Endoscopic ultrasound in cystic pancreatic lesions: operator training needs to be improved, EUS-guided sampling should be standardized, and decision-making should be multidisciplinary and evidence-based

Management strategies for cystic pancreatic lesions (CPLs) currently present a real challenge as the number of incidentally discovered lesions appears be increasing in parallel with improvements in and greater use of modern imaging methods [1]. The accompanying advances of expertise in pancreatic medicine and surgery have identified the basic criteria for categorizing CPLs into two distinct groups. Group 1 are malignant or premalignant CPLs, and are mucinous (mucinous cystadenomas or intraductal papillary mucinous neoplasms [IPMNs]), or nonmucinous (essentially comprising cystic neuroendocrine tumors and rare papillary cystic neoplasms). Group 2 are benign CPLs, and include pseudocysts, serous cystadenomas (the macrocystic type can pose diagnostic challenges [2]), cystic lymphangiomas, mesenteric/peritoneal cysts, lymphoepithelial cysts, and rare benign pancreatic cysts (isolated or associated with polycystic pancreatic disorders).

Up till recently, surgery has been recommended for all cysts with malignant potential and, indeed, for all symptomatic CPLs (these usually compress a neighbouring structure such as the pancreatic duct) regardless of their cause. The recent recognition that some small mucinous cysts (essentially small branch duct IPMNs) have a low risk of malignant transformation [3] has opened a reasonable possibility that careful surveillance strategies might be adopted for a subgroup of patients with small nonsuspicious lesions. Of course, when making a decision about surgery or surveillance, several considerations must be carefully balanced, including for example, lesion type, age and co-morbidities of the patient, presence of symptoms, location of CPL within the pancreas, and the ease, inherent consequences, and possible early and late morbidity of surgical therapy.

Most experts in pancreatic imaging recognise that the lesion morphology should be established within the clinical context (e. g., unilocular pseudocysts are best identified through bedside history-taking). Many CPLs are correctly classified using axial imaging (with or without transabdominal ultrasound); indeed, the accuracy of magnetic resonance cholangiopancreatography (MRCP) in the topographical identification of IPMNs, especially branch duct forms, has been well established. In cases where axial imaging cannot provide a diagnosis, EUS remains a potent adjunct tool for diagnosis and planning therapy. The performance of EUS in pancreatic imaging has been universally recognised: its excellent resolution allows close inspection of the cardinal features of CPLs. These include: size and shape (ductal, unilocular, multilocular, etc); location; the cyst wall; parietal thickening, nodules, or mass; septations (thick or thin); calcifications; microcysts; cyst contents (with droplets of mucus, echoic, anechoic, etc); and relationship to adjacent structures, such as the main pancreatic duct. This inspection may yield important diagnostic information and may also help in planning surgery if required. Signs suggestive of early or more advanced malignancy include parietal nodules, mass, or thickening, and the extension of these beyond the cystic cavity; the presence of these signs is extremely helpful for confidently proposing surgery in patients whose CPLs have already been categorized as group 1.

EUS-guided fine needle aspiration (EUS-FNA) with analysis of cyst fluid tumor markers and cytological examination is usually reserved for diagnostically challenging cases (< 30 % in the present authors’ experience). These are represented mainly by unilocular macrocysts where presumptive diagnosis using EUS features alone is accurate in only approximately 50 % of cases and where FNA with cyst fluid analysis has been shown by a number of leading centers to significantly increase diagnostic precision [2] [4] [5] [6].

In this issue of Endoscopy, three reports from highly experienced teams are devoted to EUS for CPLs; however they carry a disappointingly negative message with respect to overall performance.

First, a multicenter interobserver study compared how three clinician groups identified four essential EUS features in CPLs in 40 video clips [7]. The three groups comprised four experts in pancreatic EUS, four novices in this modality, and four ”semi-experts” (advanced endoscopy trainees with relatively limited EUS experience). The four EUS features – septations, parietal nodules, a solid component, and communication with the pancreatic duct – are considered to be important for both diagnostic characterization and suspicion of malignancy. Not surprisingly, the experts fared better in recognising these EUS features and this corresponds to the known learning curve generally applicable to EUS. In the expert group, concordance was good for nodules but was moderate for septations, solid components, communication with the pancreatic duct, and diagnosis, and was poor for suspicion of malignancy and decision to operate. However, diagnostic accuracy for the nature of CPLs varied considerably between observers, as, following surgical confirmation, the accuracy in the expert group ranged from 23 % to 46 % and this was only slightly better than the novices (18 % – 27 %) and the semi-experts (23 % – 32 %). Compliance with accepted criteria for surgery was mediocre overall, at 47 % for the expert group and with similar values for the other two groups.

