Most solid pancreatic tumors are reported to be ductal. Improved Differentiation of Pancreatic Tumors Using Contrast-Enhanced Endoscopic Ultrasound

Transcription

1 CLINICAL GASTROENTEROLOGY AND HEPATOLOGY 2008;6: Improved Differentiation of Pancreatic Tumors Using Contrast-Enhanced Endoscopic Ultrasound CHRISTOPH F. DIETRICH,* ANDRE IGNEE,* BARBARA BRADEN,, ANA PAULA BARREIROS,* MICHAELA OTT, and MICHAEL HOCKE *Medical Department II, Department of Pathology, Caritas Krankenhaus, Bad Mergentheim, Germany; Medical Department I, University Hospital Frankfurt, Frankfurt am Main, Germany; Medical Department, John Radcliffe Hospital, Oxford, United Kingdom; and the Department of Internal Medicine II, Friedrich Schiller University Jena, Jena, Germany See Strate T et al on page 1406 for companion article in the May 2008 issue of Gastroenterology. Background & Aims: Endoscopic ultrasound is a widely accepted imaging method for staging of ductal adenocarcinoma and the localization of neuroendocrine tumors of the pancreas. We prospectively evaluated conventional color Doppler imaging and contrast-enhanced endoscopic Doppler ultrasound (CE-EDUS) as a new imaging technique for further characterization and differentiation of solid pancreatic tumors. Methods: From 300 patients with pancreatic lesions investigated using contrast-enhanced endoscopic ultrasound we could finally include 93 patients with an undetermined, solitary, predominantly solid, lesion 40 mm or less, and a definite histologically proven diagnosis. After bolus injection of the contrast agent SHU 508A 4 g (400 mg/dl) the vascular pattern of the lesion during the arterial phase was compared with the vascularity of the residual pancreatic parenchyma. Results: Color Doppler imaging did not reveal vascularity of the pancreatic parenchyma in any of the patients, and therefore tumor hypovascularity could not be determined in contrast to all CE-EDUS examined patients revealing at least some degree of parenchymal vascularity. Fifty-seven of 62 patients with ductal adenocarcinoma of the pancreas showed a hypovascularity of the tumor using CE-EDUS. All other pancreatic lesions revealed an isovascular or hypervascular pattern using contrastenhanced endoscopic ultrasound (20 neuroendocrine tumors, 10 serous microcystic adenomas, and 1 teratoma). Hypovascularity as a sign of malignancy in contrast-enhanced endoscopic ultrasound obtained 92% (82% 97%) sensitivity and 100% specificity (89% 100%). Conclusions: Contrast-enhanced endoscopic ultrasound is effective in differentiating small solid pancreatic tumors of different origin in most cases. Hypovascularity indicates malignancy of pancreatic tumors. Most solid pancreatic tumors are reported to be ductal adenocarcinoma. 1,2 Pancreatic lesions other than ductal adenocarcinoma of the pancreas summarize a heterogeneous group of tumors with respect to the clinical presentation, biologic behavior, prognosis, and imaging features. Incidental pancreatic tumors are found in autopsy series of unselected cases in up to 1.5%. 3 8 It is important to distinguish nonductal adenocarcinoma from ductal adenocarcinoma, which are the much more commonly reported neoplasms of the exocrine pancreas because, for example, most neuroendocrine tumors are slowly progressive and imply a different therapeutic approach than ductal adenocarcinoma. Serous microcystic adenomas of the pancreas are rarely malignant and, therefore, do not need surgery in general. Other forms of solid tumors are rare. Differential diagnosis was reported to be difficult To date, there has been no ideal imaging method for the differentiation of pancreatic tumors. The vascularity pattern of pancreatic tumors has been examined in few studies mainly by using the transabdominal approach, but study population, examination technique, and tumor characteristics were inconsistent For instance, pancreatic metastases usually are not clinically challenging because they correspond to the underlying disease and often present as multiple pancreatic nodules. Primary lymphoma of the pancreas is a very rare disease. Ductal adenocarcinomas are rarely cystic and the most cystic lesions behave biologically different from adenocarcinomas of the pancreas. Endoscopic ultrasound is the imaging procedure with the highest detail resolution and was reported to exclude small pancreatic carcinoma in almost all patients without chronic pancreatitis. 25,26 Endoscopic ultrasound also is widely accepted for staging purposes in patients with ductal adenocarcinoma of the pancreas to determine vascular infiltration It is also of great value for localizing neuroendocrine pancreatic tumors. 31 Endoscopic conventional B-mode ultrasound in combination with contrast-enhanced endoscopic Doppler ultrasound (CE- EDUS) reveals not only the grade of vascularization of a pancreatic lesion but also parenchymal criteria and ductal criteria and therefore might be helpful in the differentiation of solid pancreatic tumors. The aim of this study was to analyze the vascularity pattern of solitary solid pancreatic tumors in comparison with the surrounding parenchyma using contrast-enhanced endoscopic ultrasound. Abbreviations used in this paper: CDI, color Doppler imaging; CE- EDUS, contrast-enhanced endoscopic Doppler ultrasound by the AGA Institute /08/$34.00 doi: /j.cgh

