Intraductal papillary mucinous tumors (IPMT) account for 5% of pancreatic neoplasms. Preoperative identification is important because of their frequent multifocal or diffuse involvement in pancreatic ducts, which makes extensive surgery necessary even in benign cases. To the authors' knowledge, the cytologic features of this entity in fine-needle aspiration biopsy (FNAB) specimens have seldom been described and are poorly standardized.
Eleven consecutive cases of surgically proven IPMT with previous endoscopic ultrasonography (EUS)-guided FNAB were collected for retrospective analysis. EUS-FNAB had been performed with on-site attendance of a cytopathologist in all cases. Macroscopic and microscopic appearance of mucin, cellular type and arrangement, presence of nuclear grooves, and degree of nuclear atypia were recorded.
Final diagnosis was benign IPMT (B) in four cases, borderline IPMT (Bo) in two cases, malignant IPMT (M) in one case, and IPMT associated with invasive carcinoma (Ca) in four. Retrospective analysis found moderate to high levels of extracellular mucin in 10 of the 11 cases. The other case (one Ca) showed a small amount of thick mucin. In all cases, epithelial cells were identified, although cellularity was very low in four cases (three B and one Bo). Atypia was absent in two cases (two B) slight in two cases (two B), moderate in three cases (one Bo and two Ca), and severe in four cases (one Bo, one M, and two Ca). Mucinous epithelium was found in nine cases and nonmucinous epithelium in five cases (one Bo and four Ca). Papillary structures were observed in five cases (two Bo and three Ca), sheets in eight cases (four B, one Bo, one M, and two Ca), single atypical cells in five cases (one Bo and four Ca), irregular clusters in three cases (one Bo and two Ca), and nuclear grooves in two cases (one B and one Bo).
Intraductal papillary mucinous tumors (IPMT) account for approximately 5% of pancreatic neoplasms,1 although this figure may in fact be an underestimation because the terminology has not been standardized.2 The term “IPMT” is applied to a spectrum of proliferative epithelial lesions ranging from hyperplasia to adenocarcinoma “in situ.” They have in common the production of large amounts of mucus, resulting in duct dilatation that can be detected by image techniques, particularly endoscopic ultrasonography (EUS). The involvement of the ductal system is frequently multifocal or diffuse, thus necessitating aggressive surgical procedures such as total pancreatectomy to eliminate the risk of development of an invasive adenocarcinoma.3, 4 Similarly, in cases with low-stage pancreatic adenocarcinoma in which a partial pancreatectomy is planned, the demonstration of an associated IPMT would modify the surgical approach.
Endoscopic ultrasound-guided fine-needle aspiration biopsy (EUS-FNAB) is an excellent method for procurement of diagnostic samples from the pancreas, with a diagnostic accuracy of > 90% for pancreatic adenocarcinoma.5 EUS-FNAB allows placement of the needle inside the dilated ductal system, at several levels if necessary, to distinguish IPMT from other causes of duct dilatation such as obstruction or chronic pancreatitis. Although there are several reports in the literature on EUS diagnosis of IPMT,6–9 to our knowledge the value of FNAB in this setting has not been well established.10 In fact, the evaluation of the utility of EUS-FNAB is impaired by the scarcity of descriptions of the cytologic features of this entity. Those performed to date tend to be nonsystematic or based on a limited number of cases.11–15
The aim of the current study was to provide clues for a correct preoperative diagnosis of IPMT by EUS-FNAB, based on the cytologic findings in a series of 11 histologically proven cases.
MATERIALS AND METHODS
Between January 2002 and June 2003, 147 EUS-FNABs of the pancreas were performed at the Hospital Clinic in Barcelona, Spain. Of these, 11 cases with surgical resection and with a final pathologic diagnosis of IPMT were identified and are the subject of the current study. The interval between FNAB and surgery ranged from 4 days to 6 months. There were 8 males and 3 females, with a median age of 60 years and 81 years, respectively (range, 37–77 years). The main clinicopathologic features and the original cytologic diagnoses are shown in Table 1.
Table 1. Clinical Features, Original Cytologic Diagnosis, and Final Diagnosis
EUS-FNAB was indicated to study cystic or solid masses of the pancreas with or without dilatation of the ductal pancreatic system, previously detected by other imaging techniques. EUS was carried out under conscious sedation. Patients first underwent diagnostic imaging with a radial scanning echoendoscope at 7.5 megahertz (MHz) and 12 MHz (GF UM20, Olympus America Inc., Melville, NY), and further EUS FNAB was performed using a linear echoendoscope (GF UC30P, Olympus America) with a 22G, 8-cm long, Wilson Cook needle catheter assembly (Wilson-Cook Medical Inc., Winston-Salem, NC). FNABs were performed on cystic lesions in five cases, on dilated pancreatic duct in four cases, and on solid and cystic masses associated with duct dilatation in two cases. Because an attendant cytopathologist was present throughout the procedure for immediate evaluation of the sample, several needle passes were performed when necessary.
