| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Commentary |
1 Department of Radiology, Winthrop-University Hospital, Mineola, NY
11501.
2 Division of Gastroenterology, Department of Internal Medicine,
Winthrop-University Hospital, Mineola, NY.
3 Department of Radiology, Long Island Jewish Medical Center, New Hyde Park,
NY.
Received June 27, 2007; accepted after revision June 29, 2007.
This article is a commentary on the preceding article by Visser et al.
Mineola, NY 11501.
Keywords: CT • cystic lesions • MRI • mucinous cystic neoplasms • neoplasms • pancreatic cancer • pancreatic masses • pancreatic mucinous neoplasms • pseudocysts
We don't invent our natures. They're issued to us along with our lungs, our pancreas, and everything else.—Michael Mann, The Silence of the Lambs [1]
Cystic pancreatic lesions are commonly detected, reflecting both better imaging resolution and the increased utilization of abdominal radiologic studies in our aging population [2]. They are most frequently incidental findings that are not related to the reason or reasons for imaging the patients in whom they are discovered [3–7]. The radiologic and clinical challenges are to differentiate the more common pseudocysts from mucinous pancreatic neoplasms, including mucinous cystic neoplasm and intraductal papillary mucinous neoplasm, and from serous cystadenomas and to determine the benign or malignant nature of the lesion and its potential resectability. These challenges have a reward in that the preinvasive stages of intraductal papillary mucinous neoplasm and mucinous cystic neoplasm are macroscopic and detectable on cross-sectional imaging.
Cystic degeneration of pancreatic adenocarcinoma and of solid and cystic pseudopapillary tumors, which have a variety of names, is less commonly encountered, and there are other rare cystic tumors of the pancreas, including cystic islet cell tumors and lymphoepithelial cysts, among others. Pseudocyst can most likely be inferred by a history of acute or chronic pancreatitis; alcoholism; complicated gallstone disease; and findings on pancreatic imaging including parenchymal calcifications, visualization of pancreatic duct side branches, and atrophy.
Intraductal papillary mucinous neoplasms are found most commonly in elderly men and occur most often in the pancreatic head, whereas mucinous cystic neoplasms are usually noted in the pancreatic tail of middle-aged women [8]. Unfortunately, the more recent literature has shown that the age and sex of the patient with a cystic pancreatic lesion and the location of a cystic pancreatic lesion are not particularly reliable indicators for the accurate prospective diagnosis of an individual lesion [4].
Characterization of cystic pancreatic lesions has been problematic since their initial modern classification by two pathologists at the Armed Forces Institute of Pathology in 1978, Compagno and Oertel [9, 10]. Cystic pancreatic lesions continue to be problematic for the radiologist and clinician alike for numerous reasons. They have been the subject of several excellent recent reviews from the Massachusetts General Hospital [11–14]. The article by Visser and colleagues [15], from the Departments of Surgery, Radiology, and Epidemiology and Biostatistics at the University of California at San Francisco, in this issue of the AJR, highlights these troubling problems and provides an opportunity to examine the current status of imaging and clinical evaluation and management of these lesions. Numerous strides have been made in the fields of radiology, gastroenterology, abdominal surgery, and pathology in the diagnosis and management of cystic pancreatic lesions. We believe that cystic pancreatic lesions are optimally evaluated and managed by a multidisciplinary team of physicians in all these fields.
Despite these multidisciplinary strides, the literature reviewed by Visser et al. [15], their results, our independent literature review, and our radiologic and clinical experiences at two tertiary care centers are all consistent with the sobering conclusion that a substantial subset of cystic pancreatic lesions, especially the relatively small and simple-appearing cystic lesions, are very difficult to diagnose accurately; can be premalignant or occasionally malignant; and pose dilemmas for appropriate patient management. Anecdotally, during the week before the preparation of this commentary, we noted multiple small cystic pancreatic lesions on CT scans of one of our two busy cross-sectional imaging services: a probable low-grade intraductal papillary mucinous neoplasm in the uncinate in an elderly man; several small pancreatic body cysts in older patients that could not be further characterized on CT alone; two patients with multiple small pancreatic cysts; and a 2-cm simple-appearing pancreatic body cyst in a patient with multicentric mucinous adenocarcinoma of the lung, multiple metastases elsewhere, and very high metabolic activity on a PET scan that therefore almost certainly was a mucinous metastasis. The only patient in whom the cystic pancreatic lesion was not an incidental finding was the latter one.