The overall results of that study are disappointingly unimpressive, and one might ask whether the 40 video clips selected were representative of the difficulties encountered in daily practice. It might have been better to use videos from 40 nonselected consecutive patients. A second potential bias was that 17 % of the cases were pseudocysts: these, when small, are difficult to distinguish from small single unilocular mucinous lesions, without recourse to the clinical context and use of other features suggestive of chronic pancreatitis. Finally, and perhaps most importantly, the EUS features studied in this report were not verified at macroscopic examination and the overall performance at the center could be debatable when there is no gold standard. Nonetheless, this study is important in that it highlights the low level of uniformity in training for EUS in relation to CPLs, and it may prompt the creation of an international expert educational image bank for CPLs, for incorporation into national training and continuing medical education (CME) programs. Such an educational image bank would help to standardize image analysis, as exemplified by the interpretation of Fig. 1b in the article by De Jong et al. [7] as a mural nodule when it may also be taken to represent a mucus droplet in a mucinous cyst.

In the second paper in this issue coming from the same prestigious group [8], the authors conclude that EUS-FNA does not provide adequate material for cytology and laboratory diagnosis in CPLs. The objectives of this prospective study appear to be flawed due to considerations related to the methods. First, the authors chose to privilege sending fluid for cytology to the detriment of analysis of tumor markers: 38 % of samples did not go for laboratory analysis after cytology and this represents an enormous bias. The methods, including cutoffs, used for biochemical and tumor marker analysis would also be pertinent.

Secondly, cytological examination was done using smears from liquid aspirates on slides, despite the fact that liquid-based cytology (e. g., using ThinPrep), known to concentrate and maximize sampling, has been in use for many years [9]. In the absence of solid components within small CPLs (nodules, thick septae, or wall), the lesion can be aspirated first for biochemical and tumor marker analysis of cyst fluid (which we usually privilege); the needle contents at the end are then rinsed into a medium for preparing a cytology specimen (slides can be prepared in addition if the endosonographer has been trained in this area) [5]. As small CPLs collapse upon aspiration, cellular yields may improve as the collapsed walls are in contact with the needle. Furthermore, the diagnostic yield from small cysts (< 15 mm) without parietal thickening or nodules is likely to be low. In this study, 10 % of CPLs were between 1 and 10 mm and a further 27 % were between 11 and 20 mm and this almost certainly impacted on the results.

Thirdly and finally, the authors chose to prospectively perform FNA in all the consecutively encountered indeterminate CPLs without discrimination, and it is highly likely that a number of these corresponded to branch duct IPMNs, as reflected in daily practice and in the literature (approximately 30 % of incidentally discovered CPLs [1], but this is almost certainly an underestimate). However MRCP is extremely useful in confirming that diagnosis, as is EUS by identifying other dilated branch ducts; these are typical duct-like cystic structures, often found to communicate with the main pancreatic duct. It has been previously demonstrated that cytology alone provides a poor diagnostic yield for small IPMNs without a solid component [10]; however, for discriminating larger benign and malignant IPMNs, intracystic CEA and CA 72 – 4 have shown excellent sensitivities (> 90 %) and, importantly, a high negative predictive value (95 %) [11].

The only way to correctly interpret the results of this ongoing prospective study is to wait for study completion as the absence of gold standards does not allow for meaningful analysis. While definitive results are awaited, it cannot be concluded that the overall diagnostic value of FNA in this setting is limited as final diagnosis was not established here; for example, a low intracystic CEA and a negative cytology result may be highly informative in the correct clinical setting and, in our opinion, isolating a single component of the data in the context of management of CPLs is not useful. Modifications of the methods in this continuing study may allow for appropriate readjustments that at least increase the quantity of fluid reaching the biochemistry laboratory for tumor marker analysis.

The third study from North America [12] addresses the excellent question as to whether information obtained at EUS and/or EUS-FNA influences the eventual management in CPLs. The authors conclude that EUS-FNA does not add to the information found at EUS alone that relates to the decision on whether to operate or not. In theory, the relatively large number of patients in the study (n = 194) coupled with the exceptional prestige and experience (2500 EUS/year) of the center, and the expertise there for interpreting EUS images and performing FNA (the technical aspects of the biopsy are excellent) should have led to the production of pertinent information for the gastroenterology community. However, the conclusions reached in this retrospective study are considerably weakened because of some important factors.

First, the reasons for performing EUS and EUS-FNA (70 %) are not given and this obviously affects the results. It is imperative to emphasize the necessity for EUS after classical axial imaging, in the following circumstances:

CT or MRI have failed to accurately define the nature of a macrocystic CPL. EUS allows for a more careful inspection of the minute details of the lesion and permits the use of FNA with cyst fluid analysis. Tumor marker analysis is especially important here, since the decision for surgery is highly dependent on the additional information gained by this investigation. Presence of a CPL with obvious criteria for malignancy, either locally advanced or metastatic, or occurring in a patient where surgery is impossible or high risk due to co-morbidities or advanced patient age. Here, EUS-guided biopsy and FNA allows the cytohistological documentation needed before other therapies can be proposed. Patients with a documented asymptomatic branch duct IPMN. The use of EUS allows for careful inspection of signs of early or established malignant transformation. In the presence of suspicious signs at EUS, the utility of FNA with tumor marker analysis appears very interesting 11, and it also seems that the diagnostic cytological yield here is higher as the lesion advances 10.

