Abstracts Abstracts

ABSTRACTSABSTRACTS S1

Am J Gastroenterol 2017; 112:S1–S45; doi:10.1038/ajg.2017.295

ACCEPTED: BILIARY/PANCREAS

Enhanced Recovery in Acute Pancreatitis (RAPTor): A Randomized Controlled Trial

2017 Fellows-in-Training Award (Biliary/Pancreas Category)

Elizabeth Dong, MD1, Jonathan I. Chang, MD1, Dhruv Verma, MD1, Michael Batech, DrPH2, Cecia Villarin1, Karl K. Kwok, MD1, WanSu Chen, PhD3, Bechien U. Wu, MD, MPH1. 1Kaiser Permanente Los Angeles Medical Center, Los Angeles, CA; 2Kaiser Permanente Division of Research, Pasadena, CA; 3Kaiser Permanete Division of Research, Pasadena, CA

Introduction: Acute pancreatitis (AP) remains a leading cause of hospitalization in the US. Despite the frequency of this disease there have been relatively few attempts to optimize approaches to routine care for patients with AP. We hypothesized that enhanced approaches to recovery may lead to earlier restoration of gut function in patients hospitalized for acute pancreatitis. Methods: We performed a single-blind randomized-controlled trial of patients admitted directly from the emergency department between July 2016-April 2017. Patients with evidence of organ failure or the systemic inflammatory response syndrome at the time of enrollment were excluded. All patients were enrolled within 24-hours of hospitalization and received standard fluid resuscitation. Participants were randomly assigned to receive either enhanced recovery consisting of patient-directed oral intake, early [2] Kaplan Meier Survival Curve by Race/Ethnicity. ambulation and non-opioid analgesia versus standard treatment with opioid analgesia and physician- directed diet as well as nursing parameters. The primary study endpoint was time-to-oral refeeding. Secondary endpoints included differences in pancreatitis activity scores (PASS), length-of-stay and 30-day re-hospitalization. All analyses were conducted on an intent-to-treat basis. Results: A total of 46 participants were enrolled. Etiologies were as follows: 61% gallstone, 15% alcohol, the incidence rates of m-IPMN. Tumor stage was analyzed for each subgroup analysis. Survival outcomes 13% hypertriglyceridemia, 11% other. Median age of the cohort was 53.1 years, 54.3% were female. There were evaluated using Kaplan Meier methods and log rank testing. was a significant reduction in time to successful oral re-feeding in the enhanced recovery vs. standard Results: We identified 3,867 cases of malignant IPMN from 1992-2011. Among men and women care group, median 13.8 vs. 124.8 hours, p < 0.001. Pancreatitis activity scores were lower at 48-72 hours combined and among all cancer stages, African-Americans had the highest incidence of m-IPMN, among patients assigned to enhanced recovery (mean 43.5 vs. 72.1, p < 0.001). There were no significant compared to non-hispanic whites (0.7 vs. 0.5 per 100,000/year, p < 0.005). Among all races, African- differences in length-of-stay or frequency of 30-day readmission. Americans presented with more distant disease (0.3 vs 0.2 (whites), P=0.04). From 1992-2011, the Conclusion: In this pilot randomized-controlled trial, enhanced recovery approaches were safe and yearly incidence of m-IPMN has been rising from 0.5 to 0.7 per 100,000 (P=0.07). Survival outcomes effective in promoting earlier restoration of gut function in patients hospitalized for acute pancreatitis for the patients developing m-IPMN showed significantly poorer survival in African-Americans (NCT02813876). when compared to non-hispanic whites (5 year survival: 9.7% vs 12%, p < 0.005). Asians had higher survival rates compared to whites (20.3% vs 12%, p < 0.005). Multivariate cox proportional hazard models adjusting for age, sex, treatment received, and cancer stage demonstrated Asian race to be a significant independent predictor of improved survival outcomes compared to whites (HR 0.828, CI= 0.74-0.93, p < 0.04), while African-Americans had an increased risk fof mortality (HR 1.306, CI=1.19-1.42, p < 0.04). Survival rates improved from 1992-2011 (HR 0.66 in 2007-11, CI=0.61-0.72, p < 0.04). Conclusion: Our study demonstrated a higher tumor burden in African-Americans with an increasing incidence rate. Stage-specific analysis showed a rise in more distant disease for African Americans from 1992-2011. Asian race was an independent predictor for survival while African-Americans showed increased risk for mortality. The disparities observed may reflect differences in underlying genetics and disease physiology or access to care. Delineating high risk groups is important in the early diagnosis and treatment of m-IPMN.

Volatile Organic Compound Biomarker Signature in Bile to Distinguish Pancreatic Cancer From Chronic Pancreatitis: A Single Blinded Study

2017 ACG Auxiliary Award (Member)

Udayakumar Navaneethan, MD, FACG1, Chad Spencer, MD2, David Grove, PhD3, Xiang Zhu, MS2, Mansour Parsi, MD4, John Vargo, MD4, Muhammad Khalid Hasan, MD2, Robert Hawes, MD, FACG2, Ji Young Bang, MD, MPH2, Shyam Varadarajulu, MD2, Raed Dweik, MD4. 1Florida Hospital, University of Central Florida College of Medicine, Orlando, FL; 2Center for Interventional Endoscopy, Florida Hospital, Orlando, FL; 3Cleveland Clinic Lerner Research Institute, Cleveland, OH; 4Cleveland Clinic Foundation, Cleveland, OH

Introduction: Early and accurate diagnosis of pancreatic cancer is important, particularly in at-risk populations with chronic pancreatitis. Volatile organic compounds (VOCs) are chemicals with high vapor pressures which allow them to easily diffuse and enter their gaseous forms.Our aim was to identify [1] Kaplan Meier Plot - time to oral refeeding by treatment arm; A=standard care, potential VOCs in the headspaces (gas above the sample) in bile that may help distinguish pancreatic B=enhanced recovery. cancer from chronic pancreatitis. Methods: In this prospective cross-sectional multi-center study, bile was obtained in 57 patients (47 pancreatic cancer; 11 chronic pancreatitis) undergoing ERCP for biliary obstruction. Selected ion flow tube mass spectrometry was used to analyze the concentration of 22 prevalent VOCs in bile samples. All the analysis was performed by one single investigator who was blinded to the final diagnosis. Logistic regression analysis was performed to build a predictive model for diagnosis of pancreatic cancer. In order to conduct internal model validation, we used bootstrap procedure for model discovery. Another independ- 2 ent cohort of bile samples (19 pancreatic cancer; 12 chronic pancreatitis) was employed as test set for validation of the biomarker signature. Increasing Incidence of Malignant IPMN (m-IPMN) and Poorer Survival Among African Results: A biomarker signature (three VOCs) was identified for the differential diagnosis between pancre- Americans atic cancer and chronic pancreatitis. The biomarker signature in the discovery set distinguished between pancreatic cancer and chronic pancreatitis with an area under the curve (AUC) of 0.98 (95% CI 0.95- 2017 Category Award (Biliary/Pancreas) 1.00). The biliary model cut-off of 1.33 [0.158* age + 9.747* log (ammonia) - 3.994* log (acetonitrile) + 5.044* log (trimethyl amine)] had a sensitivity of 93.5%, specificity of 100% and a diagnostic accuracy 1 2 2 1 Felix H. Lui, MD , Richard Shaw, PhD , Lauren B. Gerson, MD, MA . University of California Irvine of 96.5%. (Figure 1) Internal validation regarding model’s calibration showed that the model correctly 2 Medical Center, Anaheim, CA; California Pacific Medical Center, San Francisco, CA predicted pancreatic cancer and chronic pancreatitis rates at different risk levels P( =0.93) In the validation set, an AUC of 1.00 and, using the same cut-off, a sensitivity of 100% and a specificity of 100% were Introduction: Intraductal papillary mucinous neoplasms (IPMNs) are precancerous lesions. While achieved, successfully validating the biomarker signature. racial/ethnic disparities are well reported among other cancers of the digestive tract, limited data exists Conclusion: The measurement of VOCs in biliary fluid may be useful to diagnose pancreatic cancer in regarding racial/ethnic disparities for stage-specific malignant IPMN (m-IPMN) incidence and survival high risk chronic pancreatitis patients with high accuracy. A larger study with a longitudinal study design in the United States. is required to confirm our observations to diagnose pancreatic cancer early in patients with chronic pan- Methods: We used the Surveillance, Epidemiology, and End Results (SEER) database from 1992-2011 to creatitis. examine incidence rates for m-IPMN by ICD coding. We examined the impact of age, sex, and race on

DAPP resulted in an increase in the utilization of abdominal US (58.3% v 83%, p < 0.001) and CXR (35.5% v 72.4%, p < 0.001). There was a 50.4% (24.6% v 51.8%, p < 0.01) reduction in the number of CT scans in the ER. Subgroup analysis of the prospective arm showed no difference in AP severity regardless of whether a CT scan was performed in the ER (P=0.079). Final severity of AP was improved after initiation of DAPP (Table 1). There were no deaths in the prospective group compared to 12 (5.5%) in the retrospective group. Compared to the prospective group, the retrospective cohort was more likely to have severe AP compared to mild AP (OR = 6.91; 95% CI 1.49-64.11) or moderately severe AP (OR = 4.67; 95% CI 0.90-45.54). Conclusion: Data from AP patients managed prior to published guidelines suggested a need to improve initial imaging utilization. Implementation of our programmatic quality initiative with DAPP reduced the use of CT scans by 50.4% overall without an increase in severity outcomes.

Utility of HIDA Scan in Predicting Outcome of Cholecystectomy in Patients With Chronic Acalculous Gallbladder Disease? A Systematic Review

Nazar Hafiz, MD1, Seth D. Crockett, MD, MPH2. 1Louisiana State University Health Sciences Center, Shreveport, LA; 2University of North Carolina School of Medicine, Chapel Hill, NC

Introduction: HIDA scan with gallbladder ejection fraction (GBEF) is used to evaluate chronic acalculous gallbladder disease (CAGBD), in which abdominal pain is attributable to biliary dyskinesia or chronic acalculous cholecystitis. In current practice, cholecystectomy is recommended when the GBEF is abnormal. We conducted a systematic review to evaluate the value of HIDA to predict patient response [3] ROC curves of the biomarker (biomarker signature) results on bile samples from after cholecystectomy. all patients with pancreatic cancer versus patients with chronic pancreatitis. Methods: In consultation with librarian, a literature search was done in PubMed for studies from January 1st 1990 – October 31st, 2016 to identify manuscripts with patients with abdominal pain without gallstones who underwent HIDA followed by cholecystectomy, and where follow-up data on pain resolution following surgery was reported. After pooling data, summary statistics were calculated for positive predictive value (PPV), negative predictive value (NPV), sensitivity and specificity of HIDA with abnormal GBEF in detecting patients whose symptoms resolve after cholecystectomy. 4 Results: Initial literature search identified 74 articles. 15 studies met the inclusion criteria. Only 1 randomized controlled trial was identified (Yap et al.), which was small and low quality. 722 patients under- Initial Imaging in Patients With Acute Pancreatitis: Impact of Quality Improvement went cholecystectomy in the abnormal GBEF group, out of which 658 reported improvement (PPV = 91.1%). In the normal GBEF group, 132 patients underwent cholecystectomy and 107 patients reported Stephen Steele, MD1, Heather L. Branstetter, MD1, Jimmy Shah, MHA1, Priyanka Acharya, MSc, MPH1, improvement (NPV = 18.9%). Overall the sensitivity of HIDA EF was fairly high at 86.0% but the specific- Nathaniel Avila, MD1, Huda Khan, MD1, Rathan Reddy, MD1, Matthew Muller, MD2, Prashant Kedia, ity was low (28.1%), corresponding to a false positive rate of 71.9%. The overall positive likelihood ratio MD1, Paul Tarnasky, MD1. 1Methodist Dallas Medical Center, Dallas, TX; 2Methodist Dallas Medical (LR) is 1.20 and negative LR is 0.50. Center, Fort Worth, TX Conclusion: While the sensitivity and PPV of HIDA for detecting patients who have cholecystectomy responsive CAGBD is fairly high, the specificity is alarmingly low. This means that the majority of patients who don’t improve following cholecystectomy have abnormal HIDA scans preoperatively. Also, a rela- Introduction: Guidelines for management of acute pancreatitis (AP), including appropriate imag- tively high proportion of patients with normal HIDA scans report improvement in symptoms following ing, were published (2013). The AP published guidelines (APPG) recommend chest x-ray (CXR) and surgery, which undermines the value of test. These findings indicate that HIDA scans may be less valuable abdominal ultrasound (US) for initial imaging. Emergency room (ER) CT scans are discouraged unless in selecting patients for cholecystectomy, as a test with a high specificity would be better for this purpose. the diagnosis is unclear. The aims of this study were to report the effects of a quality initiative evaluating adherence to imaging recommendations and to determine outcomes in patients managed per APPG. Methods: The Dallas Acute Pancreatitis Protocol (DAPP) entails computer physician orders (e.g. fluid resuscitation, enteral feeding), physician and nursing education, and patient navigator to optimize adherence to APPG. Retrospective data from 8/2011-12/2014 and prospective data from 1/2015-12/2016 were analyzed. Patients were included if admitted for AP and at least 2 of 3 following criteria were met: (i) pain consistent with AP, (ii) serum amylase and/or lipase >3X ULN, and (iii) characteristic imaging finding. Exclusion criteria included: age < 18 years, transfers, post-ERCP AP, and trauma. Measured outcomes included initial diagnostic imaging, adherence to the APPG, and clinical outcomes (local and systemic complications). Final outcomes were defined as mild, moderately severe and severe as defined by the APPG. Results: Retrospective data of 220 patients and prospective data from 228 patients were evaluated. The most commonly identified etiologies (Table 1) in both cohorts were biliary and alcohol. Initiation of

[4] Etiology, Imaging, and Outcomes in Acute Pancreatitis

Retrospective Cohort Prospective Cohort P-value (n=220) (n=228) [5A] Test characteristics of HIDA scan. Etiology

Biliary 42.7% 42.1%

Unexplained 29.5% 26.3%

Alcohol 12.3% 22.8%

Hypertriglyceridemia 9.1% 6.1%

Autoimmune 2.3% 1.8%

Neoplasia 2.3% 0%

Medication 1.8% 0%

Pancreas Divisum 0% 0.4%

SOD 0% 0.4%

Imaging

CT scan 51.8% 24.6% <0.01

Abdominal US 58.3% 83.0% <0.001

CXR 35.5% 72.4% <0.001

Outcome [5B] 2x2 table with summary data for all patients with preoperative HIDA scans undergoing cholecystectomy. Mild 78.2% 86.8%

Moderately severe 16.4% 12.3%

Severe 5.5% 0.9%

The American Journal of GASTROENTEROLOGY VOLUME 112 | SUPPLEMENT 1 | OCTOBER 2017 www.nature.com/ajg ABSTRACTSAbstracts S3

[5C] Studies included in review.

We suggest that physicians perform comprehensive evaluations to rule out other causes of abdominal pain fluid(s) infused during and after ERCP (periprocedural) were evaluated for their impact on the incidence before proceeding to HIDA, and to interpret findings of HIDA with caution, as well as educate patients of PEP using univariable logistic regression analysis. regarding the limitations. Results: A total 959 patients (mean age of 52.33±14.96 years; 551 (57.4%) females) were randomized during the trial, of whom 476 (49.6%) received periprocedural fluids (mean fluid administered 1245±629mL). The incidence of PEP was reduced in patients who received periprocedural fluid vs those who did not receive periprocedural fluid (5.2% vs 8.0%,P =0.079; OR, 0.65; 95% CI 0.38-1.09). Patients who developed PEP received a lower mean volume of fluid compared to those who did not develop PEP (752±783mL 6 vs. 1012±725mL, P=0.036). There were 174 patients (37%) who received lactated Ringer’s (LR). Patients who developed PEP received a lower mean volume of LR compared to those who did not develop PEP Periprocedural Fluid Type and Volume Reduce the Risk of Post-ERCP Pancreatitis in High-Risk (329±356 vs. 570±559mL, P=0.006). The use of LR was associated with a lower risk of PEP compared to Patients: A Secondary Analysis of a Randomized Controlled Trial those who received all other types of fluid (5.8% vs 9.8%, P=0.047; OR, 0.56; 95% CI 0.31-0.99, P=0.047). Conclusion: Higher mean volume of periprocedural fluids and use of lactated Ringer’s solution further 2017 ACG Governors Award for Excellence in Clinical Research reduces the incidence of post-ERCP pancreatitis beyond rectal indomethacin based on this subgroup analysis of a pharmacological prophylaxis PEP trial conducted in high risk patients. Venkata Akshintala, MD1, Ayesha Kamal, MD2, Rupjyoti Talukdar, MD3, Rajesh Goud, BSc3, Mouen Khashab, MD1, Vikesh K. Singh, MD1. 1Johns Hopkins University Hospital, Baltimore, MD; 2Johns Hopkins University School of Medicine, Baltimore, MD; 3Asian Institute of Gastroenterology, Hyderabad, Telangana, India 7 Introduction: Aggressive hydration with lactated Ringer’s was recently shown to be efficacious in reducing the incidence of post-ERCP pancreatitis (PEP) in a randomized controlled trial (RCT) of average risk patients A Potential Screening Strategy of Pancreatic Cancer in New Onset Diabetes Patients Using Serum undergoing ERCP. The impact of fluid on the incidence of PEP has not been evaluated in high risk patients. CA19-9 in Combination With Age and BMI Methods: We conducted a secondary analysis of the effect of volume and type of intravenous fluid administered on the incidence of PEP in high risk patients who underwent ERCP as a part of a double blinded Xiangyi He, PhD, MD, Yaozong Yuan, PhD, MD. Shanghai Ruijin Hospital, Shanghai, China (People's RCT comparing the efficacy of rectal indomethacin versus a combination of papillary spray of epineph- Republic) rine and rectal indomethacin for the prevention of PEP. High risk patients were defined on the basis of prospectively validated patient and procedure-related risk factors. Patients who underwent planned therapeutic pancreatic stenting and those with suspected sphincter of Oddi dysfunction (SOD) type 3 Introduction: Type 2 diabetes mellitus is widely considered to be associated with pancreatic can- were excluded from the study. PEP was defined as per the consensus criteria. The volume and type of cer. New-onset diabetic mellitus (NODM) is recently recognized as the early symptom of pancreatic

[6] .

[7] Characteristics of the study subjects

PC with new-onset DM New-onset DM P value

N 30 30

Gender 1.0

male 19 20

female 11 10

Age,y 62.85±9.86 52.76±9.94 0.001

Stage

I 3

II 9

III 3

IV 15

Location

Head 17

Tail & body 13

Differentiation [8A] Extrapancreatic necrosis. I 0

II 10

III 20

BMI (mean± SD ) 21.63±3.03 24.77±3.80 0.001

Duration of DM (Mon, mean± SD) 7.2±11.89 16.89±9.65 0.001

CA19-9 (U/L, median, range) 211.5(20-9701) 8.7(0.08-31.5) 0.007

FBG (mean±SD) 7.65±1.99 7.94±3.92 0.688

cancer (PC). We aimed to establish screening strategy based on NODM for PC using serum CA19-9 and clinic characteristics. Methods: Preoperative serum levels of CA19-9 and clinicopathological characteristics were retrospectively analyzed in consecutive 30 with or 30 without pancreatic ductal adenocarcinoma (PDA) in new- onset diabetes (less than 24 months) patients. Results: Patients studied are described in more detail in Table 1. The gender distribution and Fasting Blood Glucose (FBG) was similar in both groups. The average age of the 2 groups was significantly different (P= 0.01) with the NODM group being younger than the PC with NODM group, the average age (SD) are 62.85(9.86) year and 76(9.94) respectively. The BMI of malignant group was lower than NODM group (they are 21.63±3.03 Kg/M^2 and 24.77±3.80 Kg/M^2 respectively), and the duration of DM was also shorter in PC group. The ROC curve based on CA19-9 resulted in an AUC of 0.975 (95% CI, 0.943-1.0), at the cut-off value of 35u/l, the sensitivity and specificity were 82.8% and 100% respectively. Furthermore, in subgroup of patients BMI>22, 11 NODM with PC and 21 NODM were included, AUC of serum CA19-9 was 1, at the cut-off value of 35u/l, the sensitivity and specificity were both 100% and 100% respectively. Conclusion: The NODM patients with old age and low BMI are more likely to be PC. Screening for pancreatic cancer (PC) based on NO-DM using serum CA19-9 have low sensitivity but high specificity, [8B] Walled off necrosis formed from acute necrotic collection (Extrapancreatic but in BMI>22 patients, the diagnostic accuracy of serum ca19-9 is more high. However, our hypothesis necrosis). needed more data to support.

(82.5% vs 28.6%) than patients with limited EPN. However mortality (27.5% vs 9.5%) and need for inter- 8 vention (62.5% vs 57.1%) was comparable in both groups. Conclusion: Patients of EPN alone have a clinical course that is severe than AIP but milder than PN and To Prospectively Study the Outcome of Adult Patients With Acute Interstitial Pancreatitis, Acute hence should be considered as a separate category. Patients with extensive EPN despite having higher Necrotizing Pancreatitis and Acute Pancreatitis With Extra-pancreatic Necrosis Alone frequency of multi-organ failure, persistent organ failure, ascites and pleural effusion have comparable mortality or need for intervention compared to patients with limited EPN. Gaurav Muktesh, MBBS, MD, DM, Surinder S. Rana, MBBS, MD, DM, Ravi Sharma, BSc, MSc, Lovneet Dhalaria, BSc, MSc, Rajesh Gupta, MBBS, MS, MCh, Mandeep Kang, MBBS, MD. Post Graduate Insti- tute of Medical Education and Research, Chandigarh, India 9 Introduction: The correlation between extent of extra pancreatic necrosis with disease outcome in patients with acute pancreatitis remains unclear and was assessed in this prospective study that looked Risk for Extrapancreatic Malignancies Among Patients With Intraductal Papillary Mucinous at the natural course of patients with acute interstitial pancreatitis (AIP), acute necrotizing pancreatitis Neoplasms of Pancreas (IPMNs) (ANP) and Extra-pancreatic necrosis (EPN) alone. Methods: Seventy eight patients of acute pancreatitis [55 (70.5%) males; mean age: 41.8 [range (16-77) Nikola Panic, PhD1, Federico Macchini, MD2, Sonia Solito, MD2, Debora Berretti, MD2, Sara Pevere, years] were prospectively observed till clinical recovery or death. Contrast-enhanced computerized MD2, Salvatore Francesco Vadalà di Prampero, PhD2, Marco Marino, MD2, Maurizio Zilli, MD2, Milutin tomography of the abdomen was done after day 5 of onset of symptoms and patients were categorized Bulajic, PhD2. 1Faculty of Medicine, University of Belgrade, Belgrade, Serbia; 2Academic Hospital Center into 3 groups: AIP, ANP and EPN alone. EPN was termed extensive if it extended to paracolic gutters "Santa Maria della Misericordia", Udine, Friuli-Venezia Giulia, Italy and/or pelvis. All groups were compared with respect to outcomes (surgery/intervention, organ failure and mortality). Results: Ten (12.8%) patients had interstitial pancreatitis, fifty one (65%) had pancreatic necrosis (PN) Introduction: Recent evidences suggest that patients with intraductal papillary mucionous neoplasms alone or combined with EPN and ten patients (12.8%) had EPN alone. Forty four patients (56.4%) devel- of pancreas (IPMNs) are at the increased risk for extrapancreatic malignancies (EPMs). We have con- oped organ failure, whereas 42 (53.8%) had persistent organ failure. Forty patients (51.3%) underwent ducted a study in order to assess the prevalence of EPMs in a cohort of IPMN patients from North- an intervention and thirteen patents (16.7%) expired. Patients with EPN alone had higher occurrence Eastern Italy. of organ failure (20 % vs 0%), persistent organ failures (20% vs 0%), and need for intervention [30% vs Methods: A single-centre study was conducted in hospital Santa Maria della Misericordia, Udine, Italy. 0 %] as compared to patients with AIP. Patients in PN group had higher frequency of ascites (64.7% vs Hospital records were screened in order to identify newly diagnosed IPMN cases in period January 1st 50%), pleural effusion (88.23% vs 70%), multi organ failure (27.4% vs 0%), persistent organ failure (72.5% 2009–December 31st 2015. Data were extracted on demographics, clinical characteristics and treatment. vs 20%), need for intervention (66.7% vs 30%) and mortality (25.5% vs 0 %) than patients with EPN We searched for any EPMs diagnosed previous, synchronous or after the IPMN diagnosis. The ratio of alone. Similarly patients with extensive EPN (n = 40) had significantly higher frequency of ascites (75% the observed (O) number of patients with EPMs to the expected (E) number was calculated along with vs 38.1%), organ failure(82.5% vs 28.6%), multi-organ failure( 35% vs 0%) and persistent organ failure 95% confidence interval (CI).

The American Journal of GASTROENTEROLOGY VOLUME 112 | SUPPLEMENT 1 | OCTOBER 2017 www.nature.com/ajg ABSTRACTSAbstracts S5

[8C] Flowchart Depicting Local Complications and Intervention Amongst Pancreatitis Patients.

[9] Observed on January 1, 2016 and expected (O/E) number of patients with Results: Number of 198 patients with IPMN was included in the study. Patients were predominantly extrapancreatic malignancies in 60 patients with intraductal papillary mucinous neoplasms female (59.1%) and age 70-79 (47.2%). IPMNs were predominantly brunch duct (79.2%), multifocal (66.2%) and were not subjected to surgical treatment (89.9%). We identified 72 EPMs in 63 patients (IPMNs) of pancreas with IPMN (31.8%). Eight patients have had 2 EPMs and 2 patients have had 3 EPMs. Among identified EPMs, 51 (70.8%) were diagnosed previous to IPMN, 17 (23.6%) synchronous to IPMN and 4 (5.6%) Observed Expected O/E CI 95% metachronous to IPMN. Among most frequently diagnosed were colorectal cancer (12 patients, 6.1%), breast cancer (8 patients, 4.0%), renal cell cancer (8 patients, 4.0%) and prostate cancer (7 patients, 3.5%). All 26 7,096 3,66 2.39-5.37 O/E ratios for EPMs were significantly increased for cancer in general (3.66, CI 95%: 2.39-5.37) as well as for renal cell carcinoma (9.62, CI 95%: 1.98-28.10) and prostate cancer (4.91, CI 95%: 1.59-11.45), while Prostate cancer 5 1,019 4,91 1.59-11.45 borderline significance was observed for CRC (3.59, CI 95%: 0.98-9.19) and breast cancer (3.16, CI 95%: Breast cancera 5 1,583 3,16 1.03-7.37 1.03-7.37) (Table 1). Conclusion: We report an increased risk for EPMs in Italian patients with IPMN, especially for renal cell, Colorectal cancer 4 1,115 3,59 0.98-9.19 prostate, colorectal and breast cancer.

Renal cell cancer 3 0,312 9,62 1.98-28.10

aCalculated for females.

