have been resected 6/13 (46.2%) 4/5 (80%) 1/1 (100%) NA Proportion of cysts (all sizes) that would have been resected 10/17 (58.8%) 8/9 (88.9%) 3/3 (100%) 2/2 (100%) NA, not applicable. a Worrisome features: symptoms from cyst, presence of mural nodule, or pancreatic duct dilation greater than 6 mm.

July 2008 BD IPMN AND SENDAI CONSENSUS GUIDELINES 817 these patients (median follow-up period, 26 mo; range, 1 82 mo). None of this group underwent pancreatic cyst surgical resection or appeared to have developed invasive pancreatic cancer on follow-up evaluation. Figure 1. Flow diagram of diagnostic accuracy of Sendai Consensus Guidelines. Abnormal result: cyst greater than 3 cm or cyst less than 3 cm with worrisome features; high-risk lesion: CIS, invasive cancer. TP, true positive; FP, false positive; FN, false negative; TN, true negative; PPV, positive predictive value; NPV, negative predictive value. 3 cm in size. Applying the Sendai Consensus Guidelines to our patients, 74.2% (23 of 31) BD-IPMNs would have been recommended for surgical resection. For BD-IPMNs greater than 3 cm with any one of the worrisome features, 46.2% (6 of 13) adenomas, 80% (4 of 5) borderlines, and 100% (1 of 1) CIS would have been recommended for resection per Sendai Consensus Guidelines. The 2 patients with BD-IPMN showing invasive cancer had lesions greater than 3 cm. For both BD-IPMNs greater than 3 cm and BD-IPMNs less than 3 cm with any one of the worrisome features, 58.8% (10 of 17) adenomas, 88.9% (8 of 9) borderlines, 100% (3 of 3) CIS, and 100% (2 of 2) invasive cancers would have been resected per the guidelines. Hence, implementation of the Sendai Consensus Guidelines would have led to resection of 69.2% (18 of 26) of histologically low-risk lesions (adenoma, borderline) and 100% (5 of 5) of histologically high-risk lesions (CIS, invasive cancer). All 8 cases of BD-IPMN that would have been recommended for nonsurgical management based on these guidelines were histologically low-risk lesions, with the majority being adenomas (Figure 1). The positive predictive value of the Consensus Guidelines (number of histologically high-risk lesions among the lesions recommended for resection) was 21.7% (95% confidence interval, 9.7% 41.9%). The negative predictive value (number of histologically low-risk lesions among the lesions not recommended for resection) was 100% (95% confidence interval, 67.6% 100.0%) (Figure 1). During the period from 2000 to 2007, there were 758 patients evaluated at our institution for pancreatic cyst based on International Classification of Diseases-9th revision billing codes. A total of 140 patients (including 29 with BD-IPMN) underwent surgical resection during this period, leaving 618 patients in the nonsurgical group. Of these 618 patients in the nonsurgical group, further review of records showed that 54 did not have evidence of pancreatic cysts based on a review of clinical records or radiologic studies during this time period. Of the remaining 564 patients, 351 were classified as likely to have BD-IPMN. The mean size of the pancreatic cyst in this group was 14 mm (range, 4 45 mm), and there was no focal mass seen by noninvasive imaging. Long-term follow-up evaluation at our institution after initial presentation was available only on 96 of Discussion Between 1995 and 2006, there were 61 IPMN patients including 31 cases of BD-IPMN identified in 204 patients who underwent surgical resection of pancreatic cysts at our institution. Application of the Sendai Consensus Guidelines to our retrospective group of BD-IPMN patients would have led to surgical resection of 74.2% (23 of 31) of BD-IPMNs, including 69.2% (18 of 26) of the histologically confirmed low-risk lesions (adenoma, borderline) and 100% (5 of 5) of the histologically confirmed high-risk lesions (CIS, invasive cancer). Retrospective implementation of the guidelines would have captured all of the histologically high-risk lesions (CIS and invasive cancer) for surgical resection. All the cases of BD-IPMN that would have been recommended for nonsurgical management were histologically low-risk lesions. The negative predictive value based on the number of histologically confirmed low-risk lesions that would have been recommended for conservative management is 100%. Furthermore, a natural history study by Lévy et al estimated the 5-year actuarial risk for BD-IPMNs progressing to high-grade dysplasia to be 15%, which was significantly lower than that observed for MD-IPMN. 11 Multiple case series also have reported similar findings. 