Sa1550 L1CAM Expression Correlates With Metastatic Spread on Endoscopic Ultrasound - Guided - Fine - Needle Aspiration of Pancreatic Cancers Sun-Chuan Dai* 1,2 , ADAM Weinberg 1,2 , Shelly Nigam 2 , Ashish Sharma 2 , Qin Huang 2 , Hiroshi Mashimo 3,2 1 Boston University Medical Center, Boston, MA; 2 Gastroenterology, Boston Veterans Affairs Healthcare System, West Roxbury, MA; 3 Harvard Medical School, Boston, MA Background: Pancreatic cancers carry a poor prognosis and are the 4th leading cause of cancer deaths in the United States. At time of diagnosis these cancers may already have lymph node involvement or metastatic spread, precluding the possibility of curative surgery and ensuring poorer prognosis. The 5-year survival rate of resectable pancreatic cancer is approximately 20% compared to less than 5% in unresectable cancers. L1 cell adhesion molecule (L1CAM) is a transmembrane glycoprotein that is not expressed in normal pancreas, but its expression in surgically-resected pancreatic cancers has been shown to correlate with prognosis, lymph node involvement, and metastatic disease. Prognostic markers of pancreatic cancer found before surgical resection with endoscopic ultrasound-guided-fine -needle aspiration (EUS-FNA) may guide patient treatment. This study aims to explore the clinical significance of L1CAM in pancreatic cancers diagnosed by EUS-FNA. Methods: Pathology specimens from 13 consecutive patients diagnosed with pancreatic cancer by EUS-FNA between 2010-2011 were retrieved for immunohistochemical staining of L1CAM (Abcam, Cambridge, MA) through retrospective review of patient electronic medical records. Records were re-examined at the end of 2011 to confirm these diagnoses were correct. The immunostain of formalin-fixed, paraffin-embedded tissues was performed according to recommendations by the L1CAM antibody manufacturer, including overnight incubation at a 1:50 dilution at 4 degrees Celsius, final development with 3,3’ diaminobenzidine substrate (Abcam), and counterstain using hematoxylin (Sigma-Aldrich, St. Louis, MO). Appropriate positive and negative controls were conducted simultaneously. Once stained, the slides were scored by a blinded experienced gastrointestinal pathologist. Expression was considered when immunoreactivity of neoplastic cells was detected in over 10% of the total number of estimated neoplastic cells. The clinical course of each patient was retrospectively examined for lymph node involvement and presence of metastasis detected by imaging, endoscopy, or surgery performed within one month from time of initial diagnosis. Results: L1CAM expression was identified in 5 patients and absent in 8 patients (Table 1). At time of diagnosis 3 patients with L1CAM expression had lymph node involvement and 4 had metastatic spread. All 5 patients had either lymph node involvement or metastasis. In the group with absent L1CAM expression no patients had lymph node involvement and only 1 patient had metastasis. Conclusion: L1CAM expression in pancreatic cancers diagnosed by EUS-FNA correlates with lymph node involvement and metastasis at the time of diagnosis, and may serve as a clinical prognosticator. Future large prospective studies are warranted to evaluate L1CAM’s role in management of pancreatic cancers. Results: L1CAM Status, Lymph Node Involvement, and Metastasis. Table 1. Lymph Node Involvement Metastasis Lymph Node Involvement and/or Metastasis L1CAM Expression (n = 5) 3 4 5 L1CAM Absent (n = 8) 0 1 1 Sa1551 The Diagnostic Value of Repeat Endoscopic Ultrasound-Guided Fine Needle Aspiration (EUS-FNA) for Solid Pancreatic Lesions: A Tertiary Referral Center Experience Rei Suzuki*, Manoop S. Bhutani, Srinivas Ramireddy, Somashekar G. Krishna, Brian R. Weston, William A. Ross, Jeffrey H. Lee GHN, MD Anderson Cancer Center, Houston, TX Background: EUS-FNA is increasingly recognized as an indispensable tool for the diagnosis and staging of solid pancreatic lesions (SPLs). However, initial attempt at EUS-FNA can be non-diagnostic. The aim of this study was to evaluate the diagnostic value of repeat EUS-FNA for SPL by expert endosonographers at a tertiary medical center. Methods: We retrospectively analyzed subjects who underwent repeat EUS-FNA for SPLs between 2006 and 2011 in UT MD Anderson Cancer Center (MDA). Among them, we selected the cases with prior EUS-FNA results as either non-diagnostic, negative for malignancy, atypical cells and suspicious for malignancy. We excluded cases which were diagnosed as cystic lesions. Main outcome measurement was diagnostic EUS-FNA specimen. Results: A total of 140 patients underwent repeat EUS-FNA out of which we identified fifty patients with SPLs meeting the above criteria. Average age was 63.8 years and 56.0 % of them were males (28/50). Final diagnosis was adenocarcinoma in 32, pancreatic neuroendocrine tumor (PNET) in 5, metastasis in 2 (melanoma, adrenal cortical carcinoma) and lymphoma