10 11

Temporal Trends in Incidence and Outcomes of Acute Pancreatitis in Hospitalized Patients in the Outcomes of Anticoagulation for Portal and/or Splenic Vein Thrombosis in Setting of Acute United States Pancreatitis

2017 Presidential Poster Award Wesley Anderson, MD, Blake Niccum, BS, Maithili Chitnavis, MD, Dushant Uppal, MD, R. Ann Hays, MD. University of Virginia, Charlottesville, VA Elizabeth Brindise, DO, Aneesh Kuruvilla, DO, Rogelio Silva, MD. University of Illinois at Chicago, Advocate Christ Medical Center, Oak Lawn, IL Introduction: Acute pancreatitis (AP) is a severe inflammatory condition capable of causing local and systemic complications. Splanchnic vein thrombosis complicates up to 10% of AP. Clinicians advise Introduction: Acute pancreatitis is one of the most common reasons for hospitalization with a gastro- anticoagulation (AC) in some patients developing portal vein thrombosis (PVT) and/or acute splenic intestinal condition accounting for a significant health care burden. Previous studies have demonstrated vein thrombosis (SpVT) following AP, but evidence-based guidelines are lacking. We aim to compare an increase in admissions for acute pancreatitis worldwide over the past 10 years. We sought to exam- the rate of gastrointestinal (GI) bleeding, thrombus resolution, and 90-day mortality between patients ine temporal trends in incidence and outcomes of acute pancreatitis in hospitalized adult patients in the who did and did not receive AC in the setting of AP and concomitant acute PVT and/or acute SpVT. United States. Methods: Retrospective cohort study (January 2007 to January 2017) at tertiary care hospital. Inclusion Methods: Subjects within this study were obtained from the HCUP-Nationwide Inpatient Sample (NIS) criteria: diagnosis of AP by revised Atlanta classification and acute PVT or SpVT on imaging (CT, MRI, database for the years 2002-2013. ICD-9-CM codes were used to identify patients hospitalized with acute US) within six months following AP. Acute thrombus was defined as lacking cavernous transformation pancreatitis. All adults over the age of 18 were included in this study. Incidence of acute pancreatitis and absent on imaging performed two months before discovery. and in-hospital all-cause mortality as well as mean total cost of hospitalization and length of stay were Results: Our Clinical Data Repository reported 4980 patients with diagnostic codes for AP; 463 of 4980 assessed. (9.3%) had concomitant PVT and/or SpVT. 128 (27.6%) met study criteria: mean age was 51.4 (SD 13.7), Results: We identified 4,791,802 cases of acute pancreatitis in hospitalized adult patients during the study 55 were female (43%). Etiologies of pancreatitis were gallstones 18%, alcohol 30%, hypertriglyceridemia period. 50.1% of the entire cohort was female. A significant increase in the incidence of acute pancreatitis 12%, other/unknown 58%. Seventy-one (55.5%) patients had a PVT, 80 (62.5%) had a SpVT, and 23 was observed from 9.48 cases per 1,000 hospitalizations in 2002 to 12.19 in 2013 (P < 0.001) (Figure 1). (18.0%) had both a PVT and SpVT. Fifty-seven (44.5%) patients received anticoagulation and 71 (55.5%) During the same period, incidence of in-hospital mortality decreased from 2.99 cases per 100 cases of did not. 14/57 (24.6%) anticoagulated patients experienced a GI bleed compared with 7/71 (9.9%) non- acute pancreatitis in 2002 to 2.04 cases in 2013 (P < 0.001) (Figure 2). Male patients had a higher incidence anticoagulated patients (OR 2.98, 95% CI 1.11-7.98, Fischer’s exact P=0.032). Among the 108 patients for of acute pancreatitis and in-hospital all-cause mortality compared to the female patient. Additionally, which thrombus resolution status was known, 23/39 (59.0%) anticoagulated patients experienced resolu- mean length of stay decreased from 6.99 days in 2002 to 6.35 in 2013 (P < 0.001), while mean cost of tion compared with 25/69 (36.2%) non-anticoagulated patients (OR 2.53, 95% CI 1.13-5.66, Fischer’s hospitalization increased from $27,827 to $49,772 (P < 0.001). exact P=0.027). Lastly, 9/57 (15.8%) anticoagulated patients died within 90 days of pancreatitis diagnosis Conclusion: Hospital admissions for acute pancreatitis in adults, increased significantly in the US during compared with 4/71 (5.6%) non-anticoagulated patients (OR 3.14, 95% CI 0.91-10.80, Fischer’s exact the last twelve years. Despite this increase, incidence of in hospital all-cause mortality and mean duration P=0.078). of hospitalization have significantly dropped during the last twelve years, while total hospital charges rose. Conclusion: This is the first cohort study assessing outcomes of AC for PVT and/or SpVT thrombo- Male sex is associated with higher incidence of acute pancreatitis and higher incidence of in-hospital sis developing in the setting of AP. As expected, anticoagulation increased the risk for GI bleeding and mortality. thrombus resolution. The long-term risk-benefit for AC in AP with splanchnic thrombosis remains unknown. We hope this study serves as a step toward development of evidence-based use of AC in AP.

[11] Sources of GI Bleeding

Source of GI Bleed Anticoagulated Patients Non-anticoagulated Patients

Pancreas, n 5 1

Duodenum, n 1 0

Splenic Artery, n 2 0

Stomach, n 0 1

Retroperitoneum, n 1 0

JP Drain, n 0 1

J-arm Erosion, n 0 1

Unidentified Upper GI Bleed, n 2 3

Unidentified Lower GI Bleed, n 2 0

Unidentified GI Location, n 1 0 [10A] Incidence of Acute Pancreatitis (per 1,000 discharges).

Proposal of a Macroscopic Classification for Tissular Lesions of the Bile Duct Detected During Peroral Cholangioscopy (POCS)

Carlos Robles-Medranda, MD, Miguel Soria-Alcivar, MD, Manuel Valero, MD, Miguel Puga-Tejada, MD, MSc, Roberto Oleas, Haydee Alvarado-Escobar, MD, Jesenia Ospina-Arboleda, MD, MSc., Hannah Pitanga-Lukashok, MD. Ecuadorian Institute of Digestive Diseases, Guayaquil, Guayas, Ecuador

Introduction: Macroscopic aspects to determine bile duct malignancy during per oral cholangioscopy (POCS) are: irregular surface with bleeding and drooling or tortuous vessels. For benign lesions, typical aspects are lesions with smooth surface “without vessels or mass”. However, many misdiagnosis are made due to a lack of correlation between the macroscopic aspects and histology. Moreover, masses are usually benign, and reported data shows 78% of sensitivity for visual impression of malignancy. AIM: To propose a macroscopic classification of bile duct tissular lesions for differentiation between benign and malignant lesions. Methods: Ambispective study. Retrospective stage (Sept/2013 - Sept/2015) was performed in patients evaluated by POCS (SpyGlass® legacy and DS). Exclusion criteria: absence of histology confirmation and/or absence of POCS at 6 months follow-up (for benign lesions). To determine macroscopic classification, all patients records were evaluated. 315 images were analyzed and classified as benign or malignant by an expert with >140 POCS cases, and compared to histology. Based on morphological [10B] Incidence of Mortality in Acute Pancreatitis (per 100 cases). and vascular pattern, lesions were classified as follow on table 1. Inter-observer and intra-observer agreement was calculated using 40 random lesions images (1 expert, 2 non-expert in POCS). Pro- spective stage (non randomized, double blind, Oct/2015 - April/2016) evaluated diagnostic accuracy (sensitivity, specificity, PPV, NPV) using new classification in consecutive tissular lesions, correlated with histology. Results: 130 patients were studied (retrospective 87, prospective 43); 30 female, 61 (50-74) yo. Retro- spective stage: 46/87 patients evaluated (21 female; 61 [52-73] yo). 18/46 malignant cases. Overall inter- observer agreement was substantial when lesions were classified as benign or malignant (K=0.75; CI 0.46-0.89) and when lesions were classified by sub-types (K=0.71; CI 0.39-0.88). Intraobserver agreement was almost perfect when lesions were classified as benign and malignant (K=0.88; CI 0.66-1.0) and when

The American Journal of GASTROENTEROLOGY VOLUME 112 | SUPPLEMENT 1 | OCTOBER 2017 www.nature.com/ajg ABSTRACTSAbstracts S7

[12] POCS macroscopic classification proposed, from benign and malignant lesions. The 13 lesions were classified into benign or malignant based on the morphological and vascular pattern Appraisal of American Gastroenterology Association (AGA) Pancreatic Cyst Guidelines to Detect Advance Neoplasia: A Systematic Review and Meta-Analysis

BENIGN LESIONS Emmanuel E. Ugbarugba, MD, MPA1, Rohan M. Modi, MD1, Darwin L. Conwell, MD, MS2, Somashekar G. Krishna, MD, MPH1. 1The Ohio State University Wexner Medical Center, Columbus, OH;2 The Ohio Type 1 Villous pattern Micronodular or villous pattern without vascularity. State University College of Medicine, Columbus, OH Type 2 Polypoid pattern Adenoma or granuloma pattern without vascularity.

Type 3 Inflammatory pattern Regular or irregular fibrous and congestive pattern with Introduction: Pancreatic cystic lesions are increasingly detected with advancement in cross sectional regular vascularity. imaging. Initial guidelines such as Sendai 2006 and Fukuoka 2012 guidelines were developed for evaluation of mucinous cystic lesions. AGA 2015 guidelines were developed for asymptomatic pancreatic MALIGNANT LESIONS cystic neoplasms (PCN) and suggest that individuals with dilated pancreatic duct and a solid component or concerning features on EUS-FNA should undergo surgical resection. A meta-analysis was Type 1 Flat pattern Flat and smooth or irregular surface with irregular or performed to appraise the role of AGA guidelines in the detection of advanced neoplasia in PCNs. spider vascularity. Objective:Determine pooled diagnostic parameters of the AGA guidelines to detect advance neoplasia in PCNs and comparing pooled diagnostic characteristics of the AGA and Fukuoka guidelines in detect- Type 2 Polypoid pattern Polypoid or mass shape with fibrosis and irregular or spider vascularity. ing advanced neoplasia. Methods: Database search was performed to identify studies that evaluated detection of advance neopla- Type 3 Ulcerated pattern Irregular ulcerated and infiltrative pattern with or with- sia in PCN utilizing AGA guidelines. Pathology of surgically resected PCN was used as the gold standard out fibrosis and with irregular or spider vascularity. for diagnosis. Data was extracted into a 2 x 2 contingency table and used to calculate the sensitivity (SN), specificity (SP), negative predictive value (NPV), and positive predictive value (PPV) to detect advance Type 4 Honey-comb pattern Fibrous honey-comb pattern with or without irregular or neoplasia. Random model meta-analysis was used for all the analysis. spider vascularity. Results: Only 5 studies which utilized surgical histopathology were included in the final analysis. These were all observational retrospective studies and included 777 patients from 1995 to 2016. The pool diagnostic parameters for AGA guidelines included, SN: 83% (95% CI, 77-89), SP: 67 % (95% CI, 63-71), PPV: lesions were classified by sub-types (K=0.90; CI 0.71-1.0).Prospective stage: 23/43 patients evaluated 41.8% (95% CI, 31.4-53.0), and NPV: 92.2 %( 95% CI, 85.1-96.0). Subsequent pooled analysis of three (9 female, 61 [48-72] yo). 13/23 malignant cases. studies that compared AGA and Fukuoka guidelines revealed a higher sensitivity in detecting advanced Accuracy was 86.9, sensitivity 100, specificity 70, PPV 81.3, NPV 100%. neoplasia for Fukuoka guidelines (SN: 96% vs. 87% p = < 0.01) but at the expense of a lower specificity Conclusion: Proposed macroscopic classification could help physicians to distinguish benign from (SP: 57% vs. 65%; p = 0.018) (Table 1). malignant lesions with a good inter and intra-observer concordance.

[13A] Characteristics of Studies used in the Systematic Review and Meta-analysis.

[13B] Forest plot showing pooled sensitivity of AGA guideline to detect advance neoplasia in PCN.

[13C] Forest plot showing pooled specificity of AGA guideline to detect advance neoplasia in PCN.

Conclusion: AGA guidelines did not perform better than Fukuoka guideline in detecting advance neo- based on medical insurance coverage. The majority (90 %) of surgeries were performed in tertiary plasia in PCNs in observational studies utilizing surgical histopathology as reference standard. While con- academic medical centers. Geographically, TP were done predominantly in the South, while TPAIT was tinued evaluation of novel cyst diagnostics is encouraged, an individualized approach is recommended more common in the Midwest. Comorbidities were similar but TP had a longer length of stay (24 vs 15 for management of PCNs. days, P= < 0.001) and a higher mortality (8 vs 0 %, P= < 0.001). On multivariate analysis, length of stay was longer for TP compared to TPAIT with length of stay difference of 9 days (95% CI, 4.51-13.60, p < 0.001] and cost was higher for TP compared to TPAIT with cost difference of $19,050 (95% CI, 2,408- 35,691, P=0.025]. 14 Conclusion: The number and percentage of TPAIT surgeries have increased over the past 10 years. Over- all, TPAIT, compared to TP alone, has a shorter length of stay resulting in less hospital costs. Further Comparison of Healthcare Resource Utilization and Outcomes in Patients Undergoing Total investigation of the short- and long-term healthcare burden, resource utilization and patient benefits of Pancreatectomy With or Without Autologous Islet Transplantation in the United States TPAIT surgery is warranted.

Emmanuel E. Ugbarugba, MD, MPA1, Darwin L. Conwell, MD, MS2, Amer Rajab, MD, PhD1, Philip Hart, MD1, Sylvester M. Black, MD, PhD1, Kenneth Washburn, MD1, Kristen Kuntz, PhD1, Shumei Meng, MBBS, PhD1, Somashekar G. Krishna, MD, MPH1, Alice Hinton, PhD3, Luis Lara, MD1. 1The Ohio 15 State University Wexner Medical Center, Columbus, OH; 2The Ohio State University College of Medicine, 3 Columbus, OH; The Ohio State University College of Public Health, Columbus, OH The Predictor Factors of Portal Vein Thrombosis in Pancreatitis Patients

Introduction: Chronic pancreatitis patients with refractory pain and recurrent attacks of acute pan- Yasir Al-Azzawi, MD, Mohamed Mahmoud, MD, Erik A. Holzwanger, MD, Wahid Wassef, MD, MPH. creatitis are often considered for total pancreatectomy (TP). Autologous islet transplantation can be UMass Memorial Medical Center, Worcester, MA done with total pancreatectomy (TPAIT) to lessen the severity of postoperative diabetes. Given the escalating cost of healthcare and the increasing interest in TPAIT, there is a need to better assess health- Introduction: Portal vein thrombosis(PVT) is partial or complete occlusion of the portal vein. The most care resource utilization and patient outcomes associated with these surgical interventions for CP. The common causes of PVT are cancers, cirrhosis or/and severe inflammatory processes like pancreatitis. objective of this study was to compare inpatient demographics, hospital characteristics, comorbidities Our aim in this study is to determine the predicator factors that associated with the development of and differences in surgical outcome between patients undergoing TP versus TPAIT for acute or chronic PVT in pancreatitis patients. Other objective of the study is to assess the recanalization rate after treating pancreatitis. with the anticoagulation. Methods: Utilizing the Nationwide Inpatient Sample, a retrospective analysis of inpatient admission of Methods: A single center, retrospective chart review and analysis of patients with pancreatitis who had adults (≥18) who had a TP or TPAIT between 2002 to 2013 for a diagnosis of Acute/Chronic Pancreatitis PVT between the age of 18 and 80 who were admitted to our institute between 2006 and January 2016. (AP 577.0; CP 577.1) was performed. Surgical procedure trend, demographics, hospital characteristics, PVT diagnosis was confirmed by CT scans, MRI or ultrasound Doppler and been followed by images 3-6 comorbidities and differences in surgical outcome over the study period were analyzed. Patients who were months later to confirm the resolution or the extension of the thrombus. pregnant or had a pancreatic neoplasm were excluded. Results: Eighty patients with PVT-related pancreatitis were initially screened, 47 were excluded second- Results: There were a total of 2535 total pancreatectomies (TP= 1,705 and TPAIT = 830) performed ary to the presence of cirrhosis, cancers or lost in the follow up. A total of 33 patients, 24 acute and 9 between 2002 and 2013. The percentage of TPAIT done in the total TP/TPAIT cohort during the analysis chronic pancreatitis, were included in our cohort. The mean age of the cohort is 59 +/-16. Males repre- time period increased 2.5 fold (Figure: 20 vs. 53%, p < 0.001). On univariate analysis, TPAIT cohorts sented 55% of the cohort and 45 % were females. Caucasian race represented 90% of the cohort. Seventeen were younger (41 vs 50 years P= < 0.001) predominantly female (69 vs 50%, P= < 0.001), with a racial patients (51%) had complications with the pancreatitis (fluid collections, pseudocyst or wall-off necrosis). and income disparity favoring Caucasian and high/very high income, but no differences were observed Complete occlusion of the portal vein was found in 57.5% of the cohort while partial occlusion was found in 42.5%. Acute PVT was found in 89% of the cohort. Table 1 20(60%) patients received anticoagulation therapy, 5 patients had resolution of the PVT and 8 patients had an extension of the thrombus. 13(40%) patients did not receive treatment, 4 patients had resolution of the PVT and 5 had an extension of the thrombus. No deference was found between the group who treated with anticoagulation in terms PVT resolution or extension when compared to the non- treated group. Conclusion: Pancreatitis-related portal vein thrombosis is associated with acute and complicated forms of pancreatitis. Treatment with anticoagulation does not increase the chances of PVT resolution or prevention of the thrombus extension.

Clinical Effects of Marijuana on Severity and Outcomes of Acute Alcoholic Pancreatitis

Hemant Goyal, MD, FACP. Mercer University School of Medicine, Macon, GA

Introduction: Marijuana/cannabis is most widely used illicit drug in world and is most commonly used drug of abuse in alcohol drinkers. Experimental studies have shown conflicting results of effects of marijuana on severity of acute pancreatitis (AP). There are no studies which have examined the clinical effects of cannabis in AP. Objective of this study is to ascertain the clinical effects of simultane- ous alcohol and cannabis use on severity at presentation and outcomes of acute alcoholic pancreatitis (AAP). Methods: A retrospective review was conducted on patients discharged with principle or secondary diagnosis of AP using ICD-9 & ICD-10 codes during Jan 2006 to Dec 2015 from our hospital. Only adult patients with their first admission during the study perioed who had positive drug screen for cannabis were included. Patients with AAP with cannabis (CB+) and without cannabis (CB-) use were identified and matched with age and sex in ratio of 1:2. Results: 114 subjects were included in study; 38 patients in CB+ and 76 in CB- group. The median BUN was higher for CB- compared to CB+ (12.00 vs. 10.00, P=0.033). Only 3 patients (7.89%) in CB+ had BUN >20 mg/dL at admission and 6 (15.79%) had elevation in BUN in 24 hours while 17 (22.37%) patients in CB- had BUN >20 mg/dL at admission and 23 (30.26%) had rise in BUN within 24 hours. A greater per- [14] Comparison of autologous islet transplantation among Pancreatitis patients centage of CB- had a BISAP score of two or three compared to CB+ (21.05% vs. 5.26%, P=0.031). A greater receiving total pancreatectomy. percentage of CB- had SIRS compared to the CB+ patients (47.37% vs. 26.32%, P=0.043). There were no

The American Journal of GASTROENTEROLOGY VOLUME 112 | SUPPLEMENT 1 | OCTOBER 2017 www.nature.com/ajg ABSTRACTSAbstracts S9

[16_A] Severity comparisons between alcoholic acute pancreatitis patients using cannabis (CB+) and not using cannabis (CB-)

Severity Measure Overall (N=114) N(%) CB+ (N= 38) N (%) CB- (N= 76) N (%) p-value*

BUN Level

Median (IQR) 11.00 (7.00,16.00) 10.00 (7.00,12.00) 12.00 (7.00,18.00) 0.033

BISAP Score

0,1 96 (84.21) 36 (94.74) 60 (78.95) 0.031

2,3 18 (15.79) 2 (5.26) 16 (21.05)

SIRS Score

Yes 46 (40.35) 10 (26.32) 36 (47.37) 0.043

No 68 (59.65) 28 (73.68) 40 (52.63)

Balthazar Index^

A,B 27 (24.77) 11 (33.33) 16 (21.05) 0.29

C 46 (42.20) 14 (42.42) 32 (42.11)

D,E 36 (33.03) 8 (24.24) 28 (36.84)

Severity of Pancreatitis

Mild 57 (50.00) 22 (57.89) 35 (46.05) 0.41

Moderate 50 (43.86) 15 (39.47) 35 (46.05)

Severe 7 (6.14) 1 (2.63) 6 (7.89)

*p-value generated from Fisher’s exact test or Chi-square test for proportions and Wilcoxon rank-sum test for continuous measures ^5 missing a Balthazar Index score due to not having a CT scan aMedian (Interquartile Range) reported; Wilcoxon rank-sum test produces mean scores: Mean score for BUN level: 48.08 CB+, 62.21 CB-

[16_B] Outcomes comparisons between alcoholic acute pancreatitis patients using cannabis (CB+) and not using cannabis (CB-)

Outcome Measure Overall (N=114) N (%) CB+ (N= 38) N (%) CB- (N= 76) N (%) p-value*

Length of Stay (Daysa)

Median (IQR) 4.00 (3.00, 6.00) 3.50 (2.00, 6.00) 4.00 (3.00, 7.00) 0.15

ICU Care

Yes 12 (10.53) 1 (2.63) 11 (14.47) 0.059

No 102 (89.47) 37 (97.37) 65 (85.53)

Charlson’s Comorbidity Index

0 95 (83.33) 32 (84.21) 63 (82.89) 1.00

>1 19 (16.67) 6 (15.79) 13 (17.11)

IQR = Interquartile Range *p-value generated from Fisher’s exact test for proportions and Wilcoxon rank-sum test for continuous measures aMedian (Interquartile Range) reported; Wilcoxon rank-sum test produces mean scores: Mean score for length of stay: 51.20 CB+, 60.65 CB-

statistically significant differences in the Balthazar Index (Table 1). There was nonsignificant trend for greater percentage of CB- preceiving ICU care as compared to CB+ (14.47% vs. 2.63%, P=0.059). There were no significant differences in length of stay or Charlson’s co-morbidity index. Overall, the median length of stay was 4 days and majority of patients had a Charlson’s score of zero, indicating that they did not have any co-morbidities that could increase the risk of mortality. No patient underwent pancreatitis related procedure in either group. Only two patients died in the selected cohort and both of them were in CB- group (Table 2). Conclusion: Based on these findings it appears that CB+ patients had less severe presentation of AAP indicating that cannabis could modulate the inflammatory effect of alcohol on the pancreas. Further large [15A] Figure 1. scale prospective studies are needed to confirm our results.

Is CEA an Accurate Predictor of Pancreatic Cyst's Malignant Potential? A Retrospective Review of Outcomes and Molecular Assessment of Cystic Lesions

Michele Barnhill, MD1, Shannon Morales, MD1, Tenzin Choden, MD1, Heather Hopkins, MD1, Alice Lee, MD1, Tanuj Sharma, MD2, Mohammed Albugeaey, MD1, Sofia M. Boilini, BS3, Tien Nguyen3, Nadim G. Haddad, MD1. 1MedStar Georgetown University Hospital, Washington, DC; 2MedStar Washington Hospital Center, Washington, DC; 3Georgetown University School of Medicine, Washington, DC

[15B] Figure 2. Introduction: Assessment of pancreatic cysts involves obtaining cyst fluid via endoscopic ultrasound (EUS)-guided fine needle aspiration (FNA). Standard of care typically is to send the pancreatic cysts

[17_A] Comparison of pancreatic cyst characteristics in all patients with a repeat FNA 18 Incidence of Post-Endoscopic Retrograde Cholangiopancreatography Pancreatitis in Liver Prior FNA =62 Index FNA =62 P value n n Transplant Patients: A Meta-Analysis and Literature Review

Age, mean (+/-SD), y 68.95+/−11.6 68.95+/−11.6 - Amaninder Dhaliwal, MBBS, MD1, Rajani Rangray, MD1, Harmeet Mashiana, MBBS, MD2, Ishfaq Bhat, MD1, Harlan Sayles, MS1, Shailender Singh, MD1. 1University of Nebraska Medical Center, Omaha, NE; Sex, male/female, % 35/65 35/65 - 2University of Nevada School of Medicine, Las Vegas, NV Multiple cysts, % 22.6 21.0 1.000

Head/body/tail, % 59/28/13 59/28/13 - Introduction: Endoscopic retrograde cholangiopancreatography (ERCP) is a commonly performed therapeutic procedure for various pancreaticobiliary disorders. Post-ERCP pancreatitis (PEP) is the Cyst diameter, mean (+/-SD), mm 21.8 +/- 15 26.3 +/- 6.3 0.199 most common complication with generally accepted risk of 5 to 15% in general population. The exact cause and mechanism of Post-ERCP pancreatitis is unclear, however, the initial events occur at the level Multilocular/unilocular, % 22.6/77.4 25.8/74.2 0.834 of acinar cells and lead to activation of several immune mediated inflammatory signaling pathways that lead to progression of the disease. Liver transplant patients on immunosuppression regimens often Connection to PD on EUS, % 50.0 54.8 0.719 require ERCP, however, the effect of the immunosuppression on incidence post-ERCP pancreatitis is Mural nodule, % 1.6 6.5 0.365 not well established. The aim of this study was to determine the risk of PEP in select population of liver transplant patients on immunosuppression, undergoing ERCP. CEA, median (range), ng/mL 60 (0-211520) 30.5 (0-54093) 0.735 Methods: We performed a systematic literature search for studies reporting ERCP outcomes among liver transplant patients. A total of 61 observational studies between 1991 and 2017 were included in our meta- CEA>192 ng/mL, % 29.0 32.3 0.846 analysis. Overall incidence of PEP in liver transplant patients was estimated using meta-analysis methods for binomial outcome data with random effects for study. Amylase, median (range), IU/L 2289 (0.3-332102) 2234 (0-278879) 0.630 Results: Among the 61 studies, there were 7,730 ERCP procedures performed on 3,980 patients, with 183 instances of PEP. The overall percentage of procedures with PEP, as estimated from our meta-analysis is Cytology test performed, % 95.2 96.8 1.000 1.53% (95% CI: 0.90% - 2.28%). Forest plot (graph 1) shows all the studies included in our meta-analysis Malignancy on cytology, % 0.0 1.7 1.000 with total number of ERCP procedures and incidence of PEP. Conclusion: In our meta-analysis of 61 studies, the overall incidence rate of post-ERCP pancreatitis in Mucin on cytology, % 4.8 8.1 0.717 liver transplant patients on immunosuppression (1.53%) was found to be lower than the reported risk in the general population. Further studies need to be done to investigate the mechanism of this beneficial P values for categorical variables were calculated using a Fisher exact test. P values for continuous variables effect of immunosuppressive medications. were calulated using t test or Wilcoxon rank-sum test. P<0.05 was considered significant.