3 10 Our study suggests that the Sendai Consensus Guidelines are highly predictive of BD- IPMNs that could be managed by surveillance based on the low likelihood of CIS or invasive cancer. In addition, our study suggests that the guidelines also would be able to capture the higher-risk BD-IPMNs that contain invasive cancer or CIS. In our study, using size alone ( 3 cm) as the criterion for surgical resection would increase the positive predictive value to 33.3%, resulting in proportionally more high-risk lesions being recommended for surgical resection, while decreasing the negative predictive value to 96.3% by recommending a high-risk lesion (CIS) for surveillance. The strength of our investigation includes pathologic evaluation and classification of all the resected pancreatic cysts at a single institution, and limiting analysis to BD-IPMN. Because we believe we can identify BD-IPMN on preoperative imaging studies, these guidelines are applicable to clinical practice. However, this study also had several limitations: the retrospective evaluation of data, analysis of only patients who underwent surgery and not all patients presenting with BD-IPMN, difficulty in truly differentiating the MD- and BD-IPMN (mixed IPMN has been described and used for classification of lesions in some IPMN studies). Thus, it often is quite difficult to compare our results with other studies that used all 3 classifications of IPMN. IPMN can be categorized by imaging studies preoperatively, or by gross and histologic evaluation postoperatively, into MDor BD-predominant. 1 Differentiating BD-IPMN from the MD subtype is important in the management because careful surveillance, instead of surgical resection, may be appropriate for BD-IPMNs with a lower risk of malignancy. 3 10 In our study, all cases of BD-IPMN were confirmed by pathologic examinations using the World Health Organization criteria. 16 The finding of multiple pancreatic cysts communicating with the main pancreatic duct without main duct dilatation on various imaging

818 TANG ET AL CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 6, No. 7 studies (eg, CT, MRI, MRCP, endoscopic retrograde cholangiopancreatography, and EUS) was used to identify BD-IPMN radiographically and was consistent with criteria used in a prior radiographic study of the natural history of BD-IPMN, in a study of radiographic predictors of invasive cancer in BD- IPMN, and in the prospective surgical series of BD-IPMN by Salvia et al. 17 19 In the prospective study of BD-IPMN by Salvia et al, 19 the presence of main pancreatic ductal dilatation greater than 5, 4, and 3 mm in the pancreatic head, body, or tail, respectively, would exclude the diagnosis of BD-IPMN. Of note, 20 of our 31 BD-IPMN patients underwent EUS fine-needle aspiration in conjunction with other preoperative imaging studies whereas Salvia et al used only EUS to diagnose BD- IPMN when there was no clear communication between the cystic lesion and the main pancreatic duct on MRCP or a suspicion of main pancreatic duct dilatation. 19 EUS was reported by Kubo et al 20 to have an accuracy of 86% in differentiating between malignant and benign IPMN. Diagnosing and risk-stratifying BD-IPMN with only CT and MRI imaging are often difficult; the addition of EUS fine-needle aspiration may provide additional morphologic and cyst fluid information in the preoperative evaluation. 