Kaplan-Meier survival curves comparing patients with noninvasive intraductal papillary mucinous neoplasm (IPMN) (n = 43) and invasive carcinoma derived from IPMN (invasive IPMC) (n = 11).
Customize your JAMA Network experience by selecting one or more topics from the list below.
Hirono S, Tani M, Kawai M, et al. Treatment Strategy for Intraductal Papillary Mucinous Neoplasm of the Pancreas Based on Malignant Predictive Factors. Arch Surg. 2009;144(4):345–349. doi:10.1001/archsurg.2009.2
Noninvasive intraductal papillary mucinous neoplasms (IPMNs) have a favorable prognosis; however, the prognosis of invasive intraductal papillary mucinous carcinoma (invasive IPMC) is poor. Identification of predictive factors for differentiating IPMC from benign IPMNs would assist in providing appropriate treatment.
Retrospective study (1999-2006).
Wakayama Medical University Hospital, Wakayama, Japan.
Fifty-four patients with IPMN who underwent surgery; histologic examination showed benign adenomas in 29, carcinoma in situ in 14, and invasive carcinoma in 11 patients.
Main Outcome Measures
Clinical data, preoperative imaging findings, cytologic findings, tumor markers in serum and pancreatic juice, and overall survival.
Age of 70 years or older, presence of mural nodules, mural nodule size of 5 mm or larger, and carcinoembryonic antigen (CEA) level in pancreatic juice of 110 ng/mL or higher (as obtained by preoperative endoscopic retrograde pancreatography) were predictive of a malignant IPMN by univariate analysis, and a CEA level of 110 ng/mL or higher in pancreatic juice was identified as the only independent predictive factor for the malignant entity. The presence of jaundice or body weight loss, main pancreatic duct type, presence of mural nodules, mural nodule size of 5 mm or larger, and CEA level in the pancreatic juice of 110 ng/mL or higher were all predictive of invasive IPMCs by univariate analysis.
Measurement of the CEA level in pancreatic juice should be considered in the diagnosis of IPMC.
Intraductal papillary mucinous neoplasm (IPMN) of the pancreas was first described by Ohashi et al in 1982.1 Since the World Health Organization clarified the nomenclature and pathological characteristics of this tumor,2 it has been recognized increasingly often and has been reported to account for 17% to 25% of surgically resected pancreatic neoplasms.3-5The IPMNs show a wide spectrum of histologic characteristics, ranging from hyperplasia to invasive intraductal papillary mucinous carcinoma (invasive IPMC), with different degrees of aggressiveness.6-8 In comparison with common ductal pancreatic adenocarcinoma, IPMNs grow slowly and have a low malignancy rate. Therefore, benign IPMNs may be carefully followed up without surgical treatment. However, IPMNs sometimes show invasive proliferation, and patients with invasive IPMCs, like those with common ductal pancreatic adenocarcinoma, frequently have a poor outcome,6,9,10 thus indicating that the treatment recommendations for an invasive IPMC may be the same as those for a common ductal adenocarcinoma. Therefore, an accurate assessment of malignancy is required for the appropriate management of IPMNs, although a clear distinction remains difficult between benign and malignant IPMNs, even with modern imaging.11-13
We retrospectively analyzed the clinical data, imaging findings, cytologic findings, and a tumor marker in the pancreatic juice and, moreover, defined the clinical and pathological factors associated with the outcome by a survival analysis. The aim of the present study was to determine the predictive factors of malignancy in IPMNs to provide a strategy for their treatment.
Between July 1, 1999, and January 31, 2006, 54 patients with IPMN were surgically treated at Wakayama Medical University Hospital. There were 31 men and 23 women, with a mean age of 69 years (range, 44-81 years). All 54 patients were examined for tumor location, tumor size, maximum diameter of the main pancreatic duct (MPD), and presence of a mural nodule by preoperative ultrasonography, contrast-enhanced computed tomography, endoscopic ultrasonography, and endoscopic retrograde pancreatography.