19 fluid for molecular analysis including the carcinoembryonic antigen (CEA) level. Historically, there has been a distinction between serous and mucinous cysts, with mucinous considered to have increased Exosomal Glypican-1 in Patients With Pancreatic Adenocarcinoma and Associated Risk Groups malignant potential. This distinction is often unclear and heavily influenced by the CEA level of cyst fluid Pedro Moutinho-Ribeiro, MD, MSc1, Soraia Silva, MSc2, Barbara Adem, MSc2, Marco Silva, MD1, with increased levels pushed towards surgery. This has recently been called into question with a recent 1 3 1 4 study showing large variability in levels with multiple samplings. Herein, we review our preliminary Rosa Coelho, MD, MSc , Susana Lopes, MD, MSc , Filipe Vilas-Boas, MD , Helena Barroca, MD , Jose Machado, PhD2, Fatima Carneiro, PhD4, Sonia Melo, PhD2, Guilherme Macedo, MD, PhD, FACG3. 1De- results with repeat sampling of pancreatic cystic lesions via EUS-FNA. 2 Methods: This is a retrospective study conducted at a tertiary care institution of patients who under- partment of Gastroenterology - Centro Hospitalar de São João, Porto, Portugal, Porto, Portugal; Instituto de Investigação e Inovação em Saúde – i3S, Universidade do Porto, Porto, Portugal; 3Centro Hospitalar de went two EUS-guided FNA procedures (“index” and “prior”) of a pancreatic cyst between 2009- 4 2017. Our primary outcome was a comparison of CEA levels between index and prior exams. The São João, Porto, Portugal; Department of Pathology - Centro Hospitalar de São João, Porto, Portugal secondary outcomes included identifying any significant differences in characteristics of the cyst, including size, location, number of cysts, and presence of mural nodule. We also assessed for cases Introduction: A recent study has identified a cell surface proteoglycan, glypican-1 (GPC1), specifi- in which a categorical change in the CEA level relative to the commonly used cutoff of 192ng/mL cally enriched on cancer-cell-derived circulating exosomes (crExos). Levels of GPC1+ crExos correlate was identified. with tumour burden and the survival of patients with pancreatic adenocarcinoma (PDAC). The authors Results: We identified 62 patients who had undergone two EUS-FNA procedures for cystic pancreatic report the initial experience in the characterization of GPC1+ crExos levels in patients with PDAC and lesions. “Index” cases were typically performed about 1 year after “prior” cases. Our data revealed no associated risk groups. significant differences between the index and prior procedures in terms of CEA level, size, number of Methods: Unicentric prospective cohort study, including patients with PDAC (cytological/histological cysts, or presence of a mural nodule (Table 1). Six cases revealed a change in the CEA category, with an confirmation), cystic mucinous neoplasm (CMN), chronic pancreatitis (CP) or family history of PDAC increased CEA in 5 of these cases (Table 2). (FHPDAC). In all cases, endoscopic ultrasonography, blood tests (determination of total CrExos and lev- Conclusion: In the vast majority of our patients, the CEA levels did not change significantly. These pre- els of GPC1+ crExos) and demographic and clinical data collection was performed. Two GPC1 antibodies liminary results support the findings of Nakai et al, which asserted that serial EUS-guided FNA on pan- were used to preclude possible bias (Sigma-Aldrich® and Abnova®). creatic cysts with repeated CEA measurements may not have a significant impact on the management of Results: Included 31 patients (55% women, median age of 61 (IQR: 51.5 - 76.5) years). Two patients were these patients. Further studies are required to identify appropriate criteria to guide treatment decisions for excluded: one due to poor sample preservation and another whose final diagnosis was neuroendocrine patients with cystic lesions of the pancreas. Nakai, Y et al. Role of serial EUS-guided FNA on pancreatic tumor. Cases analyzed: 14 (48%) PDAC, 9 (31%) CP, 4 (14%) CMN, and 2 (7%) FHPDAC. The size cystic neoplasms: a retrospective analysis of repeat carcinoembryonic antigen measurements. GIE. 2016; and levels of the crExos were significantly higher in patients with PDAC (77.5 (IQR: 63.8 - 89.3) vs. 58 84(5): 780-784. nm (IQR: 54 - 63) and 4E+10 (1.3E+10 - 1.1E+11) vs. 2.2E+10 particles/ml (IQR: 1.1E+10 - 3.2E+10); P=0.005 and P=0.033, respectively). PDAC cases presented higher levels of GPC1+ crExos comparing with other groups, regardless of the antibody used: 73.1% (IQR: 47.2-86.9) vs. 2.3% IQR: 1.6 – 36.7) with Sigma-Aldrich® antibody (p = 0.006); 85.3% (IQR: 62 - 98.5) vs. 3% IQR: 1.5 - 72.7) with Abnova® antibody (p = 0.002). The levels of GPC1+ crExos presented good accuracy for the diagnosis of PDAC: AUROC of 0.85 (p = 0.014). For a cut-off of 23.6%, presented sensitivity = 100%, specificity = 75%, positive predictive value = 70%, and negative predictive value = 100%. Conclusion: These preliminary data indicate that levels of GPC1+ crExos are significantly higher in patients with PDAC compared to the risk groups studied.

[17_B] Characteristics of patients and associated pancreatic cysts with a repeat FNA and a categorical change in the CEA level

Age Sex Multiple Location Diameter, prior/ Multilocular Connection Mural nodule, CEA, prior/index Interval Pathology, prior/index cysts index (mm) to PD prior/index (ng/mL)

Increased CEA level

Case 1 81 M No Head --/24 Yes Yes No/No 184/606 422 Benign/Benign

Case 2 76 F Yes Head 44/19 Yes No No/No 71/1278 1207 Pauci-immune/Benign

Case 3 77 F No Head 36/25 No Yes No/No 52/260 208 Pauci-immune/Benign

Case 4 51 M No Head --/20 No No No/No 23/383 360 Benign/Benign

Decreased CEA level

Case 1 83 F No Body --/-- No Yes No/No 112660/83 112577 Pauci-immune/Mucinous

A catgeorical change in CEA level was defined as a shift from one side of the cutoff of 192ng/mL to the other.

The American Journal of GASTROENTEROLOGY VOLUME 112 | SUPPLEMENT 1 | OCTOBER 2017 www.nature.com/ajg ABSTRACTSAbstracts S11

[18] .

20 (with electrocautery enhanced delivery system) was used in the rest (87.5%). 38/40 (95.0%) patients had a successful and uneventful cystogastrostomy procedure. One patient required emergency laparotomy due to major bleeding from a partially thrombosed pseudoaneurysm that was not noted on the MRI or Doppler. Lumen-apposing Metal Stents in Pancreatic Fluid Collections Stent failed to deploy in one patient.15 patients (37.5%) required further direct endoscopic necrosectomy Nayana George, MD1, Abhilash Perisetti, MD, FACP2, Debdeep Banerjee, MD3, Saikiran Raghavapuram, (DEN). Stent removal was performed within four weeks after placement in 22/40 (55%) patients, except in MD1, Mohammad Al-Shoha, MD1, Sameer Siddique, MD, MRCP4, Benjamin Tharian, MD, MRCP, DEN where it was delayed (6-20 weeks with a mean of 10.5 weeks). Follow-up ranged from 2 to 15-month FACP, FRACP1. 1University of Arkansas for Medical Sciences, Little Rock, AR; 2Texas Tech University duration. No major complications were noted other than stent migration during necrosectomy in 2 patients. Health Sciences Center, Lubbock, TX; 3University of Florida College of Medicine, Gainesville, FL; 4Einstein Conclusion: EUS-guided cystogastrostomy is a safe and effective procedure for management of sympto- Healthcare Network, Philadelphia, PA matic Pancreatic fluid collection.

Introduction: Endoscopic ultrasound (EUS) guided cystogastrostomy is the preferred intervention for symptomatic Pancreatic Fluid Collections (PFC). Lumens apposing metal stents (LAMS) are increasingly used for cystogastrostomy. Objective: Retrospective chart review of EUS-guided cystogastrostomy using LAMS. 21 Methods: The data of consecutive patients with symptomatic PFC referred for endoscopic management at a tertiary referral center from January 2016 to June 2017 were analyzed. All patients underwent MRI. A single Higher Fluid Volumes at 12 Hours May Be Associated With Poorer Outcomes in Acute operator linear-array echoendoscope (Olympus UCT140) under general anesthesia performed procedures. Pancreatitis: A Retrospective Analysis All patients were administered a prophylactic antibiotic and were admitted for overnight observation. Results: A total of 40 patients were included in the study with male predominance (M: F, 24:16). The mean Peter C. Nauka, BS1, Joaquin Cagliani, MD2, Guillaume Stoffels, MS3, Alex Castaneda, MD2, Rachel age was 42.5 years (14- 83 yrs). Symptoms were abdominal pain in 30 (75%), vomiting in 9 (22.5%) and Gray, BS1, Benjamin Villacres, BS1, Tabia Santos, BS1, Gene Coppa, MD1, Jeffrey Nicastro, MD1, Horacio jaundice in 1 patient (2.5%). The PFC was located predominantly in the body in 28 (70%), followed by L. R. Rilo, MD4. 1Hofstra Northwell School of Medicine, Hempstead, NY; 2Center for Pancreatic Diseases, the head in 4 (10%), and in the tail of pancreas in 8 patients (20%), with some overlap of territories. The Northwell Health, Great Neck, NY; 3Biostatistics Unit, The Feinstein Institute for Medical Research, Man- size ranged from 5.5 cm to 22cm. LAMS were placed in the stomach in 90% cases (36/40) followed by the hasset, NY; 4Hofstra North Shore-LIJ School of Medicine, Great Neck, NY duodenal bulb in 3 (7.5%) and esophagus in 1 (2.5%) (Figure A-D, stent location).15mm stent was used in all except in the duodenum where a 10mm stent was used. The stent was dilated in 28 patients (70.0%) Introduction: Acute pancreatitis (AP) is a common gastrointestinal disorder associated with a high using a radial expansion balloon. First generation LAMS was used in 5 (12.5%), and the second generation morbidity and mortality rate. Current guidelines advocate the need for aggressive fluid resuscitations

[20A] A- EUS with pseudocyst B, C- Cyst drianage.

[20B] D- Lumen Apposing Metal Stent.

within the initial 24 hours to prevent patient deterioration. This study aims to explore whether fluid volume by 12 hours post presentation is associated with poorer outcomes. [20C] Stent location. Methods: This retrospective cohort study was conducted at Long Island Jewish Medical Center (LIJ) and North Shore University Hospital (NSUH) between April and September 2015. 214 non-transferred, adult patients without organ failure were identified using the ICD-9 discharge code for AP (577.0). All patients met the criteria for AP on having two out of three criteria: (1) lipase or amylase >3 times the upper limit of normal, (2) radiological findings of AP, (3) presence of epigastric pain. IV fluid volume at 12 hours was treated as a continuous variable. A univariate followed by a multivariate analysis was [20] Etiology of Pseudocyst conducted to determine if there was an association between increasing fluid volumes and various outcome variables. Results: Multivariate analysis revealed that fluid volume was independently associated with the Etiology of pseudocyst development of necrosis (P<0.01), length of stay (P=0.03) and the development of pleural effusion (P=0.01). Specifically, a 100 mL increase in volume was associated with a 10% increase in the odds Number of Patients Alcoholic pancreatitis Gallstone pancreatitis Idiopathic type Others developing necrosis and a 6% increase in the odds of developing pleural effusion. The multivariate analysis also revealed that hypertension was strongly associated with the risk of developing organ 17 14 6 3 failure (P<0.001). Conclusion: These results further support the hypothesis that necrosis is an early phenomenon, and aggressive fluid resuscitation may not suppress it. The risk of pleural effusion development was higher with increasing fluid volumes. Current guidelines on fluid therapy in AP may not be appropriate for every patient.

The American Journal of GASTROENTEROLOGY VOLUME 112 | SUPPLEMENT 1 | OCTOBER 2017 www.nature.com/ajg Abstracts S13

22 Results: A total of 72 subjects with GFR <60 were included in the analysis; 57% of these were stage 3, 8% were stage 4 and 35% were dialysis dependent. Matched controls consisted of 67% stage 2 and 33% stage 1 CKD patients. There were no significant differences in any of the admission characteristics. Subjects Evaluation of Scoring Systems in the Early Prediction of Outcomes in Acute Pancreatitis with GFR <60 had significantly higher BUN levels and lower Hct levels compared to those with GFR>60. Rachel Gray, BS1, Joaquin Cagliani, MD2, Peter C. Nauka, BS1, Benjamin Villacres, BS1, Tabia Santos, BS1, Subjects with GFR <60 were more likely to have moderately severe to severe pancreatitis compared to Joanna Fishbein, MPH3, Alex Castaneda, MD2, Nibras Ahmed, BS4, Gene Coppa, MD1, Jeffrey Nicastro, the GFR>60 group (33 vs 17, P=0.004). In the GFR <60 group, BUN>20 at admission was not associated MD1, Horacio L. R. Rilo, MD5. 1Hofstra Northwell School of Medicine, Hempstead, NY; 2Center for Pancre- with severity. However, admitting BUN >20 was predictive of severe pancreatitis in subjects with GFR>60 atic Diseases, Northwell Health, Great Neck, NY; 3Biostatistics Unit, The Feinstein Institute for Medical Re- (P=0.007). In subjects with GFR <60, worse severity ratings and longer LOS were significantly more com- search, Manhasset, NY; 4Center of Pancreatic Diseases, Northwell Health, Great Neck, NY; 5Hofstra North mon in subjects with an upward trend in BUN at 24 hours (14 vs 19, P=0.006; 5.0 vs 7.0, P=0.008). Neither Shore-LIJ School of Medicine, Great Neck, NY elevated hematocrit levels on admission nor upward trend in Hct were associated with severity or LOS in either group. Conclusion: An upward trend in BUN at 24 hours may be a useful predictor of clinical severity in patients Introduction: Various classification and scoring systems have been proposed to predict severity and with CKD. Larger study populations and further investigation are needed to evaluate the link between outcomes in acute pancreatitis (AP). However, each of these systems have shortcomings, and no single these prognosticators and other outcomes such as pancreatic necrosis in the CKD population. classification system has effectively predicted severity and outcomes for AP. This study aims to compare the effectiveness of both past and current classification systems in predicting outcomes in AP, including: Original Atlanta Criteria (OAC), Revised Atlanta Criteria (RAC), Determinant based classification (DBC), PANC 3, Harmless Acute Pancreatitis Score (HAPS), Japanese Severity Score (JSS), Symptoms 25 Nutrition Necrosis Antibiotics and Pain (SNNAP), and Beside Index of Severity for Acute Pancreatitis (BISAP). Methods: This retrospective cohort study was conducted at Long Island Jewish Medical Center (LIJ) and The Impact of History of Bariatric Surgery on Acute Cholangitis Mortality and Other Outcomes: North Shore University Hospital (NSUH) between January 2015 and February 2016. Patients met the A 10-Year Nationwide Analysis criteria for AP by having two out of three criteria: (1) lipase or amylase >3 times the upper limit of normal, 1 2 3 (2) radiological findings of AP, (3) presence of epigastric pain. Scores for BISAP, PANC 3, HAPS, SNNAP, Paul T. Kroner, MD, MSc , Marwan Abougergi, MD , Juan E. Corral, MD , Christopher C. Thompson, MD, MHES, FACG4. 1Mount Sinai St. Luke's and Mount Sinai Roosevelt, New York, NY; 2Catalyst Medical OAC, RAC and DBC were calculated for 222 consecutive adult patients. Receiver Operating Character- 3 4 istic curve analysis and Akaike Information Criteria were used to compare each systems effectiveness at Consulting, Simpsonville, SC; Mayo Clinic Health System, Jacksonville, FL; Brigham & Women's Hospital, predicting need for surgery, ICU admission rate, readmission within 30 days, and length of hospital stay. Boston, MA Results: No single system showed significantly superior capacity to predict need for surgery or hospital readmission. However, SNNAP showed the best predictive capacity for readmission within 30 days. Both Introduction: Gallstone formation following rapid weight loss after bariatric surgery (BS) has been RAC and DBC strongly predicted hospital length of stay (P<0.0001 for both systems) and ICU admission observed, with subsequent occurrence of acute cholangitis (AC). However, the complex post-surgical (P<0.0001 for both systems) without any significant differences between them. Additionally, both BISAP anatomy limits the possibility of performing an ERCP as part of AC treatment. Therefore, the aim of this and PANC 3 showed weak predictive capacity at identifying length of stay and ICU admission. study was to assess the impact of bariatric surgery on mortality and resource utilization among patients Conclusion: We suggest that BISAP and PANC3, which can be easily obtained within the initial 24 hours with AC using a national database. of patient presentation, offer an early prediction of length of stay and ICU admission. Subsequently, RAC Methods: This was a case-control study using the National Inpatient Sample 2004-2013, the largest publi- and DBC can offer further information about prognosis later during the disease course. Further research cally available inpatient database in the US. All patients with an ICD-9 CM code for a principal diagnosis is needed to help clarify which combinations of scoring systems are optimal for outcome prediction. of AC were included. There were no exclusion criteria. Patients with a past history of BS were identified using the appropriate ICD-9CM codes. The primary outcome was all cause mortality. The secondary outcome was resource utilization: use of ERCP, cholecystectomy, length of hospital stay (LOS), total hospitalization charges and costs (adjusted for inflation). Multivariate regression analyses were used to adjust 23 for the following confounders: Age, sex, race, income in patients’ zip code, Charlson Comorbidity Index, hospital region, location, size and teaching status. Severe Acute Pancreatitis and Shock: Are High Angiopoetin-2 Levels Causing Endothelial Cell Results: A total of 106,500 patients with AC were included in the study, of which 1,361 (1.3%) had BS. The Dysfunction and a Vascular Leak Syndrome? mean patient age was 61 years and 49% were female. After adjusting for confounders, patients with and without history of bariatric surgery did not display a statistically significant difference in adjusted odds of Annette Wilson, PhD1, Anna Phillips, MD1, Kelley Woods, RN2, Kimberly Stello1, David Binion, MD1, mortality (aOR: 0.55, 95% CI: 0.08-3.91, P=0.55). In terms of resource utilization, patients with bariatric David C Whitcomb, MD, PhD3. 1University of Pittsburgh School of Medicine, Pittsburgh, PA; 2University of surgery had an expectedly lower adjusted odds of ERCP (aOR: 0.36, 95%CI: 0.25-0.52, P<0.01), but simi- Pittsburgh Medical Center, Pittsburgh, PA; 3University of Pittsburgh, Pittsburgh, PA lar odds of cholecystectomy (aOR: 1.45, 95%CI: 0.88-2.42, P=0.14). Both patient groups had similar LOS (adjusted mean difference: -0.33 days, 95% CI: -0.93-0.26,P =0.27), total hospitalization costs (adjusted mean difference: -$692, 95% CI: -$2512 - $1128, P=0.46), and charges (adjusted mean difference: -$2865, Introduction: Acute pancreatitis is an acute inflammatory syndrome originating in the pancreas with 95%CI: -$9472 – $3742, P=0.40). variable progression from injury, to local inflammation, to systemic inflammation to vascular leak Conclusion: A history of bariatric surgery was not associated with different odds of inpatient all-cause syndrome and multi-organ dysfunction syndrome (MODS). Systemic inflammation, measured as mortality among patients who develop acute cholangitis, despite its association gallstone acute pancrea- the systemic inflammatory response syndrome (SIRS), appears to be necessary, but not sufficient to titis and limited ERCP performance. In addition, bariatric surgery does not affect resource utilization in cause MODS. We previously demonstrated that angiopoetien-2 (Ang-2), an endothelial cell paracrine this patient population as measured by length of stay and total hospitalization costs and charges. hormone associated with local vascular leak following injury, was significantly higher on admission in patients who developed persistent organ failure compared with those who did not (PMID: 20461065). This suggested that endothelial cell dysfunction linked SIRS with MODS via VLS with pulmonary edema and hypovolumea/hemoconcentration as primary clinical signs. However, it was not clear whether [25] Adjusted means and odds ratio of evaluated parameters in patients with cholangitis Ang-2 was a consequence, or mediator of endothelial cell injury. Methods: In a follow-up clinical study, clinical information and blood was collected from patients with that had a past surgical history of bariatric surgery, compared to patients with no history severe AP at admission and for 7 days. Serum was assayed for Ang-2 levels (MSD). Human vascular of bariatric surgery endothelial cells were cultured and treated with serum from severe AP patients and synthetic Ang-2 (R&D Systems) at concentrations of 0 (control) to 50,000 pg/ml. Cell morphology and viability were Adjusted Odds Ratio 95% Confidence Interval P-value measured by lactate dehydrogenase release assay, MTT assay, phalloidin fluorescent staining. Caspase 3/7 activation was also measured after 4 hour treatments. Mortality 0.56 0.08–3.91 0.55 Results: Ang-2 levels in patients with severe AP ranged from 3,000 to 260,000 pg/ml, with normal values in our controls being <2,000 pg/ml. The addition of serum from patients with severe, but not mild AP Shock 1.90 0.68–5.28 0.22 induced in endothelial cell stress and death after 24 hour treatments. Synthetic Ang-2 began decreasing cell viability at concentrations of 5,000 pg/ml with IC 50 of 42,500 pg/ml as measured by the MTT ICU 0.80 0.19–3.31 0.76 assay. Caspase 3/7 was not activated by Ang-2. Ang-2 alone did not appear to cause an endothelial stress Multi-Organ Failure 1.18 0.69–2.02 0.55 response, or apoptosis. Conclusion: This is the first study to demonstrate that Ang-2, in levels seen during severe AP in humans, TPN 2.22 1.06–4.66 0.03 result in endothelial cell death. These results provide insight into possible mechanism of severe AP with VLS and prolonged recovery. The mechanism of cell death, and the potential contributions of other fac- ERCP 0.36 0.25–0.52 <0.01 tors in this process are important for future development of effective interventions to prevent or limit MODS. Cholecystectomy 1.45 0.88–2.42 0.15 Adjusted Mean Value Change P-value

Total Charges -$2,865 -$9,472–$3,742 0.40 24 Total Costs -$692 -$2,512–$1,128 0.46

BUN and Hematocrit as Single-marker Prognosticators for Acute Pancreatitis in Chronic Kidney Length of Stay -0.33 days -0.93–0.26 0.27 Disease Patients

Tiffany Y. Chua, MD1, Peter Lee, MD2, Matthew Hoscheit, MD1, Sevag Demirjian, MD1, Tyler Stevens, MD1. 1Cleveland Clinic Foundation, Cleveland, OH; 2University Hospitals Cleveland Medical Center, Cleveland, OH 26 Introduction: There are few if any published studies investigating the validity of BUN and hematocrit as prognostic indicators in patients with chronic kidney disease (CKD). In this study, we investigated Pancreatic Duct Stents Without Internal Flaps Spontaneously Migrate in Most Patients When the validity of BUN and hematocrit (Hct) as prognosticators of severity in acute pancreatitis (AP) in the Inserted for Prevention of Post-ERCP Pancreatitis CKD population. Methods: A retrospective cohort study was carried out comparing AP patients with GF <60 (stages 3, Michael J. Clores, DO1, Eduardo J. Quintero, MD1, Ioannis Papayannis, MD1, Brandon Yim, MD1, Anoop 4, ESRD) to matched subjects with GFR>60 (stages 1 and 2). Patient demographics, labs, severity of AP, Appannagari, MD1, Demetrios Tzimas, MD2, Juan Carlos Bucobo, MD2, Jonathan M. Buscaglia, MD1. organ failure, and length of stay (LOS) were extracted from our single-center database of confirmed AP 1Stony Brook University Hospital, Stony Brook, NY; 2Stony Brook University School of Medicine, Stony admissions. Dialysis records were obtained from GAIA Software, LLC for patients with stage 4 or ESRD Brook, NY baseline renal function. Subjects who underwent dialysis (intermittent, peritoneal or continuous) within the first 24 hours of their AP admission were excluded. Stratified analysis was performed to see if elevated BUN (admission BUN>20 or increased BUN at 24h) or hematocrit levels (Hct≥44% on admission) were Introduction: Pancreatic duct (PD) stents are often placed for the prevention of post-ERCP pancreatitis associated with worse outcomes for either of the two GFR groups. (PEP) in high-risk individuals. Stent insertion within the main PD frequently mandates a 2nd procedure

Racial and Socioeconomic Disparities Among Inpatients With Acute Pancreatitis

Stephanie D. Pointer, MD1, Darwin L. Conwell, MD, MS1, Alice Hinton, PhD2. 1The Ohio State University College of Medicine, Columbus, OH; 2The Ohio State University College of Public Health, Columbus, OH

Introduction: Acute pancreatitis hospitalizations are a significant burden on US healthcare costs. The aim of this study was to assess disparities in healthcare resource utilization among races hospitalized with acute pancreatitis (AP). Methods: Using the 2013 Nationwide Inpatient Sample (NIS), all patients (≥18 years of age) were extracted if they had a primary diagnosis of AP. Healthcare resource utilization was compared between under-represented minorities (URMs) (blacks, hispanics) and white race based on income, type of insurance and type of hospital. Outcome measures included mortality, length of stay (LOS) (days) and cost (dollars). Patients were excluded if they had missing values for race or were listed as a race other than white, black or hispanic. Statistical analyses - SAS 9.4, Cary, NC. Results: A weighted sample of 241,390 inpatients were extracted (White N=167,595; Black N=43,075; Hispanic N=30,720). URM AP. Black and Hispanic patients are more likely to have Medicaid (26.48% and 24.65%; P<0.001, respectively). Black and Hispanic patients are more likely to present with diabetes (27.77% and 29.25%; P<0.001, respectively). Black patients are more likely to be admitted to an urban teaching hospital (57.50%; P<0.001). Black patients are more likely to present with comorbid alcohol abuse, drug abuse, hypertension, and renal failure (40.23%; P<0.001, 12.71%; P<0.001, 68.11%; P<0.001, 14.05%; P<0.001, respectively). Hispanic patients have higher costs ($10,484; P<0.001). White AP. White patients are more likely to be associated with high income (19.74%; P<0.001) and private insurance (34.27%; P<0.001). On multivariate analysis, after adjusting for patient and hospital characteristics, white patients have higher rates of mortality (OR 1.64; P=0.040), longer LOS (0.55 days; P<0.001) and higher costs ($1,021; P<0.001) as compared to black [26] Dislodged pancreatic duct stent at the splenic flexure. patients. In addition, there was no statistically significant difference in mortality, LOS, and cost based on the type of insurance. Conclusion: Racial disparities are evident among the different races hospitalized with AP. Blacks hospitalized with AP are more likely to be on Medicaid. Hispanic hospitalizations are associated with higher costs. While URMs hospitalized with AP are more likely to be associated with comorbid illness, white for stent removal within 2-4 weeks. More recently, PD stents without internal anti-migration flaps have patients have longer LOS and a higher mortality. Further investigation of the etiology for these disparities been utilized to allow for spontaneous migration, thus obviating the need for a repeat procedure. Newly is warranted. developed single-pigtail PD stents without internal flaps have been designed with radiopaque markers, in order to easily document spontaneous passage with a simple XRAY. The aim of this study was to systematically evaluate the spontaneous migration rate of the new Advanix PD stent (external pigtail, no internal flap, single radiopaque marker) when inserted for the prevention of PEP. Secondary aims were to evaluate technical feasibility of stent insertion, and overall safety of the stent. Methods: This was a retrospective analysis of consecutive patients undergoing ERCP between February 2015 and April 2017 who had the Advanix PD stent inserted to prevent PEP. All patients were ordered to undergo abdominal XRAY at 2 weeks following insertion. The presence or absence of the PD stent on follow-up XRAY imaging was recorded. Results: Fifty-three PD stents were placed in 53 patients. Technical success rate of stent insertion was 98.1% (52/53). Median stent length was 3 cm (range, 3-5 cm); and median stent diameter was 5 Fr (range, 4-5 Fr). Overall, 42/52 stents (80.7%) spontaneously migrated; 35/42 confirmed by XRAY, and 7/42 confirmed by alternative imaging prior to XRAY (CT, MRI, endoscopy). In the remaining 10 patients, 2 had their stents removed endoscopically before 2 weeks, and 8 were lost to follow-up. Of those that spontaneously migrated, 19/42 (45.2%) migrated by 2 weeks, 36/42 (85.7%) by 4 weeks, and 40/42 (95.2%) by 3 months. Eighteen patients had a biliary stent positioned adjacent to the PD stent; 77.8% of these patients [28] . (14/18) had spontaneous stent migration, 2/18 had their stents removed endoscopically, and 2 were lost to follow-up. There were no cases of inward stent migration or ductal perforation. Four of 53 patients (7.5%) developed PEP. All were given rectal indomethacin during ERCP. Conclusion: Small size, single-pigtail PD stents without internal flaps appear to migrate spontaneously, as desired, within 4 weeks after insertion in the majority of cases. Even when placed next to a biliary stent, spontaneous PD stent migration seems likely. 29