in 1. In 10 cases, EUS-FNA was negative for malignancy. EUS-FNA was performed for pancreas only in 41, pancreas with lymph node in 9. Among them who underwent EUS- FNA for lymph node at the same time, 3 patients with PNET were positive for metastasis. The overall clinical impact of repeat EUS FNA was 80.0 % (40/50) by providing a conclusive diagnosis. The mean interval between outside EUS-FNA and EUS-FNA in MDA was 48 days. Comparing outside EUS-FNA to EUS-FNA in MDA, there were no statistical differences in size (29.5mm vs. 30.2 mm, P- value=0.73) and number of needles passes (4.0 vs. 3.4, P-value=0.20). No complications were seen in all cases. Conclusion: A repeat EUS-FNA for SPLs at a tertiary-referral center had a clinical impact in 80.0 % of patients when performed by experts for a similar clinical indication. Sa1552 A Novel Technique Using a Structure-to-Probe (STP) Ratio to Do Post-Hoc Measurements of Stored EUS Images Brian A. Brunson*, William F. Marsteller, Brenda J. Hoffman, Robert H. Hawes, Joseph Romagnuolo Medicine/Gastroenterology, Medical University of South Carolina, Charleston, SC Background: EUS images are increasingly utilized in multidisciplinary settings such as tumor board; new (re-)measurements are sometimes needed. Static images or cine clips can be stored in PACS (picture archiving and communication system), but secondary measurements of structures can then only be done in “pixels”. As images often vary in magnification, it is difficult to standardize a pixels:mm calibration. We aimed to pilot a novel technique for calculating measurements of structures in static EUS images, using the probe in the image as a reference, and assess robustness across varying magnifications. Methods: 40 consecutive EUS’s with a stored static image with a measured structure were included. 20 images were a “calibration” group, 20 a “test” group. Low/high magnification images and radial/linear probe cases were divided equally between the groups. In the calibration group, structures measured with calipers (in mm) at EUS were re-measured (in pixels) on the PACS image; a pixel-to-mm ratio was calculated. The probe size was measured (in pixels), and using that pixel:mm ratio, the size of the probe was back-calculated and averaged. This was repeated with a second reviewer. Calculated probe measurement consistency across reviewers (noting magnifications used) was assessed. In the test group, 2 reviewers, blinded to a structure’s recorded measurements, measured the probe and an unlabelled calipered structure in pixels. This resulted in a structure:probe (STP) ratio, which was then multiplied by the calculated mean probe size (from the calibration group), to arrive at an estimated measurement of the calipered structure. The structure’s original caliper measurement(in mm) was used as a reference standard to assess the technique’s accuracy. Results: In the calibration group, calculated mean probe size was 11.05 mm (radial) and 10.40 mm (linear). The mean absolute discrepancy between reviewer 1 and reviewer 2 was 0.31 mm (range -0.49 to +1.01 mm), or 3.5% (0.2%-9.1%). This precision occurred despite a wide range (50-163 pixels) in original probe measurement,due to variable magnification. In the test group, mean recorded structure caliper measurement was 15.87 mm (reference). Reviewer 1’s blind mean back-calculated structure measurement (using probe as reference) was 15.89 mm: absolute discrepancy 0.174 mm (range -0.37 to +0.46; +/- 2.9%); mean relative discrepancy 1.3%. For reviewer 2, mean calculated structure measurement was 16.25 mm; absolute discrepancy of 0.38 mm (range -0.05 to 1.32; +/- 8%), or 2.7%. Conclusion: In this study, we piloted a novel technique of post-hoc calculated measurements of structures in saved EUS PACS images using the STP ratio, and preliminarily validated its high accuracy. This technique will allow real-time new (re-)measurements of saved pixel-based images in multi-disciplinary settings such as tumor board. Sa1553 Development of Endoscopic Ultrasound-Guided Fine-Needle Aspirate (EUS-FNA) Derived Pancreatic Cancer Cell Lines From Advanced Stage Cancer Shiro Urayama* 1 , Ramez Saroufeem 2 1 Internal Medicine, University of California, Davis, Sacramento, CA; 2 Pathology, University of California, Davis, Sacramento, CA Background: The investigation of primary pancreatic cancer cells are becoming more critical in studying various solid cancers which are comprised of heterogeneous groups of cells including the tumor initiating or cancer stem cells. As only 10 to 15% of all pancreatic cancers (PDAC) are operable due to the advanced stage discovery, endoscopic ultrasound (EUS) provides as a unique tool to acquire tissue specimen for developing primary cell cultures. With our success from previous pilot study in culturing primary cells of PDAC from EUS- Abstracts www.giejournal.org Volume 75, No. 4S : 2012 GASTROINTESTINAL ENDOSCOPY AB199