AIM: To elucidate the role of contrast-enhanced endoscopic ultrasonography (CE-EUS) in the diagnosis of branch duct intraductal papillary mucinous neoplasm (BD-IPMN).

METHODS: A total of 50 patients diagnosed with BD-IPMN by computed tomography (CT) and endoscopic ultrasonography (EUS) at our institute were included in this study. CE-EUS was performed when mural lesions were detected by EUS. The diagnostic accuracy for identifying mural nodules (MNs) was evaluated by CT, EUS, and EUS combined with CE-EUS. In the patients who underwent resection, the accuracy of measuring MN height with each imaging modality was compared. The cut-off values to diagnose malignant BD-IPMNs based on MN height for each imaging modality were determined using receiver operating characteristic curve analysis.

RESULTS: Fifteen patients were diagnosed with BD-IPMN with MNs and underwent resection. The remaining 35 patients were diagnosed with BD-IPMN without MNs and underwent follow-up monitoring. The pathological findings revealed 14 cases with MNs and one case without. The accuracy for diagnosing MNs was 92% using CT and 72% using EUS; the diagnostic accuracy increased to 98% when EUS and CE-EUS were combined. The accuracy for measuring MN height significantly improved when using CE-EUS compared with using CT or EUS (median measurement error value, CT: 3.3 mm vs CE-EUS: 0.6 mm, P < 0.05; EUS: 2.1 mm vs CE-EUS: 0.6 mm, P < 0.01). A cut-off value of 8.8 mm for MN height as measured by CE-EUS improved the accuracy of diagnosing malignant BD-IPMN to 93%.

CONCLUSION: Using CE-EUS to measure MN height provides a highly accurate method for differentiating benign from malignant BD-IPMN.

Intraductal papillary mucinous neoplasm (IPMN) is defined as an intraductal, grossly visible (typically ≥ 1.0 cm) epithelial neoplasm of mucin-producing cells that arises in the main pancreatic duct (MPD) or its branches. The neoplastic epithelium is usually papillary, and the degrees of mucin secretion, duct dilation (cyst formation), and dysplasia are variable[1]. IPMN can be subdivided into main duct IPMN (MD-IPMN) and branch duct IPMN (BD-IPMN) depending on the location of the primary lesion[2]. Most BD-IPMNs are less invasive and can be monitored; thus, the differential diagnosis of benign and malignant BD-IPMN must be accurate to appropriately indicate surgical resection[3-6]. In 2006, an international panel of experts published the International Consensus Guidelines for the Management of IPMN (ICG2006)[7]. These guidelines were updated in 2012 (ICG2012)[8]. According to the most recent guidelines, all BD-IPMNs diagnosed with mural nodules (MNs) by computed tomography (CT), magnetic resonance imaging (MRI) or endoscopic ultrasound (EUS) are recommended for resection. However, certain studies have indicated that both the presence of MNs and their height may be risk factors for malignancy[9-13]. If MN height is included in the factors used to determine resection, superior criteria for BD-IPMN resection might be created. Therefore, in pre-surgical examinations of BD-IPMN, the identification of MNs as well as accurate measurements of MN height is important.

Recently, many reports have noted the effectiveness of contrast-enhanced EUS (CE-EUS) in the diagnosis of pancreatic tumors[14-17]. With regard to BD-IPMN, it has been reported that CE-EUS is effective at differentiating MNs from mucinous clots[18,19]. Furthermore, CE-EUS is effective at identifying the mucosal fluid attached to MNs, thus enabling more accurate measurements of MN height than when using EUS alone. However, to the best of our knowledge, no study has determined the ability of CE-EUS to accurately measure MN height.

Therefore, in this study, the utility of CE-EUS to accurately evaluate the presence and height of MNs was determined. The purpose of this study was to elucidate the role of CE-EUS in the differential diagnosis of benign and malignant BD-IPMN.

Although the frequency of IC in MD-IPMN is high (43.1%; 11%-81%), the frequency of IC in BD-IPMN is relatively low at only 17.7% (1.4%-36.7%). Deliberation is necessary before performing a resection for BD-IPMN because these lesions mostly occur in elderly patients, and the annual malignancy rate is only 2%-3%[8]. ICG2006 suggests that when encountering suspicious findings for malignant BD-IPMN, surgery is recommended in cases with (1) an MPD diameter of 6 mm or greater; (2) a cyst size of 30 mm or greater; or (3) the presence of MNs as determined by diagnostic imaging[7]. However, a recent meta-analysis demonstrated that rather than cyst size and MPD diameter, the presence of MNs is strongly indicative of malignant BD-IPMN[20]. In this study, all BD-IPMN cases in which MNs were pathologically confirmed were ultimately diagnosed with either high-grade dysplasia (HGD) or IC. Along with diagnostic imaging, the methods for diagnosing malignant BD-IPMN include pancreatic fluid cell examination and cyst fluid examination; however, both methods have disadvantages. Pancreatic fluid cell examination has the advantage of high specificity, but the sensitivity can vary widely from 11%-92%[21-23], leading to a high risk of false negatives. Although EUS-guided fine needle aspiration cytology and laboratory analysis of cyst fluid have provided excellent results in certain studies[24-27], the safety of this method remains unclear because other reports have indicated the possibility of peritoneal dissemination due to the leakage of the cyst contents[28,29]. Therefore, we believe that diagnosing malignant BD-IPMN by evaluating MNs is both accurate and safe.