The aspirated material was smeared onto slides, part of which were air dried and stained with a rapid Romanowsky stain (Diagnostic Grifols SA, Barcelona, Spain) and the rest fixed in 96% alcohol for Papanicolaou stain. Fluid samples of > 1 mL were processed after centrifugation. Tissue fragments or clots were recovered and processed as cell blocks.
A retrospective analysis was performed on cytologic smears to examine the following features reported in previous descriptions of IPMT12, 14: quantity and density of mucus, amount of cellularity, presence or absence of mucinous and nonmucinous epithelium, papillae, sheets, irregular clusters, discohesiveness, nuclear grooves, psammomas, and degree of nuclear atypia.
Significant findings from the cytologic analysis are shown in Table 2. The quantity and thickness of the mucin varied. In most cases, it was evident on macroscopic inspection at the time of aspiration. It tended to be more fluid in benign cases, resulting in a slightly stained, diffuse background on smears.
Cellularity was more abundant in malignant cases. In most cases, cells were arranged in monolayered sheets. Well formed papillae with fibrovascular cores were not observed in any of our cases. Micropapillary clusters were seen only in borderline and malignant cases. Similarly, irregular clustering and discohesiveness with single atypical cells were features of borderline and malignant lesions (Fig. 1).
Mucinous type epithelium was found in all but two cases (Fig. 2). In five cases, nonmucinous epithelium was observed, corresponding to malignant (four) or borderline (one) lesions. In three of these cases, the epithelium was distinctively eosiniphilic, with a similar appearance to duodenal mucosa. To rule out the possibility of needle track contamination, nonmucinous epithelium was recorded only in cases with cellular atypia or abnormal arrangement, usually micropapillary (Figs. 1A and 3).
Cellular atypia correlated well with the final histopathologic diagnosis. Two benign cases demonstrated slight atypia. Nuclear grooves were prominent only in two cases. Neither nuclear inclusions nor psammoma bodies were found.
In summary, benign cases presented abundant but less thick mucin, and usually little mucinous cellularity. High cellularity, high mucus density, and micropapillary structures were features of malignant and borderline cases.
Intraductal proliferative lesions of the pancreas are controversial entities. Currently, terms such as pan-in, mucinous ectasia, mucinous cystic neoplasia, and IPMT still overlap.1, 2 Fortunately, the debate is usually confined to the definition of a given lesion in the surgical specimen. From a practical point of view, there is general agreement on the need to characterize lesions detected preoperatively as potentially dangerous or not. This approach would allow us to classify mucinous lesions as a group,16 regardless of the final diagnosis. Indeed, even with the whole specimen, the distinction of a macrocystic IPMT from a mucinous cystic neoplasm could ultimately be based on clinical data such as age, gender, and cyst location.2, 17
IPMT poses major clinical problems. Even when associated with an invasive component, the prognosis is better than for standard invasive adenocarcinoma.17, 18 Cure of these lesions requires radical surgery. A ductal dilatation detected by imaging techniques may be a benign change that does not need therapy, as is the case in chronic pancreatitis, or a lesion that will require total pancreatectomy. Previous reports on IPMT suggest that the issue is a simple one for the cytopathologist, because typical papillae and abundant thick mucin will be found.12, 14, 15 However, in the current series, the amount of mucin was small in some cases, or was very fluid and difficult to demonstrate in others. Furthermore, the papillary pattern was not obvious in many cases, and the epithelium was not always mucinous.
Some previous reports assumed that the cytologic features of IPMT parallel those of mucinous cystic neoplasms, and so did not systematically analyze the specific features of IPMT.12, 14 Others present somewhat conflicting results because of the small number of cases.11, 15 The largest series published so far identifies features of malignancy and invasion with a varied spectrum of sampling techniques,13 an approach of only limited use in a clinical setting such as EUS-FNAB. In a more recent series of EUS-FNAB, Maire et al.10 reported very poor results for specimens collected from dilated ducts. In the current series, most specimens were also obtained from cysts or duct dilatations and were adequate for diagnosis. The methods used by Maire et al. for specimen collection and processing are confusing, and may be the reason for the disappointing results. The lack of well defined diagnostic criteria also explains the low precision of the original cytologic diagnoses in our series (Table 1).
