Pancreatic juice cytology in the diagnosis of intraductal papillary mucinous neoplasm of the pancreas
Significance of sampling by peroral pancreatoscopy
The examination of pancreatic juice cytology could hypothetically contribute to the establishment of a definite diagnosis of malignant intraductal papillary mucinous neoplasm of the pancreas (IPMN), but to the authors' knowledge, its significance has not been confirmed to date. The current study was conducted to assess the diagnostic value of pancreatic juice cytology in IPMN and to examine the usefulness of peroral pancreatoscopy (POPS) in sampling pancreatic juice.
The study subjects were comprised of 103 patients with IPMN who underwent surgical resection of pancreatic tumors (adenoma in29 patients, borderline in17 patients, carcinoma in situ in 25 patients, and invasive carcinoma in 32 patients). Pancreatic juice was collected with a catheter in 71 patients and by POPS in 32 patients. Patients with pancreatic carcinoma (n = 81) and chronic pancreatitis (n = 76) also were investigated.
The cytologic diagnosis was found to be of nondiagnostic value in only one patient with an IPMN, whereas it was of no diagnostic value in 14 of the patients with pancreatic carcinoma (17.3%), a difference that was statically significant (P < 0.001). The location of the IPMN (either in the pancreas or the pancreatic ducts) was not found to significantly affect the diagnostic value of the test. The sensitivity for IPMN was 62.2% when pancreatic juice was collected by POPS, and was 38.2% when it was collected using a catheter. In the case of pancreatic carcinoma, the sensitivity of pancreatic juice cytology was found to be 25.4%, which was significantly lower than that for IPMN when the pancreatic juice was collected by POPS (P < 0.001).
Pancreatic juice cytology was found to have better diagnostic value in the patients with IPMNs compared with those with pancreatic carcinoma. POPS was found to be useful for the collection of pancreatic juice. Cancer 2005. © 2005 American Cancer Society.
Intraductal papillary mucinous neoplasms (IPMNs) of the pancreas are often difficult to diagnose even after the exclusion of benign and malignant tumors based on findings obtained with different imaging techniques. Therefore, it is not easy to determine the best treatment option.1–3 There have been various attempts to overcome this situation by investigating the pancreatic juice: examination of its cytology,4, 5 analysis of K-ras gene mutations,6 or telomerase activity.7 Of these, cytology plays a central role in the evaluation of IPMN and there have been some very promising reports,4, 8 but unfortunately, these were without confirmation in a large number of patients. Therefore, to our knowledge, the usefulness of pancreatic juice cytology for the differential diagnosis of benign IPMNs from malignant ones has not been clarified as yet.
Peroral pancreatoscopy (POPS) has been reported to be useful for the differential diagnosis between benign IPMNs and malignant ones.9 POPS allows a direct view of the lesion and the sampling of relatively large amounts of pancreatic juice from a site close to the lesion. Uehara et al. reported the high diagnostic value of the cytologic examination of pancreatic juice collected by POPS in the case of pancreatic ductal carcinoma,10 but to our knowledge there have been no reports concerning the usefulness of this method in the diagnosis of IPMN.
The objective of the current study was to evaluate the usefulness of pancreatic juice cytology in the diagnosis of IPMN in a large number of patients, with special reference to the role of pancreatic juice collected by POPS. In addition, the cytology results in patients with an IPMN were compared with those from patients with ordinary pancreatic carcinoma and chronic pancreatitis.