CT has been reported as effective for diagnosing MNs in addition to providing the information on BD-IPMN morphology, specifically the location and the presence of any communication with the MPD[30,31]. Nakagawa et al[32] reported that using CT to detect MNs in BD-IPMN yielded a sensitivity, specificity and accuracy of 68%, 100%, and 77%, respectively. In our study, using CT to diagnose MNs in BD-IPMN yielded a sensitivity, specificity, and accuracy of 71%, 100%, and 92%, respectively, which were superior values to those reported by Nakagawa et al[32] Despite this, out of 14 pathologically confirmed BD-IPMN cases with MNs, CT failed to detect four cases (29%) of MNs during the pre-surgical examination. Furthermore, of these four cases, two were eventually diagnosed with IC based on pathological examination, thus indicating that there are limits to differentiating benign and malignant BD-IPMN using CT alone.

EUS is an invaluable modality for evaluating pancreatic diseases because of its high spatial resolution. Compared with CT and MRI, EUS is excellent at detecting small pancreatic lesions and is also useful for diagnosing IPMN[33]. Thus far, EUS has demonstrated a high level of sensitivity at diagnosing MNs in BD-IPMN in addition to producing very few false negatives[34,35]. In this study, no false negatives were obtained with EUS. The 22 BD-IPMN patients diagnosed without MNs using EUS at the initial examinations did not have detectable MNs during the entire follow-up period. However, because it is difficult to differentiate MNs from mucinous clots using EUS alone, the specificity of this methodology is low, and there is a high risk of false positives[32]. In this study, 13 of the 28 mural lesions (46%) detected by EUS alone were, in fact, mucinous clots. The accuracy of EUS alone (72%) was insufficient. Zhong et al[36] reported that ascertaining the features of MNs can improve the ability to differentiate mucinous clots from MNs; however, this only increases the accuracy to 79%, which is not ideal. Therefore, it is difficult to accurately diagnose the presence of MNs using EUS alone.

Originally, a contrast-enhanced imaging technique was not available for EUS because the transducer of the echoendoscope was too small to produce enough acoustic power to perform contrast-enhanced imaging using a first-generation ultrasound contrast agent. A recently developed second-generation contrast agent now allows for the production of harmonic signals, even at lower acoustic power, making CE-EUS available for clinical use[14]. To date, there have been relatively few reports of CE-EUS being used to diagnose pancreatic cystic tumors, and only a few reports exist on the utility of CE-EUS for diagnosing MNs in BD-IPMN[18,19]. In this study, we looked not only at cases in which MNs were diagnosed using CE-EUS and resected but also at cases in which mucinous clots were diagnosed using CE-EUS and followed up. The results indicated that 14 of the 15 cases (93%) diagnosed with MNs using CE-EUS were subsequently pathologically confirmed to have MNs. In addition, the 13 cases diagnosed with mucinous clots by CE-EUS showed no malignant findings during follow-up, and they presented no contradictions regarding the absence of MNs. The accuracy of diagnosing MNs in BD-IPMN increased from 72% to 98% when EUS was combined with CE-EUS. The accuracy of EUS combined with CE-EUS was better than that for CT or EUS alone. As a result, we believe that CE-EUS is the most appropriate method for detecting the presence of MNs in BD-IPMN.

Certain studies have indicated the possibility that MN height is a risk factor for malignant BD-IPMN[9-13]. Therefore, accurately detecting the presence of MNs and accurately measuring MN height may contribute to the differential diagnosis of benign and malignant BD-IPMN. To the best of our knowledge, no previous studies including CE-EUS have determined the optimal imaging modality for measuring MN height. In this study, the measurement error values, calculated as the absolute values of the difference between the MN height measured by specific imaging modalities and that measured on pathological specimens, were compared. The results indicated that the measurement error values for CE-EUS were significantly lower than those for CT or EUS, supporting our conclusion that CE-EUS is the optimum imaging modality for measuring MN height in BD-IPMN. CE-EUS most likely both differentiates the mucosal fluid attached to MNs and clarifies the structure of the cystic wall, making it possible to accurately measure MN height. The threshold value of 8.8 mm, which was determined from a ROC curve related to the diagnosis of malignant BD-IPMN based on MN height as measured by CE-EUS, gave an excellent accuracy of 93%. CE-EUS accurately measures MN height and has potential applicability in differentiating benign and malignant BD-IPMN.

There were several limitations to our research. Firstly, because this study was a retrospective study with a limited cohort, there is the possibility of selection bias. Secondly, the study involved patients who underwent resection as well as those who were merely monitored. As a result, it was not possible to pathologically confirm whether MNs were actually present in each case. Despite these limitations, CE-EUS demonstrated effectiveness at diagnosing the presence of MNs and in accurately measuring MN height. Recent studies have already begun to adopt MN height to inform treatment policy in BD-IPMN cases[37]. According to our results, CE-EUS is the optimum modality for measuring the height of MNs. If indications of resection in BD-IPMN cases can be determined from both the presence of MNs and their height as measured by CE-EUS, many unnecessary surgeries can be avoided.

In conclusion, CE-EUS offers the potential to accurately diagnose the presence of MNs in BD-IPMN. According to our results, in cases of a diagnosis of BD-IPMN with MNs, it is highly probable that the pathological diagnosis will be HGD or IC, in which case resection is indicated; this is consistent with ICG2012. Furthermore, if the MN height, as measured by CE-EUS, is 8.8 mm or greater, it is highly probable that the pathological diagnosis will be indicative of IC, resulting in a strong positive recommendation for resection. CE-EUS not only diagnoses the presence of MNs but also facilitates the measurement of MN height, thereby playing an important role in differentiating benign and malignant BD-IPMN. CE-EUS should therefore be considered necessary for the accurate pre-surgical evaluation of BD-IPMN.