Fax: (011) 81-6-6981-8055
Role of intraoperative cytology combined with histology in detecting continuous and skip type intraductal cancer existence for intraductal papillary mucinous carcinoma of the pancreas
Article first published online: 19 OCT 2006
Copyright © 2006 American Cancer Society
Volume 107, Issue 11, pages 2567–2575, 1 December 2006
How to Cite
Eguchi, H., Ishikawa, O., Ohigashi, H., Sasaki, Y., Yamada, T., Nakaizumi, A., Uehara, H., Takenaka, A., Kasugai, T. and Imaoka, S. (2006), Role of intraoperative cytology combined with histology in detecting continuous and skip type intraductal cancer existence for intraductal papillary mucinous carcinoma of the pancreas. Cancer, 107: 2567–2575. doi: 10.1002/cncr.22301
- Issue published online: 17 NOV 2006
- Article first published online: 19 OCT 2006
- Manuscript Accepted: 6 SEP 2006
- Manuscript Revised: 27 AUG 2006
- Manuscript Received: 1 JUL 2006
- intraductal papillary mucinous neoplasms;
- intraductal cancer extension;
- skip lesions;
- pancreatic juice cytology
Intraductal papillary mucinous neoplasm (IPMN) is a recently discovered pancreatic tumor that has continuous or discontinuous (skip) lesions. Recent reports suggest a higher frequency of cancer recurrence in the remnant pancreas after surgical resection of IPMN. It is therefore important to precisely detect intraductal cancer extension and skip lesions when resecting IPMN.
Both intraoperative histologic examination of the surgical margin and cytologic examination of the pancreatic juice from each pancreatic segment (head, body, or tail) were performed on 43 IPMN patients. In addition to the preoperatively planned resection, 1 or 2 pancreatic segment(s) were additionally resected if the pancreatic juice tested positive in cytology. When a surgical margin was positive but the cytology in the remaining segment was negative, a subsegment (2-cm slice in width) was additionally resected until a negative margin was confirmed.
Twenty-five patients (58%) demonstrated negative results in both histology and cytology obtained from the segment(s) that were not initially intended to be removed. In contrast to the preoperative estimation, 5 patients were found to have a positive surgical margin and negative cytology, 5 patients demonstrated a negative surgical margin and positive cytology, and 8 patients demonstrated a positive surgical margin and positive cytology. Investigations of the resected specimens revealed that 8 patients (19%) had skip lesions in addition to the main lesion. Logistic regression analysis revealed that patients with a dilated main pancreatic duct, or those with cancerous lesions in the main tumors, were at high risk for positive histology and/or cytology.
Using intraoperative frozen-section histology and pancreatic juice cytology, 18 out of 43 patients in the current study (42%) required additional resection of the pancreas. A necessary and sufficient range of resection should be determined by intraoperative examination. Cancer 2006. © 2006 AmericanCancer Society.
Due to the recent advances in magnetic resonance pancreatography (MRP) and the widespread use of endoscopic retrograde pancreatography (ERP), it has become easier to delineate duct trees of the pancreas. Consequently, we have encountered an increasing number of intraductal papillary mucinous neoplasm (IPMN) cases in the last decade. This can be explained by the appearance of the tumor, which may display diffuse dilation of the main pancreatic ducts and/or cystic dilation of the branching ducts. Occasionally, the tumor reveals intraductal polypoid mass and copious mucin storing. The World Health Organization (WHO) has classified tumors histopathologically into the following 4 categories: slight dysplasia (intraductal papillary mucinous adenoma [IPMA]), moderate dysplasia (borderline malignancy [borderline IPMN]), severe dysplasia (intraductal papillary mucinous carcinoma in situ [noninvasive IPMC]), and invasive carcinoma (invasive IPMC).1 IPMC differs from the common type of invasive ductal carcinoma of the pancreas and has a lower potential for cell proliferation and invasion.2–5 However, IPMN is more likely than ductal adenocarcinoma to extend intraductally into the adjacent portions of the duct trees than to invade the pancreatic parenchyma.6–9
Accordingly, recent studies report cancer recurrence in patients who had IPMN extending into the margin of resection.6, 10 Therefore, when a curative resection is intended for patients with IPMN, surgeons need to be careful in determining the appropriate cut line of the pancreatectomy. Nonetheless, it is nearly impossible to predict the range of intraductal extension of neoplastic lesions beforehand by preoperative radiologic imaging or intraoperative macroscopic inspection.11 To determine an adequate cut line for the pancreatectomy of IPMN, an intraoperative frozen-section histology is needed.12–14
Several authors have proposed that IPMN has a multicentric origin, a concept based on reports of intrapancreatic cancer recurrence after surgical resection. This is true even of patients in whom the surgical margins of the resected pancreas were histologically confirmed as negative for cancer cells.4, 6, 15 Likewise, previous studies have reported that 18% to 21% of IPMN patients have skip lesions.16, 17 These findings suggest that a negative surgical margin does not always guarantee the complete resection of cancers. To solve this problem, an alternative method is needed to detect such lesions.