MATERIALS AND METHODS
The subjects were 103 consecutive patients with IPMN who underwent surgical resection between May 1989 and December 2004. They were 72 men and 31 women, with a mean age of 63.3 years (range, 37–86 yrs). The tumor was located in the head of the pancreas in 76 patients and in the body and/or tail of the pancreas in 27. The tumor was located in the main pancreatic duct (MPD) in 27 patients (26.2%) and in a branch duct (BD) in 76 patients (73.8%). Based on World Health Organization (WHO) criteria,11 IPMNs were histopathologically diagnosed as adenoma in 29 patients, borderline in 17 patients, carcinoma in situ (CIS) in 25 patients, and invasive carcinoma in 32 patients. With regard to the type of IPMN, there were 3 BD-type IPMN cases of 26 adenomas, 5 BD-type cases of 12 borderline tumors, 10 BD-type cases of 15 CIS, and 9 BD-type cases of 23 invasive carcinomas. These patients were divided into two groups for the purpose of data analysis. One group was comprised of 46 patients with benign tumors (adenomas and borderline tumors) and the other group consisted of 57 patients with malignant tumors (carcinoma).
The patients initially were suspected of having an IPMN based on diagnostic imaging including ultrasonography (US), computed tomography (CT), magnetic resonance cholangiopancreatography (MRCP), and endoscopic ultrasonography (EUS). A definite diagnosis was established based on findings in combination of these images as well as those of endoscopic retrograde cholangiopancreatography (ERCP). The diagnostic criteria of IPMNs were the presence of a filling defect in the dilated ductal system, a patulous ampulla, efflux of mucin, and opacification of a mucinous substance noted on pancreatography.2, 12
Pancreatic juice was sampled before surgical resection in all patients for the cytologic examination. In 71 patients, pancreatic juice was collected for 10–15 minutes during ERCP through a catheter measuring 1.8 mm in greatest dimension by the intravenous administration of 50 U of secretin (Eizai, Tokyo, Japan). In the remaining 32 patients, the pancreatic juice was collected using POPS (BP-30 scope; Olympus, Tokyo, Japan) through a working channel (measuring 1.2 mm in greatest dimension) while observing the lesion or from a position close to the lesion.
POPS was performed using the “mother–baby” endoscopic method. TJF10 scope (Olympus Optical Co. Ltd., Tokyo, Japan) was used as the mother endoscope and BP-30 was used as the baby endoscope that was inserted directly through the papillary orifice. When insertion was difficult, a guidewire was used through the working channel. The pancreatic duct was examined from the distal to proximal end.
The tumor was found to be located in the MPD in 8 patients and in a BD in 63 patients from the group whose pancreatic juice was collected by catheter. In the group whose pancreatic juice was sampled by POPS, the tumor was found to be located in the MPD in 19 patients and in a BD in 13 patients.
The results of cytology and the diagnosis by POPS were compared. POPS findings indicative of malignancy were as follows: a fish egg-like protruding lesion with a vascular image, a villous protruding lesion, and a vegetative protruding lesion.9
To improve the visualization by POPS, physiologic saline was flushed into the pancreatic ducts, and pancreatic juice mixed with saline was used for the cytologic examination. The samples were immediately centrifuged at 1000 × gravity for 10 minutes. Smears were fixed in 95% ethyl alcohol and stained by the Papanicolaou method as well as with Giemsa stain. Cytologic findings were categorized to be nondiagnostic or diagnostic (benign, atypical, and malignant).
The samples with cytology as well as histology were reviewed blindly by two pathologists who provided a consensus opinion with regard to the diagnosis without having any clinical information.
The cytological criteria used to establish the diagnosis of IPMN were the presence of mucinous epithelial cells that were either isolated or arranged in small papillary clusters. Furthermore, IPMNs were graded based on these cytologic criteria and the degree of cell differentiation as follows: benign (tall columnar cells arranged in cohesive folds with mucinous hypertrophy in the apical region, and small and uniform nuclei), atypical (an increased nuclear-to-cytoplasmic ratio, nuclear hyperplasia, nuclear crowding, and stratification), and malignant (irregular projections, prominent anisonucleosis, large nuclei, nuclear irregularities, and loss of polarity, with or without neoplastic infiltrating glands).13
The 81 patients with pancreatic carcinoma were comprised of 56 men and 25 women with a mean age of 63 ± 9.6 years. The final diagnosis of pancreatic carcinoma was obtained by surgical resection in 32 patients, by fine-needle aspiration biopsy of the tumor in 42 patients, and by both imaging diagnosis and clinical follow-up in 7 patients. The 78 patients with chronic pancreatitis were comprised of 57 men and 21 women with a mean age of 58.1 ± 11 years. The definitive diagnosis of chronic pancreatitis was made using various imaging methods as well as by clinical follow-up for longer than 1 year (mean, 7.4 yrs;, range, 1–14 yrs); none of these patients died or developed apparent malignancy during the follow-up period. Data from consecutive patients with pancreatic carcinoma and chronic pancreatitis were collected between May 1989 and December 2004, which is the same period during which data from patients with IPMN were collected. In the group of patients with pancreatic carcinoma and chronic pancreatitis, the pancreatic juice was sampled using a catheter in the same manner as in the IPMN group, and the criteria for the cytologic diagnosis also were the same.
The sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) (percentage) of the pancreatic juice cytology in differentiating benign IPMNs from malignant IPMNs were calculated based on the histopathologic diagnosis of the resected specimens. These patients also were compared between those with malignant IPMNs and those with pancreatic carcinoma as well as between the catheter collection method and the POPS collection method.
Comparisons regarding discrete values were evaluated using the chi-square test or Fisher exact test. Statistical significance was set at a P value < 0.05.
Informed consent was obtained from all patients and the study was approved by the institutional review board of Chiba University.
Comparison between Cytologic Diagnosis and Histopathologic Diagnosis
The results of the cytologic diagnosis of the pancreatic juice and the histopathologic diagnosis of the resected specimen are shown in Table 1. Pancreatic juice cytology was found to be of nondiagnostic value in only one patient with an IPMN, whereas it was found to be of nondiagnostic value in 14 of the 81 patients with pancreatic carcinoma (17.3%), which is a significant difference (P < 0.001).
Table 1. Comparison between Cytologic and Histopathologic Diagnoses
|IPMN|| || || || |
| Adenoma (n = 29)||0||16||13||0|
| Borderline (n = 17)||0||8||8||1|
| Carcinoma in situ (n = 25)||0||6||8||11|
| Invasive carcinoma (n = 32)||1||5||9||17|
|Pancreatic carcinoma (n = 81)||14||46||4||17|
|Chronic pancreatitis (n = 78)||2||70||6||0|
The results of the cytologic examination were analyzed in relation to the location of the tumor (Table 2). The sensitivity, specificity, PPV, and NPV according to the tumor location in the pancreas were as follows: 50%, 97.1%, 95.5%, and 61.1%, respectively, for tumors in the head of the pancreas, and 50%, 91.7%, 87.5%, and 61.1%, respectively, for tumors in the body and/or tail of the pancreas. This difference was not statistically significant between the two groups. Similarly, the same values according to the type of IPMN were as follows: 57.9%, 100%, 100%, and 50.0%, respectively, for MPD tumors and 47.4%, 94.7%, 90.0%, and 64.3%, respectively, for BD tumors (Table 3). Therefore, the sensitivity was better, but not significantly different, in the patients with MPD tumors.
Table 2. Correlation between Cytologic and Histopathologic Diagnoses with Regard to the Location of the IPMN in the Pancreas
|Adenoma (n = 29)||0||11||10||0||0||5||3||0|
|Borderline (n = 17)||0||7||6||0||0||1||2||1|
|Carcinoma in situ (n = 25)||0||4||5||7||0||2||3||4|
|Invasive carcinoma (n = 32)||0||5||7||14||1||0||2||3|
Table 3. Correlation between Cytologic and Histopathologic Diagnosis with Regard to the Location of the IPMN in the Pancreatic Duct
|Adenoma (n = 29)||0||1||2||0||0||15||11||0|
|Borderline (n = 17)||0||1||4||0||0||7||4||1|
|Carcinoma in situ (n = 25)||0||3||2||5||0||3||6||6|
|Invasive carcinoma (n = 32)||0||2||1||6||1||3||8||11|
The sensitivity of pancreatic juice cytology in samples obtained by POPS was found to be better compared with that obtained using the usual catheter collecting method in patients with IPMN, but the difference was not statistically significant. Similarly, there was no statistically significant difference noted between IPMN by catheter and pancreatic carcinoma; however, the sensitivity of POPS in patients with IPMN was found to be statistically better compared with that in patients with pancreatic carcinoma (P < 0.001) (Table 4).