2 May 2008 IMPROVED DIFFERENTIATION OF PANCREATIC TUMORS 591 Methods Patients From 1999 to 2006 we prospectively examined more than 300 patients with pancreatic lesions using CE-EDUS. Institutional Board approval and oral informed consent according to the ethical guidelines from Helsinki were obtained from all patients after informing the patient about the purpose and aim of the study before the endoscopic ultrasound examination was started. Inclusion Criteria for Study Analysis Inclusion criteria for study analysis were an undetermined, solitary, predominantly solid, lesion 40 mm or less with a definite histologically proven diagnosis. The lesions were detected by transabdominal ultrasound, computerized tomography, or magnetic resonance imaging. Exclusion Criteria Metastases. Ductal adenocarcinoma of the pancreas with metastases to the liver, lung, or any other organ were not examined by endoscopic ultrasound because metastases were determining the clinical work-up (eg, liver biopsy) and prognosis of the diseases. Tumor size. The endoscopic ultrasound range of high-quality images is restricted to about 50 mm. Therefore, we excluded lesions from study analysis with a size greater than 40 mm, implying that the outer border of a pancreatic lesion will be in the range of 50 mm to the surrounding stomach or duodenum at least at one site of the probe. The measurement was performed by transabdominal ultrasound or computerized tomography and endoscopic ultrasound, the latter was decisive in patients with discrepancy between findings. Solitary lesions. Only patients with solitary lesions were included in the study. Patients with multiple lesions were excluded from the study because the latter are mostly secondary (metastases or lymphoma) or caused by functional (hormoneproducing) neuroendocrine tumors. Predominantly solid lesions. Predominantly solid lesions were defined as presenting cysts that were less than 20 mm or cystic area(s) less than 50% of the total tumor volume. Chronic pancreatitis. Patients with chronic pancreatitis were excluded from the study because the contrast-enhancement pattern was compared with the surrounding pancreatic parenchyma. Chronic pancreatitis was assumed if any imaging modality revealed inflammatory changes of the surrounding pancreatic parenchyma. Two patients suspected to have ductal adenocarcinoma of the pancreas by computerized tomography revealed a thrombosed aneurysm of the gastroduodenal artery (28 mm) and splenic artery (25 mm), both not histologically proven and therefore excluded from study analysis as well. Autoimmune pancreatitis. Patients with a definite diagnosis of autoimmune pancreatitis and lesion size greater than 40 mm were excluded from study analysis. Neuroendocrine tumors with hormone-derived syndromes. Neuroendocrine tumors with hormone-derived syndromes (eg, insulinoma, glucagonoma, and gastrinoma) were excluded from study analysis because pancreatic tumors in patients with a clinically defined syndrome do not present problems of differential diagnosis of pancreatic tumor lesions. Pancreatic metastases. Patients with a final diagnosis of pancreatic metastases without clinical relevance for the course of the disease were excluded from study analysis. Other reasons for exclusion. Possible other reasons for exclusion have been former allergic reactions to Levovist (Schering, Berlin, Germany) (n 0), no patient consent (n 12), as well as concomitant unsuspected stenosis in the upper gastrointestinal tract (n 5) leading to inconclusive endoscopic ultrasound results. Inadequate visualization as a result of other reasons was not observed. Data from patients examined by contrast-enhanced endoscopic ultrasound but not included in the study analysis were documented as well. During the study period we also examined intraductal papillary mucinous neoplasia (n 6), mucinous cystadenoma/cystadenocarcinoma (n 19), solid pseudopapillary neoplasia (n 1), acinus cell carcinoma (n 2), and pseudocysts (n 90). Methods Endoscopic ultrasound was performed using an electronic linear ultrasound probe (Pentax FG 34 or Pentax FG 38 UX; [Pentax, Hamburg, Germany]; Hitachi EUB 525, Hitachi 6500, or Hitachi 8500 [Hitachi, Wiesbaden, Germany]). The pancreas was evaluated as recently described. 32 A contrast-enhancing agent (Levovist) was applied in all patients (4 g, 400 mg/dl, bolus injection) via a catheter of 1.2-mm diameter or larger into a cubital vein. 33 Levovist consists of air and a galactose palmitate shell for stabilization, which was reported to enhance Doppler signals approximately 20 to 30 db. 34 During the procedure at least 3 examiners were present. Monitoring was performed by one physician as recently described including oxygen saturation using pulse oxymetry, heart frequency, and blood pressure; side effects were recorded. 35,36 All examinations were performed by one examiner (C.F.D.) who was not aware of the histologic diagnosis. Examinations have been evaluated additionally by the 2 more (actually present) examiners and accordance was achieved during the procedure. In 5 patients repositioning of the echoendoscope during the enhancement phase was necessary because of patient movements leading to inadequate visualization and therefore Levovist was re-injected with adequate visualization in all 5 patients. Conventional B-Mode and Color Doppler Scanning After the detection of the lesion by endoscopic ultrasound the size, tumor location (pancreatic head, corpus, tail), echogenicity, morphology, and border delineation were documented routinely. By analyzing the border delineation important discrepancies were found between the examiners in more than 30% of patients (mainly in the larger lesions). Therefore, border delineation was not included in the study analysis. Contrast-Enhanced Color Doppler Scanning Machine settings. The penetration depth and focus were adjusted to the tumor location and always equal or below 50 mm, about two thirds of depth penetration. The pulse repetition frequency and wall filters were adjusted individually to enable the detection of parenchymal and simultaneously intratumoral vessels and to avoid artifacts (eg, blooming). The gain was adjusted as low as possible to get optimal signals.