The current study used an intraoperative 2-segmental or 3-segmental cytology of the pancreatic juice, originally developed for pancreatic cancer in situ, in addition to a frozen-section histology of the surgical margin.18–20 In this procedure, we collected pancreatic juice from the pancreatic head, body, and tail with the aid of a balloon catheter. The resection was conducted in the area from which the cancer cells were obtained. The current study was intended to retrospectively clarify the usefulness of this method in detecting the intraductal cancer extension and skip lesions, and to clarify the risk factors for cancer residual if intraoperative histocytology had not been used.
MATERIALS AND METHODS
Between January 1990 and January 2005, 43 patients who were preoperatively diagnosed as having IPMN with no more than minimally invasive carcinoma underwent pancreatectomy at the Osaka Medical Center for Cancer and Cardiovascular Diseases. Patients with only IPMA or macroscopically invasive IPMC were excluded from this study. The study protocol was approved by the Human Ethics Review Committee of Osaka Medical Center for Cancer and Cardiovascular Diseases. A signed consent form was obtained from each patient.
The inclusion criteria in this study were 1) patients demonstrating either diffuse dilatation of the main pancreatic ducts or cystic dilatation of its branching ducts by MRP or ERP, and 2) patients whose pancreatic juice cytology collected during preoperative ERP were diagnosed as severe dysplastic or cancer cells. To exclude macroscopically invasive IPMC patients, the patients found to have detectable low-density mass lesions in the pancreatic parenchyma based on preoperative computed tomography (CT) or ultrasound were excluded. Histologic inspection of the resected specimens confirmed that all patients included had IPMC or borderline IPMN and that no patients had only IPMA or a mucinous cystic neoplasm. The study sample included 14 cases of invasive carcinomas, 16 cases of severe dysplasias (carcinoma in situ), and 13 cases of moderate dysplasias (borderline malignancy). Patients with borderline malignancy were included because IPMN with carcinoma in situ and borderline malignancy are difficult to distinguish. Before surgery, all patients received ERP to reveal either diffuse dilatation of the main pancreatic ducts (MPD type) or cystic dilatation of its branching ducts (Branch type). Patients who were found to have dilatation in both the main duct and in a branch duct were classified as having “Combined type.”
According to the location of either intraductal tumor mass or duct dilatation, each patient was preoperatively classified into 1 of the following 4 groups: head group (H) (n = 23), body group (B) (n = 7), head plus body group (H&B) (n = 5), and body plus tail group (B&T) (n = 8). During our study period, we encountered and excluded 3 patients whose entire pancreas (3 segments of the pancreatic head, body, and tail) had been involved by either diffuse or cystic changes, and thereby total pancreatectomy had been indicated without any intraoperative examinations.
During surgery, the pancreas was divided at a right angle in relation to the main pancreatic duct, either on the line between the head and body for the H, B, and B&T groups, or on the line between the body and tail for the H&B group. From the cut margin, a thin slice (5 mm in width) was removed for the intraoperative frozen-section histology. When the cut margin was diagnosed as having normal epithelia or IPMA (with no or only slight dysplasia) by the pathologist, the surgical margin was reported as negative. A positive diagnosis was made when moderate or severe dysplasia (borderline IPMN or IPMC) was proven.
In the cytologic examination of the pancreatic juice, either a single-lumen catheter or a triple-lumen catheter with a balloon (Boston Scientific, Natick, MA) was inserted into the main pancreatic duct. In the resection of the H, B, and B&T groups, a single-lumen catheter was inserted across the cut surface into the main pancreatic duct of the cranial pancreas and a triple-lumen catheter was inserted into the caudal pancreas (Fig. 1). Once the balloon of the catheter was inflated to block the pancreatic duct between the body and tail, it was then possible to collect the pancreatic juice separately from each portion of the pancreatic body and tail. For the H&B group, a single-lumen catheter was placed into the main pancreatic duct of the pancreatic tail and a triple-lumen catheter was inserted to divide the pancreatic head and body. Twenty-five units of secretin (Secrepan; Eisai, Tokyo, Japan) were intravenously administered to aid with the collection of the pancreatic juice from each segment of the pancreas. These samples were then used for the intraoperative cytodiagnosis. The cytodiagnosis was made based on the nuclear/cytoplasmic ratio, the shape of the nucleus, the amount and distribution of nuclear chromatin, and the status of nucleoli and cytoplasmic mucin, and the result was classified as either negative or positive (Fig. 2). The cytology samples were all read by our cytologist and the surgical resection margins were by our pathologist. They diagnosed independently and the diagnoses were reported directly to the operating room.