Table 4. Comparison of the Diagnostic Value of Pancreatic Juice Cytology between Malignant IPMN and Pancreatic Carcinoma According to the Method Used to Collect Pancreatic Juice
|Positive predictive value||100%||92.9%||100%|
|Negative predictive value||58.8%||62.5%||60.3%|
With regard to the location of the IPMN in the pancreatic ducts, the sensitivity of pancreatic juice cytology in samples obtained by POPS also was found to be better in the case of MPD tumors compared with BD tumors, but the difference was not statistically significant (Table 5).
Table 5. Diagnostic Value of Pancreatic Juice Cytology in Samples Obtained by POPS in Patients with Malignant IPMN according to their Location in the Pancreatic Ducts
|Positive predictive value||100%||100%|
|Negative predictive value||57.0%||60.0%|
Results of Pancreatic Juice Cytology in Patients with Malignant IPMNs in whom Malignant Findings Could Not Be Detected in Samples Obtained by POPS
Of the 32 patients who underwent POPS, 22 were found to have malignant IPMN; no malignant findings were observed by POPS in 7 patients, of whom the results of pancreatic juice cytologic examination indicated malignancy in 4 patients (57.1%). In another 15 patients whose POPS findings met the criteria of malignancy, cytology indicated malignancy in 11 patients (73.3%). Therefore, cytology could actually compensate for the insufficient diagnosis made using the POPS examination.
IPMNs can be of various histopathologic degrees ranging from adenoma to invasive carcinoma.1 In addition, they often exhibit extensive intraductal growth in the premalignant and invasive phases of their development.14 The exact cytologic discrimination of benign from malignant tumors based on the cytologic findings of pancreatic juice is sometimes difficult. For these reasons, the usefulness of pancreatic juice cytology in the diagnosis of malignant IPMNs remains controversial.4, 5, 15–17
The accuracy of diagnosis based on pancreatic juice cytology will depend on the patients' background; that is, if the study includes a high proportion of patients with one type of cancerous lesion, then the results of the cytologic examination will be biased. The same will occur if the study involves a small number of patients with IPMNs. In the current study, there was a large number of patients and their distribution by histopathologic diagnosis was homogeneous. In addition, because the study was conducted in one institution and the cytologic examination was performed by the same pathologists, interpretation of the results was considered to have very little variation.
The results of the current study indicate that one could almost definitely diagnose malignant IPMN in approximately 50% of the patients based on cytologic findings. Cytology has the advantage of providing the final diagnosis over other diagnostic techniques. In this respect, the validity of cytology is basically different from that of imaging diagnosis.
In cytology, if the atypical diagnosis included malignancy, the sensitivity would increase to approximately 80%, which was better than that of imaging techniques.18, 19 However, at the same time, the PPV was reduced to approximately 50%, indicating that a patient could be at risk of being referred to undergo surgery for a nonmalignant IPMN. Because IPMNs may progress from adenoma to invasive carcinoma, this risk is believed to be fundamentally different from the risk of patients with noncancerous diseases such as chronic pancreatitis to be mistakenly referred to surgery for pancreatic carcinoma. In fact, some researchers have emphasized the need for an early surgical resection in patients with suspected IPMN of the pancreas because of the high frequency of invasive carcinoma and the inadequacy of preoperative imaging for assessing malignancy.1, 5, 18 Similarly, because the tumors can be considered as either premalignant or already malignant, their removal would potentially prevent or eliminate a neoplastic process that could eventually kill the patient.2
Based on the location of the tumor in either the pancreas or the ducts, the sensitivity of the cytologic examination was better in the case of MPD tumors than in BD tumors; however, the difference was not statistically significant. This was presumably because the epithelial cells of IPMNs along the entire length of the main pancreatic duct and branch ducts, together with those lining peripheral cysts communicating with the main duct, can be collected in a transpapillary manner.4
Several efforts have been made to improve the sensitivity of cytology in diagnosing malignant pancreatic lesions, including IPMNs. With regard to the diagnosis of ordinary pancreatic carcinoma, some authors have attempted to collect pancreatic juice aided by POPS, as in our study,10 but to our knowledge, the current study is the first to use POPS for the diagnosis of IPMNs. This procedure has been reported to be useful for the qualitative diagnosis of IPMNs.9, 20 In previous studies, we demonstrated that POPS had a high diagnostic value in the differential diagnosis of benign IPMNs and malignant IBMNs.17 POPS allows us to examine the lesion directly and to collect pancreatic juice under direct vision. In addition, POPS combined with cytology would contribute to increasing the diagnostic accuracy of the procedure.