3 592 DIETRICH ET AL CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 6, No. 5 Table 1. Examined Patients With Ductal Adenocarcinomas, Neuroendocrine Tumors, and Serous Microcystic Pancreatic Adenomas to Show Inclusion and Exclusion Criteria Examined patients to show exclusion criteria Included in study analysis Ductal adenocarcinoma (liver metastases with or without chronic pancreatitis) 52 (size 40 mm with or without chronic pancreatitis) Neuroendocrine tumors (liver metastases) 25 (hormone-producing syndromes) 5 (size 40 mm) Serous microcystic pancreatic adenoma (size 40 mm) Teratoma 1 1 NOTE. Inclusion criteria for study evaluation: mainly solid, tumor size 40 mm, and solitary nodule. Exclusion criteria were as follows: tumor size 40 mm, multiple lesions, hormone-producing syndromes, and metastases. In contrast to the first-generation microbubble echo-enhancers with air bubbles developed for Doppler-enhancing purposes (eg, Levovist), second-generation microbubble echo-enhancers contain gas and are used mainly with resonating low mechanical techniques. Second-generation microbubble echo-enhancers were available only recently. It was of interest to compare air-derived Levovist with newer second-generation contrast agents containing gas with respect to Doppler-enhancing properties. By comparing SonoVue (Bracco, Milan, Italy) and Levovist in 10 consecutive patients we showed similar results in the imaging pancreatic vascularity pattern using Levovist 4 g and 10 ml (400 mg/dl) and SonoVue 4.8 ml (unpublished data). In all lesions vascularity was determined in comparison with the surrounding pancreatic tissue and regarded as hypovascular, isovascular, or hypervascular using conventional and contrast-enhanced Doppler imaging. For analysis, isovascular and hypervascular lesions were compared together against hypovascular lesions. Video-documentation was performed (3 minutes) and representative images were documented. In addition, we performed blind reading and evaluation of the vascularity after contrast enhancement from the documented videos. Final Diagnosis The final diagnosis of 93 tumors was always achieved by histology (surgery [n 83] or transabdominal (percutaneous)- guided puncture [n 35], or endoscopic-guided 22G fineneedle aspiration cytology [n 22]; double numbers were possible). Reference Imaging Methods At least one reference imaging examination was performed in every patient (eg, computerized tomography, magnetic resonance imaging, transabdominal ultrasound, and in a few patients positron emission tomography) as part of the clinical work-up of the patients in several cases outside our institution and not for the purpose of study analysis in a standardized fashion. Statistics Data are given as mean and SD and range (minimum maximum) if appropriate. Vascularization assessed with EDUS and CE-EDUS was compared with histology assuming malignant lesions to be hypovascular. Sensitivity, specificity, positive and negative predictive values, and accuracy were calculated. 95% confidence intervals are given. Results Ductal Adenocarcinoma During the study period we observed 212 patients with histologically proven ductal adenocarcinoma of the pancreas. Ninety-eight patients were excluded because of liver and/or lung metastases or metastases of other organs. A total of 114 patients with undetermined lesions were examined by endoscopic ultrasound, all with solitary nodules of the pancreas. In 52 patients, lesions were greater than 40 mm with or without chronic pancreatitis and, therefore, also excluded from study Table 2. Contrast-Enhanced Color Doppler Ultrasound in 93 Patients Analyzing Inclusion and Exclusion Criteria Entity No Hypovascular (marginal tumor vessels) Isovascular/hypervascular Ductal adenocarcinoma of the pancreas (92%) 5 (8%) Pancreatic tumors of other origin Neuroendocrine tumor 20 0 (0%) 20 (100%) Microcystic serous pancreatic adenoma 10 0 (0%) 10 (100%) Teratoma 1 0 (0%) 1 (100%) NOTE. Inclusion criteria for study evaluation: mainly solid, tumor size 40 mm, solitary nodule. Exclusion criteria were as follows: tumor size 40 mm, multiple lesions, hormone-producing syndromes, and metastases.

