Factors in Intraductal Papillary Mucinous Neoplasms of the Pancreas Predictive of Lymph Node Metastasis

Background: Little is known about the frequency of lymph node metastasis (LNM) in intraductal papillary mucinous neoplasms (IPMNs), and we have not been able to determine how much lymph node dissection is necessary in individual cases. The aim of this study was to investigate the predictive factors for the LNM in IPMNs. Methods: Medical records of 120 patients pathologically diagnosed as having IPMN were reviewed, and 16 possible predictive factors regarding the LNM were analyzed. Results: LNM was observed in 7 patients (6%), all of whom were diagnosed as having mural nodules preoperatively. Sensitivity, specificity, and accuracy of preoperative imaging for detecting mural nodules of IPMNs in this study were 84, 97, and 90%, respectively. Univariate analysis using 61 patients having mural nodules preoperatively revealed that the size of mural nodules 6 10 mm and positive imaging findings for invasive tumor and possible LNM were significant predictive factors for the LNM. Multivariate analysis demonstrated that only an imaging finding for invasive tumor was an independent significant predictive factor. Positive and negative predictive values of the imaging finding of invasive IPMNs for LNM were 50 and 98%, respectively. Conclusions: Standard lymph node dissection Received: March 11, 2010 Accepted after revision: August 19, 2010 Published online: January 18, 2011 Masao Tanaka, MD, PhD, FACS Department of Surgery and Oncology, Graduate School of Medical Sciences Kyushu University, 3-1-1 Maidashi, Higashi-ku, Fukuoka 812-8582 (Japan) Tel. +81 92 642 5441, Fax +81 92 642 5458 E-Mail masaotan @ med.kyushu-u.ac.jp © 2011 S. Karger AG, Basel and IAP 1424–3903/10/0106–0720$26.00/0 Accessible online at: www.karger.com/pan D ow nl oa de d by : 54 .7 0. 40 .1 1 10 /6 /2 01 7 5: 17 :1 0 A M Lymph Node Metastasis in IPMN Pancreatology 2010;10:720–725 721 Patients and Methods Medical records of 143 patients, who underwent pancreatectomy for IPMN at the Department of Surgery and Oncology, Kyushu University Hospital between January 1990 and June 2009, were retrospectively reviewed. As the preoperative radiologic/ultrasonic examinations, contrast-enhanced computed tomography, magnetic resonance imaging/cholangiopancreatography, percutaneous ultrasonography (US), and endoscopic retrograde pancreatography were routinely performed. Endoscopic US (EUS), intraductal US, and cytologic examination of the pancreatic juice obtained during endoscopic retrograde pancreatography were also assessed, if applicable. IPMN was then classified into two subtypes: main duct type and branch duct type, as described in our previous report [2] . The presence or absence of mural nodules and the size of the cyst and mural nodules were basically examined by US and/or EUS. Invasion of the tumor into the pancreatic parenchyma or adjacent organ and LNM was also assessed by computed tomography, US, and/or EUS. Invasion of the tumor was suspected when radiologic/ultrasonic findings demonstrated that the mural nodule destroyed the cyst wall and invasion of such organs as the pancreatic parenchyma, retroperitoneum, bile duct, portal vein, artery, stomach or duodenum was observed. A suspected metastatic lymph node on radiologic images was defined as a round-shaped node 6 10 mm in size. Lymph node dissection of stations 1 and 2 (D2) according to the Japanese general rules for the study of pancreatic cancer [3] is a standard procedure for invasive ductal carcinoma in Japan. Our usual strategy of node dissection for IPMNs is as follows: in patients diagnosed as having mural nodules preoperatively, pancreatectomy with various degrees of lymph node dissection was performed (D1–D2), while in those without mural nodules, node dissection is diminished to D0–D1, depending on the suspicion of malignancy, i.e. D0 for obviously benign lesions to be resected to relieve symptoms due to acute pancreatitis or D1 for a 6 3 cm IPMN to be resected as possible malignancy. Pathologic diagnosis of the IPMN was based on World Health Organization criteria, i.e. intraductal papillary mucinous adenoma (IPMA), borderline lesion (IPMB), and carcinoma (IPMC) [4] . The lymph node station was defined according to the Japanese general rules [3] . To determine the quality of preoperative radiologic examinations in terms of the detection of real mural nodules, sensitivity, specificity, and accuracy at our institution were calculated. Possible factors predictive of LNM included gender, age ̂ 65 and 1 65, morphologic type (main duct or branch duct), tumor location (pancreatic head or body and tail), presence or absence of mural nodules, size of a mural nodule ! 10 or 6 10 mm, radiologic findings suggesting invasive carcinoma (invasion to pancreatic parenchyma or adjacent organs) and possible LNM, cyst size of branch duct IPMNs ! 30 or 6 30 mm, size of the main pancreatic duct in main duct IPMNs ! 20 or 6 20 mm, presence of symptoms, history of acute pancreatitis, recent deterioration of diabetes mellitus, presence of atypical cells in the pancreatic juice (class IV or V), elevations of tumor markers such as carcinoembryonic antigen (CEA, normal limit ! 2.3 ng/ml at our institution) and carbohydrate antigen 19-9 (CA19-9, normal limit ! 37 ng/ml). Positive and negative predictive values of significant predictive factors for LNM were also calculated. Comparisons between the two groups were performed by the 2 test. A multivariate logistic regression model was used to determine the effects of possible predictive factors on LNM in IPMNs with mural nodules. A p value ! 0.05 was considered to be statistically significant.