In the current study, samples collected using POPS were found to have better diagnostic value than those collected by catheter. In the case of BD tumors, POPS did not permit direct visualization of the lesion, and collecting pancreatic juice by POPS was believed to be less useful. In fact, the sensitivity of cytology in the case of BD tumors was comparatively low compared with that for MPD tumors. However, the proportion of MPD tumors to BD tumors differed between those patients in whom the pancreatic juice was collected by POPS and those in whom it was collected by catheter; therefore, we cannot draw a definite conclusion regarding the usefulness of POPS. A controlled randomized study will be needed to clarify this point.
There have been previous reports on the analysis of pancreatic juice cytology in specimens collected using EUS-guided fine-needle aspiration (EUS-FNA).21–25 The indication for EUS-FNA is the same as that of POPS in terms of collecting pancreatic juice and identifying the lesion; in addition, EUS-FNA may be better for patients with BD tumors. Furthermore, there is an advantage in obtaining samples directly from the mural nodule. However, thick, viscous mucin may cause some difficulties for the aspiration of pancreatic juice when using a thin needle in patients with IPMNs8; it occasionally is difficult to obtain an adequate amount of pancreatic juice, even with a large-bore catheter. In fact, Uehara et al. indicated the need for frequent flushing with saline through the catheter to obtain the sample of pancreatic juice.4 Nevertheless, additional studies will be needed to assess the significance of EUS-FNA in the differential diagnosis of IPMN in a large number of patients.
There have been other attempts to distinguish benign IPMNs from malignant ones using pancreatic juice: the examination of K-ras mutations, telomerase activity, or carcinoembryonic antigen levels (CEA). The examination of K-ras mutations was found to be useless because of its considerably low specificity.6 The examination of telomerase activity was reported to be hopeful but at the same time very complicated, requiring a special technique, making it far from useful in a clinical setting.7 Conversely, CEA has recently been reported to have a relatively high diagnostic value.24 However, to our knowledge, the significance of CEA values in IPMN has not been elucidated to date.
Considering the treatment strategies for IPMNs, the diagnostic value of pancreatic juice cytology is believed to be useful. Specifically, the majority of patients with malignant cytology must be referred to undergo surgical treatment for malignant IPMN. With regard to patients with atypical cytology, those whose imaging diagnosis is consistent with a diagnosis of malignant IPMN should be strongly referred to surgery; however, when the imaging diagnosis does not suggest a malignant IPMN, the treatment strategy is more complicated. Our policies are as follows: young patients, those who have symptoms associated with IPMN, or those who seriously desire aggressive treatment should be referred to surgical treatment; the patients whose imaging diagnosis and pancreatic juice cytology do not suggest a malignant IPMN can be followed without surgical treatment and they should be followed closely using imaging modalities such as US or EUS. Repeated cytology would be desirable if possible. Other patients can be followed periodically.
In the current study, pancreatic juice cytology detected approximately 50% of malignant tumors in patients with malignant IPMNs. POPS was useful for the collection of pancreatic juice in patients with malignant IPMNs, especially those with MPD tumors.