4 May 2008 IMPROVED DIFFERENTIATION OF PANCREATIC TUMORS 593 showed hormone-production defined syndromes including 10 patients with multiple pancreatic lesions. Five patients showed pancreatic lesions larger than 40 mm and therefore also were excluded from study analysis. Therefore, 20 patients (age, y; range, y; 8 men, 12 women) with neuroendocrine tumors fulfilled inclusion criteria. The tumor size ranged from 8 to 40 mm. Thirteen lesions were located in the head of the pancreas, 5 in the corpus, and 2 in the tail of the pancreas. Conventional CDI did not reveal pancreatic vascularity in all 20 patients. In contrast, tumor vascularity in comparison with the nonvascularized pancreatic parenchyma was seen in 12 of 20 (60%) patients. Interestingly, all 50 lesions examined by contrast-enhanced ultrasound, but excluded, revealed hypervascularity at least of tumor parts in comparison with the surrounding pancreatic parenchyma (Figure 2). It has to be taken into account that some lesions showed nonenhancing parts of the tumor (eg, necrosis, a feature similarly observed in liver metastases). Serous Microcystic Pancreatic Adenoma During the study period we examined 12 patients with histologically proven serous microcystic adenoma of the pancreas. Two patients were excluded because of size greater than 40 mm, leaving 10 patients for study evaluation (age, y; Figure 1. CE-EDUS in a patient with hypovascular ductal adenocarcinoma of the pancreas. (A) CDI and (B) contrast-enhanced Doppler ultrasound both revealed a hypovascular lesion. analysis. Sixty-two patients (age, y; range, y); 37 men, 28 women) had lesions 40 mm or less and therefore were included in the study analysis (Tables 1 and 2). Conventional color Doppler imaging (CDI) did not reveal normal pancreatic vascularity in all 62 patients and, therefore, hypovascularity could not be used as a sign of malignancy. Five of 62 (8%) patients showed isovascular or hypervascular enhancement (3 ductal adenocarcinomas with neuroendocrine differentiation and 2 undifferentiated tumors) and 57 of 62 (92%) ductal adenocarcinomas showed hypovascularization (Figure 1). The size of the tumors was between 10 and 40 mm. Forty-eight carcinomas were located in the head of the pancreas, 8 in the corpus, and 6 in the tail. It is of interest that almost all excluded larger lesions (at least the part visualized by endoscopic ultrasound) greater than 40 mm (49 of 52; 94%) also revealed hypovascularity. Neuroendocrine Tumors From 76 patients with histologically proven neuroendocrine tumors of the pancreas we excluded 26 patients because of liver metastases. Fifty patients have been examined by contrast-enhanced endoscopic ultrasound. In addition, 25 patients Figure 2. CE-EDUS in a patient with a hypervascular neuroendocrine tumor. (A) CDI and (B) contrast-enhanced Doppler ultrasound both showed a hypervascular lesion.

5 594 DIETRICH ET AL CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 6, No. 5 Table 3. Sensitivity, Specificity, Positive Predictive Value, Negative Predictive Value, and Accuracy in the Characterization of Pancreatic Tumors (Hypovascularity as a Sign of Ductal Adenocarcinoma vs Hypervascularity as a Sign of Tumors of Other Origin) Using CE-EDUS or CDI Sensitivity Specificity Positive predictive value Negative predictive value Accuracy Hypovascularity as a sign of ductal adenocarcinoma using CE-EDUS Hypervascularity as a sign of tumors of other origin using CE-EDUS Hypervascularity as a sign of tumors of other origin using CDI 92% (82% 97%) 100% (89% 100%) 100% (94% 100%) 86% (71% 95%) 95% (88% 98%) 100% (89% 100%) 92% (82% 97%) 86% (71% 95%) 100% (94% 100%) 95% (88% 98%) 61% (42% 78%) 100% (94% 100%) 100% (82% 100%) 84% (73% 91%) 87% (79% 93%) NOTE. It must be noted that hypovascularity could not be determined using conventional CDI because the normal pancreatic parenchyma did not show reproducible vascularity at all using the equipment indicated. Ninety-five percent confidence intervals are shown in parentheses. range, y; 1 man, 9 women). The tumor size ranged from 15 to 40 mm. Eight lesions were located in the corpus and 2 were located in the tail of the pancreas. Conventional CDI did not reveal pancreatic vascularity in all 10 patients, whereas tumor vascularity in comparison with the nonvascularized pancreatic parenchyma was seen in 7 of 10 (70%) patients. All patients revealed a hypervascular tumor using CE-EDUS. Specific vascularity was seen in microcystic adenoma of the pancreas showing a central or unusual arteryand-spoke-wheel appearance in 7 of 10 (70%) (Figure 3; see supplementary material online at Other Lesions During the study time we observed 1 teratoma (10 mm), 1 ganglioneuroma (45 mm), and 1 partially necrotic leiomyoma (50 mm) of the pancreas. Only the teratoma fulfilled the criteria to be included in the study analysis. CE-EDUS revealed hypervascularity only using CE-EDUS. Neither parenchymal vascularity nor tumor vascularity could be