The amount and quality of mucin are crucial for the identification of this entity.12 Some have advocated the use of special stains to identify mucin objectively.14, 16 In our experience, thick mucin is easy to demonstrate on microscopic examination with routine stains, whereas very fluid mucin will produce a diffuse background that makes special stains useless, because its behavior is similar to that of the fluid originating in the gastrointestinal lumen in EUS-FNAB procedures. Macroscopic examination of the fluid, and ideally the presence of the cytopathologist in the exploration room during the procedure, would help to clarify the issue. Nevertheless, the identification of a nearly watery fluid as mucin is too subjective and cannot be the sole basis for an important decision. Biochemical analysis of the fluid for viscosity, mucin proteins, and tumor markers is reported to be of help,14, 19–21 although the results are not entirely specific and may be controversial. We did not perform biochemical analysis systematically. In the current series, all cases yielded epithelium and we believe that, when the mucinous nature of the fluid is not absolutely obvious, the evaluation of the cellular component is the key for diagnosis.
Classic cytologic features of papillary tumors, which have been reported in IPMT,12 were seldom seen in the current study. Although true papillae with fibrovascular cores are a typical feature of IPMT, we found none in the current series, nor did we find psammoma bodies, and only two cases showed prominent nuclear grooves. Cases in the current study with papillae were exclusively small three-dimensional micropapillary clusters, and were all associated with cellular atypia corresponding to malignant or borderline cases. Monolayered sheets were the predominant pattern in this series, and indeed the only one in benign cases. This finding is no surprise because it is a standard FNAB cytology presentation for papillary tumors of any origin, probably corresponding to the lining of true papillae.
Mucinous epithelium, with wide vacuolated cytoplasm, was found in most cases, but not in all. Loss of mucin production is a sign of dedifferentiation and probably accounts for the presence of nonmucinous epithelium in borderline and malignant cases. Moreover, the cytopathologist must be aware of the different types of epithelium that may be present in IPMT, described variously as pancreatobiliary, intestinal, or oncocytic.1 A distinctive eosinophilic, nonmucinous epithelium may be representative of an otherwise typical IPMT, as can be seen in Figure 3.
An additional difficulty that we believe has not been stressed enough in cytologic reports of IPMT is the contamination by gastrointestinal epithelium and mucus that often occurs in EUS-FNAB samples. Small amounts of mucus must be regarded with caution for this reason. When a transgastric FNAB is performed for lesions in the body and tail of the pancreas, the cytopathologist has to be alert to the normal appearance of the gastric mucosa, which has a honeycombed appearance and may bear some resemblance to neoplastic mucinous epithelium. However, tumoral cells are significantly taller and cytoplasms are wider. In cases without evident cytologic atypia, the combination of a high amount of mucin that is invariably present in benign IPMT and the characteristic mucinous epithelium will be diagnostic. Nevertheless, however, the possibility of confusion with normal gastric mucosa has to be considered in every case. Similarly, duodenal epithelium may resemble ductal epithelium and the eosinophilic component of IPMT described above. Typically, duodenal epithelium contains goblet cells, but these may also be present in IPMT2 (Fig. 3B). For this reason, when only nonmucinous epithelium was present, we considered it as neoplastic when significant atypia or evident micropapillary configuration was found. Obviously, benign-appearing nonmucinous epithelium may be a more frequent component of IPMT than our data suggest, but the lack of previous references does not support this possibility.
Diagnosis in cases with benign IPMT can be particularly difficult because, as can be observed in the current series, cellularity tends to be poor, without complete papillae, and mucin, although abundant, may be fluid and difficult to recognize microscopically. An accurate microscopic examination is mandatory to detect the highly characteristic mucinous epithelium, which can establish a diagnosis on its own. With regard to malignant cases, the classic cytologic features of malignancy are usually found.
Once a mucinous neoplasm has been identified, the distinction between IPMT and a mucinous adenocarcinoma or a mucinous cystic neoplasm is a matter of correlation with clinical and image data.12 In this regard, the great advantage of being present in the exploration room is that the cytopathologist can ascertain personally that the specimen is obtained from a dilated duct or from a cystic lesion connected to the ductal system. Otherwise, a diagnosis of “mucinous neoplasm” is appropriate, whereas a cytologic diagnosis of IPMT requires a specific clinicopathologic communication.
A cytologic diagnosis of IPMT through EUS-FNAB is feasible, at least as a clinically significant orientation. The papillary nature of the lesion may be not obvious, particularly in benign cases, and the epithelium may be nonmucinous, particularly in malignant cases. Contamination with normal gastrointestinal epithelium, which occurs frequently in EUS-FNAB procedures, is a potential diagnostic pitfall in benign IPMT.