Outcomes in Intra-ductal Papillary Mucinous Neoplasms in the New York Harbor Veteran 27 Population Matthew D. Grunwald, MD1, Mohammed Nawaz, MD2, Ching-Ho Huang, MD3, Cynthia Victor, DO2, Risk Factors for 30-Day Readmission for Pancreatitis in the National Readmission Database Manuel Martinez, MD4. 1State University of New York Downstate Medical Center, Beechhurst, NY; 2State (NRD) University of New York Downstate Medical Center, Brooklyn, NY; 3SUNY Downstate Medical Center, Brooklyn, NY; 4Brooklyn VA NY Harbor Healthcare System, Brooklyn, NY Jonah Rubin, MD, MA1, Daniel Shoag, PhD2, John N. Gaetano, MD3, Dejan Micic, MD4. 1University of Chicago Medical Center, Chicago, IL; 2Harvard University, Cambridge, MA; 3Northwestern University, Chicago, IL; 4Northwestern University Feinberg School of Medicine, Chicago, IL Introduction: Intra-ductal papillary mucinous neoplasms (IPMN) have become more frequently and incidentally diagnosed in recent years due to increases in imaging procedures. With recommendations based on weak evidence, the only consistent recommendation is for surgical intervention in patients Introduction: Thirty-day readmissions have become a focus for cost reduction for select clinical condi- with main duct lesions, branch duct lesions greater than 3 cm, or a solid component of the cyst. The tions. In addition, early readmission has been an identified risk factor for mortality in acute pancrea- intent of our study is to evaluate the natural history of incidental pancreatic cysts and to determine if titis. Our aim was to utilize the novel National Readmission Database (NRD) to determine the risk these lesions have an effect on morbidity and mortality. factors for 30-day readmission in patients with pancreatitis and evaluate the predictive ability of the Methods: Retrospective chart review was performed on patients with diagnosis of pancreatic cyst, pan- created model. creatic lesion, and intra-ductal pancreatic mucinous neoplasm from the New York Harbor Veterans Methods: We utilized the Healthcare Cost and Utilization Project’s (HCUP) 2013 and 2014 NRD. Index Affairs Hospital from 2000 to 2015. Of 767 patients, 116 were randomly selected for further analysis. admissions had either a primary or secondary diagnosis of pancreatitis (ICD-9: 577, 577.0). Patients with Charts were reviewed for outcomes which included progression of IPMN into disease or death of patient an index admission death and those with an index admission in the month of December were excluded. from pancreatic or other causes. The primary outcome of interest was risk factors for 30-day readmission. Data was analyzed using Stu- Results: Of the 116 patients with pancreatic cysts, 47 patients were diagnosed with IPMN. Baseline dent’s t-test and stepwise, backward multivariate logistic regression analysis. characteristics are outlined in Table 1. In summary, all patients were male, more than half of patients Results: We identified 93,641 (2013) and 98,579 (2014) index admissions with pancreatitis. The were over age 70 at diagnosis and more than half had BMI greater than 25. There were few Hispanic average patient age was 52.3 years with an average of 4.5 chronic conditions per patient in the com- (7%) or Asian (0%) patients, with most patients self-identifying as white (42.6%) or black (40.4%). In bined data set. The average length of stay (LOS) of the index hospitalization was 4.8 days and aver- terms of patient outcomes (Table 2), only 8 of 47 patients had passed away at time of publication, with age cost was $ 40167.76. Utilizing the 2013 data set, within 30 days, 11,659 (12.5%) readmissions 3 passing from a pancreatic etiology, 2 being pancreatic cancer. Only 17 (36.2%) patients underwent were identified. Twenty-four variables were included in the final multivariate model to include: EUS to further evaluate the cysts and 4 of these 17 had FNA sampling of the cyst. Out of 47 patients, 4 Medicaid insurance (OR 1.45, 95% CI 1.36 – 1.55), malnutrition (OR 1.33, 95% CI 1.24 – 1.43), underwent surgical removal of pancreatic lesions and all 4 pathology samples showed IPMN without congestive heart failure (OR 1.33, 95% CI 1.23 – 1.45) and active malignancy (OR 1.89, 95% CI malignancy. 1.67 – 2.12). Protective risk factors included: obesity (OR 0.89, 95% CI 0.84 – 0.94), alcohol use (OR Conclusion: Currently, management of IPMNs is based on consensus rather than strong evidence. This 0.94, 95% CI 0.89 – 0.99) and the performance of a cholecystectomy (OR 0.46, 95% CI 0.43 – 0.5). study preliminarily supports the evolution of management to conservatively observing these lesions. The final multivariate model had an area under the curve (AUROC) of 0.631 (95% CI 0.626-0.636). However, as with many published studies on IPMNs, this study has limitations such as advanced age, Validation of the model utilizing 2014 data demonstrated a similar AUROC of 0.633 (95% CI 0.628- all male sample and multiple comorbidities preventing further invasive interventions. Because of this, 0.639, P=0.518). our findings cannot be applied to the general population. Therefore we recommend further larger mul- Conclusion: Utilizing a cross-sectional nationally available dataset we were able to identify plausible ticenter trials to establish the risk of IMPN’s developing into pancreatic cancer, specifically in the aging readmission risk factors among patients with pancreatitis that is replicated across data sets. However, population. the predictive ability of the final model limits the application of the model and dataset. Further study utilizing factors outside of discharge diagnosis codes is required in order to create validated predictive models.

The American Journal of GASTROENTEROLOGY VOLUME 112 | SUPPLEMENT 1 | OCTOBER 2017 www.nature.com/ajg Abstracts S15

Results: Twenty-one previously identified pancreatic cancer risk genes formed the basis of a systematic [29_A] Baseline Characteristics review of the English literature. We identified 296 germline variants among 65 informative publications. Twelve genes contained 55 variants predicted to be pathogenic, with 18 variants in 5 genes (ATM, BRCA1, Characteristic Number (percent) of Patients with IPMN (N=47) BRCA2, CDKN2A, PALB2) being highly pathogenic. Pancreatic cancer risk genes were organized into different biological mechanisms that may be important in 4 different stages of the oncogenic process. Age at IPMN diagnosis Conclusion: Widely available genotyping technologies are being applied to refine pancreatic cancer risk assessments. We identified multiple, well defined pathogenic genetic variants in pancreatic cancer patients <60 5 (10.6%) that could be used in future risk models. Annotating various pathogenic variants reported in multiple publications provides documentation of pancreatic cancer risk, and therefore facilitates pancreatic cancer 60–69 17 (36.2%) risk assessment. Combining continuously updated variant information within the framework of an oncogenic progression model may be useful for interpreting early biomarkers and directing pathway-specific 70–79 13 (27.7%) treatment for pancreatic cancer in the future. >79 12 (25.5%)

Race

White 20 (42.6%) 31

Black 19 (40.4%) Pancreatic Adenocarcinoma In Chronic Pancreatitis: Prevalence and Associated Factors

Hispanic 7 (14.9%) Shashank Garg, MBBS, Houssam S. Mardini, MD, MBA, MPH. University of Kentucky College of Medcine, Asian 0 (0%) Lexington, KY

Other 1 (2.1%) Introduction: Chronic pancreatitis (CP) is believed to be a risk factor for pancreatic adenocarcinoma HbA1c (PCa; RR 13.3). However prevalence of PCa or factors associated with PCa in patients with CP are not well studied. This study utilized a national deidentified commercial insurance database to evaluate the <5.7% 20 (42.6%) prevalence of PCa in patients with CP and factors associated with occurrence of PCa in CP. Methods: A national deidentified database of 15 million commercially insured US patients between 5.7–6.4% 14 (29.8%) January 1st 2007 and December 31st 2009 was used for the study. All patients with ICD-9 code for CP (577.1) were identified in the database. Patient’s demographics and, ICD-9 codes for PCa (157.0-157.3), >6.4% 13 (27.6%) obesity (278.00), smoking (305.1), hepatic cirrhosis (571.2, 571.5, 571.6) and bile duct obstruction BMI (576.2), and details about statin use were collected. Patients with age <18 years (n=256), unknown gender (n=2) or race (n=932), insurance duration <180 days (n=690), or medical claim for CP made >30 days <18.5 1 (2.1%) after the diagnosis of PCa (n=126) were excluded. Races in the dataset that individually constituted <2% of the total population were grouped together as other races. Categorical data were described in percent- 18.5–24.9 11 (23.4%) ages and analyzed using chi-square test. Continuous data were described as mean with standard deviation and analyzed using t-test. Multivariate logistic regression was used to compare demographics and other 25–29.5 25 (53.2%) variables in CP patients with and without PCa. Two-sided P-value of <0.05 was considered significant. 30–34.9 8 (17.0%) The analysis was performed with SAS software version 9.4 (©SAS Institute Inc., NC, USA). Results: A total of 12,785 (0.085%) patients had documentation/medical claim of CP in the database. >34.9 2 (4.3%) 2,006 patients were excluded based on the exclusion criteria. Final analysis included 10,779 CP patients and 452 (4.19%) patients had a medical claim for PCa (Table 1). Multivariate regression showed that age (OR 1.04, 95% CI 1.03-1.05; P<0.0001), male gender (OR 1.70, 95% CI 1.39-2.07; P<0.0001), smoking (OR 1.44, 95% CI 1.16-1.79; P=0.001) and bile duct obstruction (OR 7.72, 95% CI 6.29-9.50; P<0.0001) were independently associated with a subsequent medical claim for PCa in patients with CP (Table 2). Conclusion: CP was present in 0.085% of the study population. A subsequent medical claim for PCa was made in 4.19% of the CP patients. Increasing age, male gender, smoking and bile duct obstruction were associated with a diagnosis of PCa in patients with CP. [29_B] IPMN characteristics and outcomes

Outcome Number (percent) of patients

Size of largest cyst [31_A] Characteristics of final CP patient population with and without PCa

<3 cm 40 (85.1%) Variable No pancreatic adenocar- Pancreatic adenocarci- P-value cinoma (n=10327) noma (n=452) ≥3 cm 7 (14.9%)

Dilated Pancreatic Duct Mean age in years (+SD) 49.86 (13.37) 56.46 (11.54) <0.0001

Yes 9 (19.1%) Gender (%)

No 38 (80.9%) 1.Male 5116 (49.54) 275 (60.84) <0.0001

Septations/Worrisome Features 2.Female 5211 (50.46) 177 (39.16)

Yes 6 (12.8%) Race (%)

No 41 (87.2%) 1. Caucasian 8025 (77.71) 359 (79.42) 0.0108

Patient underwent EUS 2. African-American 861 (8.34) 42 (9.29)

Yes 17 (36.2%) 3. Hispanic 884 (8.56) 20 (4.42)

No 30 (63.8%) 4. Other races 557 (5.39) 31 (6.86)

Death Obesity (%) 1365 (13.22) 52 (11.5) 0.2913

Any Etiology 8 (17%) Smoking (%) 2561 (24.8) 139 (30.75) 0.0042

Pancreatic Etiology 3 (6.4%) Statin use (%) 3235 (31.33) 169 (37.39) 0.0066

Cirrhosis (%) 2371 (22.96) 119 (26.33) 0.0963

Bile duct obstruction (%) 846 (8.19) 182 (40.27) <0.0001 30

Germline Variants in Pancreatic Cancer Families: Systematic Review and Emerging Concepts

Wei Zan, Celeste Shelton, MS, CGC, Phil Greer, MS, Randall Brand, MD, David C Whitcomb, MD, PhD. [31_B] Multivariate logistic regression model of demographics and other variables in CP University of Pittsburgh, Pittsburgh, PA patients with and without medical claim for pancreatic adenocarcinoma*

Introduction: Complex disorders such as pancreatic cancer arise from multiple etiologies and risk fac- Variable Odds ratio 95% CI of Odds Ratio P-value tors that converge to generate a well-defined, acquired cancer syndrome with characteristic pathologic features. Germline mutations are believed to contribute to the risk of pancreatic cancer, but no single Age (in years) 1.040 1.032–1.048 <0.0001 factor is recognized as causative. Instead, genetic analysis from association studies and candidate gene studies reveals many germline variants that may contribute to different steps in this complex oncogenic Male gender 1.696 1.387–2.074 <0.0001 process. Identification of pancreatic cancer-associated pathogenic variants may inform both biomarker Smoking 1.441 1.159–1.791 0.001 analysis and direct precision management. Methods: We conducted a systematic review of the literature on germline pancreatic cancer risk variants Bile duct obstruction 7.719 6.286–9.479 <0.0001 and compiled variant tables with inferred pathogenicity levels. The genes and variants were assessed for functional effects and associated biological systems and pathways. *Area under receiver operator curve=0.747.

32 [32_B] Multivariate regression model of demographics and other variables in PCa patients with and without a preceding medical claim for CP* Chronic Pancreatitis Preceding Pancreatic Cancer: Prevalence and Associated Factors

2017 Presidential Poster Award Variable Odds ratio 95% CI of Odds Ratio P-value

Shashank Garg, MBBS, Houssam S. Mardini, MD, MBA, MPH. University of Kentucky College of Medcine, Age (in years) 0.983 0.975–0.991 <0.0001 Lexington, KY Male gender 1.289 1.055–1.576 0.0131

Introduction: Chronic pancreatitis (CP) is believed to be a risk factor for pancreatic adenocarcinoma Hispanic vs Caucasian 0.577 0.357–0.933 0.0248 (PCa; RR 13.3). However, prevalence of CP and/or factors associated with CP that precede PCa are not well studied. This study utilized a national deidentified commercial insurance database to evaluate the Hispanic vs AA 0.497 0.281–0.879 0.0161 prevalence of CP preceding PCa and factors associated with presence of CP preceding PCa. Hispanic vs Others 0.428 0.234–0.784 0.006 Methods: A national deidentified database of 15 million commercially insured US patients between Janu- ary 1st 2007 and December 31st 2009 was used for the study. All patients with ICD-9 code for PCa (157.0- Smoking 2.722 2.177–3.403 <0.0001 157.3) were identified in the database. Patient’s demographics and, ICD-9 codes for CP (577.1), obesity (278.00), smoking (305.1), hepatic cirrhosis (571.2, 571.5, 571.6) and bile duct obstruction (576.2), and Cirrhosis 2.41 1.907–3.045 <0.0001 details about statin use were collected. Patients with age <18 years (n=61), unknown gender (n=2) or race (n=507), or medical claim for CP made >30 days after the diagnosis of PCa n( =132) were excluded. Bile duct obstruction 3.082 2.501–3.797 <0.0001 Races in the dataset that individually constituted <2% of the total population were grouped together as * Area under receiver operator curve=0.709; AA: African-American. other races. Results: A total of 6,462 (0.043%) patients had documentation/medical claim for PCa in the dataset. 702 patients were excluded based on the exclusion criteria. Final analysis included 5,760 adult patients with a medical claim for PCa and 469 (8.14%) patients had a preceding medical claim for CP (Table 1). Multivariate logistic regression showed that PCa patients with preceding CP were younger than PCa CI 1.91-3.05; P<0.0001) and bile duct obstruction (OR 3.08, 95% CI 2.5-3.8; P<0.0001) were indepen- patients without CP (OR 0.98, 95% CI 0.98-0.99; P<0.0001). Male gender (OR 1.29, 95% CI 1.06-1.58; dently associated with a medical claim for CP preceding PCa. Hispanics with a medical claim for PCa P<0.05), and medical claims for smoking (OR 2.72, 95% CI 2.18-3.4; P<0.0001), cirrhosis (OR 2.41, 95% were less likely to have a preceding medical claim for CP as compared to Caucasians (OR 0.58, 95% CI 0.36-0.93; P<0.05), African-Americans (OR 0.5, 95% CI 0.28-0.88; P<0.05) and other races (OR 0.43, 95% CI 0.23-0.78; P<0.05; Table 2). Conclusion: In patients with PCa 7.96% had a preceding documentation of CP. Younger age, male gender, smoking, cirrhosis and bile duct obstruction were independently associated with a medical claim for [32_A] Characteristics of final PCa patient population with and without CP CP preceding PCa. Hispanics were less likely to have CP preceding PCa as compared to all other racial groups. Variable No chronic pancreatitis Chronic pancreatitis P-value (n=5291) (n=469)

Mean age in years (+SD) 58.93 (13.02) 56.49 (11.5) <0.0001 33 Gender (%) Utility of Peripheral Eosinophilia for Diagnosis of IgG4-Related Disease in Subjects With 1. Male 2776 (52.47) 285 (60.77) 0.0006 Elevated Serum IgG4 Levels

2. Female 2515 (47.53) 184 (39.23) 2017 Presidential Poster Award

Race (%) Sonmoon Mohapatra, MD1, Shounak Majumder, MD2, Raghuwansh Sah, MD2, Ayush Sharma, MBBS2, 2 2 1 2 1. Caucasian 4185 (79.1) 375 (79.96) 0.1058 Mark Topazian, MD , Suresh Chari, MD . Saint Peter's University Hospital, New Brunswick, NJ; Mayo Clinic, Rochester, MN 2. African-American 441 (8.33) 43 (9.17)

3. Hispanic 372 (7.03) 20 (4.26) Introduction: Elevated serum IgG4 (sIgG4) levels are characteristic but not diagnostic of IgG4-related disease (IgG4-RD). Peripheral eosinophilia (PE) (an increased serum eosinophil count (SEC)) has also 4. Other races 293 (5.54) 31 (6.61) been described in IgG4-RD; however, the diagnostic significance of the combination is unknown. We assessed the diagnostic utility for IgG4-RD of elevated serum IgG4 concentration alone and in combina- Obesity (%) 447 (8.45) 55 (11.73) 0.0158 tion with peripheral eosinophilia. Methods: From the Mayo Clinic, Rochester electronic medical record database we identified 585 patients Smoking (%) 627 (11.85) 147 (31.34) <0.0001 between 1/2008 and 4/2017 with above normal levels of sIgG4 (reference range 121-140 mg/dL). Of these Statin use (%) 1393 (26.33) 137 (29.21) 0.1754 the 409 patients who had serum IgG4 measured to differentiate IgG4-RD from another disease were included in this study. SEC (reference range 0.05-0.5 x 109/ L) was available in 379/409 (92.6%) patients at Cirrhosis (%) 583 (11.02) 122 (26.01) <0.0001 the time of diagnosis. The study subjects were divided into Group 1: Elevated serum IgG4 without PE and Group 2: Elevated serum IgG4 and PE. Group differences were tested with the student t test for continu- Bile duct obstruction (%) 867 (16.39) 185 (39.45) <0.0001 ous variables and chi-squared test for categorical variables. Receiver operative characteristic curve was used to illustrate the predictability of IgG4-RD using serum IgG4 and SEC levels.

[33] .

The American Journal of GASTROENTEROLOGY VOLUME 112 | SUPPLEMENT 1 | OCTOBER 2017 www.nature.com/ajg Abstracts S17

to develop alcohol withdrawal (5.9% vs. 5.7%, P=0.002), Ileus (4.1% vs. 3.1%, P=0.04), Acute Respiratory [33_A] Baseline characteristics of study patients Distress Syndrome [ARDS] (4.7% vs. 4.4%, P<0.0001) and were more likely to require endotracheal intubation (3.9% vs. 3.6%, P<0.0001). Weekend admissions were more likely to undergo cholecystec- Characteristics IgG4-RD (n=109) Non IgG4 RD (n=280) P value tomy (17.2% vs. 16.3%, P<0.001) during the same admission. In the other hand, weekday admissions were more likely to develop pancreatic pseudocyst (3.9% vs. 3.5%, P<0.0001) and required pancreatic Age (y) (median (range)) 64 (19–85) 50.5 (7–95) <0.0001 surgeries and cyst drainage (0.13% vs 0.08%, P<0.0001), biliary stenting (2.6% vs. 2.3%, P<0.0001) and ampullary surgery (0.54% vs. 0.49%, P=0.05). They were also more likely to develop End Stage Renal Gender Disease [ESRD] (2.4% vs. 2.2%) and required initiation of hemodialysis (3.1% vs. 2.9%, P=0.0002) during the same admission. Despite these differences in complication and procedure rates, there was no Male (n) % 104 (80.6) 178 (63.5) <0.0005 significant in-hospital mortality difference between the weekend and weekday admissions (2.59% vs 2.57%, P=0.54). Female (n) % 25 (22.9) 102 (36.5) Conclusion: Patients admitted during weekend for acute pancreatitis develop more severe complications IgG4 level (median (range)) 251 (121–3490) 181 (122–900) <0.0001 requiring ICU care. In spite of this, there was no weekend effect in terms of in-hospital mortality between weekend and weekday admissions for acute pancreatitis. Cut off level

>2 x ULN (n) % 70 (64.2) 58 (20.7) <0.0001

>3 x ULN (n) % 45 (41.3) 10 (3.5) <0.0001 35

>4 x ULN (n) % 26 (23.8) 2 (0.71) <0.0001 Men With Acute Pancreatitis Admitted to U.S. Hospitals Have Increased Comorbidities and SEC (median (range)) 0.26 (0.01–4.45) 0.14 (0–2.84) 1 Higher In-hospital Mortality

1 2 3 4 Elevated (n) % 36 (33) 23 (8.2) <0.000 Prabin Sharma, MD , Rodrigo Aguilar, MD , Mark Abi Nader, MD , Shaheryar Siddiqui, MD , Ramkaji Baniya, MD5, Amir Masoud, MD6. 1Bridgeport Hospital, Yale University School of Medicine, Bridgeport, Normal (n) % 65 (59.6) 235 (83.9) CT; 2Marshall University Joan C. Edwards School of Medicine, Huntington, WV; 3Kidney Care Consultants, Memphis, TN; 4University of Texas Health Science Center, Houston, TX; 5Hurley Medical Center, Michigan H/o allergy/asthma (n) % 35 (32.1) 35 (12.5) <0.0003 State University, Flint, MI; 6Yale New Haven Hospital, New Haven, CT

Introduction: Influence of gender in the epidemiology of acute pancreatitis (AP) has been previously reported. We aim to analyze gender based differences of AP hospitalizations in the US over the past decade. We hypothesize that men with AP are likely to develop severe complications and have a higher in-hospital mortality rate. [33_B] Prevalence of peripheral eosinophilia across different organ manifestation in IgG4- Methods: Between 2005 to 2012, data was extracted from the National Inpatient Sample on adult (Age>18 RD years) patients with AP (ICD 9 CM: 577.0). Exclusion criteria was applied for chronic pancreatitis and other pancreatic and biliary disorders (ICD 9 CM: 577.1, 577.8, 577.9, 157.0, 157.1, 157.2, 157.3, 157.4, Organ manifestations (n) Elevated SEC (n) % Normal SEC (n) % P value 157.8, 157.9, 155.0, 155.1, 155.2,156.0, 156.1, 156.2, 156.8. 156.9). Using multivariate logistic regression, in-hospital mortality, comorbidities, complications and procedure rates were compared between the men Orbit (25) 16 (64) 9 (36) 0.0001* and women with AP (SAS version 9.3). Results: During the study period, 293,165 women and 286,573 men with acute pancreatitis were admit- Asthma (35) 19 (54) 16 (46) 0.0001* ted to the US hospitals, with a ratio of 1.03:1. Men admitted for AP had increased number of comorbidities: DM (P<0.0001), Hypertension (P<0.0001), Hyperlipidemia (P<0.0001) and CAD (P<0.0001). Submandibular gland (20) 7 (35) 13 (65) 0.015* Men were more likely to be smokers (23.9% vs.15.6%, P<0.0001) and had history of drug abuse (33.6% vs.19.6%, P<0.0001) and alcohol abuse (19.7% vs. 7.2%, P<0.0001). Women were more likely to have Lung (11) 5 (45) 6 (55) 0.015* biliary tract disorders (39.3% vs 26.8%, P<0.0001). Men were more likely to be evaluated by CT (2.9% Lymph nodes (21) 8 (38) 13 (62) 0.005* vs 2.7%, P<0.0001) while women were more likely to undergo Ultrasound (1.6% vs. 1.5%, P=0.02) and MRCP (1.2% vs. 0.9%, P<0.0001). Men were more likely to develop septic shock (2.2% vs 1.8%, Pancreatobiliary disease (92) 20 (22) 72 (78) 0.02* P<0.0001), ARDS (5.2% vs. 3.9%, P<0.0001) and required more intubation (4.3% vs. 3.0%, P<0.0001). Women were more likely to undergo biliary intervention (cholecystectomy, biliary stenting, ampullary Renal (16) 3 (19) 13 (81) 0.62 surgery). More men received enteral nutrition (1.2% vs. 0.9%, P<0.0001) while more women received TPN (2.6% vs. 2.4%, P<0.0001). Men with AP had a higher in-hospital mortality rate (2.8% vs. 2.4%, Retroperitoneal fibrosis (11) 2 (18) 9 (82) 0.72 P<0.0001). Conclusion: Men are more likely to be hospitalised with alcoholic pancreatitis and women are more likely Receiver operating characteristic curve using any elevation of serum IgG4: x axis: the false positive rate for elevated serum IgG4; y axis: the sensitivity of elevated serum IgG4. Area under curve for serum IgG4 cut off 3.52g/l to be hospitalized with biliary pancreatitis. Men with AP are likely to develop severe complications lead- in group 1=0.74 (P<0.001) and group 2=0.77 (P<0.06) (figure 1A); Area under the curve for serum cut off level ing to higher in-hospital mortality when compared with women. Men with AP may benefit from more of IgG4 4.75g/l in group 1=0.67 (P<0.01) and group 2=0.87 (P<0.004) respectively (Figure 1B). intensive treatment at the onset for better outcome.