shown using conventional CDI. In all patients no isoechoic or hyperechoic solid lesions were found. Therefore, all lesions (ductal adenocarcinomas and pancreatic tumors of other origin) were hypoechoic in comparison with the surrounding pancreatic parenchyma. The diagnostic test qualities for hypovascularity as a sign of ductal adenocarcinoma and for hypervascularity as a sign of tumors of other origin are given in Table 3. Results of patients examined by contrast-enhanced endoscopic ultrasound but not included in the study analysis are summarized in Table 4. Blind Reading In addition, we performed blind reading with respect to tumor vascularity. In 7 patients blind reading was not possible because of technical reasons. Differentiation of hypovascular versus isovascular or hypervascular lesions in comparison with the surrounding pancreatic parenchyma was concordant also using blind reading in 82 of 86 (95%) patients. In 1 patient the originally hypovascular lesion was regarded as isovascular and in 3 patients the originally isovascular lesion was regarded as hypovascular. In contrast, the differentiation of isovascular versus hypervascular lesions using blind reading was much more inconsistent. In 70 of 86 (81%) patients blind reading agreed on the vascularity of the tumor (hypovascular, isovascular, hypervascular in comparison with the surrounding pancreatic parenchyma), whereas in 16 patients a discrepancy was found between examiners (isovascular vs hypervascular in 12 patients and isovascular vs hypovascular in 4 patients). It has to be taken into account that differentiation of isovascular and hypervascular lesions was not part of the study protocol because both groups (isovascular and hypervascular lesions) were compared together against the hypovascular lesions. The test performances for the blind reading from documented videos of the CE-EDUS are given in Table 5. Table 4. Contrast-Enhanced Color Doppler Ultrasound in 118 Patients Not Included in the Study Analysis Because of Cystic Origin, Size 40 mm, Multiple Lesions, and Signs of Chronic Pancreatitis N Explanations Intraductal papillary mucinous neoplasia 6 In all 6 patients independently of the type of the lesion contrastenhancing neoplastic nodules were identified Mucinous cystadenoma/cystadenocarcinoma 19 In 15 of 19 patients (79%) contrast-enhancing neoplastic nodules were identified whereas 4 patients revealed hypovascularity (all benign lesions with small neoplastic nodules difficult to identify) Acinus cell carcinoma 2 In both patients contrast-enhancing neoplastic nodules were identified Solid pseudopapillary neoplasia 1 Contrast-enhancing vessels were identified Pseudocysts 90 In 8 of 90 patients contrast-enhancing vessels were identified transversing the pseudocyst NOTE. Final diagnosis was achieved by surgery or fine-needle aspiration cytology (additional follow-up evaluation in patients who did not undergo surgery).

6 May 2008 IMPROVED DIFFERENTIATION OF PANCREATIC TUMORS 595 Table 5. Sensitivity, Specificity, Positive Predictive Value, Negative Predictive Value, Accuracy in the Characterization of Pancreatic Tumors (Hypovascularity as a Sign of Ductal Adenocarcinoma vs Hypervascularity as a Sign of Tumors of Other Origin) Using CE-EDUS After Blind Reading From Documented Videos Sensitivity Specificity Positive predictive value Negative predictive value Accuracy Hypovascularity as a sign of ductal adenocarcinoma using CE-EDUS and blind reading Hypervascularity as a sign of tumors of other origin using CE-EDUS and blind reading 90% (80% 96%) 88% (68% 97%) 95% (86% 99%) 78% (58% 91%) 90% (81% 95%) 88% (68% 97%) 90% (80% 96%) 78% (58% 91%) 95% (86% 99%) 90% (81% 95%) NOTE. Ninety-five percent confidence intervals are shown in parentheses. Discussion If a pancreatic tumor is diagnosed, its differentiation is mandatory for adequate therapy and important for prognosis. The role of conventional endoscopic ultrasound and other imaging methods (eg, computerized tomography and magnetic resonance imaging) in the differential diagnosis of pancreatic masses were reported to be disappointing Therefore, a method with high sensitivity and specificity for tumor differentiation is necessary. The vascularity pattern of pancreatic tumors has been investigated in few studies. Before analyzing imaging features of solitary pancreatic masses the clinical work-up and the need for further imaging have to be assessed. Patients with pancreatic tumors and metastases elsewhere as well as pancreatic metastases of tumors of other origin normally do not pose a (differential) diagnostic problem of the pancreatic tumor itself. In addition, patients with metastases to the liver, lung, or any other organ normally will not be examined by endoscopic ultrasound because the distant metastases determine the clinical work-up (eg, liver biopsy) and prognosis of the diseases, which is true for ductal adenocarcinoma and neuroendocrine carcinoma. Predominantly cystic lesions are very rarely ductal adenocarcinoma. Ductal adenocarcinoma can be virtually ruled out in cystic lesions without chronic pancreatitis, at least in tumors up to 40 mm. Therefore, we excluded patients with macrocystic masses from study analysis. It is of interest that differentiation of benign and malignant cystic lesions (Table 4) is not possible by analyzing tumor vascularity. It has to be taken into account that it is sometimes difficult to identify neoplastic nodules in small cystic lesions. Transabdominal and endoscopic ultrasound easily can identify cystic lesions as cystic. It is well known that most cystic neoplasia are hypervascularized. 