The extent of the pancreatectomy was determined by intraoperative histology and cytology. As a rule, segments with positive cytology were additionally resected. When the surgical margin was diagnosed as positive on histology but negative on cytology for the remaining segments, a subsegment (2-cm slice in width) was additionally resected until the new cut margin was confirmed as being histologically negative (Fig. 3). In cases with pancreaticoduodenectomy, a segmental resection of the pancreatic body, or the resection of the pancreatic head plus body, the cut end of the caudal pancreas was anastomosed into the posterior wall of the stomach with an invagination method. Postoperatively, both initially and additionally resected specimens were fixed in 10% formalin, sliced into 5-mm sections, and embedded in paraffin blocks. The 4-μm section from the face of each block was stained with hematoxylin and eosin for microscopic observation. Lastly, to determine whether cancer recurrence developed in the remnant pancreas, adjacent tissues, or other organs, all patients were followed every 3 to 6 months at the study site's outpatient clinic, with the use of chest X-ray, CT, or ultrasound of the abdomen; tumor marker; and physical examination.
The clinical characteristics of the 43 patients are shown in Table 1. Through the preoperative radiologic diagnosis, pancreaticoduodenectomy had been planned for 23 patients, segmental resection of the pancreatic body for 7 patients, pancreaticoduodenectomy plus pancreatic body resection for 5 patients, and body plus tail resection for 8 patients.
|Abdominal pain||15 (35)|
|Deterioration of diabetes||3 (7)|
|Back pain||1 (2)|
|Weight loss||1 (2)|
|Suspected segment(s) by preoperative imaging diagnosis|
|Head & body||5 (12)|
|Body & tail||8 (19)|
|Borderline (moderate dysplasia)||13 (30)|
|Carcinoma in situ (severe dysplasia)||16 (37)|
|Invasive carcinoma||14 (33)|
Intraoperative Histology/Cytology and Surgical Procedures
During surgery, all 43 patients were found to have positive cells in the pancreatic juice obtained from the segment(s) that were initially planned for resection. As expected, 25 patients (58%) patients (16 in the H group, 3 in the B group, and 6 in the B&T group; the negative group) were found to have negative results both in the histology of the surgical margin and in the cytology of the pancreatic juice obtained from segment(s) that were initially planned for preservation (Table 2). Consequently, none of these patients received additional pancreatic resection, resulting in 16 pancreaticoduodenectomies, 3 segmental resections of the pancreatic body, and 6 resections of the body and tail. In contrast to the preoperative estimation, the remaining 18 patients (42%) (7 in the H group, 4 in the B group, 5 in the H&B group, and 2 in the B&T group; the positive group) exhibited positive histology and/or positive cytology in the pancreatic juice from the remaining segments (Table 2).
|Preoperatively Suspected Segment(s)||Additional resection|
| His(−)/Cy(−)||Total|| His(+)/Cy(−)|| His(−)/Cy(+)|| His(+)/Cy(+)|
|Head & body||0||5||1||1||3|
|Body & tail||6||2||1||1||0|
|Total||25 (58%)||18 (42%)||5 (12%)||5 (12%)||8 (19%)|
Table 3 lists the correlations of the preoperative diagnosis of the tumor locations (planned operations) and all performed operations. In the H group, 7 patients required additional resection, including 3 patients who required a pancreaticoduodenectomy plus an additional 2-cm resection of the pancreatic body H+α, 1 patient who required a pancreaticoduodenectomy plus a pancreatic body resection H&B, and 3 patients who required entire pancreatectomies. All 3 patients who required entire pancreatectomies had positive cytology in pancreatic juices collected from both the body and the tail. In the B group, 2 patients underwent entire pancreatectomies because they demonstrated positive cytology in the pancreatic juice from both the head and tail. All 5 patients in the H&B group and 1 patient in the B&T group underwent entire pancreatectomies. In all, pancreaticoduodenectomies were performed in 16 patients, pancreaticoduodenectomy plus an additional 2-cm resection of the body was performed in 3 patients, segmental resection of the body was performed in 3 patients, pancreaticoduodenectomy plus segmental resection of the body was performed in 1 patient, resection of 2 segments of the body and tail were performed in 8 patients, resection of 2 segments plus a part of the head was performed in 1 patient, and an entire pancreatic resection was performed in 11 patients.