Over the past decade and particularly the past several years, although there has been continued controversy about the diagnosis and management of cystic pancreatic lesions, on review of the literature eight principles and guidelines have emerged that have altered the approach to the patient with a cystic pancreatic lesion. First, the concept that 90% of pancreatic cysts are pseudocysts has been challenged by numerous reports and is no longer considered accurate [4, 5, 16]. In addition, if a patient has a history of pancreatitis, although it is still more likely that the cystic pancreatic lesion is a pseudocyst, this is not the case in a substantial percentage of patients. A minority of cystic pancreatic lesions that are not pseudocysts may actually be the cause of pancreatitis, particularly intraductal papillary mucinous neoplasms, so eliciting a history of pancreatitis is not diagnostic of a pseudocyst when a cystic pancreatic lesion is discovered [5].
Second, as we already noted, there is a consensus in essentially all recent reports on cystic pancreatic lesions that they are more common than previously realized, particularly small (2–3 cm or less in maximal dimension) lesions, and that they are most often incidental findings on cross-sectional imaging examinations— typically CT, but also sonography and MRI— performed for unrelated reasons [4–6, 16–20].
Third, as with the current series reported by Visser et al. [15], almost all recent reports have identified limitations of all imaging techniques, including endoscopic sonography, for the accurate characterization of cystic pancreatic lesions based on imaging alone [11, 21–23].
Fourth, although relatively unusual, simple cysts (i.e., epithelial cysts) with no malignant potential are not as rare as previously believed and can be discovered in older patients, analogous to simple cysts in other solid abdominal organs [7, 24]. The final diagnosis of several patients in the current series by Visser et al. [15] was a simple pancreatic cyst.
Fifth, a subset of cystic pancreatic lesions, with or without associated pancreatic ductal dilatation, are intraductal papillary mucinous neoplasms; although histologically somewhat different from mucinous cystic neoplasms of the pancreas (i.e., they lack ovarian-type stroma and often communicate with the main pancreatic duct), they are similar in other respects to a broad spectrum of biologic behavior from an incidental low-grade cystic lesion in an older patient to a frankly malignant cystic neoplasm [25, 26].
Sixth, some of the "classical" pathology findings described by Compagno and Oertel [9, 10] and in the cross-sectional imaging articles that followed have not held up to further scrutiny. For example, the diagnostic finding of a serous cystadenoma with small central cysts in a honeycomb pattern, a stellate central calcification, and relatively larger peripheral cysts occurs in only a small percentage of all such tumors [27, 28]. Also, a subset of serous cystadenomas (10–25%) are composed of only a single cyst. Although recent reports have noted that such serous cystadenomas are more likely to be lobulated compared with mucinous cystic neoplasms and intraductal papillary mucinous neoplasms [29, 30], we believe, as do other authors [21, 31, 32], that reliable prospective diagnosis can be difficult. A review of Table 3 from the article in this issue by Visser et al. [15] highlights this particular problem in their retrospective review by two radiologists of the cross-sectional imaging studies of 58 patients, 52 of whom underwent CT and 18 of whom underwent MRI (including MR cholangiopancreatography [MRCP] and contrast-enhanced sequences), with histologically proven cystic pancreatic lesions. There were multiple instances in which serous cystadenoma was incorrectly diagnosed as another entity or vice versa by one or both of the radiologists.
Seventh, the dictum that all cystic pancreatic lesions should be resected in patients who are potentially surgical candidates is not correct. Patients in whom the cystic pancreatic lesion is believed to be incidental should be further evaluated, and if the lesion is thought to be benign or low-grade based on imaging (i.e., no solid components, no main pancreatic ductal dilatation [16, 18]) and on subsequent fine-needle aspiration (FNA) and cyst fluid analysis performed at endoscopic sonography, most of these patients can be managed conservatively with follow-up imaging as opposed to surgical resection. Recent work has shown that small cystic pancreatic lesions without septae or solid components are rarely frankly malignant [14], although two such tumors were included in the current series by Visser et al. [15]. Several reports documenting stability of most of these lesions on follow-up imaging have emerged over the past few years [6, 16–20]. However, the natural history of cystic pancreatic lesions is still under investigation, and there is no consensus to our knowledge about what the best follow-up imaging strategy should be (CT vs MRI vs endoscopic sonography with or without repeat aspiration or biopsy) and at what intervals follow-up imaging should be performed.