18 This is mostly true for benign mucinous cystadenoma, at least up to 40 mm, and the intraductal papillary neoplasia, both lesions with malignant potential, which therefore should be resected if possible. Malignant cystic tumors often are larger than 40 mm and show variable vascularity, but small malignant cysts also may occur. So it was surprising to see that Goh et al 41 described a malignant or premalignant condition in cystic lesions smaller than 2 cm in 38%. However, the presence of septa may help to discriminate between premalignant/malignant cysts and benign cysts. 42 We compared the grade of vascularization of solid pancreatic masses less than 40 mm in comparison with the surrounding pancreatic parenchyma to differentiate between ductal adenocarcinoma of the pancreas and other lesions in a highly selected patient population. We proved that analyzing tumor vascularity in comparison with the surrounding pancreatic parenchyma is helpful in tumor characterization in most tumors less than 40 mm that show typical vascular tumor characteristics. We showed clearly that CDI is not helpful to determine hypovascularity because the normal pancreatic vascularity cannot be displayed reliably. It can be concluded that CE-EDUS should be performed in all patients with nonvascularized lesions using conventional CDI. In contrast, tumor hypervascularity can be visualized using conventional CDI in 19 of 31 (61%) patients. It can be concluded that CE-EDUS is not of additional value if hypervascularity is shown using conventional CDI. Different examiners agreed in 81% of patients on the vascularity of the tumor after blind reading of the documented videos. Most discrepancies between reviewers were observed in the differentiation between isovascular and hypervascular lesions, which was not part of the study protocol. These discrepancies might to some extent be explained by the fact that the visualization of the central artery in the 12 serous microcystic pancreatic adenomas was not considered hypervascular by all. We showed that CE-EDUS can differentiate between mostly hypovascularized malignant ductal adenocarcinoma containing mainly small vessel size with slow blood flow and hypervascularized benign tumor entities, mostly neuroendocrine tumors, and serous microcystic adenomas of the pancreas. This differential diagnosis is of importance because serous microcystic adenomas can be observed because of their very low malignant potential. On the other hand, neuroendocrine tumors are slowly progressive and might be enucleated or otherwise less radically resected than ductal adenocarcinoma. Multiple endocrine neoplasia has to be ruled out in patients with neuroendocrine tumors as well as genetically determined diseases (eg, von Hippel-Lindau syndrome) in patients with serous cystadenoma of the pancreas. Most other very rare entities are hypervascular as well (eg, teratoma) and a biopsy is mandatory for individual therapeutic strategies, mostly resulting in resection. In larger tumors heterogeneity is more prominent and typical tumor features are overlayed by regressive changes. Ductal adenocarcinoma 40 mm or smaller rarely are confused with cystic neoplasia. Lesions greater than 40 mm also are seldom a diagnostic challenge. Most large ductal adenocarcinoma are metastasized or show local arterial vessel complications leading to inoperability. In those large tumors computerized tomography is advantageous, revealing also distant

7 596 DIETRICH ET AL CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 6, No. 5 metastases of the liver and lung in comparison with endoscopic ultrasound with limited penetration depth. The role of CE-EDUS in the differential diagnosis of chronic pancreatitis and ductal adenocarcinoma has been discussed recently. In ductal adenocarcinomas only arterial vessels can be displayed in contrast to chronic pancreatitis displaying both arterial and venous vessels. 