21 Because IPMNs are being diagnosed more often, increasing attention has been devoted to identifying radiographic or clinical predictors of malignancy. 22 MD-IPMNs are reported to have a high prevalence of cancer, ranging from 57% to 92%. 3 10 In 2004, Salvia et al 23 published a study of 140 patients from the United States and Italy with MD-predominant IPMN and found that older age, presence of jaundice, and new-onset diabetes predicted a higher risk of malignancy. These lines of evidence lead to the general belief that all MD-IPMNs should undergo resection. The management of the BD-IPMN is more controversial. Before publication of the Sendai Consensus Guidelines in 2006, different research groups attempted to identify markers for malignant degeneration in BD-IPMN. In 2003, Matsumoto et al 6 studied 57 IPMN patients retrospectively and reported an absence of malignancy in BD-IPMNs with sizes less than 3 cm and without mural nodules, indicating the possibility of nonsurgical management for BD-IPMNs fulfilling these criteria. In the same year, Sugiyama et al 10 conducted a study on 62 IPMN patients who underwent surgery, including 32 BD-IPMNs, and found that BD-IPMNs with a size greater than 3 cm and mural nodules, as well as main pancreatic duct diameter greater than 7 mm, are predictive of malignancy. 10 These findings were incorporated into the Sendai Consensus Guidelines. In a prospective study of 109 BD-IPMN patients between 2000 and 2003, Salvia et al 19 used a different set of worrisome features for malignant degeneration (presence of symptoms, cysts size 3.5 cm, mural nodules, thick walls, carbohydrate antigen 19.9 level 25 U/L, recent-onset, or worsened diabetes). BD-IPMN patients with the earlier-described worrisome features were subject to surgical resection. 19 Abdominal ultrasound and MRCP with secretin, with the addition of EUS if the former 2 imaging tests failed to diagnose BD-IPMN, were used for preoperative evaluation. 19 Of note, 5 of the 89 conservatively managed patients after a median follow-up period of 32 months had an increase in lesion size, leading to surgical resection. 19 Despite being sent for surgery, those 5 patients were later found to have low-risk lesions (3 adenomas and 2 borderlines). 19 Although fewer predictors of malignant degeneration were used in the Sendai Guidelines when compared with the protocol by Salvia et al, 19 all the BD-IPMNs recommended for nonsurgical management in our study per the Sendai Guidelines were histologically low-risk lesions (adenoma and borderline), with a negative predictive value of 100% (95% confidence interval, 67.6% 100.0%). Our study represents a single-institution attempt to evaluate the surgical characteristics of the Sendai Consensus Guidelines in pathologically reviewed and confirmed BD-IPMN. Other research groups have tried to apply parts of the Sendai Guidelines to a collection of different pancreatic cysts, but their conclusions have been mixed. In conclusion, retrospective implementation of the Sendai Consensus Guidelines to our singleinstitution, referral-based BD-IPMN patient population would have recommended resection of all the presumed histologically high-risk lesions. All the lesions recommended for nonsurgical management were confirmed histologically as low-risk lesions. For presumed BD-IPMNs less than 3 cm, application of the Consensus Guidelines may reduce resection for histologically low-risk lesions. References 1. Tanaka M, Kobayashi K, Mizumoto K, et al. Clinical aspects of intraductal papillary mucinous neoplasm of the pancreas. J Gastroenterol 2005;40:669 675. 2. D Angelica M, Brennan MF, Suriawinata AA, et al. Intraductal papillary mucinous neoplasms of the pancreas: an analysis of clinicopathologic features and outcome. Ann Surg 2004;239: 400 408. 3. Terris B, Ponsot P, Paye F, et al. Intraductal papillary mucinous tumors of the pancreas confined to secondary ducts show less aggressive pathologic features as compared with those involving the main pancreatic duct. Am J Surg Pathol 2000;24:1372 1377. 