The pancreatic juice was collected by endoscopic retrograde pancreatography with brushing in the pancreatic duct, following an intravenous injection of 50 U of secretin (Eisai Co Ltd, Tokyo, Japan). The collection of pancreatic juice was performed by one expert in endoscopic retrograde pancreatography (M.K.) in all patients. The pancreatic juice was divided into 2 portions, one for cytologic examination and another for measurement of carcinoembryonic antigen (CEA) levels by means of a CEA immunometric chemiluminescent assay kit (Bayer Medical Co, Tokyo, Japan). The cutoff level of CEA in the pancreatic juice required to detect a malignant IPMN was 110 ng/mL (to convert to micrograms per liter, multiply by 1.0), determined by constructing receiver operating characteristic curves as described previously.13
Of the 54 IPMNs, histologic examination showed 25 to be malignant, consisting of 14 cases of carcinoma in situ (5 MPD type, 4 combined type, and 5 branch type) and 11 invasive adenocarcinomas (5 MPD type and 6 combined type). The 29 benign IPMNs were all adenomas (2 MPD type, 13 combined type, and 14 branch type), according to the classification established by the World Health Organization. Pathological tumor staging of the invasive IPMCs was done by the TNM classification system of malignant tumors from the International Union Against Cancer.14 We retrospectively analyzed the clinical findings, preoperative imaging diagnostic data, and laboratory data to predict the malignant potential of IPMNs and the invasive property of IPMC.
Postoperative laboratory examinations and abdominal imaging (ultrasonography or computed tomography) were performed every 3 months after surgery. Recurrence was defined as the presence of recurrent IPMNs in the pancreatic remnant or as local, diffuse, or distant metastatic disease determined by imaging examinations in the follow-up period after resection.
The 15 potential risk factors listed in Table 1 were assessed by a univariate analysis with the χ2 test or Fisher exact test. The significant predictors of malignant IPMN in the univariate analysis were then included in a forward stepwise multiple logistic regression model. The cumulative overall survival rates were calculated by the Kaplan-Meier method, and statistical analysis was performed by log-rank test. Statistical significance was defined as P < .05.
The 4 factors found to be predictive of malignancy in the univariate analysis were age of 70 years or older (P < .001), presence of mural nodules (P = .001), mural nodule size of 5 mm or larger (P < .001), and a CEA level in pancreatic juice of 110 ng/mL or higher (P < .001) (Table 1). These predictive factors of malignant IPMNs identified in the univariate analysis were analyzed in a multivariate analysis. Only 1 factor, CEA level in the pancreatic juice, was found to be independently associated as a predictive factor for malignancy (odds ratio, 29.8; 95% confidence interval, 2.8-313.9). The P values for the 4 factors in the multivariate analysis were .39, .94, .70, and .005, respectively.
A univariate analysis demonstrated 6 factors to be significantly predictive of invasive IPMC: jaundice (P = .007), body weight loss (P = .007), main duct or combined type (P = .01), mural nodules (P = .005), mural nodule size of 5 mm or larger (P = .005), and a CEA level in the pancreatic juice of 110 ng/mL or higher (P = .02) (Table 1).
Pancreaticoduodenectomy was the most common operation and was performed in 34 patients (63%); in 29 of these patients, pylorus-preserving methods were used.15 A distal pancreatectomy was performed in 7 patients (13%), central pancreatectomy in 8 (15%), and total pancreatectomy in 5 (9%). A portal vein resection was performed in 2 patients (4%) and a gastrectomy in 4 (7%) because of invasion to the stomach and accompanying gastric cancer.
The median follow-up for all 54 patients who underwent resection was 47 months (range, 3-101 months). Table 2 demonstrates the clinical and pathological factors associated with the prognostic outcome in the 54 patients with complete resections. Of the clinical factors analyzed, the presence of positive symptoms was the factor significantly associated with a poor survival (P = .04). Of the pathological factors analyzed, the presence of a cancer component, found in 25 cases (46%), was a predictor of the prognostic outcome (P = .004), and the presence of an invasive component, found in 11 cases (20%), was a strong predictor of shorter survival (P < .001) (Figure). Except for 2 patients who died of other diseases (pneumonia and hypoglycemia), all the patients who had no invasive component were alive without evidence of disease at last follow-up, whereas the median survival of the 11 patients with an invasive component was 15 months (Table 3).
Except for 2 patients with minimally invasive IPMCs, the 11 patients with invasive IPMCs included 4 with stage IIA and 5 with stage IV, according to the TNM classification. Lymph node metastasis outside the regional area was recognized in 5 patients (45%) with invasive IPMC, and vascular invasion in 2 (18%). The median survival of the 5 patients with involved lymph nodes was 13 months. The 2 patients with vascular invasion both died of their disease within 1 year (Table 3).