To be truly relevant to daily practice, the results of this study should have been stratified according to the above real situations. In addition, the results of FNA cytology are surprising and they are classified as acellular, malignant, suspicious, or inadequate. What about benign cytology pertaining to serous cystadenomas and pseudocysts? Moreover, as many CPLs were presumably benign, the negative predictive value of cytology must not be ignored. As with the study of De Jong et al. [8], the cytological methodology may be suboptimal without the use of a liquid-based method where diagnostic yield is, not surprisingly, more effective.

The second consideration is that while the management approach appears to have been multidisciplinary, involving gastroenterologists, radiologists and surgeons, it is surprising that surgeons were left to their own devices in deciding whether to operate or not? This may reflect cultural differences but would appear to be suboptimal practice, as highlighted by the information that 28 % of patients with lesions classified as group 2 proceeded to surgery, for reasons that are not fully apparent and that may indeed represent an overtreatment strategy. Additionally, neither the final histological results nor the follow-up of patients are given, and we cannot infer whether the surgeon or the results of EUS-FNA are correct. It is also surprising that cyst fluid CEA findings did not correlate with surgical decisions. Overall, it is difficult to support the conclusion as to the futility of EUS-FNA in this situation.

In summary, two of these papers are not very encouraging with regard to support for EUS-FNA in the management of patients with CPLs, and the further study reveals mediocre interobserver agreement even amongst experienced endosonographers. The important take-home messages are that: (i) the FNA procedure, especially the processing of the material obtained, requires standardization; (ii) it is imperative that training is improved with regard to characterization of CPLs, and this training needs to be uniform, for example using standardized image banks that have been correlated with pathology results; (iii) all members of the multidisciplinary team, including surgeons, should follow the locally adopted algorithm applicable to CPLs, as there is no point in performing FNA if decisions are not influenced by this.

References

• 1 Allen P J, D’Angelica M, Gonen M et al. A selective approach to the resection of cystic lesions of the pancreas: results from 539 consecutive patients.  Ann Surg. 2006;  244 572-582
• 2 O’Toole D, Palazzo L, Hammel P et al. Macrocystic pancreatic cystadenoma: The role of EUS and cyst fluid analysis in distinguishing mucinous and serous lesions.  Gastrointest Endosc. 2004;  59 823-829
• 3 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. International Association of Pancreatology.  Pancreatology. 2006;  6 17-32
• 4 Levy M T. Pancreatic cysts.  Gastrointest Endosc. 2009;  69 110-116
• 5 Palazzo L, O’Toole D, Hammel P. Technique of pancreatic cyst aspiration.  Gastrointest Endosc. 2009;  69 146-151
• 6 Brugge W R, Lewandrowski K, Lee-Lewandrowski E et al. Diagnosis of pancreatic cystic neoplasms : a report of the cooperative pancreatic cyst study.  Gastroenterology.. 2004;  126 1330-1336
• 7 de Jong K, Verlaan T, Dijkgraaf M et al. Interobserver agreement for endosonography in the diagnosis of pancreatic cysts.  Endoscopy. 2011;  43 579-584
• 8 de Jong K, Poley J W, vanHooft J E et al. Endoscopic ultrasound-guided fine-needle aspiration of pancreatic cystic lesions provides inadequate material for cytology and laboratory analysis: initial results from a prospective study.  Endoscopy. 2011;  43 585-590
• 9 Frossard J L, Amouyal P, Amouyal G et al. Performance of endosonography-guided fine needle aspiration and biopsy in the diagnosis of pancreatic cystic lesions.  Am J Gastroenterol. 2003;  98 1516-1524
• 10 Maire F, Couvelard A, Hammel P et al. Intraductal papillary mucinous tumors of the pancreas: the preoperative value of cytologic and histopathologic diagnosis.  Gastrointest Endosc. 2003;  58 701-706
• 11 Maire F, Voitot H, Aubert A et al. Intraductal papillary mucinous neoplasms of the pancreas: performance of pancreatic fluid analysis for positive diagnosis and the prediction of malignancy.  Am J Gastroenterol. 2008;  103 2871-2877
• 12 Sahai A V, Chua T S, Paquin S, Gariepy G. Analysis of variables associated with surgery versus observation in patients with pancreatic cystic lesions referred for endoscopic ultrasound.  Endoscopy. 2011;  43 591-595