Eighth, the accuracy of endoscopic sonography for cystic pancreatic lesion characterization and the utility of FNA and cyst fluid analysis at endoscopic sonography remain somewhat controversial. In multiple relatively large recent series, the results varied depending on the particular report [23, 33–36], but most have shown that an elevated carcinoembryonic antigen (CEA) level (i.e., > 200–500 ng/mL) in a cystic pancreatic lesion aspirate is reliable evidence of either a mucinous cystic neoplasm or an intraductal papillary mucinous neoplasm, whereas a low CEA level is consistent with either a pseudocyst or a serous cystadenoma [23, 35, 37]. Similarly, identification of mucin on cystic pancreatic lesion aspiration is a significant predictor of a mucinous neoplasm, although there are occasional false-positives due to contaminant mucin [37–39]. The identification of malignant cells on FNA is also highly specific but unfortunately is not very sensitive [35, 40]. Challenges remain, for example, for the accurate characterization on FNA of serous cystadenomas [39], a benign lesion that if asymptomatic could be observed rather than resected. Among other problems, the identification of glycogen in the aspirate of a cystic pancreatic lesion, although highly specific for a serous cystadenoma, is not detected in some cases of subsequently proven serous cystadenoma [10, 12, 39].
We are not surprised that MRI did not provide any major advantages compared with CT for the characterization of cystic pancreatic lesions in the current report by Visser et al. [15]. As they correctly note, there have been few such reports in the literature retrospectively comparing CT and MRI for the same patients with cystic pancreatic lesions, and those series have been small [41]. We are not aware of any large series that has retrospectively or prospectively evaluated both CT and MRI for the evaluation of cystic pancreatic lesions in the same group of patients. However, no matter which imaging technique is applied (CT, MRI, or endoscopic sonography) and likely no matter how much the technology improves, we agree with Visser et al. and others that an intrinsic limitation in the radiologic evaluation of these lesions is that for a substantial percentage of cystic pancreatic lesions, particularly small lesions, their appearance on cross-sectional imaging is nonspecific and that only complete pathologic examination can be used to determine their true nature.
One might anticipate some advantages of MRI compared with CT—for example, for showing more subtle septae or nodules and associated pancreatic ductal abnormalities, especially in mixed-type intraductal papillary mucinous neoplasms. However, any major advantages of MRI—other than the elimination of ionizing radiation and iodinated contrast material—appear to be nullified with the use of optimized MDCT technique, including thin-section imaging with IV iodinated contrast material and supplemental reformations. In a recently published article, Sahani et al. [6] reported equivalence of MDCT and MRCP for detecting the presence or absence of malignancy in a retrospective review of 25 patients with intraductal papillary mucinous neoplasm, which included a mixture of the different types of lesions, some of which had cystic components. A prior publication with the same number of patients showed that MRI was equal to or was only slightly superior to thin-section helical CT for the evaluation of intraductal papillary mucinous neoplasms [42].
Based on the results reported by Visser et al. [15] and in conjunction with our review of the literature and our clinical experience, we believe that the main role of MRI is therefore for the follow-up of small cystic pancreatic lesions to limit cumulative radiation dose in relatively young patients for whom conservative management is chosen, or as a problem-solving secondary examination in selected patients.
We are also not surprised that the overall diagnostic accuracy and the likelihood of predicting malignancy were not significantly different between the more experienced radiologist and the less experienced radiologist in the current study by Visser et al. [15]. In the study by Procacci et al. [22] on the retrospective characterization of 100 cystic pancreatic lesions by two radiologists, the performance of the experienced abdominal imager compared with the radiologist who had just finished training in abdominal imaging was actually worse. The authors ascribed this paradoxical finding to the skilled radiologist trying more frequently to establish a more specific rather than a nonspecific diagnosis and subsequently making more errors [22].