43 Several limitations have to be considered: especially the depth of penetration, which is limited to about 50 mm. However, especially small pancreatic carcinomas often are hard to distinguish from larger focal inflammatory pancreatic masses with necroses. Functional neuroendocrine tumors are mostly less than 40 mm and sharp delineated. Hypervascularity is the predominant imaging feature of neuroendocrine tumors independent of the functional status. Hypervascularity, sharp delineation, fibrotic strands, and typical vessel architecture are the predominant feature of serous microcystic adenoma. Recent advances in technology have supported the development of new endoscopic ultrasound systems, making it possible also to use low-mi contrast-enhanced imaging techniques, but promising wide-band harmonic imaging performed with endoscopic ultrasound is currently at a preliminary stage and not yet available for routine use in contrast to the transabdominal approach. 44,45 In conclusion, the successful management and treatment of pancreatic tumors requires highly sensitive and specific imaging techniques. Contrast-enhanced endoscopic ultrasound is effective in differentiating small solid pancreatic tumors of different origin in most cases. However, histology is still the standard of reference for the differentiation of pancreatic lesions. We suggest biopsy in patients with hypervascular lesions up to 40 mm and suspected neuroendocrine tumors or serous microcystic adenoma. In patients with isovascular and hypervascular lesions biopsy is of importance because isovascular and hypervascular lesions imply a different therapeutic approach in comparison with the ductal adenocarcinoma of the pancreas. In patients with serous cystadenoma follow-up evaluation may be recommended whereas neuroendocrine tumors will require surgery because of their malignant potential. Hypovascular pancreatic lesions in patients with no contraindication (liver metastases, peritoneal metastases) primarily should undergo surgery because they are indicative for ductal adenocarcinoma of the pancreas. In case of nonoperability, a biopsy yielding at least cytology before palliative chemotherapy is mandatory. Supplementary Data Note: To access the supplementary material accompanying the article, visit the online version of Clinical Gastroenterology and Hepatology at References 1. Yoshimori M, Nakamura K, Ozaki H. [Malignant tumor of the pancreas. A clinical study based on autopsy cases from 1963 to 1975 in National Cancer Center Hospital (part 3) (author s transl)]. Nippon Shokakibyo Gakkai Zasshi 1978;75: Ishii T, Maeda K, Nakamura K, et al. Cancer in the aged: an autopsy study of 940 cancer patients. J Am Geriatr Soc 1979; 27: Grimelius L, Hultquist GT, Stenkvist B. Cytological differentiation of asymptomatic pancreatic islet cell tumours in autopsy material. Virchows Arch 1975;365: Compagno J, Oertel JE. Microcystic adenomas of the pancreas (glycogen-rich cystadenomas): a clinicopathologic study of 34 cases. Am J Clin Pathol 1978;69: Ishii T, Maeda K, Nakamura K, et al. Cancer in the aged: an autopsy study of 940 cancer patients. J Am Geriatr Soc 1979; 27: Nakamura E, Shimizu M, Itoh T, et al. Secondary tumors of the pancreas: clinicopathological study of 103 autopsy cases of Japanese patients. Pathol Int 2001;51: Stamm B. [Cystic neoplasms of the pancreas. Pathologico-anatomic study of 15 pancreatic tumors]. Schweiz Med Wochenschr 1984;114: Kimura W, Nagai H, Kuroda A, et al. Analysis of small cystic lesions of the pancreas. Int J Pancreatol 1995;18: Adsay NV, Basturk O, Klimstra DS, et al. Pancreatic pseudotumors: non-neoplastic solid lesions of the pancreas that clinically mimic pancreas cancer. Semin Diagn Pathol 2004;21: Adsay NV, Bandyopadhyay S, Basturk O, et al. Chronic pancreatitis or pancreatic ductal adenocarcinoma? Semin Diagn Pathol 2004;21: Barthet M, Hastier P, Buckley MJ, et al. Eosinophilic pancreatitis mimicking pancreatic neoplasia: EUS and ERCP findings is nonsurgical diagnosis possible? Pancreas 1998;17: Kim YH, Saini S, Sahani D, et al. Imaging diagnosis of cystic pancreatic lesions: pseudocyst versus nonpseudocyst. Radiographics 2005;25: Klimstra DS, Heffess CS, Oertel JE, et al. Acinar cell carcinoma of the pancreas. A clinicopathologic study of 28 cases. Am J Surg Pathol 1992;16: Klimstra DS, Lonardo F. Microglandular carcinoma of the pancreas. Am J Clin Pathol 1997;107: Klimstra DS, Adsay NV. Lymphoplasmacytic sclerosing (autoimmune) pancreatitis. Semin Diagn Pathol 2004;21: Klimstra DS. Cystic, mucin-producing neoplasms of the pancreas: the distinguishing features of mucinous cystic neoplasms and intraductal papillary mucinous neoplasms. Semin Diagn Pathol 2005;22: Kloppel G, Luttges J, Sipos B, et al. Autoimmune pancreatitis: pathological findings. JOP 2005;6(Suppl): Rickes S, Randhan W, Malfertheiner P. Differentiation of cystic pancreatic lesions by echo-enhanced sonography with pulse inversion imaging presentation of case reports. Z Gastroenterol 2004;42: Rickes S, Unkrodt K, Ocran K, et al. Differentiation of neuroendocrine tumors from other pancreatic lesions by echo-enhanced power Doppler sonography and somatostatin receptor scintigraphy. Pancreas 2003;26: Chen CH, Yang CC, Yeh YH, et al. Contrast-enhanced power Doppler sonography of ductal pancreatic adenocarcinomas: correlation with digital subtraction angiography findings. J Clin Ultrasound 2004;32: D Onofrio M, Malago R, Vecchiato F, et al. Contrast-enhanced ultrasonography of small solid pseudopapillary tumors of the pancreas: enhancement pattern and pathologic correlation of 2 cases. J Ultrasound Med 2005;24: D Onofrio M, Caffarri S, Zamboni G, et al. Contrast-enhanced ultrasonography in the characterization of pancreatic mucinous cystadenoma. J Ultrasound Med 2004;23: Itoh T, Hirooka Y, Itoh A, et al. Usefulness of contrast-enhanced transabdominal ultrasonography in the diagnosis of intraductal papillary mucinous tumors of the pancreas. Am J Gastroenterol 2005;100: Ueno N, Tomiyama T, Tano S, et al. Contrast enhanced color

