The role of EUS for diagnosis of pancreatic cysts: observe, needle, or brush?

It is probably fair to say that most endosonographers are generally “pro” cyst fluid analysis (CFA) for cystic pancreatic lesions (CPLs). On the face of it, the ability of endoscopic ultrasonography (EUS) to provide detailed imaging, tissue, and cyst fluid for analysis makes it a seemingly powerful diagnostic tool for these potentially malignant lesions.

In this issue of Endoscopy, Haddad et al. assess the safety and efficacy of a cytobrush device compared to standard EUS-guided CFA for CPLs. They conclude that “cytology brushings are more likely to provide an adequate mucinous epithelium specimen than standard FNA.” However, they also report a complication rate of 19 % (8 % relatively severe). Despite this, they conclude that in appropriately selected patients it “appears to be a safe technique.”

Is the cytobrush more likely to provide an adequate mucinous epithelium specimen than standard fine-needle aspiration (FNA)? There are a few issues that may limit the clinical applicability of the reported results. Specifically, the full potential of standard FNA was not exploited and, perhaps more importantly, the patients included may not have been truly appropriate candidates for CFA in the first place.

If the aim is to compare the cytology yield of two techniques, an attempt should be made to obtain cytology by both methods. In this study, standard FNA was used to aspirate fluid, but no attempt was made to sample the wall or mural nodules with the FNA needle. Also, the appearance of the cyst fluid (e. g., viscosity, color, clarity) was ignored (at least for study purposes). Presumably, if the fluid is mucoid or gelatinous, the lesion should be considered mucinous, whether or not mucinous cells are found by microscopic examination. By contrast, watery, brownish fluid is typical of pseudocysts and is extremely unusual in mucinous lesions. By not exploiting the full potential of standard FNA to provide tissue and to allow macroscopic assessment of cyst fluid appearance, the results are clearly biased in favor of the cytobrush – which makes the validity of any conclusions regarding the superiority of one method over the other debatable.

Many of the lesions in this study appear to be intraductal papillary mucinous neoplasms (IPMNs). At least 6/39 lesions (15 %) were surgically confirmed IPMNs; the total number was likely higher. This brings into question the true clinical relevance of the study population. Based on current guidelines, detection of mucinous epithelium (with or without dysplasia) is useful in cases of suspected mucinous cystic neoplasms only – not IPMNs [1]. The diagnosis of IPMNs can usually be made reliably using imaging criteria alone: when there is obvious, globular dilation of a side branch with obvious communication with the main pancreatic duct. When this is present, there is no need for FNA to confirm the diagnosis (even less if there is obvious main duct involvement and/or mural nodules, and even less if there is the classic “fish mouth” papilla). Surgical decision making for IPMNs also depends on lesion appearance (mural nodules, size, main duct involvement) and not tumor marker levels – again making FNA unnecessary. In other cases with main duct involvement (at least two patients in this study), it is probably safer to aspirate fluid via the papilla than by FNA (if for some reason fluid cytology or confirmation of mucinous nature is desired, and if the main duct involvement is not too far from the ampulla). Therefore, one could argue that suspected IPMNs are not appropriate study lesions and should be excluded.

Cyst fluid sampling may also be inappropriate for small lesions (< 1–2cm), for cysts in patients who are not surgical candidates, or for lesions that clearly have worrisome features – particularly in younger patients (i. e., unlikely to change management anyway, risk of tumor seeding if malignant, etc.). The median patient age in the study population was more than 70 years; one patient was 94! It is hard to imagine that all these patients were truly surgical candidates. Moreover, patients who are not surgical candidates do not represent a clinically relevant study population. Presumably, the patients who did eventually go to surgery did so for a reason; if these lesions had imaging features that made them highly suspect for malignancy, the value of FNA is questionable. The authors provide little information that allows one to determine the pre-test likelihood of malignancy or whether CFA was truly likely to influence management. It is therefore difficult to determine whether there was a clinically relevant spectrum of disease severity.

In summary, it is inappropriate to include suspected/probable IPMNs or lesions with a very high or very low pre-test likelihood of malignancy in a study of CFA techniques. The information provided leads one to suspect that the study population was not entirely clinically relevant.

The authors also claim that the procedure is relatively safe, in selected patients. Which patients? Clearly it was safe in those that did not have complications, but other than avoiding brushings in anticoagulated patients, the study provides no way to select out those at lower risk of complications.

This raises the question as to whether any risk of complications is justified … . Does CFA of any type influence the management of CPLs? Have you ever noticed that at multidisciplinary symposiums on CPL management the only person who seems to spend any time discussing the value of CFA results is the endosonographer? Although it appears that CFA results should influence CPL management, whether it actually does remains unproven. To take it further, if we show that CFA does influence management, it remains to be shown whether CFA-based management decisions improve outcomes – which is the ultimate measure of their value. Studies indirectly addressing these issues have shown marginal value of EUS-guided CFA in managing pancreatic cysts [2] [3]. Due to the absence of published data addressing these issues, I can offer only some personal observations after performing CFA for approximately 1000 cystic lesions. We have seen patients with what appear clearly to be serous (benign) lesions (by appearance and by CFA results) still be sent for surgery due to vague symptoms or lingering doubts about the long-term risk of malignant transformation. Conversely, patients with malignant cytology have not always been sent for surgery, due to co-morbidity or to lack of clarity about the survival benefits in older patients. If the results of EUS-guided CFA are ignored, what is the point of doing it? As we have seen, CFA is not without risk – especially if repeat CFA is required. Bleeding is usually alarming but inconsequential. Infections are fortunately rare, but when they occur can be life-threatening. We have also found that complete CFA (for all tumor markers and cytology) is possible in less than 75 % of cases (due to insufficient volume, excessive viscosity, technical difficulties). The tremendous overlap in CFA marker profiles [4] not infrequently leads to diagnostic confusion more than anything else (especially when the cyst appearance on EUS/CT is very discordant with CFA results).

So, in response to the question at hand, “The role of EUS for diagnosis of pancreatic cysts: observe, needle, or brush?”, the absence of a clear answer suggests that further work is clearly warranted. However, instead of focusing of “what?” and “how?” to sample, I believe the focus should be more on “why?”

Competing interests: None

References

• 1 Tanaka M, Chari S, Adsay V, Fernandez-del Castillo C. et al . International consensus guidelines for management of intraductal papillary mucinous neoplasms and mucinous cystic neoplasms of the pancreas.  Pancreatology. 2006;  6 17-32
• 2 Volmar K E, Creager A J. Fine needle aspiration of pancreatic cysts: use of ancillary studies and difficulty in identifying surgical candidates.  Acta Cytol. 2006;  50 647-655
• 3 Ajay V, Maker A V, Lee L S. et al . Cytology from pancreatic cysts has marginal utility in surgical decision-making.  Ann Surg Oncol. 2008;  15 3187-3192
• 4 Hamel P, Levy P, Voitot H. et al . Preoperative cyst fluid analysis is useful for the differential diagnosis of cystic lesions of the pancreas.  Gastroenterology. 1995;  108 1230-1235