Of the 54 patients with IPMN, there were 8 disease recurrences (15%), all of which were in patients with an invasive IPMC. Six of 8 patients with a recurrence had disease-related death. Three of the recurrences occurred in the first year after surgery, and all except 1 occurred within 2 years. There were 2 patients with a distant recurrence alone (25%) (Table 3). In 1 patient (13%), a diffuse intra-abdominal recurrence occurred in conjunction with a distant metastasis. There were 4 cases (50%) of locoregional nodal recurrence. There was 1 pancreatic stump recurrence (13%). That patient had minimal invasion with a negative margin and had a recurrence in the pancreatic stump. The patient's tumor was re-resected 35 months after the first surgery, and he was alive without evidence of disease 71 months after the first resection (Table 3).
The surgical outcome of IPMNs has been widely accepted to be better than that of ductal adenocarcinoma of the pancreas.10,16-18 The possible reasons for a favorable prognosis of IPMNs might be their indolent biological behavior, including growing slowly and metastasizing late, and early diagnosis and detection using advanced imaging techniques. However, IPMNs sometimes show invasive proliferation and progress rapidly with a poor prognosis.6,9,10 Therefore, it is essential to select the appropriate treatment strategy for IPMNs by accurately differentiating malignant from benign IPMNs.
In this study, we predicted 4 factors for the preoperative diagnosis of malignant IPMNs, which were age of 70 years or older, the presence of mural nodules, mural nodule size of 5 mm or larger, and a CEA level in the pancreatic juice of 110 ng/mL or higher. These results were similar to those of various previous reports,11-13,18 except for the CEA level in the pancreatic juice. Moreover, a CEA level in the pancreatic juice of 110 ng/mL or higher was the only useful predictive factor by a multivariate analysis, and the accuracy for differentiating malignant from benign IPMNs was 84%.
The presence of symptoms, a cancer component, and an invasive component were found to be the significant factors influencing the survival of patients with IPMNs in our study. Among them, the presence of an invasive component was the most serious prognostic factor (P < .001), indicating that, once an invasive IPMC develops, the treatment required may be the same as that for a common ductal adenocarcinoma of the pancreas, such as an extended operation including a lymph node dissection and adjuvant chemotherapy, although further studies of patients with invasive IPMC are still necessary.
Some researchers have reported that resection should be performed whenever IPMNs are diagnosed, even if they are considered to be benign by the imaging findings,10 whereas others reported that no enlargement of the IPMNs occurred during several years of follow-up19,20 and long-term survival could be expected in some patients without a resection.21 Our results suggest that benign IPMNs should be strictly followed up without resection because IPMNs generally progress slowly, occur in elderly patients, and sometimes occur multifocally, thus leading to an extensive resection and higher postoperative morbidity. However, malignant IPMNs should be resected because, once they show invasive proliferation, they become aggressive and carry a poor prognosis. Our criteria for surgical resection of IPMN are as follows: (1) presence of symptoms, (2) main duct type or combined type, (3) presence of mural nodules of 5 mm or larger, (4) cytologic findings of class IV or V in the pancreatic juice, and (5) CEA concentration in the pancreatic juice of 110 ng/mL or higher. We have been strictly following up 57 patients with IPMNs diagnosed as benign who did not initially undergo surgery. So far, 3 patients later underwent resection because they met the criteria for surgery during the follow-up period; none of the 3 patients had invasive IPMC, 2 had carcinoma in situ, and 1 had adenoma.
Our results strongly suggest that measurement of the CEA level in the pancreatic juice is a useful diagnostic method to distinguish malignant from benign IPMNs.
Correspondence: Hiroki Yamaue, MD, Second Department of Surgery, Wakayama Medical University, School of Medicine, 811-1 Kimiidera, 641-8510 Wakayama, Japan (firstname.lastname@example.org).
Accepted for Publication: April 7, 2008.
Author Contributions:Study concept and design: Hirono and Yamaue. Acquisition of data: Hirono, Ina, Nishioka, Miyazawa, and Fujita. Analysis and interpretation of data: Hirono, Tani, Kawai, and Uchiyama. Drafting of the manuscript: Hirono, Tani, Kawai, Ina, Nishioka, Miyazawa, Fujita, and Uchiyama. Critical revision of the manuscript for important intellectual content: Yamaue. Statistical analysis: Tani and Yamaue. Obtained funding: Not applicable. Administrative, technical, and material support: Tani, Kawai, Uchiyama, and Yamaue. Study supervision: Yamaue.
Financial Disclosure: None reported.