36 Results: Among 409 patients with any elevation in sIgG4 levels 129 (31.5%) had a definite diagnosis of IgG4-RD. Compared to prevalence of IgG4-RD in subjects with sIgG4 1-2 XULN (n=281, IgG4-RD The Increasing Diagnostic Value for Endoscopic Ultrasonography in the Setting of CT-negative 20.9%), the prevalence of IgG4RD was higher (P<0.0001) at 2X (n=128, 54.7%), 3 X (n=55, 81.8%) and 4 Patients With Pancreatic Adenocarcinoma: A Presentation of 18 Cases X ULN (n=28, 92.8%) of sIgG4. Age, gender, serum IgG4 levels, SEC, and history of allergy were significantly different in IgG4-RD vs non IgG4-RD groups (Table 1). In subjects with elevated sIgG4, PE was Neil R. Sharma, MD1, Colin Linke, DO2, Akshay Sharma, MD3, Alexander Perelman, DO, MS4, Kevin more likely to be seen in subjects with IgG4-RD vs non-IgG4-RD at 1X (29.7% vs 8.9%, P<0.0001), 2X Lowe, MD, PhD5, Christina M. Zelt, RN, MSN6. 1Parkview Health System, Fort Wayne, IN; 2Loyola (35.9% vs 9.2% P<0.0004) and 3X (42.8% vs 0% P<0.002) of sIgG4 respectively. PE was more common University Medical Center, Maywood, IL; 3Sinai Grace Hospital/Detroit Medical Center, Wayne State in IgG4-RD manifestations above the diaphragm compared to intrabdominal manifestations (60.8% vs School of Medicine, Detroit, MI; 4Yale New Haven Hospital, New Haven, CT; 5Parkview Cancer Center, 21.2%, P<0.001, Table 2). The ability of group 1 versus group 2 to distinguish IgG4-RD from non IgG4-RD Parkview Health System, Fort Wayne, IN; 6Parkview Regional Medical Center, Fort Wayne, IN is demonstrated in the ROC curves (Figure 1). Conclusion: Even in subjects in whom IgG4-RD is suspected, only a minority (~30%) with elevated sIgG4 levels have IgG4-RD; sIgG4 is more specific at higher levels, though never diagnostic. Peripheral eosino- Introduction: The purpose of this article is to present eighteen CT-negative pancreatiuc adenocarci- philia increases with increasing sIgG4 and adds diagnostic value at higher sIgG4 levels. noma cases in which tumors were found on EUS. Methods: From the months of March 2014 to November 2016, eighteen patients were examined retrospectively in which CT with contrast was unable to detect a discrete pancreatic mass. EUS with core needle biopsy was subsequently performed on these patients, resulting in 18 total pancreatic adeno- carcinomas. The patients were analyzed based on tumor location, type, size, and clinical staging. These 34 cases were presented and reviewed at a multi-disciplinary tumor board that include radiology, oncology, pathology, radiation oncology, surgical oncology, interventional endoscopy, and other physicians. Our inclusion criteria required that the CT not have any evidence of a discrete pancreatic mass, there must No Weekend Effect for Hospitalization for Acute Pancreatitis in the United States have been findings of a defined pancreatic mass on EUS imaging, and the confirmation of carcinoma be made by two pathologists Prabin Sharma, MD1, Rodrigo Aguilar, MD2, Mark Abi Nader, MD3, Shaheryar Siddiqui, MD4, Ramkaji Results: During the time period, 127 cases of pancreatic adenocarcinoma that met inclusion criteria Baniya, MD5, Amir Masoud, MD6. 1Bridgeport Hospital, Yale University School of Medicine, Bridgeport, were presented at our large tertiary referral centers’ tumor board; 18 of which were found upon CT; 2Marshall University Joan C. Edwards School of Medicine, Huntington, WV; 3Kidney Care Consultants, EUS to have a pancreatic tumor that was not visible on contrasted CT imaging. These patients were Memphis, TN; 4University of Texas Health Science Center, Houston, TX; 5Hurley Medical Center, Michigan determined to have CT-negative results and EUS-positive results for a total of 18 pancreatic adeno- State University, Flint, MI; 6Yale New Haven Hospital, New Haven, CT carcinoma diagnoses. Of the 18 total masses, eleven were found to be less than 3.0 cm in diameter, while 16 were located in the head of the pancreas. The most common finding on CT scan was bile Introduction: Weekend effect refers to the difference in mortality rate between a weekend and weekday duct dilatation without clear obstruction, which was the prompting factor for performing an EUS. hospital admission. It may owe to reduction in staffing and reduced access to specialists during the week- It should be noted that most patients exemplified clinical signs of pancreatic cancer (weight loss, ends. Mortality rates and clinical outcome for acute pancreatitis (AP) are worse with delay in receiving jaundice, abdominal pain). appropriate treatment. We hypothesize that weekend admissions for acute pancreatitis are associated Conclusion: Of the 127 cases of pancreatic adenocarcinoma with contrasted imaging presented at mul- with poorer outcome compared to weekday admissions. tidisciplinary tumor board, 14% had no measurable mass on contrasted CT imaging. Upon EUS, 39% of Methods: Between 2005 to 2012, data was extracted from the National Inpatient Sample on adult (Age>18 the 18 cases in our study were found to have tumors ≥3.0 cm while 61% had tumors <3.0 cm. In addition, years) patients with Acute Pancreatitis (ICD 9 CM:). Exclusion criteria was applied for chronic pancrea- 72 % of the 18 cases showed biliary ductal dilation but no visible mass. Endoscopic ultrasound appears to titis and other pancreatico-biliary disorders (ICD 9 CM: 577.1, 577.8, 577.9, 157.0, 157.1, 157.2, 157.3, have a diagnostic advantage over CT in the setting of small pancreatic-head adenocarcinoma. If clinical 157.4, 157.8, 157.9, 155.0, 155.1, 155.2,156.0, 156.1, 156.2, 156.8. 156.9). In-hospital mortality, comorbidi- presentation for a patient warrants, EUS should be considered despite a negative high resolution con- ties, complications and procedure rates were compared between the weekend and weekday admissions. trasted CT to evaluate for occult pancreatic carcinoma. Analyses were performed using SAS version 9.3. Results: During the study period, there were a total of 432,303 weekday admissions and 147,435 weekend admissions for AP in the US hospitals. Patients with AP admitted during weekend were more likely

[36] Patient Demographics and Clinical Characteristics

Case Presenting Symptoms CT Results Tumor Location Tumor Size (cm) Nodal Involvement Clinical Stage

1 RUQ abd pain, Jaundice Intrahepatic and extrahepatic biliary dilitation, no mass seen pancreatic head 1.3 x 1.6 0 T1N0M0

2 jaundice Intrahepatic, extrahepatic and pancreatic duct dilitation, no mass seen pancreatic head 2.4 x 3.8 0 T3N0M0

3 epigastric and RUQ abd pain intra and extrahepatic biliary ductal dilitation pancreatic head 2.4 x 2.5 peripancreatic and porta T2N1M1 hepatis (liver mets)

4 jaundice intrahepatic biliary ductal dilitaion otherwise unremarkable pancreas panreatic head 3.2 x 2.9 0 (liver mets) T3N0M1

5 abd pain, weight loss low density fluid collection adjacent to the pancreatic head pancreatic body/tail 4.4 x 4.5 celiac axis and peripancreatic T3N2M0

6 jaundice intrahepatic and extrahepatice biliary dilitaion with significant gallbladder pancreatic head 2.2 x 2.0 peripancreatic T2N1M0 dilitation

7 RUQ abd pain, jaundice dilitation of intra and extrahepatic biliary tree, dilated common bile duct, pancreatic head 1.0 x 1.2 peripancreatic T2N2M0 no mass seen

8 upper abd pain, jaundice intrahepatic and extrahepatic biliary ductal dilitation pancreatic head 3.0 x 3.0 0 T3N0M0

9 epigastric pain, pancreatitis pancreatic body and tail are grossly normal, mild edema involving pancreatic head 3.4 x 3.5 0 T3N0M0 pancreatic head

10 weight loss, jaundice no discrete pancreatic mass is identified, peripancreatic edema present pancreatic head 2.0 x 3.6 porta hepatis (liver mets) T3N2M1

11 elevated liver enzymes prominence of the main pancreatic duct in the pancreatic body and head pancreatic head 1.6 x 1.8 peripancreatic T2N1M0 without evidence od calculus or mass

12 abd pain pancreas unremarkable pancreatic head 2.7 x 1.9 porta hepatis(liver mets) T3N1M1

13 jaundice biliary ductal dilitaion of the intra nd extrahepatic bile ducts, no definite pancreatic head 2.4 x 2.0 0 T3N0M0 obstruction lesion identified

14 weight loss, elevated liver enzymes pancreatic ductal dilitation pancreatic head 1.6 x 1.2 peripancreatic T2N1M0

15 abd pain dialted common bile duct pancreatic head 1.9 x 1.8 porta hepatis T2N1M0

16 weight loss, epigastric pain distal pancreatic duct is dilate pancreatic head 1.6 x 0.25 peripancreatic T3N1M0

17 abd pain, jaundice dilated common bile duct pancreatic head 2.5 x 3.6 peripancreatic T3N1M0

18 follow-up for hx of pancreatic cancer no evidence of mass or adenopathy pancreatic head 1.9 x 1.7 0 T1N0M0

Test, chi-square analysis, and multivariable logistic regression were conducted with IBM SPSS 37 Statistics 23. 0 Results: Among 26,414 patients, 23,313 underwent LC, 2,395 underwent ERCP, and 706 underwent A Systematic Review of Solid Pseudopapillary Neoplasms: An Update Since 2012 both procedures. ERCP had a significantly higher in-hospital mortality rate (3.7% vs 0. 5% P<0.001) compared to the other two groups combined. ERCP was also an independent predictor of mortality 1 1 2 2 Nathaniel H. Kwak, MD , Shivakumar Vignesh, MD , Tomas DaVee, MD, MCSI , Jeffrey H. Lee, MD . when controlling for age, COPD, type II diabetes, cirrhosis, and coronary artery disease (OR=5.02, 95% 1 2 State University of New York Downstate Medical Center, Brooklyn, NY; University of Texas MD Anderson CI=3.78 to 6.68). The ERCP group had a longer LOS (8.1 days) compared to LC (4.24 days P<0.001) and Cancer Center, Houston, TX ERCP-LC (6. 3 days P=0.001), with 74% in the ERCP group with a LOS of 4 or more days compared to 48% in the other two groups combined (P<0.001). ERCP had more patients discharged to a skilled nursing facility or long term care facility (SNF/LTAC) compared to the other two groups combined Introduction: Solid-pseudopapillary neoplasms (SPNs) were first described by Frantz in 1959 and (12.7% vs 4.5% P<0.001) which was significant when controlling for other comorbidities. Within the defined by the World Health Organization in 1996. SPNs are rare cystic tumors of the pancreas which ERCP group having zero or even up to three comorbidities (COPD, DM, cirrhosis, and CAD) did not usually present as large, solitary, well-circumscribed lesions. While the majority of patients have local- affect morbidity or mortality. ized disease, there has been a reported incidence of malignant transformation of 9-15%. An initial lit- Conclusion: We found that ERCP increases the risk for morbidity and in-hospital mortality when com- erature review identified 718 cases of SPN that had been reported in the English language up to 2003. A pared to an invasive procedure, laparoscopic cholecystectomy. Gastroenterologists are often asked to per- follow up systematic review was then performed which identified 2,744 cases of SPN published in the form ERCP in a patient who is “too sick” to undergo surgery, especially when the suspicion for biliary English literature up to 2012. The aim of this study was to determine if there had been any change in the obstruction is equivocal. Although the ERCP group did seem to have more comorbidities, a subgroup number of SPN cases detected and their evaluation or management over time since 2012. analysis did not show that this affected outcome, and rather ERCP itself was an independent predictor Methods: A systematic review of SPNs was performed of all articles published in English in PubMed and for morbidity and mortality. ERCP is not a lower risk procedure than LC and careful consideration and SCOPUS. Predetermined search strings that included the terms "pancreas" and "pseudopapillary" were perioperative risk should be considered. used. A review of the bibliography of all studies was performed as part of the systematic review. Results: 1,170 patients with SPN were identifiedin 49 studies published between September 19, 2012 (end date of the previous systematic review) to May 1, 2017. 1,103 (94.3%) patients were female and the mean age was 33 years (S.D. ± 10.7). The most common symptom was abdominal pain in 52.7% with incidental detection in 28.1%. There were 1,114 patients who underwent pancreatic resection. The mean tumor size was 8 cm (S.D. ± 3.7). Follow up was reported for 870 patients, with mean follow-up of 44.7 months (S.D. ± 48.9). Disease-free survival was documented in 1,069 (91.4%) patients with recurrence in 101 (8.6%). The mean time to recurrence was 35.1 months (S.D. ± 30.7). Chemotherapy for treatment of recurrence was administered in seven cases, while transarterial chemoembolization (TACE) was used to treat liver metastases in two cases. Conclusion: The number of SPNs reported in the literature has increased by just over 40% in nearly five years. These tumors continue to be found primarily in young women and are frequently found with nonspecific symptoms, such as abdominal pain, or are found incidentally. Surgery remains the mainstay of treatment with an excellent long term prognosis.

ERCP Is Associated With Greater Morbidity and Mortality When Compared to Laparoscopic Cholecystectomy

Ankush Sharma, MD1, David Mossad, MD2, Ronald J. Markert, PhD3. 1Wright State University Boon- shoft School of Medicine, Beavercreek, OH;2 Wright State University, Dayton, OH; 3Wright State University Boonshoft School of Medicine, Dayton, OH

Introduction: Endoscopic Retrograde Cholangio-pancreatography(ERCP) is performed to treat biliary disease and as an adjunct to laparascopic cholecystectomy(LC). It is well established that ERCP is associated with risk of pancreatitis, perforation, and bleeding, but because its an endoscopic procedure, can be labeled as “low risk.” We compared morbidity and mortality in patients undergoing ERCP vs LC to better understand the risks of performing ERCP Methods: Using ICD-9 procedure codes from the 2001-2010 National Hospital Discharge Survey, we identified patients who underwent ERCP, LC, or both ERCP and LC. The three groups were compared [38] . on demographics, comorbidities, hospital length of stay (LOS), and disposition. The Mann-Whitney

The American Journal of GASTROENTEROLOGY VOLUME 112 | SUPPLEMENT 1 | OCTOBER 2017 www.nature.com/ajg Abstracts S19

[38_A] Multivariable logistic regression for mortality in ERCP group

95% confidence interval

P value Odds Ratio Lower Upper

age <0.001 1.06 1.05 1.07

COPD 0.080 1.48 .96 2.29

cirrhosis <0.001 4.26 2.50 7.24

ERCP <0.001 5.02 3.78 6.68

CAD <0.001 5.02 2.21 11.39

DM 0.002 2.34 1.35 4.04

Mortality

alive dead Total

ERCP only Count% 2287 89 2376

96.3% 3.7% 100%

others Count% 23760 119 23879 [40A] Proportion of Discharges with Post-ERCP Pancreatitis from 2004 to 2014. 99.5% 0.5% 100%

[38_B] Multivariable logistic regression for mortality in ERCP group

95% confidence interval

P value Odds Ratio Lower Upper

age <.001 1.06 1.05 1.06

COPD .080 1.48 .96 2.29

cirrhosis <.001 4.26 2.50 7.24

ERCP <.001 5.02 3.78 6.68

CAD <.001 5.02 2.21 11.39

DM .002 2.34 1.35 4.04

Drain Placement After Pancreatic Resections: A Retrospective Review

Karina Fatakhova, MD1, Poppy Addison, BS2, Peter C. Nauka, BS2, Leo Amodu, MD3, Nina Kohn, MA, 3 4 1 2 MBA , Horacio L. R. Rilo, MD . Feinstein Institute for Medical Research, Great Neck, NY; Hofstra North [40B] Proportion of ERCP Discharges with Cannabis Use disorder from 2004 to 2014. Shore-LIJ School of Medicine, Hempstead, NY; 3Feinstein Institute for Medical Research, Manhasset, NY; 4Hofstra North Shore-LIJ School of Medicine, Great Neck, NY

Introduction: Post-operative intra-abdominal drains have historically been used for early detection of hemorrhages, anastomotic leaks and pancreatic fistulas. However, growing literature suggests drains may receptor agonist, to mice increases the severity of pancreatitis. We hypothesized that cannabis use is increase the risk of infection, causing mechanical damage to tissues, leading to major complications associated with increased incidence of PEP. The purpose of this study was to investigate the impact of and longer hospital stays. This study examines the number of drains and its impact in postoperative cannabis use on PEP and hospital resources in patients after ERCP. outcomes post pancreatic resections. Methods: The NIS database was queried to identify patients who had an ERCP with or without PEP Methods: A retrospective chart review of 132 adults who underwent pancreatic resections between July from 2004 to 2014. We adopted the previously validated definition for PEP from the NIS database. The 2009 and September 2015 was performed. Fisher’s exact test or chi-square test was used to compare cat- primary outcome was PEP and secondary outcomes were in-hospital death, length of stay and hospital egorical variables and Mann-Whitney for continuous variables. costs. Cannabis use was identified on the basis of ICD-9 codes 304.3* and 305.2* in adults aged 18 Results: Of 132 patients, 6 (4.55%) did not receive a drain, 49 (37.12%) received one drain, and 77 years or more. We excluded cases coded "in remission". Poisson regression models were used to derive (58.3%) two or more. Median drain duration was 27 days and 72 patients (57.14%) were discharged adjusted incidence risk ratios (IRR) for outcomes in patients with Cannabis use compared to those with the drains in place. Out of all patients, 7 (5.3%) developed a fistula, 12 (9.30%) developed a drain without Cannabis use. infection, 31 (24.6%) developed a new intra-abdominal fluid collection and 6 (4.55%) developed acute Results: Of 381,288 discharges (60.8% females) for patients who underwent ERCP, we found a total of pancreatitis post operatively. Additionally, 40 patients (30.53%) were readmitted within a year of the 37,712 discharges for PEP (9.9%). Among the included patients, 1,479 had a cannabis use disorder. The resection. There was no association between number of drains and the development of acute pan- median age was 60 (IQR 46-75) years. From 2004 to 2014, the rate of PEP increased by 23.6 % (8.9% to creatitis, fluid collection, drain infection, fistula, readmission rate, need for reoperation or DVT inci- 11.0%, p-trend <0.01, Fig 1). Cannabis use significantly increased over the study period (0.20 % to 0.70 dence post operatively. There were significant associations between number of drains, ICU admission %, P<0.01, Fig 2). Univariate Poisson regression analysis showed that cannabis use was associated with a (P=0.0005) and hospital length of stay (P<0.0001). 67% increased risk of PEP (IRR 1.67; 95% CI 1.47–1.90). In a multivariate Poisson analysis, adjusting for Conclusion: The number of drains placed after pancreatic resections is not significantly associated with the age, sex, race, Elixhauser Comorbidities, hospital teaching status, diagnostic vs therapeutic indication for development of complications, including pancreatitis, intra-abdominal fluid collections, drain infections ERCP, biliary vs pancreatic indications for ERCP and ERCP related interventions such as sphincterotomy and pancreatic fistulas. Due to the small number of patients without drains, further studies are needed. and pancreatic stent placement, cannabis use remained an independent predictor of PEP (IRR, 1.39; 95% CI 1.21–1.59). Cannabis use was not associated with in-hospital death (IRR, 0.44; 95% CI 0.10–3.15), but was associated with a shorter length of stay (IRR, 0.80; 95% CI 0.76–0.84) and lower hospital costs (IRR, 0.761; 95% CI 0.760–0.762). 40 Conclusion: Cannabis use is on the rise among patients undergoing ERCP. Cannabis use was associated with significant increase in post-ERCP pancreatitis without significant increase in mortality in “real- world” U.S. clinical practice. Cannabis Use Disorder Is Associated With Increased Risk of Post-Endoscopic Retrograde Cholangiopancreatography (ERCP) Pancreatitis: Analysis of the U.S. Nationwide Inpatient Sample (NIS) Database, 2004-2014

Basile Njei, MD, MPH1, Thomas R. McCarty, MD1, Prabin Sharma, MD2, Manpreet Singh, MD3, Lamia Haque, MD1, Harry Aslanian, MD1, Priya Jamidar, MD1, Thiruvengadam Muniraj, MD, PhD1. 1Yale University School of Medicine, New Haven, CT; 2Bridgeport Hospital, Yale University School of Medicine, Bridgeport, CT; 3Brooklyn Hospital Center, Brooklyn, NY

Introduction: Post-ERCP pancreatitis (PEP) is the most common complication following ERCP. Studies in mice with cerulein-induced pancreatitis have shown that administration of anandamide, a cannabinoid

Results: A total of 181,221 patients had index hospitalization with acute cholecystitis where 178,095 [41_A] Analysis of National Readmission Database 2013: Mulitvariate Analysis of (98.3%) patients underwent cholecystectomy and 3167 (1.7%) patients were managed with CCY- Baseline Characteristics for Cholecystostomy-tube Placement in Gallstone-related Acute tubes. Among patients with CCY-tubes, 1,196 (37.8 %) underwent eventual CCY in 2013 where as Cholecystitis 1,971 (62.2%) did not. Patients receiving CCY-tube incurred a 21.4% 30-day readmission rate with poor outcomes during index hospitalization.(Table 2). For patients with index admission with acute cholecystitis; on multivariate analysis, increasing age (OR 1.20, 95% CI 1.16-1.25), men (OR 1.35, Variable Multivariable Odds Ratio P-value (95% CI) 95% CI 1.18- 1.56), coronary artery disease (OR 1.54, 95% CI 1.28-1.85), cirrhosis (OR 1.65, 95% CI 1.20-2.27), atrial fibrillation (OR 1.48, 95% CI 1.27-1.74), diastolic CHF (OR 1.55, 95% CI 1.13- 2.14) and sepsis (OR 4.53, 95% CI 3.53, 5.82) were associated with CCY tube placement (Table 1). Age (OR for 5 year increase) 1.20 (1.16, 1.25) <0.001 Following placement of CCY-tube; on multivariate analysis, the following factors were associated Sex, Male 1.35 (1.18, 1.56) <0.001 with not undergoing a CCY (standard of care): older age (OR 1.16, 95% 1.09-1.23), Elixhauser score 3-4 (OR 1.92, 95% CI 1.03-3.57), cirrhosis (OR 3.23, 95% CI 1.59-6.67) and diastolic CHF (OR 2.44, Elixhauser Comorbidity Index 95% CI 1.33-4.55). Conclusion: Nearly 2 in 3 patients who receive CCY-tube for acute cholecystitis failed to get standard of 0 reference care CCY during one-year longitudinal hospital follow-up. Apart from improving immediate outcomes of index admission, focused healthcare delivery efforts to ensure prompt CCY is necessary. 1–2 1.45 (1.12, 1.87) 0.01

3–4 2.27 (1.70, 3.05) <0.001 ≥5 3.76 (2.70, 5.23) <0.001 42 Type of Insurance Predictors of Severity in Acute Pancreatitis in Native Americans Private reference Khalil Aloreidi, MD1, Chencheng Xie, MD1, Jamal Dudin, MD1, Prince Sethi, MBBS1, Jing Zhao2, Robert Medicaid 1.69 (1.33, 2.14) <0.001 T. Lapp, MD1 1University of South Dakota Sanford School of Medicine, Sioux Falls, SD; 2Sanford Medical Center, Sioux Falls, SD Income

$1–$37,999 reference Introduction: Acute pancreatitis (AP) is the leading gastrointestinal cause of hospitalizations in the United States. The bedside index for severity in AP (BISAP), is a method to assess AP severity. Native $48,000–$63,999 1.30 (1.07, 1.59) 0.01 Americans (NA) constitute a unique population who have not been investigated for predictors of sever- $64000 or more 1.29 (1.03, 1.62) 0.02 ity in AP. The aim of this study is to use the BISAP score to identify NAs with severe AP and predict markers of severity in this population. Secondary aim is determining whether NAs have an increased Hospital type severity of AP compared with Caucasians. Methods: We conducted a retrospective chart review of NA patients who were admitted for AP between Metropolitan teaching 2.87 (1.98, 4.15) <0.001 Jan 2008 and Nov 2016. A randomly selected group of Caucasian patients were also selected for comparison. Patient demographics, clinical, laboratory data and BISAP score were recorded. The associations Non-metropolitan teaching Reference between BISAP and other measurements were evaluated using correlation coefficient and Fisher’s exact Hospital bed size test. Logistic regression was performed to build the predictive model of the severity of AP against multiple variables. small reference Results: There were 92 NA patients with AP. BISAP score ≥ 3 defines severity. Five NA patients (5.4%) developed severe AP. BISAP score correlates with BMI, initial BUN, initial WBC, and systolic blood Large 1.50 (1.07, 2.12) 0.02 pressure (SBP) (P<0.05). There is no significant correlation between BISAP score and length of hospital stay, admission costs and total I.V. fluids. Age ≥60 years was associated with severe AP (P<0.05) using Hypertension 0.80 (0.70,0.92) 0.002 Fisher’s exact test. Multivariate logistic regression demonstrates BUN and SBP associated with severe AP (P<0.05). One mg/dL increase of BUN would result in an increase of odds of severity by 6%. Similarly, CAD 1.54 (1.28, 1.85) <0.001 one mmHg increase of systolic blood pressure would lead to 4% increase in odds of severity. NAs and Cirrhosis 1.65 (1.20, 2.27) 0.002 Caucasians have no difference in severity of AP (OR=1.52; 95% CI, 0.50-4.81). No deaths were recorded in this cohort secondary to AP. Atrial Fibrillation 1.48 (1.27, 1.74) <0.001 Conclusion: A positive correlation between patient’s BMI, WBC, BUN and SBP and BISAP score were found, suggesting increased severity in AP with these factors. In contrast, BISAP score didn’t appear to Sepsis 4.53 (3.53, 5.82) <0.001 correlate with length of stay, cost or total of IV fluids in NA as previously suggested in other studies. Only five NA patients developed severe AP and no deaths were recorded, which may limit conclusions able CHF diastolic 1.55 (1.13, 2.14) 0.01 to be drawn from this study. There is no significant difference in severity between NAs and Caucasians indicating similar management is recommended in both groups.