4. Kobari M, Egawa S, Shibuya K, et al. Intraductal papillary mucinous tumors of the pancreas comprise 2 clinical subtypes: differences in clinical characteristics and surgical management. Arch Surg 1999;134:1131 1136. 5. Doi R, Fujimoto K, Wada M, et al. Surgical management of intraductal papillary mucinous tumor of the pancreas. Surgery 2002;132:80 85. 6. Matsumoto T, Aramaki M, Yada K, et al. Optimal management of the branch duct type intraductal papillary mucinous neoplasms of the pancreas. J Clin Gastroenterol 2003;36:261 265. 7. Sohn TA, Yeo CJ, Cameron JL, et al. Intraductal papillary mucinous neoplasms of the pancreas: an updated experience. Ann Surg 2004;239:788 797. 8. Choi BS, Kim TK, Kim AY, et al. Differential diagnosis of benign and malignant intraductal papillary mucinous tumors of the pancreas: MR cholangiopancreatography and MR angiography. Korean J Radiol 2003;4:157 162. 9. Kitagawa Y, Unger TA, Taylor S, et al. Mucus is a predictor of better prognosis and survival in patients with intraductal papillary mucinous tumor of the pancreas. J Gastrointest Surg 2003;7: 12 19. 10. Sugiyama M, Izumisato Y, Abe N, et al. Predictive factors for malignancy in intraductal papillary-mucinous tumours of the pancreas. Br J Surg 2003;90:1244 1249. 11. Lévy P, Jouannaud V, O Toole D, et al. Natural history of intraductal papillary mucinous tumors of the pancreas: actuarial risk of malignancy. Clin Gastroenterol Hepatol 2006;4:460 468. 12. Spinelli KS, Fromwiller TE, Daniel RA, et al. Cystic pancreatic neoplasms. Observe or operate. Ann Surg 2004;239:651 659. 13. Kalra MK, Maher MM, Mueller PR, et al. State-of-the-art imaging of pancreatic neoplasms. Br J Radiol 2003;76:857 865. 14. Brugge WR, Lewandrowski K, Lee-Lewandrowski E, et al. Diagno-

July 2008 BD IPMN AND SENDAI CONSENSUS GUIDELINES 819 sis of pancreatic cystic neoplasms: a report of the cooperative pancreatic cyst study. Gastroenterology 2004;126:1330 1336. 15. Tanaka M, Chari S, Adsay V, et al. International consensus guidelines for management of intraductal papillary mucinous neoplasms and mucinous cystic neoplasms of the pancreas. Pancreatology 2006;6:17 32. 16. Kloppel G, Luttges J. WHO-classification 2000: exocrine pancreatic tumors. Verh Dtsch Ges Pathol 2001;85:219 228. 17. Irie H, Yoshimitsu K, Aibe H, et al. Natural history of pancreatic intraductal papillary mucinous tumor of branch duct type: follow-up study by magnetic resonance cholangiopancreatography. J Comput Assist Tomogr 2004;28:117 122. 18. Kawamoto S, Lawler LP, Horton KM, et al. MDCT of intraductal papillary mucinous neoplasm of the pancreas: evaluation of features predictive of invasive carcinoma. AJR Am J Roentgenol 2006;186:687 695. 19. Salvia R, Crippa S, Falconi M, et al. Branch-duct intraductal papillary mucinous neoplasms of the pancreas: to operate or not to operate? Gut 2007;56:1086 1090. 20. Kubo H, Chijiiwa Y, Akahoshi K, et al. Intraductal papillary-mucinous tumors of the pancreas: differential diagnosis between benign and malignant tumors by endoscopic ultrasonography. Am J Gastroenterol 2001;96:1429 1434. 21. Brugge W. Evaluation of pancreatic cystic lesions with EUS. Gastrointest Endosc 2004;59:698 707. 22. Fernandez-del CC, Targarona J, Thayer SP, et al. Incidental pancreatic cysts: clinicopathologic characteristics and comparison with symptomatic patients. Arch Surg 2003;138:427 434. 23. Salvia R, Fernández-del Castillo C, Bassi C, et al. Main-duct intraductal papillary mucinous neoplasms of the pancreas: clinical predictors of malignancy and long-term survival following resection. Ann Surg 2004;239:678 685. Address requests for reprints to: James J. Farrell, MD, Division of Digestive Diseases, UCLA David Geffen School of Medicine, Box 951693, 200 Med Plaza, Suite 365A, Los Angeles, California 90095. e-mail: jfarrell@mednet.ucla.edu; tel: (310) 267-4664; fax: (310) 794-9718.