8 May 2008 IMPROVED DIFFERENTIATION OF PANCREATIC TUMORS 597 Doppler ultrasonography in diagnosis of pancreatic tumor: two case reports. J Ultrasound Med 1996;15: Klapman JB, Chang KJ, Lee JG, et al. Negative predictive value of endoscopic ultrasound in a large series of patients with a clinical suspicion of pancreatic cancer. Am J Gastroenterol 2005;100: 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: Rosch T. Staging of pancreatic cancer. Analysis of literature results. Gastrointest Endosc Clin N Am 1995;5: Chang KJ, Nguyen P, Erickson RA, et al. The clinical utility of endoscopic ultrasound-guided fine-needle aspiration in the diagnosis and staging of pancreatic carcinoma. Gastrointest Endosc 1997;45: Maire F, Sauvanet A, Trivin F, et al. Staging of pancreatic head adenocarcinoma with spiral CT and endoscopic ultrasonography: an indirect evaluation of the usefulness of laparoscopy. Pancreatology 2004;4: Powis ME, Chang KJ. Endoscopic ultrasound in the clinical staging and management of pancreatic cancer: its impact on cost of treatment. Cancer Control 2000;7: Rosch T, Lightdale CJ, Botet JF, et al. Localization of pancreatic endocrine tumors by endoscopic ultrasonography. N Engl J Med 1992;326: Jenssen C, Dietrich CF. [Endoscopic ultrasound in chronic pancreatitis]. Z Gastroenterol 2005;43: Dietrich CF, Ignee A, Trojan J, et al. Improved characterisation of histologically proven liver tumours by contrast enhanced ultrasonography during the portal venous and specific late phase of SHU 508A. Gut 2004;53: Rickes S, Unkrodt K, Ocran K, et al. Differentiation of neuroendocrine tumors from other pancreatic lesions by echo-enhanced power Doppler sonography and somatostatin receptor scintigraphy. Pancreas 2003;26: Schmitt T, Seifert H, Dietrich CF, et al. [Propofol sedation in endoscopic manometry of Oddi s sphincter]. Z Gastroenterol 1999;37: Seifert H, Schmitt TH, Gultekin T, et al. Sedation with propofol plus midazolam versus propofol alone for interventional endoscopic procedures: a prospective, randomized study. Aliment Pharmacol Ther 2000;14: Meining A, Rosch T, Wolf A, et al. High interobserver variability in endosonographic staging of upper gastrointestinal cancers. Z Gastroenterol 2003;41: Meining A, Dittler HJ, Wolf A, et al. You get what you expect? A critical appraisal of imaging methodology in endosonographic cancer staging. Gut 2002;50: Allescher HD, Rosch T, Willkomm G, et al. Performance, patient acceptance, appropriateness of indications and potential influence on outcome of EUS: a prospective study in 397 consecutive patients. Gastrointest Endosc 1999;50: Wargo JA, Warshaw AL. Surgical approach to pancreatic exocrine neoplasms. Minerva Chir 2005;60: Goh BK, Tan YM, Cheow PC, et al. Cystic lesions of the pancreas: an appraisal of an aggressive resectional policy adopted at a single institution during 15 years. Am J Surg 2006;192: Sahani DV, Saokar A, Hahn PF, et al. Pancreatic cysts 3 cm or smaller: how aggressive should treatment be? Radiology 2006; 238: Hocke M, Schulze E, Gottschalk P, et al. Contrast-enhanced endoscopic ultrasound in discrimination between focal pancreatitis and pancreatic cancer. World J Gastroenterol 2006;12: Dietrich CF, Ignee A, Frey H. Contrast-enhanced endoscopic ultrasound with low mechanical index: a new technique. Z Gastroenterol 2005;43: Dietrich CF, Braden B, Hocke M, et al. Improved characterisation of solitary solid pancreatic tumours using contrast enhanced transabdominal ultrasound. Journal of Cancer Research and Clinical Oncology 2007; epub ahead of press October 19, Address requests for reprints to: Professor Christoph F. Dietrich, MD, Innere Medizin 2, Caritaskrankenhaus Bad Mergentheim, Uhlandstr. 7, D Bad Mergentheim, Germany. Christoph.dietrich@ckbm.de; fax: (49) (0)

9 May 2008 IMPROVED DIFFERENTIATION OF PANCREATIC TUMORS 597.e1 Figure 3. CE-EDUS in a patient with a hypervascular serous microcystic adenoma of the pancreas. (A) CDI was not conclusive but (B D) contrast-enhanced Doppler ultrasound revealed definite hypervascularity. In the initial arterial phase stimulated acoustic emission might be misleading. (E and F) Nonvascularized lesions also can be displayed using this method shown in a patient with thrombosed pseudoaneurysm. Note the avascular parts of thrombosis.