[41_B] Univariate Analysis for 30-day Readmission Rates, In-hospital Mortality and 43 Hospital Resources for Patients With and Without Cholecystostomy Tube Percutaneous Trans-hepatic Gallbladder Aspiration (PTGBA) was Applicable as a First-line Outcome Cholecystostomy-tube No Cholecystostomy-tube P-value Treatment for Acute Cholecystitis: Retrospective Chart Review (n=3,167) Mean(SE), 95% CI (n=178,095) Mean(SE), 95% CI Tomohito Morisaki, PhD1, Kei Ikeda, MD2, Yuki Takeuchi, MD1, Daisuke Yamaguchi, PhD1, Naoyuki 3 1 4 1 1 Readmissions within 678 (21.4%) 11,938 (6.7%) <0.001 Tominaga, PhD , Keisuke Ario, PhD , Yasuhisa Sakata, MD, PhD , Seiji Tsunada, PhD . Ureshino Medical 2 3 30 days Center, Ureshino, Saga, Japan; Nagasaki University Hospital, Nagasaki, Japan; Saga-Ken Medical Center Koseikan, Saga, Japan; 4Saga Medical School, Saga, Japan In-hospital Mortality 89 (2.8%) 510 (0.3%) <0.001 (Index Admission) Introduction: Several studies, including the clinical guidelines, have recommended early or emer- Total Charges 66,399 (4,526);(57,522, 47,126 (603);(45,943, 48,310) <0.001 gency cholecystectomy for management of acute cholecystitis. In practice, emergency cholecys- 75,277) tectomy is sometimes hazardous for high-risk patients due to several causes such as inadequate equipment for early cholecystectomy. First-line treatments might include conservative managements Length of Stay 7.87 (0.26); (7.37, 8.37) 3.60 (0.02); (3.55, 3.65) <0.001 with fasting, intravenous fluids, antibiotics, and/or percutaneous gallbladder drainage. Percutane- ous trans-hepatic gallbladder drainage (PTGBD) is most standard approach to gallbladder drainage, and it has not been clearly demonstrated whether percutaneous trans-hepatic gallbladder aspiration (PTGBA) provides sufficient benefit in the clinical settings. We conducted a retrospective analysis to evaluate the treatment outcomes of PTGBA in order to evaluate whether it was available for the 41 first-line treatment of acute cholecystitis. Methods: Retrospective chart review was performed for 188 patients with acute cholecystitis treated Cholecystostomy Tube Placement in Gallstone-Related Acute Cholecystitis: Implications From a in our hospital between 2010 and 2017. The patients divided into the three groups based on the sever- National Survey ity grading in the Guidelines of the Japanese Society of Gastroenterology; 105 patients: Grade 1 (mild) cholecystitis, 71 patients: Grade 2 (moderate), 12 patients: Grade 3 (severe). The treatment effectiveness Ravi Babu Pavurala, MD1, Rohan M. Modi, MD2, Vimal K. Narula, MD2, Kyle Porter, MAS3, Darwin L. was analyzed in the 67 patients who underwent PTGBA, which compared to the patients who received Conwell, MD, MS1, Somashekar G. Krishna, MD, MPH2. 1The Ohio State University College of Medicine, conservative management and/or PTGBD. Columbus, OH; 2The Ohio State University Wexner Medical Center, Columbus, OH;3 The Ohio State Uni- Results: Treatment response rate in the PTGBA was 100% (28/28 patients) in the patients with Grade 1 versity, Columbus, OH cholecystitis, 88.2% (30/34) in those with Grade 2, and 100% (5/5) in those with Grade 3. Conservative management was effective as high as 84.3% (70/83) in patients with Grade 1, and the effective response rate of conservative therapy was only 24.0% (6/25) in the patient with Grades 2 & 3, which was sig- Introduction: While cholecystectomy (CCY) is the standard of care for gallstone-related acute nificantly less than that of PTGBA. The overall PTGBD response rate was 100% (29/29) in the patients cholecystitis, cholecystostomy (CCY) tube is an alternative option in patients with significant comorbid with all severity levels, and there was no statistically significant difference in the response rate between conditions. We sought to determine predictors, immediate and longitudinal hospital outcomes of PTGBD and PTGBA in the present evaluation. Four patients, who could not be treated with PTGBA, were patients who underwent CCY-tube placement on a national level in the United States (US). successfully treated by followed PTGBD. Methods: We identified all adults (age ≥18 years) with a primary diagnosis of acute cholecystitis from Jan- Conclusion: PTGBA could reasonably apply as the first-line treatment for acute cholecystitis because Nov 2013 in the Nationwide Readmission Database. Patients with concurrent diagnosis of acute pancrea- the treatment outcome was comparable to PTGBD as demonstrated in the present study, as PTGBA titis, choledocholithiasis, hepatobiliary, and intestinal malignancy were excluded. Outcomes of patients required the simple techniques, easy patient-management after the procedure, and the low burden undergoing CCY and CCY-tube were compared. Univariate and multivariate analyses were performed to for the patient. identify (a) predictors of CCY-tube placement and (b) predictors of not undergoing standard of care CCY following CCY-tube during longitudinal inpatient follow-up of 1 year.

The American Journal of GASTROENTEROLOGY VOLUME 112 | SUPPLEMENT 1 | OCTOBER 2017 www.nature.com/ajg Abstracts S21

44 Methods: We developed a patient-level Monte Carlo simulation model of 10,000 patients; all patients were 55 years old and followed for up to 15 years. Model inputs including initial cyst size, growth rates, presence of a solid component, and pancreatic duct size were distributed as per Table 1. The percentage Comparison of Missed Cancers and Surveillance Imaging Between Pancreatic Cyst Guidelines in of malignant transformation over 15 years was varied from 3.5% (estimated at 0.24% per year as per Patients With Intraductal Papillary Mucinous Neoplasm (IPMN) AGA guideline) to 25%. The patients were followed at 6-month intervals and outcomes were compared between 3 guidelines: 1) 2015 AGA Pancreatic Cyst Guideline (strict), 2) 2015 AGA Pancreatic Cyst 2017 Presidential Poster Award Guideline (modified to proceed to endoscopic ultrasound (EUS) with any solid component on imaging), and 3) 2012 International Consensus Guidelines (Fukuoka Guidelines). The primary outcome was Jennifer M. Lobo, PhD, Vanessa M. Shami, MD, Victor M. Zaydfudim, MD, Bryan G. Sauer, MD, MSc the number of missed cancers for each guideline. Secondary endpoints included number of MRIs and (Clin Res). University of Virginia, Charlottesville, VA EUS procedures. Results: Based on the rate of malignant transformation over the 15 year period (3.5%, 15%, and 25%), Introduction: Current management of pancreatic cystic neoplasms relies on consensus recommenda- 360, 1510, and 2511 cancers developed, respectively. Table 2 presents the results for cancer diagnoses, tions published as pancreatic cyst guidelines. In the absence of clinical trials comparing management cancer misses, and number of imaging studies. The proportion of missed cancers ranged from 2.3 to strategies, we propose using a patient-level computer simulation model to compare guidelines for man- 2.5% using Fukuoka guidelines and 44 to 48% using the AGA guidelines. The slightly modified AGA agement of intraductal papillary mucinous cystic neoplasms (IPMNs). The objective of this study is to guidelines missed 20-24% of cancers. Imaging was more common with Fukuoka guidelines (12-14 compare rates of missed pancreatic cancer and use of surveillance imaging between different manage- studies) compared to AGA guidelines (6 studies). Figure 1 presents the comparison of missed cancers ment guidelines. (bars) to the average number of imaging studies (dots) for each guideline based on rate of malignant transformation. Conclusion: Management that follows the AGA Pancreatic Cyst Guidelines leads to a significant number of missed pancreatic cancers in comparison to the Fukuoka Guidelines. Further assessment of performance of pancreatic cyst guidelines is warranted to establish recommendations that optimize the rate of cancer diagnosis and the use of imaging.

Characteristics of Significant Fibrosis by Transient Elastography and Evaluation of Non-Invasive Fibrosis Tests in Patients With Primary Biliary Cholangitis

Ahlin Jacob, MB1, Dhuha Alhankawi, MD1, Negar Niknam, MD1, Prasun Shah, MD2, James Park, MD1. 1New York University Langone Medical Center, New York, NY; 2Maimonides Hospital, New York, NY

Introduction: Primary biliary cirrhosis (PBC) affects up to 40 thousand persons in the United States of America. Liver fibrosis can be assessed by transient elastography (TE) and serum tests such as Fibro- sis-4 index (FIB- 4), aspartate transaminase to platelet ratio index (APRI) and aspartate transaminase to alanine transaminase ratio (AST/ALT). Comparison of serum tests with TE in PBC patients has not been studied. The aim was to characterize the population of PBC patients and assess liver fibrosis using transient elastrography (TE) and validated serum tests.

[44] Comparison of Missed Cancers and Surveillance Imaging.

[44_A] Model Inputs

Input Value

Percent Male 70%

Probability of Death from Other Causes Age and sex dependent values

Initial Cyst Size Uniform distribution from 3 mm to 25 mm

Benign Growth Rate 50%: Minimal linear growth

50%: Moderate linear growth

Malignant Growth Rate 60%: Linear growth, 1-2 mm per 6 months

40%: Exponential growth, r=10% per 6 months

Solid Component Can be present up to 1 year before cyst is defined as malignant

Pancreatic Duct 15% of patients can have growth

Growth of 0.5 mm per 6 months [45] Correlation of non significant fibrosis (F0-F1) and significant fibrosis Starts growing up to 3 years before malignancy (F2-F4) to validated serum tests in patients with primary biliary cholangitis (PBC).

[44_B] Results for Cancer Diagnoses, Cancer Misses, and Number of Imaging Studies

3.5% Malignant 15% Malignant 25% Malignant

AGA Modified AGA Fukuoka AGA Modified AGA Fukuoka AGA Modified AGA Fukuoka

Total Number of Cancers 360 360 360 1510 1510 1510 2511 2511 2511

Total Diagnosed 170 258 338 759 1126 1413 1215 1804 2364

Percentage Diagnosed 47.2% 71.7% 93.9% 50.3% 74.6% 93.6% 48.4% 71.8% 94.1%

Total Missed 172 84 9 670 313 34 1161 599 61

Percentage Missed 47.8% 23.3% 2.5% 44.4% 20.7% 2.3% 46.2% 23.9% 2.4%

Average Number of MRIs 6.2 7.8 8.4 6.0 7.4 7.8 5.8 7.1 7.3

Average Number of EUSs 0.2 2.7 5.4 0.2 2.7 4.9 0.2 2.6 4.6

Average Total Imaging Studies 6.3 10.5 13.7 6.2 10.1 12.7 6.1 9.7 11.9

[45] General characteristics and mean values for common lab tests

Variable Mean or Percentage

Male 9.7%

Age 62±9.1

BMI 27.3±6.3

AST 31.6±12.4

ALT 36.2±20.3

ALP 131.3±52.9

platelet 226±46.5

Methods: This is a cross-sectional retrospective study focusing on patients with PBC. A total of 31 patients who have PBC and had TE from March 2014 to March 2017 were included. The correlation between the presence of significant fibrosis by TE and data of other non-invasive tests were performed with crosstab analysis. Chi-square tests were performed to determine statistical significance. Results: The study group was consisting mainly of patients with Caucasian ethnicity at 71 %, and the predominant gender was female at 90.3%. The results of LSMs showed a mean value of 5.6 ± 2.9 kPa. The LSM values corresponding to F2-F4 were seen in 4 of 31 patients (12.9%). A total of 85.7% were anti mitochondrial antibody (AMA) positive, 29.4 % were anti-nuclear antibody (ANA) positive. Twelve of 31 patients (39.3%) with available data were alcohol users at the time of the TE procedure. Among the four with F2-F4 fibrosis per TE measurements, 3 were females (75%) and the mean age was 66±7.1 years. The mean BMI was 27.9±8.3 kg/m2. Fibrosis-4-index tests showed statistical correlation with data from serum tests. For Fib-4-index 100% of patients with F2-F4 by LSM had a value of >1.45 and 64% of patients with F0-F2 had Fib-4 <1.45. For APRI score no patients had a positive value of >1 and no crosstab evaluation could be performed. Conclusion: Our study population were predominantly elderly Caucasian females consistent with other studies but other ethnicity groups are also represented. In PBC patients, Fib-4 was more predictive of [47A] Endoscopic image of circumferential narrowing of duodenum in a patient with significant fibrosis than APRI and AST/ALT, Fib-4 was better for excluding PBC patients with significant annular pancreas, with minor papilla noted (yellow arrow) on rim of annulus. fibrosis than including. Considering the fact that liver biopsy is invasive. Use of these serum tests can be questionable if used alone but It can be a useful tool taking into consideration overall clinical context as well as additional tests used.

The Incidence, Risk Factors and Impact of 30-Day Readmission After Whipple Procedure for Pancreatic Cancer on Mortality and Healthcare Resource Utilization: A Nationwide Analysis

Heather Peluso, DO1, Marwan Abougergi, MD2, Chebli Mrad, MD3, Wesley Jones, MD1. 1University of South Carolina, Greenville, SC; 2Catalyst Medical Consulting, Simpsonville, SC; 3Icahn School of Medicine at Mount Sinai/Mount Sinai West, New York, NY

Introduction: We sought to determine the 30-day readmission rate after a radical pancreaticoduodenectomy for patients with a malignant tumor of the pancreas and its impact on mortality and healthcare utilization in the United States using the largest national readmission database. Methods: This is a cohort study using the 2014 National readmission Database. Discharges were included if they had a procedure code for a radical pancreaticoduodenectomy and had any ICD-9 CM code for pancreatic cancer. Exclusion criteria were age <18 years. A readmission was defined as the first admission to any hospital for any non-trauma diagnosis within 30 days of the index admission. Same day admissions and discharges were excluded. The primary outcome was 30-day readmission. Secondary outcomes were 30-day mortality rate, most common reasons for readmission, readmission mortality rate and resource utilization (length of stay and hospitalization total charges). Independent predictors of readmission were identified using multivariate regression analysis. Results: 6,216 patients with pancreatic cancer had a Whipple procedure in 2014. The mean age was 65.9 years (65.5 - 66.5) and 46.0% were female. The all cause 30-day readmission rate was 22.0%. The in- hospital and 30-day mortality rate after a whipple procedure were 2.8% and 3.2%. The in-hospital mortality rate for readmitted patients was 3.7%. The most common 5 reasons for readmission after a Whipple procedure were sepsis, dehydration, intra-abdominal abscess, pancreatic cancer-related admission and intestinal obstruction. A total of 7,134 hospital days were associated with readmission, with a total healthcare in-hospital economic burden of $50.1 million. Chalrson score predicted readmission (aOR:1.05, CI: 1.00-1.09, P=0.02), while private insurance was associated with lower odds of admission (aOR:0.69, CI: 0.49-0.96, P=0.03) and age,sex, income, ICU admission, shock, hospital volume, teaching status, location and size had no impact. Conclusion: The in-hospital and 30-day mortality after a Whipple procedure for patients with pancreatic cancer are 2.8% and 3.2%, respectively. However, almost a fifth of all patients are readmitted within 30 days after the procedure. Mortality rates after readmission increase to 3.7%, with an associated healthcare in-hospital economic burden of $5.21 billion in 2014. Although mortality from Whipple procedure seems to be decreasing, the healthcare economic burden and resource utilization is still significant after readmision. [47B] Pancreatogram demonstrating annular ring (yellow arrows), with otherwise normal ductal anatomy (i.e. no pancreas divisum).

Annular Pancreas: Endoscopic and Pancreatographic Findings From a Tertiary Referral ERCP Center Methods: This is a retrospective observational study. Our institutional prospective ERCP database was queried for cases of annular pancreas. The electronic medical record was searched for patient Mark A. Gromski, MD, Stuart Sherman, MD, Glen A. Lehman, MD, James L. Watkins, MD, Lee McHenry, and procedure-related data. Technical success was defined as completion of intended imaging and MD, Jeffrey J. Easler, MD, Ihab I. El Hajj, MD, Evan L. Fogel, MD. Indiana University School of Medicine, therapy. Indianapolis, IN Results: From 1/1/94 - 12/31/16, 46 patients with annular pancreas underwent ERCP at our medical center. Indications for ERCP were: chronic abdominal pain (suspicion of sphincter of Oddi dysfunc- Introduction: Annular pancreas is a congenital anomaly whereby pancreatic tissue encircles the duo- tion, n=20, 42.6%), history of acute pancreatitis or recurrent acute pancreatitis (n=16, 34.8%), elevated denum. Current knowledge of endoscopic findings of annular pancreas has been limited to small case liver tests (n=10, 21.7%) or other (n=11, 23.9%). The majority of patients were female n( =33, 71.7%) and series. The aim of this study is to describe the endoscopic and pancreatographic findings of patients with Caucasian (n=40, 87.0%). The mean age at index ERCP was 46.3 ± 17.2 years. Index ERCP was technically annular pancreas at a large tertiary care ERCP center. successful in 40 patients (87.0%) and technical success was achieved in 45 patients (97.8%) after two

The American Journal of GASTROENTEROLOGY VOLUME 112 | SUPPLEMENT 1 | OCTOBER 2017 www.nature.com/ajg Abstracts S23

attempts. Pancreatobiliary neoplasia was identified in 7 patients (17.9%). A duodenal narrowing or 49 ring (Fig. 1) was found in the majority of patients (n=39, 84.8%), yet only 2 (4.3%) had retained gastric contents. Pancreas divisum was found in 21 patients (45.7%), 18 being complete divisum. The annular branch was visualized on pancreatogram in 34 patients (73.9%), arising from the ventral pancreatic duct Incidence and Outcomes of Acute Respiratory Distress Syndrome in Patients With Acute (Fig. 2) in 30 of those 34 patients (88.2%). There were findings of chronic pancreatitis in 15 patients Pancreatitis

(32.6%) noted at the index ERCP. 1 2 2 2 Conclusion: This is the largest series to date describing the endoscopic and pancreatographic findings Harika Balagoni, MD , Vikas Koppurapu, MD , Kshitij Chatterjee, MD , Pooja Gurram, MD , Abhinav Goyal, MD3, Chakradhar Reddy, MD1, Mark F. Young, MD1. 1East Tennessee State University, Johnson City, of patients with annular pancreas. It demonstrates that patients with annular pancreas who undergo 2 3 ERCPs are mostly adults. While children with annular pancreas may undergo surgery for upper GI TN; University of Arkansas for Medical Sciences, Little Rock, AR; Einstein Medical Center, Philadelphia, obstructive symptoms, adults appear to present with pancreatobiliary pathology. We found that 45.7% PA of patients had concurrent pancreas divisum, consistent with prior literature. Nearly one-third of patients had findings of chronic pancreatitis at the time of index ERCP and incidence of pancreato- Introduction: Acute pancreatitis is the leading gastrointestinal cause of hospitalization in the United biliary neoplasia was nearly 18%. It is unclear whether these may be features of the natural history States. Severe pancreatic damage leads to systemic inflammatory response syndrome (SIRS) and affects of annular pancreas. multiple organs including lungs, the severe form of which is acute respiratory distress syndrome (ARDS). We seek to analyze the incidence and in-hospital mortality in ARDS associated with pancreatitis. We also analyzed the predictors associated with the development of ARDS and the predictors of in-hospital mortality in these patients. Methods: Using the National Inpatient Sample 2003-2013, we identified adults (age>18 years) admit- 48 ted for acute pancreatitis with ICD-9-CM code 577.0 Among these patients, ARDS was identified by ICD-9 code 518.82. Patients with missing data for age and gender were excluded. To identify the predictors associated with development of ARDS, we created a multivariate logistic regression model Frequency of Appropriate Use of Pancreatic Enzyme Replacement Therapy (PERT) and with covariates including demographics, Elixhauser comorbidities, and etiology of pancreatitis (alco- Symptomatic Response in Pancreatic Cancer Patients hol/gallstone/hypertriglyceridemia). In-hospital mortality and length of stay (LOS) among these patients was studied. Predictors of in-hospital mortality were examined using similar multivariate Lola Rahib, PhD1, Amy Westermann, PhD1, Jodie A. Barkin, MD2, William Hoos, MBA1, Cassadie logistic regression. Moravek, BS1, Lynn Matrisian, PhD, MBA1, Hongwei Wang, MS3, Lynn Shemanski, PhD3, Jamie S. Results: Between 2003-2013, there were a total of 2,787,590 hospitalizations for acute pancreatitis. Barkin, MD, MACG, MACP, AGAF, FASGE2. 1Pancreatic Cancer Action Network, Manhattan Beach, Among these, 6,163 (0.2%) developed ARDS. The independent predictors for development of ARDS CA; 2University of Miami Miller School of Medicine, Miami, FL; 3Cancer Research and Biostatistics, in these patients were age >65 (OR 1.52, P<0.001), male sex (OR 1.45, P<0.001), alcohol use (OR 1.18, Seattle, WA P<0.001), hypertriglyceridemia (OR 1.56, P<0.001) and obesity (OR 1.43, P<0.001) Among patients with acute pancreatitis and ARDS, the mean age (SD) was 55.6 (17.6) years and 60.4% were males. Introduction: Pancreatic cancer (PC) and its treatments can result in pancreatic exocrine insufficiency Median LOS was 11 days and the in-hospital mortality was 11.4%. The independent predictors of in- that requires pancreatic enzyme replacement therapy (PERT). Appropriate PERT usage is during meals hospital mortality were age >65 (OR 5.93, P<0.001), male sex (OR 1.80, P<0.001) and underlying liver and snacks. The aim was to determine the frequency of appropriate use of PERT and its impact on symp- disease (OR 1.79, P<0.001) tom alleviation in PC through a patient reported outcomes online platform. Conclusion: We noticed that the incidence of ARDS is low among individuals with acute pancreatitis Methods: Users who enrolled in the Pancreatic Cancer Action Network’s Patient Registry (launched and the mortality of ARDS in pancreatitis is 11%, which is lower than the mortality of ARDS from other January, 2016) were prompted to answer a standalone 25 part questionnaire about their experience disease processes.We also noticed that older males are more likely to develop ARDS and have higher inci- with PERT. In May 2016, 5 supplementary questions were added to capture additional symptoms and dence of in-hospital mortality. Furthermore, alcohol use, obesity and hypertriglyceridemia are associated side effects. with increased incidence of ARDS and underlying liver disease is associated with increased in-hospital Results: 136 users completed the enzymes questionnaire by March 2017 (33 prior to addition of the mortality. 5 questions). The median age at registry enrollment was 63 (Range 23-86). 62 (46%) were female, while 14 did not report gender. Seventy percent of patients reported having adenocarcinoma, 9% neuroendocrine, 21% reported other types or unknown. 85 (63%) had surgery; 59 (43%) had radiation therapy, and 112 (82%) had chemotherapy. 115/136 (84%) reported speaking to a healthcare 50 professional about PERT and 104/115 (90%) were prescribed PERT. 68/104 (65%) reported that PERT was prescribed with all meals and snacks, of which 44/68 (65%) reported compliance. There Surgery versus Conservative Management After Endoscopic Retrograde were 63 responses about PERT timing, of which 40 (63%) took PERT with meals, 19 (30%) prior to Cholangiopancreatography (ERCP) for Biliary Stone Disease in High-Risk Veterans meals, and 4 (6%) after meals. Patients who reported taking PERT with meals had higher alleviation of symptoms. “Feeling of indigestion” and “Increased or foul smelling flatus” were significantly Samarth Patel, MD1, Divyanshoo R. Kohli, MD1, Jeannie Savas, BS, MD2, Pritesh Mutha, MD2, Alvin decreased when taking PERT with meals compared to those taking PERT prior to or after meals Zfass, MD, MACG2, Tilak Shah, MD3. 1Virginia Commonwealth University, Richmond, VA; 2Hunter (P=0.005 & P=0.04 respectively). Though not statistically significant, there was a trend that patients Holmes McGuire VA Medical Center, Richmond, VA; 3Hunter Holmes McGuire VA Medical Center, taking PERT with meals were less likely to report “Frequent stools”, “Loose stools”, and “Visible food Richmond, VA particles in stool” compared to taking PERT prior to or after meals. Patients taking PERT with meals reported weight gain and less weight loss. Conclusion: Of the 76% of PC patients prescribed PERT, 65% were prescribed PERT appropriately with Introduction: In otherwise healthy patients, randomized trials demonstrate reduced mortality with all meals & snacks. Overall compliance with PERT administration guidelines was low (38%; 44/104). cholecystectomy (CCY) when compared to conservative management after endoscopic retrograde chol- Improvement in symptoms significantly correlated with appropriate use of PERT. Increase in PC patient angiopancreatography (ERCP) for biliary stone disease. These studies generally exclude older patients and provider education about appropriate PERT usage and administration is warranted. with comorbidities at increased risk of post-operative complications. Our study assessed the benefit of CCY among older veterans with significant comorbidities.

[50] CCI Charlson comorbidity index; ASA American Society of Anesthesiologists; NSQIP National Surgery Quality Improvement Program.

hours. The primary endpoint was the development of moderately severe and severe pancreatitis as defined [50_A] Comparison of baseline characteristics by the revised Atlanta Classification. Results: During the study period 450 patients presented with acute pancreatitis. The mean admission Surgery (n=71) No surgery (N=81) P-value PASS score was significantly higher, 187+9, among the 78 patients with moderately severe or severe pancreatitis versus 131+3 among the 372 with mild disease (P<0.001). We determined that a cutpoint Age (years; mean±SD) 65.07±11.63 72.49±11.64 <0.001 PASS score of 145 optimized sensitivity and specificity and correctly classified 2/3 of the patients (Fig- ure 1). Univariate analysis revealed that admission PASS>145, comorbidities, altered mental status, Males 68 (95.7%) 75 (92.6%) 0.46 Charlson index, and BUN>20 predicted moderate or severe pancreatitis though the latter two variables were collinear with comorbidities. We did not find a correlation between severe pancreatitis and other CCI (Median±IQR) 1±1.7 3±2.2 <0.001 variables including etiology, hematocrit, or BMI. Subsequent multivariate analysis (Table 2) including significant univariate and relevant demographic factors demonstrated that the development of mod- ASA Classification (Median±IQR) 2±0.6 3±0.6 <0.001 erately severe and severe pancreatitis was predicated by a PASS score >145 (OR 3.5 [95% CI 2.0-6.1]). SD standard deviation; CCI Charlson comorbidity index; IQR inter-quartile range; ASA American Society of Admission PASS score >145 was also associated with development of SIRS (excludes SIRS at admission) Anesthesiologists. (OR 3.2 [2.1-4.9]), local complications (OR 2.6[1.4-4.7]), ICU admission (OR 4.8 [2.7-8.4]), intubation, pressors, or initiation of dialysis (OR 5.8 [1.2-28.2]) and they were hospitalized for a mean of 2(+0.6) days longer. Conclusion: A pancreatitis activity scoring system score >145 predicts the development of moderately severe and severe acute pancreatitis and associated adverse clinical outcomes.