We compliment Visser et al. [15] for including pancreatic pseudocysts in their retrospective review. Correct identification of pseudocysts is a continuing problem in a minority of cases, and pseudocysts have been resected because their true nature was not known before surgery [28]. Even the measurement of amylase levels within cystic pancreatic lesions has its limitations [38]. We agree, as noted by Visser et al., that the elimination of pseudocysts from most retrospective CT series of cystic pancreatic lesions adds bias. Regardless of whether pseudocysts were included, the accuracy of CT for characterizing cystic pancreatic lesions—even with current imaging techniques and with experienced radiologists—is less than ideal [22, 43]. However, there are other unavoidable biases in a retrospective series such as the one reported by Visser and colleagues, which they did acknowledge. Only a minority of cystic pancreatic lesions in their study were discovered incidentally. Although appropriate management of cystic pancreatic lesions that are not pseudocysts and that are believed to be causing symptoms would generally be surgical regardless of the underlying histopathology, the resultant distribution of lesions in their series may not reflect the distribution of cystic pancreatic lesions identified prospectively in other radiology practices. Regardless, we agree with the conclusion of Visser et al. that "heterogeneity among cystic lesions and overlap in imaging characteristics" should cause radiologists to "approach the specific characterization of cystic pancreatic masses with a substantial degree of humility." Recommending an appropriate management approach based on imaging findings, in conjunction with clinical information and clinical consultation, may therefore be more important than attempting to assign a specific diagnosis to a cystic pancreatic lesion.
Visser et al. [15] did not examine the role of endoscopic sonography in the characterization and management of cystic pancreatic lesions and state that "CT and MRI remain the primary means of assessment" and that "endoscopic sonography with cyst aspiration... [has] been used...at major referral centers." Although, as already noted, we agree that endoscopic sonography continues to have its limitations in the evaluation of cystic pancreatic lesions, it plays a major role in the workup of cystic pancreatic lesions at both of our institutions.
We believe that Visser et al. may underestimate the current role of endoscopic sonography for cystic pancreatic lesion evaluation in the gastroenterologic community, although we agree it is not nearly as available as CT or MRI. We believe that endoscopic sonography complements CT (and, if performed, MRI) evaluation of cystic pancreatic lesions, but it is highly operator-dependent, requires sedation, and carries a small risk of complications. As other investigators have noted, aspiration may not yield adequate tissue or fluid for analysis [33], and morphologic stratification of a cystic pancreatic lesion by endoscopic sonography has pitfalls similar to those of CT and MRI [44]. However, a promising, very recent pilot study [45] on endoscopic sonography cytology brushings showed that technique to be superior to FNA in terms of determining cellularity and confirming mucinous or dysplastic epithelium in seven of 10 patients, although the accuracy and safety of this approach in a larger series of patients remain to be seen. In another recent report, investigators noted improved results for endoscopic sonography–guided FNA (correct cytologic diagnosis in 28 of 30 cystic pancreatic lesions, with accurate diagnosis of 10 of 11 cases of malignancy and 100% accuracy for diagnosis of mucinous cystic neoplasms), which the authors attributed in part to the presence of an experienced cytopathologist at the time of the endoscopic procedures [46]. Endoscopic sonography remains inferior to CT and MRI for the global assessment of lymphadenopathy or metastases in more advanced lesions.
CT, MRI, and endoscopic sonography are complementary, but as suggested by Visser et al. [15], all three techniques are rarely necessary for evaluation of a cystic pancreatic lesion in an individual patient. In addition, in their recent study, Mansour et al. [43] concluded that PET is not essential for the evaluation of cystic pancreatic lesions. Integration of clinical data is also paramount. A younger woman, with no history of pancreatitis who presents with a 6-cm pancreatic tail cystic lesion should reasonably proceed directly to surgery without further imaging. An elderly man with chronic obstructive pulmonary disease should not immediately undergo surgery for a 2-cm pancreatic head cystic lesion found incidentally on a CT scan obtained for suspected pulmonary embolism. The subsequent evaluation and management of such a patient remain controversial, although cost-effective imaging approaches and algorithms have very recently been proposed [5, 12, 17, 47]. Further evaluation should be thoughtful and practical, and management of these cystic pancreatic lesions should involve input from radiologists, gastroenterologists, and abdominal surgeons.
We commend Visser and his radiologic, surgical, and epidemiologic colleagues for their outstanding addition to the literature on the characterization of cystic pancreatic lesions using CT and MRI, which remains somewhat problematic, but the nature of which has become somewhat clearer over the past several years.
References
CiteULike 
Complore 
Connotea 
Del.icio.us 
Digg 
Reddit 
Technorati What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