Detection of Skip Lesions
According to the intraoperative positive cytology, 5 patients underwent additional pancreatectomies of at least 1 pancreatic segment, even when the surgical margin was negative (Group 3 in Table 2). To clarify whether the additionally resected segments of the pancreas with a positive cytologic examination truly possessed skip lesions, the resected specimens of all 43 patients were sliced into 5-mm sections and investigated histologically. Thorough investigation revealed that 8 patients (19%) had skip lesions; all 5 patients who demonstrated positive cytology and negative surgical margins had skip lesions in additionally resected specimens (Fig. 4), and 3 patients who were found to have positive cytology and positive surgical margins also were found to have skip lesions in the additionally resected specimens. Among 8 skip lesions, 1 lesion was diagnosed as moderate dysplasia and remainder were diagnosed as severe dysplasia. These skip lesions were found in the main pancreatic ducts in 3 patients and another 5 were found in their branch ducts.
Background Factors Compared between Positive and Negative Groups
It is important to distinguish the high-risk patients who required additional resection of the pancreas. A comparison of background factors between the positive group (n = 18) and the negative group (n = 25) is shown in Table 4. Dilatation of the main pancreatic duct was seen in 17 of 18 positive patients (94%), which was significantly higher than the 11 of 25 patients (44%) observed in the negative group (P = .0006). In the positive group, 17 patients (94%) had carcinoma (IPMC) with or without parenchymal invasion in the main lesion, whereas 13 patients (52%) in the negative group had carcinoma (IPMC) with or without parenchymal invasion in the main lesion (P = .003). The number of patients who had preoperative symptoms was slightly lower in the positive group (P = .04). Logistic regression analysis revealed that the dilation of MPD and the histology of the main lesion (carcinoma) were independent risk factors for the positive group (Table 5).
|Age, y (±SD)||66.0 ± 6.9||63.8 ± 7.8||.35|
|Location of suspected segment(s)|
|Head/head and body||12||16||.09|
|Body/body and tail||6||9|
|No. of suspected segments|
|Dilatation of MPD|
|Carcinoma in situ/invasive carcinoma||17||13|
|λ2||P||Hazards ratio||95% CI|
|Type (with dilated MPD)||8.64||.003*||40.1||3.42–470.7|
|Presence of symptoms||2.97||.085||5.4||0.79–36.4|
Histologic examination revealed that 8 patients had skip lesions, as described above. These 8 patients included 5 MPD-type patients, 2 Combined-type patients, and 1 Branch-type patient. These numbers represent 33% of MPD patients, 15% of Combined patients, and 7% of Branch patients, respectively.
Postoperative Histology and Follow-Up Data
All patients survived surgery and were followed for 57.6 ± 44.8 postoperative months. The 5-year recurrence-free survival rate was 54% in the invasive carcinoma group (n = 14), whereas it was 100% in the carcinoma in situ and the borderline groups (n = 29) (Fig. 5A). Of 14 patients with invasive carcinoma, 3 were histologically proven to have minimally invasive lesions and demonstrated no recurrences during their follow-up periods. The remaining 11 patients had macroscopically invasive (more than minimally invasive) lesions, including 7 ductal carcinomas and 4 noncystic mucinous carcinomas.2, 7 Among these 11 patients with macroscopic invasion, the recurrences presented as metastatic diseases in 6 patients, including 3 in the adjacent tissues, 2 in the liver, and 1 in the peritoneal dissemination. These recurrences were observed within 2 postoperative years. No recurrences in the remnant pancreas were observed in any patients in this study.
The postoperative survival of each type is shown in Figure 5B. No significant differences were observed in the survival rates between the MPD-type, Branch duct-type, or Combined-type patients.
In the current study, we enrolled patients who were preoperatively diagnosed as having IPMN with no more than minimally invasive carcinoma. This inclusion criterion was established because the prognoses of patients with invasive carcinoma do not depend on recurrences in the remnant pancreas, but rather on metastatic diseases in the adjacent tissue, lymph node, or liver. Our 5-year survival rate was slightly better than previously reported,2, 4, 6 even in patients with invasive IPMC, and may be a result of this inclusion criterion. It also may be an additional reason why our intraoperative histocytology reduced the recurrences in the remnant pancreas.