[50_B] Comparison of estimated risk of serious post-operative complications to actual risk of serious biliary events in patients managed conservatively

NSQIP (%) Cohort (%) P value

All patients (n=81)

Laparoscopic Cholecystectomy

Risk of death 0.7 3.7 0.33

Risk of complications 4.6 28.4 <0.001

Open Cholecystectomy

Risk of death 1.9 3.7 0.66

Risk of complications 15.7 28.4 0.047

ASA ≥3 and CCI ≥3 (n=43)

Laparoscopic Cholecystectomy

Risk of death 0.89 6.98 0.08

Risk of complications 5.2 25.6 <0.001

Open Cholecystectomy

Risk of death 2.47 6.98 0.16

Risk of complications 17.5 25.6 0.36 [51] Figure 1.

Methods: Medical records of all patients undergoing ERCP for biliary stone-related diseases from July 2008 to December 2016 at a single tertiary-care Veterans hospital were reviewed. Patients with prior CCY were excluded. Among patients who did not undergo CCY, risk of serious post-operative complications or death with open or laparoscopic CCY was estimated using the American College of Sur- geons National Surgery Quality Improvement Program (NSQIP) risk calculator. Charlson comorbidity [51_A] Pancreatitis Activity Index Scoring System index (CCI) and American Society of Anesthesiologists classification system (ASA) were used to assess patient’s functional status. The primary outcome was incidence of serious biliary events or death after Parameter Weight ERCP with conservative management, compared to estimated risk of serious post-operative complications or death with surgery. Organ Failure (Modified Marshall Score>2) X 100 per system Results: Of the 152 patients meeting inclusion criteria, 81 did not undergo CCY (Figure 1). 89% of patients managed conservatively had a sphincterotomy. Patients managed conservatively were signifi- Systemic Inflammatory Response Syndrome X 25 per criterion cantly older and less medically fit than patients who underwent CCY (Table 1). Open CCY was performed in 23/71 patients managed surgically (32%). Serious biliary events occurred in 23 patients (28%) managed Intravenous Morphine Equivalents (mg) X 5 conservatively, including 3 deaths due to cholangitis. Serious post-operative complications occurred in 8 Abdominal Pain (0-10 scale) X 5 patients in the CCY group (11%), including 2 deaths due to post-operative complications. Among patients with CCI ≥3 and ASA ≥3 who were managed conservatively (n=43), the risk of serious biliary events was Tolerating Solid Diet (no=1, yes=0) X 40 significantly higher than estimated risk of serious post-operative complications with laparoscopic CCY (26% vs. 5%, P<0.001). The corresponding estimated risk with open CCY did not differ significantly (28% vs. 17%, P=0.4). (Table 2) Conclusion: Our study suggests that the benefit of laparoscopic CCY for biliary stone disease may out- weigh risks even among older patients with significant comorbidities. Prospective randomized trials that focus on patients with higher ASA or CCI are warranted. [51_B] Predictors of Moderately Severe and Severe Pancreatitis

Categorical Variables N (%) OR Wald 95% Confidence Limit 51 PASS Score >145 171 (38) 3.5 2.0 6.1 PASS (Pancreatitis Activity Scoring System) Predicts Moderately Severe and Severe Acute Pancreatitis Female Gender (versus Male) 216 (48) 0.7 0.3 1.3

Bradford Chong, MD1, Chung Yao Yu, MD1, Michael Quezada, MD1, Ian Roy, MD1, Ben Da, MD2, Hispanic Ethnicity 366 (81) 0.9 0.5 1.8 Negar Yaghooti, MD1, Christopher Ko, MD3, Ira Shulman, MD1, Bechien U. Wu, MD, MPH4, James L. Buxbaum, MD3. 1LAC+USC Medical Center, Los Angeles, CA; 2National Institute of Diabetes and Etiology 3 Digestive and Kidney Diseases, National Institutes of Health, Baltimore, MD; Keck School of Medicine Alcohol 116 (26) Reference Category of the University of Southern California, Los Angeles, CA; 4Kaiser Permanente Los Angeles Medical Center, Los Angeles, CA Gallstones 207 (46) 1.2 0.6 2.5

Other 132 (29) 1.4 0.7 3.1 Introduction: Acute pancreatitis can lead to organ failure in 15-20% of patients. We have previously reported that PASS (Pancreatitis Activity Scoring System) predicts readmission and provides a dynamic Comorbidities 174 (39) 1.4 0.8 2.4 measurement of clinical activity. Our aim is to determine the role of PASS in forecasting the development of moderately severe and severe pancreatitis. Altered Mental Status 19 (4) 4.7 1.5 14.5 Methods: We evaluated consecutive patients who presented with acute pancreatitis from March 2015 to Continuous Variables Mean (SD) OR Wald 95% Confidence Limit April 2017. Only a patient’s first admission for pancreatitis in the study time period was included in the analysis. We prospectively collected data on analgesic administration, etiology, and more than sixty addi- Age 44.9 (15.1) 1.01 0.99 1.04 tional clinical variables. The PASS (Table 1) score was determined at admission, discharge, and every 12

The American Journal of GASTROENTEROLOGY VOLUME 112 | SUPPLEMENT 1 | OCTOBER 2017 www.nature.com/ajg Abstracts S25

52 Our aim was to evaluate the role of plasmapheresis vs standard therapy in a large case-matched series of patients with severe HTG-AP. Methods: This is a retrospective cohort study of 142 patients with HTG-AP who were divided into Revisiting the Management of Hypertriglyceridemia-Induced Pancreatitis plasmapheresis and non-plasmapheresis group. Inclusion criteria was serum TG level equal to or greater

1 2 3 1 than 1000mg/dl, elevated serum lipase and/or abdominal pain. We compared the epidemiologic charac Neethi Paranji, MD , Palashkumar Jaiswal, MBBS , Bashar M. Attar, MD, PhD , Yuchen Wang, MD . teristics, clinical severity and end-points between the groups directly and after successful propensity score 1Cook County Health and Hospitals System, Chicago, IL; 2John H. Stroger, Jr. Hospital of Cook County, 3 match. The clinical trajectory of plasmapheresis and post-match non-plasmapheresis group were plotted Chicago, IL; Cook County Health and Hospital Systems, Chicago, IL and compared. Results: Of the 142 patients, 127 patients were in the non-plasmapheresis group and 15 patients in Introduction: Hypertriglyceridemia (HTG) is an infrequent, although well-recognized cause of acute the plasmapheresis group. Mean serum lipase levels on admission were equivalent between groups pancreatitis (AP). Plasmapheresis has been used in severe cases of HTG-AP to actively and rapidly cor- (781 unit/L vs 1167 unit/L; P=0.391) as were mean triglyceride levels on admission (7411 mg/dL vs 7193 rect elevated TG levels. However, due to the heterogeneity in severity of AP, clinical outcomes used to mg/dL; P=0.848). APACHE II scores (11.4 vs 11.4; P=1.00) and Ranson’s score were equally matched determine efficacy and lack of well-balanced control groups, the role of Plasmapharesis in HTG-AP (3.70 vs 3.73; P=0.928). Around 60% (n=9) of patients in the non-plasmapheresis group required ICU remains in question. admission vs 100% (n=15) in the plasmapheresis group (P=0.017), although there were no statistically significant differences regarding in-hospital mortality, total LOS, complications of acute pancreatitis, or re-admission rate during an average follow-up of 893 days. Conclusion: Our study supports the notion that plasmapheresis is highly efficacious in reducing serum triglyceride level and improving APACHE II score in HTG-AP. However, its benefits and cost- effectiveness compared to standard therapy still remains in question. Larger randomized studies are needed.

Tumor Growth Rate of Pancreatic Serous Cystadenoma: Endosonographic Follow-up With Volume Measurement to Predict Cyst Enlargement

Federico Salom, MD1, William Piedra, MD1, Herbert Burgos, MD2. 1Hospital Mexico, Uruca, San Jose, Costa Rica; 2Gastroclinica, Tibas, San Jose, Costa Rica

Introduction: Serous cystadenomas are benign lesions of the pancreas. They are found mainly on cross- sectional imaging studies as an incidental diagnosis. Most of them are asymptomatic. Endoscopic ultrasound is a valuable tool in the diagnosis and follow-up of these cystic lesions. Given its benign nature, surgical resection is advised only in symptomatic patients. The interval and length of surveillance is not [52A] Average Serum Triglyceride in Plasmapheresis and Non-Plasmapheresis groups well established. up to 96 hours after Admission. Methods: A retrospective single center study with endosonographic evaluation of pancreatic serous cystadenoma, between December 2008 and December 2015 was done. The aim was to search for pancreatic cystic features that can predict a higher risk of cyst enlargement, to define an adequate interval and lenght of surveillance of these lesions. All patients were asymptomatic, had a polycystic lesion, microcystic or oligocystic with a carcinoembryonic antigen level in cystic-fluid of less than 5 ng/dl; had no mural nod- ules and no pancreatic duct dilation. All the lesions were follow-up endosonographically at least once, in a 12 months interval. Volume of the cyst was measured with the formula Pi/6 x (d1 x d1 x d2). Two groups were evaluated. Those with a volume of less than 10 ml and those with a volume of 10 ml or more at presentation. Growth rate measurement between these two groups, as well as between microcystic and oligocystic lesions were compared. Results: Thirty-one patients were evaluated, all had a lesion classified as a serous cystadenoma. Twenty- seven patients (87.1%) had microcystic lesions with a mean growth of 1.89 ml per year. Only 4 patients had oligocystic lesions (12.9%), with a mean growth of 11.2 ml per year, being statistically significant (P=0.0007). After evaluating volume growth rates, those lesions with less than 10 ml at presentation (24 patients) had mean growth of 0.67 ml per year, where as those lesions with 10 ml or more had a mean growth of 9.8 ml per year (P=0.0001). Conclusion: Patients with a diagnosis of serous cystadenoma who are asymptomatic should be follow-up for growth evaluation. Oligocystic lesions and cystic lesions of 10 ml of volume or more at presentation have a more rapid enlargement during surveillance. Cysts of less than 10 ml in volume at presentation had practically no increase in their volume, compared with higher volume cysts, with statistical significance. Cyst pattern (oligo or microcystic) and volume at presentation are features that should be evaluated when deciding interval of surveillance of serous cystadenomas.

[52B] Average APACHE II score in Plasmapheresis and Non-plasmapheresis groups up to 96 hours After Admission. Abbreviation: APACHE, acute physiology and chronic health evaluation.

[52] Pre-match and Post-match Comparison Between Plasmapheresis and Non- plasmapheresis Group

Variables Pre-match Post-match

Apheresis Non-Apheresis P-value Non-Apheresis P-value (n=15) (n=127) (n=15)

Triglyceride(mg/dl) 7193(2812) 3511(2465) <0.001 7411(3385) 0.848

APACHE II Score 11.4(4.3) 9.1(5.6) 0.06 11.4(5.7) 1.00

Ranson's Criteria 3.73(1.98) 3.07(1.69) 0.08 3.7(2.0) 0.928

Length of NPO (days)1 5.1(3.2) 3.9(3.4) 0.039 4.1(1.5) 0.773

Length of stay (days)1 10.3(4.9) 6.9(4.9) 0.003 7.7(3.2) 0.115

SIRS 6(40.0%) 70(55.1%) 0.267 8(53.3%) 0.715 [53] .

1 Presented as mean value (standard deviation). Abbreviations: APACHE, acute physiology and chronic health evaluation; NPO, nil per os; SIRS, systemic inflammatory response syndrome.

54 ences were observed in age and BMI- adjusted HOMA-IR and HOMA-BCF between PDAC-DM,CP-DM and T2DM(Table 1).Notably,patients with PDAC-NOD(n=41) had lower age- and BMI-adjusted HOMA- BCF as compared to those with Type 2 NOD(n=39)(43.89% vs 74.61%; P=0.0163*),with similar HOMA- Comparison of Insulin Resistance and Beta Cell Function in Patients With Type 3c IR (3.86 vs 6.34; P=0.1388). (Pancreatogenic) and Type 2 Diabetes Mellitus Using Homeostatic Model Assessment (HOMA) Conclusion: IR and BCF are similar in CP-DM,PDAC-DM and T2DM.However,lower HOMA-BCF in patients with Type 3c DM and PDAC-NOD might suggest a significant role ofβ -cell dysfunction 2017 Presidential Poster Award in their pathogenesis, as opposed to T2DM which is driven primarily by insulin resistance, with β -cell dysfunction at later stages.Identification of biomarkers responsible for these differences might allow Sajan J. Nagpal, MBBS, Rita Basu, MD, William Bamlet, MS, Suresh Chari, MD. Mayo Clinic, Rochester, MN for earlier detection of PDAC by distinguishing PDAC-NOD from Type 2NOD and other forms of Type 3c DM. Introduction: Diabetes mellitus (DM) resulting from chronic pancreatitis(CP) and pancreatic ductal adenocarcinoma (PDAC) is classified by the American Diabetes Association as a form of pancreatogenic DM, which has also been referred to as type 3c DM.However, there are no data on the comparison of insulin resistance(IR) and β -cell function(BCF) among patients with Type 3c versus Type 2 DM(T2DM). Also,new-onset DM (< 2 yrs from diagnosis) resulting from PDAC(PDAC-NOD) may have a distinct pathophysiology from CP-DM.The Homeostatic Model Assessment (HOMA) estimates IR and BCF using fasting serum glucose and insulin values. Using HOMA,we aimed to distinguish:a) Type 3c DM from T2DM,b) Among CP-DM,PDAC-DM and T2DM, and c)between PDAC-NOD and new-onset 55 T2DM(Type2NOD). Methods: A total of 265 patients with PDAC,112 with CP and 118 controls seen at our center between Risk of Progression of Pancreatic Cysts in Solid Organ Transplant Recipients: A Systematic 1996 and 2011 had fasting levels of glucose,insulin and c-peptide measured as part of other studies. Review and Meta-Analysis Patients with unknown glycemic status, impaired fasting glucose,no DM and those on insulin were excluded. Overall, 90 patients with PDAC-DM, 24 with CP-DM and 46 with T2DM were included in the 2017 Presidential Poster Award

final analysis. IR and BCF were measured using standard formulae for HOMA. 1 2 1 1 Results: Mean age of CP patients(53.35 years) was lower than that of PDAC(66.22 years) and T2DM Jodie A. Barkin, MD , Elie Donath, MD , Jatinder Goyal, MD , Zsuzsanna Nemeth, MLIS , Enrico O. Souto, MD1, Paul Martin, MD, FACG1, Jamie S. Barkin, MD, MACG, MACP, AGAF, FASGE1. 1University (66.10 years) respectively (P<0.0001*).Patients with T2DM had higher BMI(29.24 kg/m2) vs CP(24.52 2 kg/m2) and PDAC (26.28 kg/m2) respectively;P<0.0001*.Among patients with Type3c and T2DM, while of Miami Miller School of Medicine, Miami, FL; University of Miami Miller School of Medicine, Atlantis, there was no difference in age- and BMI-adjusted mean HOMA-IR(4.00 vs 6.21 respectively,P =0.5051), FL HOMA-BCF was lower in patients with Type3cDM(45.09% vs 73.90% respectively,P=0.0356*).No differ- Introduction: Pancreatic cysts are a common incidental finding in liver transplant candidates. Immunosuppressive medications used in solid organ transplant recipients (SOTRs) increase the risk of development of malignancies. The risk of progression of pancreatic cysts in SOTRs remains unknown. [54] Comparison of Mean Glucose, C-peptide, Insulin, HOMA-IR and HOMA-BCF values The aims of this study were to establish the incidence of pancreatic cyst progression in SOTRs and to establish the relative risk of progression in SOTRs compared to controls. between patients with PDAC-DM, CP-DM and Type 2 DM Methods: A systematic search was performed using PubMed, Embase, and Cochrane databases to April 21, 2017, without language or year limitations, with a medical reference librarian. Search terms PDAC-DM CP-DM Type 2 DM Unadjusted Adjusted were "pancreatic cyst" or "intraductal papillary mucinous neoplasm" (IPMN) and "solid organ trans- (n=90) (n=24) (n=46) P value (age and bmi) plant" with all associated permutations. Studies of adult SOTRs with pancreatic cysts or IPMNs fol- value P lowed to assess cyst progression were included for analysis. Case reports/series (≤3 patients) were excluded. A recursive search was performed of the references of included articles. The primary Glucose (mg/dl) 166.78±72.18 138.67±44.74 147.57±23.85 0.0504 0.0686 outcome measure was a composite outcome of new development of worrisome features defined by Insulin (microU/mL) 10.49±10.55 6.58±5.73 16.72±10.13 <0.0001* 0.0750 the Fukuoka criteria, high-risk features, consensus indication for surgical resection or malignancy. Overall incidence and relative risk was calculated using a meta-analysis with fixed and random effects C peptide (nmol/L) 1.00±0.72 0.75±0.54 1.40±0.62 <0.0001* 0.1855 models. Results: There were 120 total search results. All results were reviewed in abstract form by 2 authors (JAB HOMA-IR 4.47±6.45 2.22±1.80 6.21±4.24 <0.0001* 0.2239 & JG). Nine studies of 446 patients with pancreatic cysts who underwent solid organ transplantation met inclusion criteria, dating from 2009 to 2016. Of the 9 studies, 4 had control groups (98 cases vs. 488 HOMA-BCF (%) 46.50±37.72 39.79±42.63 73.90±42.97 <0.0001* 0.10 controls with cysts). The overall incidence of cyst progression to the primary outcome in SOTRs was 4% via fixed & 7% via random effects models (30/446 SOTRs) in 9 studies, compared to 13.5% (66/488) in controls in 4 studies, with relative risk of 1.23 (95%CI 0.65-2.31; P=ns) in cases vs. controls (Figures 1-2).

[55A] Forest Plot of Incidence of Progression to Primary Outcome in all Solid Organ Transplant Recipients.

[55B] Forest Plot of Relative Risk of Progression to Primary Outcome in Studies of Solid Organ Transplant Recipients Compared to Controls.

The American Journal of GASTROENTEROLOGY VOLUME 112 | SUPPLEMENT 1 | OCTOBER 2017 www.nature.com/ajg Abstracts S27

There were 2 total cases of development of pancreatic cancer (0.45%) in SOTRs and 1 in the control group 58 (0.2%). No attributable deaths were reported. Conclusion: This is the first systematic review and meta-analysis examining risk of progression of pancreatic cysts in SOTRs. The risk of pancreatic cyst progression in SOTRs remains low with no significantly Pancreatic Pseudocysts and Parenchymal Necrosis in Patients With Autoimmune Pancreatitis: A increased risk compared to controls. Presence of pancreatic cysts should not be considered a contraindi- Systematic Review cation to solid organ transplantation. Jean Donet, MD, Jodie A. Barkin, MD, Tara Keihanian, MD, MPH, Zsuzsanna Nemeth, MLIS, Jamie S. Barkin, MD, MACG, MACP, AGAF, FASGE. University of Miami Miller School of Medicine, Miami, FL

Introduction: Autoimmune Pancreatitis (AIP) is being increasingly recognized as a cause of acute pancreatitis (AP). The resulting potential complications of AP such as pancreatic pseudocysts (PC) and 56 parenchymal necrosis and their natural history and management are unknown in patients with AIP. The aim of this systematic review was to examine the prevalence of the AP complications of PC and Cocaine-induced Acute Pancreatitis: A Systematic Review pancreatic necrosis in patients with AIP as well as their characteristics, management and outcomes. Methods: A systematic search using PubMed/Medline, Embase, Scopus, and Cochrane was performed Jodie A. Barkin, MD1, C. Roberto Simons-Linares, MD, MSc2, Zsuzsanna Nemeth, MLIS1, Jamie S. Barkin, without language or year limitations to October 1, 2016 with a medical reference librarian. Search terms MD, MACG, MACP, AGAF, FASGE1. 1University of Miami Miller School of Medicine, Miami, FL; 2John H. were “Autoimmune pancreatitis”, “pancreatic pseudocyst”, “acute fluid collection” and “pancreatic necro- Stroger, Jr. Hospital of Cook County, Chicago, IL sis” with all permutations. AIP was defined according to international guidelines. The search yielded 271 results. 17 case reports dating from 2004 to 2015 were included in this systematic review. Results: 12/17 (70.6%) patients were men, with a mean age of 54.82 ± 10.97 years (Range 26-75 years). Introduction: Cocaine is a commonly consumed, highly dangerous, illicit drug responsible for multiple In 8/17 cases, the PC were noted concurrently with the AIP diagnosis, while in the other half, the PC medical problems. Cocaine was initially postulated as a cause of Acute Pancreatitis (AP) in 1990 (Bernad appeared several months or even years after the initial AIP diagnosis. In 10/17 cases, the PC appeared on M, et al. An Med Interna 1990). Presently, the etiology of AP remains unknown in approximately 20% imaging as solitary, while in the remaining 7 there were multiple PC. The location of the PC was variable, of cases even after exhaustive investigations. The aim of this systematic review is to investigate cocaine being present in any segment of the pancreas. PC ranged in size from small (3cm) in 13 cases. 9cm was as a potential etiology of AP. the maximal diameter reported. 6/17 (35%) cases had a normal pancreatic duct on imaging, while 9/17 Methods: A systematic search using PubMed/Medline, Embase, Scopus, and Cochrane was performed (53%) had a narrowed/stenotic pancreatic duct. 14/17 cases received steroid therapy (4/4 small, and 10/13 with a reference librarian, without language or year limitations through June 1, 2017. Search terms were large PC). All 4 small PC responded completely to steroid therapy, while only 4 of 10 (40%) with large PC “Cocaine” and “Acute Pancreatitis” with all permutations. AP was defined by meeting 2 of 3 Revised treated with steroids had response. 9 of the 13 cases of large PC eventually went on to undergo endoscopic Atlanta Classification criteria (Pain consistent with AP; amylase or lipase >3x upper limit of normal; or surgical interventions. None of the 17 cases had pancreatic necrosis. imaging consistent with AP). Cocaine induced AP was defined by preceding use of cocaine, and exclusion Conclusion: The development of PC in AIP is a rare phenomenon with only 17 cases reported. PC can be of common causes of AP when reported (significant alcohol use, biliary, hypertriglyceridemia, medica- single or multiple with variable size and location. Small PC all responded to steroid therapy. Conversely, tions, hereditary). Two authors reviewed each study for eligibility. The search yielded 171 results. Eight large PC had poor response to steroid therapy, and required endoscopic or surgical drainage. AIP has not case reports met inclusion criteria dating from 1996 to 2017. been reported to be associated with development of pancreatic necrosis. Results: There were 8 total cases of cocaine induced AP. Seven of 8 (87.5%) were men, with an average age of 26.4 years (range 19-53 years), with 7/8 cases under age 30. Development of AP occurred within 48 hours of cocaine intake in 7/8 patients. Three patients had a history of, but no current use of alcohol prior to the onset of AP. There were no further episodes of AP after cocaine cessation in all 5 patients for whom data 59 was available. Conclusion: Cocaine should be considered as a potential cause of toxin induced AP. Its mechanism remains unknown, but is likely due to a combination of its physiologic effects which include vasoconstric- Update on the Effect of Region-Wide Incorporation of an Algorithm Based on 2012 International tion, micro-arterial thrombosis, and ischemia. Cocaine induced AP occurs primarily in young patients Consensus Guideline on the Practice Pattern for the Management of Pancreatic Cystic Neoplasms below age 30. Toxicology screens should be considered as standard of care in all patients under age 30 in an Integrated Health System with AP. Inclusion of cocaine as an etiology of AP may reclassify idiopathic AP cases. Further studies are needed to investigate this association. 2017 Presidential Poster Award Brian Lim, MD, MCR1, Agathon Girgis, MD2, Emilio Alonso, MD2, Andrew Nguyen, MD, MBA3, Timnit Tekeste, MD2, Karen Chang, DO2, Mopelola Adeyemo, MD2, Armen Eskandari, MD4, Priya Yaramada, MD5, Charles Chaya, MD6, Edmund Burke, MD5, Albert Ko, MD5, Rebecca Butler, ScM7, Aniket Kawatkar, 57 PhD, MS8, Isaiah Roggow, MS2. 1Kaiser Permanente Riverside Medical Center, UC Riverside School of Medicine, Riverside, CA; 2UC Riverside School of Medicine, Riverside, CA; 3University of California Online Case Challenge With Educational Reinforcement: Does it Improve Physicians’ Knowledge Riverside School of Medicine, Moreno Valley, CA; 4UC Davis Medical Center, Sacramento, CA; 5Kaiser and Competence in EPI? Permanente Riverside Medical Center, Riverside, CA; 6Kaiser Permanente Riverside Medical Center, Riverside, AR; 7Department of Research and Evaluation, Kaiser Permanente Southern California, Jovana Lubarda, PhD1, Julia Muino2, Keith Johnson, MBA3, Scott Tenner, MD, MPH, JD4. 1Medscape, LLC, Pasadena, CA; 8Kaiser Permanente Southern California, Pasadena, CA Burlington, ON, Canada; 2Medscape LLC, New York, NY; 3Medscape, LLC, New York, NY; 4State University of New York, Brooklyn, NY Introduction: Pancreatic cystic neoplasms (PCN) have become a challenging entity for primary/acute care physicians, gastroenterologists, radiologists and surgeons. International consensus guideline was Introduction: The goal of this study was to determine if online, case-based continuing medical educa- developed to assist in the management of PCNs. Based on the 2012 revision, Kaiser Permanente Southern tion (CME) could improve knowledge and competence of physicians managing patients with exocrine California (KPSC) published an algorithm in October 2013, including a summary of recommendations pancreatic insufficiency (EPI). in radiology reports. We examined the practice pattern in KPSC, i.e. gastroenterology (GI)/surgery refer- Methods: rals and endoscopic ultrasound (EUS), for PCNs after the region-wide dissemination of this algorithm. • Physicians participated in a text-based online CME activity composed of 2 patient cases with Methods: Retrospective review was performed; patients with PCN diagnosis given between April 2012 multiple-choice knowledge or competence questions on diagnosis, assessment, and tailoring and April 2015 (18 months before and after the publication of the algorithm) in KPSC (integrated health therapies for EPI system with 15 hospitals and 202 medical offices in Southern California) were identified. • Educational designed included a “test, then teach” approach to elicit cognitive dissonance with Results: 2694 (1087 pre- and 1607 post-algorithm) received a new diagnosis of PCN in the study period. evidence-based feedback provided following each learner response There was no difference in the mean cyst size (pre- 19.2 mm vs post- 18.6 mm, P=0.38). A smaller per- • Four questions were selected to be repeated immediately after participation and serve as the centage of PCNs resulted in EUS after the implementation of the algorithm (pre- 43.8% vs post- 34.9%, instrument to assess the impact of the education using a repeated pairs, pre- to post-assessment P<0.0001). A smaller proportion of patients were referred for GI (pre- 62.84% vs post- 53.29%, P<0.0001) study design where individual each participant served as his/her own control and surgery consultations (pre- 21.53% vs post- 15.71%, P=0.0003) for PCN after the implementation. • The analysis plan included: There was no significant change in operations for PCNs. Cost of care was reduced after the implementa- ˚˚ For all questions combined, McNemar’s chi-square test assessed the response differences tion by 20%, 16%, and 27% for EUS, GI, and surgery consultations, respectively (table 1). All calculations from pre- to post-assessment were based on expected cost for 1,000 patients (KPSC has approximately 1,000 new PCN cases each year). ˚˚ P values measured significance;P values <.05 were considered statistically significant The incidence rate of pancreatic cancer after diagnosis of PCN was 4.79 per 1000-person years (95% CI: ˚˚ Effect size was calculated using Cramer’s V by determining the change in proportion of 2.93, 7.43) for pre-algorithm, 5.98 per 1000-person years (95% CI: 3.71, 9.17) for post-algorithm, and 4.95 participants who answered questions correctly from pre- to post- assessment per 1000-person years for combined (95% CI: 4.95, 9.46). • Survey data were collected from July 7, 2016, to August 8, 2016 Conclusion: In current healthcare climate, there is increased need to optimize resource utilization. Dis- Results: semination of an algorithm for PCN management in an integrated health system resulted in fewer EUS • Data set included responses from the 269 gastroenterologists and 229 PCPs who completed all and GI/surgery referrals, possibly due to increasing the confidence level of physicians ordering imaging assessment questions during the study period studies. This translated to cost saving of 20%, 16%, and 27% for EUS, GI, and surgical consultations, • Gastroenterologists: respectively. Incidence rate of pancreatic cancer after cyst diagnosis was 4.95 per 1000-person years in • Knowledge/competence improved (P <.05; V=0.324; large educational effect) following partici- this population. pation in the CME activity • While 20% answered all 4 questions correctly on pre-assessment, 85% answered them all correctly on post-assessment • PCPs: Knowledge/competence improved (P <.05; V=0.317; large educational effect) following participation in the CME activity [59] Cost of care for pancreatic cystic neoplasms pre- and post-algorithm implementation • While 11% answered all 4 questions correctly on pre-assessment, 60% answered them all correctly on post-assessment EUS GI consultation Surgery consultation • Specific improvements for both gastroenterologists and PCPs were demonstrated in selection of appropriate diagnostic tests for EPI, monitoring therapy for effectiveness and complications, and optimizing pancreatic enzyme replacement therapy (PERT) dosing and administration Pre- $678,823 $62,346 $21,772 Conclusion: Online CME presented in case-based format and using the test, then teach approach can Post- $539,973 $52,450 $15,834 improve knowledge and competence of physicians managing EPI, which may translate to practice improvements in patient management. Cost saving $138,850 $9,896 $5,938