Previous reports concerning the adequate range of surgical resection of IPMN revealed that 36% to 50% of the surgical margins were histologically positive,2, 10 suggesting the difficulty in determining adequate cut lines for IPMN. More important, IPMN is believed to have a multicentric origin and intrapancreatic recurrences were observed after the surgical resection of IPMN in patients with negative surgical margins. These reports indicate that preoperative imaging techniques are insufficient in detecting the microscopic spread of cancerous lesions or the existence of skip lesions.
In the current study, we therefore employed intraoperative investigations to determine suitable cut lines. Intraoperative cytodiagnosis has been reported to be useful even for occult neoplasms in common types of pancreatic cancer.18–20 Using this method, we found that 8 patients (19%) had skip lesions that could potentially be missed because they were located in the remnant pancreas, which preoperatively had been planned to be reserved. Using both histologic and cytologic examinations, we determined that 42% of all patients required additional resection. It is impossible to know whether recurrences occurred in those patients in whom the additional surgical resection had not been performed; however, the intrapancreatic recurrence rate in our study was extremely low compared with that of previously reported studies. This suggests that at least some of the skip lesions detected were precursors of clinically detectable lesions.
It is still unclear how many years are needed for the development of clinically visible tumors from a microscopic lesion. Salvia et al.3 reported that 5% of patients who underwent resection of main ductal IPMN presented with recurrence in the remnant pancreas, at a median of 50 months after their initial surgery. Given that the chance of recurrence in the remnant pancreas is acceptably low and that the duration of tumor development is more than several years, avoiding entire pancreatectomy, especially for older patients, might be a good idea even when the surgical margin contains a dysplastic lesion or carcinoma in situ. However, a better strategy would be to find patients who have a high risk of recurrence in the remnant pancreas, utilizing the intraoperative histology and cytology procedures described in the current. Through the identification of high-risk patients using intraoperative investigations, entire pancreatectomies were performed in 24% of patients (a rate that is as low as the rate of 10%–23% previously reported2–4, 6, 8). As a result, no subsequent recurrences in the remnant pancreas were observed in any patients in the current study.
One of the final objectives of the current study was to sufficiently identity high-risk patients who develop recurrences in the remnant pancreas. Because intraoperative cytodiagnosis is available only at hospitals in which experienced cytologists can make a precise diagnosis during surgery, surgeons should be aware of patients who have a high risk for residual tumor if intraoperative cytodiagnosis is not available.
This study's logistic regression model revealed that the patients who needed additional resection of the pancreas were those with a dilated main pancreatic duct, along with the presence of cancerous lesions. In other words, IPMC patients with dilated MPD have a higher risk for developing recurrence in the remnant pancreas if the intraoperative examination is not available and additional resection is not performed. The current study results also indicate that patients with MPD dilatation had accompanying skip lesions slightly more frequently than those without MPD dilation. Although several previous reports revealed that pancreatic intraepithelia neoplasia (PanIN) lesions can be found in patients with IPMN, it is still unclear whether this kind of lesion can be referred to as a skip lesion of IPMN or whether it should be referred to as an accompanying PanIN lesion.21, 22 Regardless, these skip lesions are precursors of pancreatic cancer and patients should undergo resection when they are detected. Interestingly, a univariate analysis of our study data indicated that patients who did not require additional resection (the negative group) had a higher incidence of accompanying symptoms. However, this was not a significant finding of a multivariate analysis of our study data. To date, we do not fully understand this paradoxical finding. However, given that Salvia et al.3 reported a higher incidence of accompanying symptoms (abdominal pain) in patients with benign IPMN, this may indicate the difficulties in finding adequate surgical procedures that are determined only by the patient's symptoms.
Through intraoperative examination, we determined that approximately half of the patients in the current study required additional resection of the pancreas and that continuous or skip lesions could be detected in the additionally resected specimens. Risk factors for residual tumor are dilatation of the main pancreatic duct and the presence of a cancerous lesion. Ultimately, surgeons should avoid leaving tumors in the remnant pancreas when treating patients with IPMC given the favorable prognosis of the disease and the ability to perform a curative resection, especially in patients with relatively early-stage IPMC.
- 1Intraductal papillary-mucinous neoplasms of the pancreas. In: HamiltonSR, AaltonenLA, editors. World Health Organization Classification of Tumours. Pathology and Genetics of Tumours of the Digestive System. Lyon: IARC Press; 2000: 237–240., , , .
- 13Surgical management of intraductal papillary mucinous tumors of the pancreas: the role of routine frozen section of the surgical margin, intraoperative endoscopic staged biopsies of the Wirsung duct, and pancreaticogastric anastomosis. Arch Surg. 2001; 136: 1256–1262., , , et al.