% cost saving 20% 16% 27%

60 Methods: In this case-control study,gallstones, bile and gallbladder mucosa were collected from 25 patients without gallstone biliary disease, 24 with acute cholecystitis and 28 with chronic cholecystitis. The presence of Helicobacter pylori (H. pylori), Helicobacter bilis (H. billis), Helicobacter hepaticus The Incidence of Infection of Walled-off Pancreatic Necrosis (WON) and Implications for (H. hepaticus) and Helicobacter pullorum (H. pullorum) was investigated by PCR using species-specific Antibiotic Prophylaxis primers. Also, serum samples were tested for presence of H. pylori-specific immunoglobulin G by Nadav Sahar, MD1, Shingo Chihara, MD2, Shayan Irani, MD3, Andrew S. Ross, MD3, Seng-Ian Gan, MD3, ELISA. Michael Larsen, MD3, Richard A. Kozarek, MD, FACG3, Michael Gluck, MD3. 1Digestive Diseases Institute, Results: In this study, 77 subjects of acute and chronic cholecystitis and control groups with a Virginia Mason Medical Center, Seattle, WA; 2Virginia Mason Medical Center, Seattle, WA; 3Digestive Dis- mean age of 46.85±14.53, including 58 (67.25%) women and 19 (32.75%) men were included. eases Institute, Virginia Mason Clinic, Seattle, WA DNA of 10 Helicobacter spp. were detected in bile of patients with cholecystitis including 8 H. pylori and 2 H. bilis. However, we could not detect H. hepaticus and H. pullorum DNA in samples. Moreover, there was an association between H. pylori and acute cholecystitis (P=0.048), which was Introduction: Antibiotics are frequently utilized in patients with walled-off pancreatic necrosis (WON), found to be stronger in 31-40 years old group (P=0.003).The increased level ofHelicobacter pylori- a complication of severe acute pancreatitis (SAP). The aim of this study was to determine the frequency specific immunoglobulin G was shown in 87.5% patients with acute cholecystitis, 89.2% patients with of infection in WON, the type of organisms, and the utility of antibiotics. chronic cholecystitis and 80% patients in the control group. But no Helicobacter species were grown Methods: All patients in IRB approved, prospectively maintained database between 2008 and 2017 who from the bile. underwent a combined endoscopic and percutaneous treatment (DMD) of symptomatic WON were Conclusion: In summary, we found association between presence of H. pylori DNA and acute chole- included. Both clinical and microbiological data were extracted and statistically compared, including cystitis with gallstone, especially in 31-40 years old group. Moreover, there is not statistically significant logistic regression to assess variables predicting infection. correlation between three enterohepatic Helicobacter species (H. Bilis, H. Hepaticus, H. pullorum) and Results: One hundred eighty-two patients underwent DMD for WON with a median age of 56 and calculos cholecysistitis. Given the low sample size of patients, more studies are required to clear the clini- predominately male (67%). Seventy-eight patients (41%) did not receive empiric antibiotics before cal role of Helicobacter spp. in the gallstone disease and cholecystitis. undergoing DMD, of which 69 (88% of group) harbored sterile cultures from WON. One-hundred and four patients (59%) that were treated empirically had higher ASA scores (3 vs. 2.6, P<0.05), higher BMI (31 vs. 28 kg/m2, P<0.05), higher modified CTSI scores (8.4 vs. 7.6, P<0.05), a shorter interval from the start of SAP to drainage (57 days vs. 127 days, mean, P<0.05) and a longer duration of antibiotic treatment after DMD (19 days vs. 10 days, mean,P <0.05) in comparison to patients not treated empirically. Cultures were infected in 65 patients treated empirically, with similar numbers of monomicrobial and polymicrobial cultures.The most common organisms isolated wereEnterococcal species (18%) and 62 coagulase negative Staphylococcus (15%). In patients treated with antibiotics empirically, 20 cultures (11%) were infected with Candida species and 5 cultures (3%) with multi-drug resistant bacteria (VRE Post-ERCP Pancreatitis After Minor Papilla Intervention in Pancreas Divisum: The Incidence, the and MRSA). Multivariate analysis showed an age greater than 60 to be a risk factor for infection of Risk, and Prevention WON (P=0.001) and the presence of a disconnected pancreatic duct a risk factor for candida and MDR infection (P=0.03). Parit Mekaroonkamol, MD, Kara L. Raphael, MD, Rushikesh Shah, MD, Sunil Dacha, MD, Qiang Cai, Conclusion: Cultures of WON taken at the time of initial treatment demonstrate a considerable amount MD, PhD, FACG, Field F. Willingham, MD, MPH, Steven A. Keilin, MD, FASGE. Emory University School of sterile samples in select patients. The majority of organisms isolated were colonizers of the skin or from of Medicine, Atlanta, GA the intestinal tract and empiric treatment may result in more resistant organisms. Decisions on use of antibiotics should be based on the clinical setting. Introduction: Endotherapy on the minor papilla is a technically challenging procedure with significant potential complications. Data on Post-ERCP Pancreatitis (PEP) in patients with pancreas divisum undergoing minor papilla intervention is variable and limited. This study aimed to assess risk factors associated with PEP in this unique population. Methods: All patients with pancreatic divisum who developed PEP after therapeutic endo 61 therapy on minor papilla from April 2009 to November 2016 were included in this retrospective case control study. Control group of interventions, procedural complexity, endoscopist, and year Is There Correlation BetweenHelicobacter pylori and Enterohepatic Helicobacter Species With performed-matched procedures without PEP in 2:1 match ratio were compared using bivariate Gallstone Cholecystitis? analysis. Results: Among 174 ERCP performed for patients with pancreas divisum during study period, Seyed Masih Fatemi, MSc1, Abbas Doosti, PhD2, Dariush Shokri, PhD3, Sadegh Ghorbani-Danili, PhD4, 5.17% had PEP. 9 ERCPs with PEP performed in 8 patients with pancreas divisum and 18 matched- Morteza Molazadeh, MSc1, Hossein Tavakoli, BSc5, Mohammad Minakari, MD3, Hamid Tavakkoli, MD3. procedures without PEP performed in 18 patients were analyzed. All procedures were ASGE com- 1Islamic Azad University, Shahrekord, Chahar Mahall va Bakhtiari, Iran; 2University of Michigan, Ann plexity category 3. In the PEP group, 8 (89%) were male, 6 (67%) were Caucasian, 7 (78%) had Arbor, MI; 3Isfahan University of Medical Sciences, Isfahan, Esfahan, Iran; 4Islamic Azad University, complete divisum, and the mean age was 37.8+/-16.1 years old. Regarding endoscopic interven- Jahrom, Fars, Iran; 5University of British Columbia, Kelowna, BC, Canada tion, 4 (44%) had minor papillotomy with stent placement, 2 (22%) had unsuccessful cannulation, 1 (11%) had stent exchange, 1 (11%) underwent epinephrine injection and argon plasma coagulation for post papillotomy bleeding, and 1 (11%) had stent extraction. 2.3%, 1.7%, and 1.1% had mild, Introduction: Cholecystitis is a common surgical condition. Recently, several authors have reported moderate, and severe PEP, respectively with mean hospitalization of 6.8+/-4.1 days. Compared with that bile tolerant Helicobacter spp. DNA has been found in the human bile colonizing the biliary tract. the non-PEP group, indomethacin use was significantly lower (x2=5.85; P=0.01). After adjusting for The aim of this study is evaluation of association between the presence of Helicobacter spp. and gallstone indomethacin use in post-hoc analysis, smoking and chronic opioid use were significantly more cholecystitis.

[60] Analysis of the microbiology of walled-off pancreatic necrosis (WON) in patients undergoing dual-modality drainage (DMD).

The American Journal of GASTROENTEROLOGY VOLUME 112 | SUPPLEMENT 1 | OCTOBER 2017 www.nature.com/ajg Abstracts S29

common in the PEP group (x2=3.7; P=0.05 and x2=10.4; P=0.001). There was no statistically significant difference in age, gender, ethnicity, complete or partial divisum, or native papilla (P=0.08, 1.0, [64] Baseline characteristics, demographics and outcomes 0.72, 0.76, 0.59, respectively). Conclusion: Patients with pancreas divisum undergoing minor papilla intervention had a low rate of PEP Mean (standard deviation) or % Total CDI positive CDI negative P-value in this cohort. Smoking and chronic opioid use were associated with increased risk, while indomethacin group group appeared to be protective. Even in a small single institution study, there appeared to be heterogeneity in the endoscopic approach for minor papilla therapeutic intervention. Age 51.9 (16.6) 51.8 (17.7) 52.0 (16.6) 0.97

Gender male (%) 80 77.8 82.9 0.66

BISAP Score (%) 0.55

63 0 34.1 44.5 31.4

Pancreatic Duct Stent Exchanges via the Minor Papilla for Recurrent Acute Pancreatitis in 1 15.9 0.00 20.0 Pancreas Divisum: How Many is Enough? 2 27.3 33.3 25.7

Parit Mekaroonkamol, MD, Rushikesh Shah, MD, Kara L. Raphael, MD, Qiang Cai, MD, PhD, FACG, 3 22.7 22.2 22.9 Field F. Willingham, MD, MPH, Steven A. Keilin, MD, FASGE. Emory University School of Medicine, Atlanta, GA Charlson Comorbidity score (%) 0.24

Introduction: Pancreas divisum is the most common congenital anomaly of the pancreas and can result 0 25.0 44.4 20.0 in recurrent acute pancreatitis (RAP) and chronic pancreatitis. Minor papilla stricture and intraductal 1 15.9 0.00 20.0 hypertension have been proposed to underlie the pathophysiology in this setting. Minor papillotomy with pancreatic duct stenting has shown variable outcomes in patients with divisum and RAP. The spe- 2 11.4 22.2 8.57 cific techniques and protocol are not well defined or homogeneous. This study aims to evaluate procedural factors associated with a favorable outcome. 3 15.9 22.2 14.29 Methods: We conducted a retrospective study on pancreas divisum patients with RAP who underwent minor papillotomy with pancreatic duct stenting from April 2009 to October 2016. Records were 4 15.9 0.00 20.0 reviewed from the index procedure and at subsequent stent exchanges until 6 months after extraction 5 6.82 0.00 8.57 of all stents. Procedural data and stent-related parameters were collected and examined using bivariate analysis. 6 9.10 11.1 8.57 Results: Of 74 patients with pancreas divisum and RAP who underwent minor papillotomy and pancreatic duct stent placement, a total of 163 ERCPs with a mean of 2.3+/-1.6 stent exchanges were per- Number of DEN sessions(IQR) 6 (1,12) 7(3,10) 6(1,12) 0.27 formed. 4.29% had post ERCP pancreatitis. Median size and length of the initial stent were 5Fr and 7 cm, while the median final size and length at removal were 7Fr and 8 cm. The mean interval between Length of Antibiotic use in days 35.1(31.0) 58.8(36.3) 29.0(26.8) 0.01 stent exchanges was 33.7+/-17 days. 53 patients (71.6%) achieved clinical success with no further epi- (SD) sodes of pancreatitis during the study period. Comparing between the responder and the non-responder Length of hospitalization in 40.2(35.3) 61.7(40.2) 34.6(32.1) 0.04 groups, there was no statistically significant difference in age (P=0.64), gender (P=0.91), race (P=0.81), days (SD) divisum anatomy (complete vs partial; P=0.66), smoking history (P=0.16), alcohol use (P=0.89), or opioid use (0.72). Regarding procedural variables, there was no significant difference in number of stent Positive Pancreatic fluid culture 70.5 77.8 68.6 0.70 exchanges (P=0.24), stent configuration (pigtail vs straight;P =0.4), stent diameter (P=0.31), or length (%) of stent (P=0.14). Conclusion: Pancreatic duct stenting for the treatment of RAP in patients with divisum in this cohort Positive blood cultures (%) 9.10 0.00 11.43 0.57 yielded a favorable success rate with a low rate of post-ERCP pancreatitis. While an average of 2 stent exchanges per patient with an upsized stent from 5Fr to 7Fr were observed, number of exchanges, type, Previous History of CDI (%) 14.3 11.1 15.5 0.62 size, or length of the stent were heterogeneous. The technical aspects of therapeutic stent placement in Clinical Resolution (%) 84.1 77.8 85.7 0.61 this setting are not standardized, and it would be of benefit to evaluate the practice parameters in a large group of providers. Follow up duration in Months (SD) 11.3(11.2) 9.00(5.78) 11.8(12.1) 0.33

Mortality (%) 6.98 0.00 8.57 1.00

Values are reported as mean (standard deviation), median (interquartile range [IQR]) for continuous variables, or % for categorical variables. Statistical analysis was done with unpaired T-test or Fisher’s exact test depending 64 on the variable.

Antibiotic Therapy in Infected Walled-Off Pancreatic Necrosis: Impact of Duration of Therapy on Resolution and the Development of Clostridium difficile Infection

Enad Dawod, MD1, Aleksey Novikov, MD1, Najib Nassani, MD, MSc2, Carl V. Crawford, Jr., MD3, Reem Z. Sharaiha, MD, MSc4. 1Weill Cornell Medical College, Cornell University, New York, NY; 2Staten Island University Hospital, Staten Island, NY; 3Weill Cornell Medicine, New York, NY; 4New York-Presbyterian/ 66 Weill Cornell Medical Center, New York, NY Racial Disparities in the Diagnosis and Treatment of Pancreatic Cancer Introduction: Walled-off pancreatic necrosis (WON) is a severe complication of acute pancreatitis. Emmanuel Ofori, MD1, Kavitha Vemuri, MD1, Gaurav Parhar, MD1, Madhavi Reddy, MD, FACG2, Ghula- WON is frequently infected and is associated with a substantial morbidity and mortality. Antibiotic 1 1 2 therapy is an important part of an infected WON management, but there is no consensus on adequate mullah Shahzad, MD . Brooklyn Hospital Center, Brooklyn, NY; The Brooklyn Hospital Center, Brooklyn, antibiotic strategy in such patients. Direct endoscopic necrosectomy (DEN) is a new minimally invasive NY endoscopic technique that allows WON drainage into the gastric or duodenal lumen. WON patients frequently require prolonged hospitalizations. Antibiotic exposure and prolonged hospital stay are Introduction: Pancreatic cancer (PC) is the 4th leading cause of cancer death in the United States with known risk factors for Clostridium difficile infection (CDI). The aim of this study is to determine the very poor overall survival. African Americans (AA) belong to a subset of the population with the highest impact of antibiotic treatment on WON and the associated adverse effects of prolonged antibiotics incidence of pancreatic cancer with incidence of 14.8 per 100,000 among AA men and 8.8 per 100,000 therapy with respect to CDI. persons in the general population. Many population studies have found that AA and those with low Methods: We included all patients who underwent DEN with placement of fully covered self-expanding socioeconomic status have an independent risk factor for pancreatic cancer mortality. This study aimed metal stents (FCSEMS), including lumen apposing metal stents (LAMS) between 2011 to 2016. We col- to investigate the degree of disparity in the diagnosis and treatment of pancreatic cancer among our lected demographic information, Bedside Index for Severity in Acute Pancreatitis (BISAP) score, Charl- urban minority patient population. son comorbidity score, number of DEN sessions, and mortality along with length of stay, and a length of Methods: This was an observational retrospective chart review of pancreatic cancer patients at the Brook- antibiotic treatment. Clinical success was defined as complete resolution of WON both radiographically lyn Hospital Center from 01/01/1990 to 11/30/2016, who were all of low socioeconomic status and pre- and clinically. dominantly African American ethnicity. Exclusion criteria were subjects less than 18 years of age, and Results: 44 patients were included. The mean patient age was 51.9 (16.5) years, and 80% were male. those with incomplete medical documentation. Patients diagnosed with pancreatic cancer were assessed Pancreatitis etiology included gallstones (n=18, 40.9%), alcohol (n=7, 15.9%), medications (n=3, 6.8%), for race, stage of diagnosis, and treatment option received. post-ERCP (n=4, 9.1%). All patients received at least one antibiotic during their hospitalization. The Results: 103 patients were analyzed in which 76% (78/103) were AA and 24% (25/103) were non-AA. mean duration of antibiotic therapy was 35.1 days (31.0) and the mean length of hospitalization was 45% (35/78) AA patients were diagnosed with pancreatic cancer at stage 4 with distant metastases. All 40.2 days (35.3). The median number of DEN sessions was 6 (1, 12). Clinical resolution of WON was other pancreatic cancers were found at earlier stages in a loco-regional area of the pancreas. Among achieved in 84.1% of patients. There was no significant association between total duration of antibiotic non-AA, 48% (12/25) were detected at stage 4 with distant metastases. Patients were also analyzed for therapy and clinical resolution (P=0.36). However, CDI positive patients were significantly more likely treatment disparity. AA diagnosed with pancreatic cancer received treatment 62% (48/78) of the time to have longer duration of antibiotics therapy (P=0.01) as well as a significantly longer length of stay compared to non-AA who received treatment 70% (18/25) of the time with either surgery, chemotherapy, (P=0.04) (Table). radiation, or surgery with adjuvant chemotherapy. Conclusion: We identified CDI as a significant adverse event in patients undergoing DEN treatment for Conclusion: Our study showed no major difference in the late stage detection of pancreatic cancer WON associated with an increase in duration of antibiotics and a longer length of stay. The type and dura- amongst AA and those of other races. In fact, AA patients had a slightly lower rate of late stage diag- tion of antibiotics in WON patients should be given with caution and routinely reviewed for the narrowest nosis. This data did not show much of a diagnostic disparity, which is different from documented stud- spectrum of activity required. ies in the literature. Our data however suggests a relative treatment disparity of about 8% between the two groups. Studies have shown that AA patients often refuse treatment for unknown reasons. Various socioeconomic or lifestyle causes could be reasons for treatment refusal. Our study showed that if low socioeconomic status populations have accessible quality healthcare, pancreatic cancer treatment dispari- 65 ties can be improved.

WITHDRAWN

67 [67] Baseline Characteristics and In-hospital Outcomes with Acute Pancreatitis Stratified by Gender Male Gender Is Associated With Increased In-patient Mortality and Healthcare Utilization Among Acute Pancreatitis Patients: Analysis of National Inpatient Samples From 2010-2014 Variable Name Male (N=227,282) Female (N=221,787) P value

Dhruvan Patel, MD1, Yash Shah, MD2, Pranav D. Patel, MD3, Palashkumar Jaiswal, MBBS4, Kalpit De- Age (Years, SD) 52.1±16.9 53.3±19.8 <0.001 vani, MD5, Rajkumar P. Doshi, MD, MPH6. 1Drexel University College of Medicine, Hahnemann University Hospital, Philadelphia, PA; 2VA Medical Center, Philadelphia, PA; 3Montefiore Medical Center, Bronx, NY; Race*: (%) <0.001 4John H. Stroger, Jr. Hospital of Cook County, Chicago, IL; 5James H. Quillen College of Medicine, East Ten- nessee University, Johnson City, TN; 6North Shore University Hospital, Jersey City, NJ White 60.5% 60% Black 15.3% 14.9% Introduction: Acute pancreatitis (AP) has a higher rate of inpatient admissions in the United States. Other 24.2% 25.1% There are several score systems available to determine the mortality in patients with AP. However, none of the scoring system includes gender as one of the factors. There is a paucity of data on the impact Admission Type$: (%) <0.001 of gender on AP-related in-hospital mortality in population-based studies. We sought to evaluate the impact of gender on AP-related in-patient mortality and health-care utilization using a nationwide data- Emergent/Urgent 94.2% 93.3% base. Methods: Our study cohort was derived from a nationwide inpatient sample between 2010-2014 using Elective 5.8% 6.7%

International Classification of Diseases (ICD-9) diagnostic codes for AP. Patients below 18 years of age @ were excluded. The patient population in our cohort was divided based upon gender. Age, race, admis- Primary Payment Method : (%) <0.001 sion type, primary payment method and charlson’s comorbidity index (CCI) were adjusted in this model. Medicaid/Medicare 48.2% 56.3% Outcomes of interest were in-hospital mortality, the length of hospitalization stay and discharge to home. Multivariate analysis was performed to analyze in-hospital outcomes. Private Insurance 31.2% 30.2% Results: Total 449,069 patients with AP were identified in our nationwide cohort over the span of 5 years. Of these, 50.6% were male, 49.4% were female. Age at admission and race were equally distributed Other 20.6% 13.5% amongst both genders. (Table 1) Medicaid/Medicare insurance rate was higher in females (56.3 vs 48.2%, f P<0.001). Hospitalizations due to AP were similar in both gender in the year of 2010, however, there is Charlson’s Comorbidity Index <0.001 a consistently increasing trend of hospitalization in males compared to females from 2011-2014. (Fig 1) 0 46.1% 49.6% Similarly, males had consistently higher in-patient mortality rate due to AP compared to females (2.3% vs 1.9%, P<0.001) from 2010-2014 (Fig 2). Males with AP had higher comorbidities compared to females 1 26.5% 24.8% with AP (CCI of ≥3,10.3% vs 8.3%, P<0.001). Additionally, length of hospital stay was higher in male population compared to female population (6.0 ± 8.8 vs 5.8 ± 7.7 days). 2 10.9% 10.8% Conclusion: In AP inpatient admissions, male gender had overall higher mortality rate compared to female gender. Although average female population was older, male population had higher CCI in our 3 6.2% 6% cohort. Comorbidities with AP may be associated with an increased in-patient mortality rate in males. >3 10.3% 8.8% Medicaid/Medicare insurance was higher in females, which may had a beneficial impact on in-patient mortality. Further studies are warranted to evaluate the impact of gender on in-patient mortality in In-hospital Outcomes&: patients with AP. In-hospital mortality 2.3% 1.9% <0.001

Length of Stay (Days) 6.0±8.8 5.8±7.7 <0.001

Discharge to home 77.8% 77.4% <0.001

Discharge to Other facilities 19.9% 20.7% <0.001

Died in-hospital 2.3% 1.9% <0.001

In this multivariate model, age, race, admission type, primary payment method and Charlson’s comorbidity index were included. f- Charlson's/Deyo comorbidity index was calculated as per Deyo classification.

Patient and Procedural Factors Associated With Increased Islet Cell Yield in Total Pancreatectomy With Islet Autotransplantation

Katherine V. Trinh, BS, Kerrington D. Smith, MD, Timothy B. Gardner, MD, MS. Dartmouth-Hitchcock Medical Center, Lebanon, NH [67A] Trends of Hospitalization with Acute Pancreatitis- Stratified by Gender Introduction: Total pancreatectomy with islet autotransplantation (TP-IAT) provides pain relief to Jonckheere-Terpstra Trend Test: <0.001. highly select patients with recurrent acute and/or chronic pancreatitis. However with variable outcomes and no standardized guideline for patient selection, it is important to refine islet manipulation procedures and patient selection characteristics to optimize outcomes. Success of the procedure depends on a high number of isolated islet equivalents. This study explores the patient and procedural characteristics associated with high islet cell yield. Methods: This study evaluated patients who underwent TP-IAT at Dartmouth Hitchcock Medical Center from 2012 to 2016. 38 patients met inclusion criteria. Odds ratios with 95% confidence intervals were found for various patient and procedural characteristics listed in Table 1. The primary clinical outcome was the number of isolated islet equivalents per kilogram body weight (IEQ/Kg), defined as IEQ/Kg >2,500.

[68] Factors Associated with Successful Islet Cell Yield During TPIAT

Characteristic Odds Ratio 95% Confidence Interval

Age at Evaluation 0.65 0.14-2.93

CT or MRI Findings of Chronic Pancreatitis 28.91 1.53-546.69

Pre-operative HgA1c >5.6 0.13 0.03-0.68

Lack of Pancreatic Duct Stones 23.00 1.77-298.46

Lack of Pancreatic Parenchymal Stones 55.00 4.13-732.75 [67B] Trends of In-hospital Mortality with Acute Pancreatitis- Stratified by Gender Jonckheere-Terpstra Trend Test: <0.001. Islet Cell Purification 0.12 0.01-1.06 Positive Islet Cell Culture 0.05 0.00-0.55

The American Journal of GASTROENTEROLOGY VOLUME 112 | SUPPLEMENT 1 | OCTOBER 2017 www.nature.com/ajg Abstracts S31

Results: All patient factors and procedural fields evaluated are listed in Table 1. Patients with no CT/ 70 MRI evidence of chronic pancreatitis showed statistically significant higher odds of success (OR=29, P=0.02). Patients without pancreatic duct stones or parenchymal stones were associated with higher odds of success (OR=23, P=0.02 and OR=55, P=0.002, respectively). Islet cell suspensions positive for cultures or positive gram stains were associated with lower chances of success (OR=0.06, P=0.02, and OR=0.48, P=0.01, respectively). Patients with preoperative HgbA1c greater than 5.6 were associated with lower odds of success (OR=0.13, P=0.02). Conclusion: This investigation found that patients without CT/MRI evidence of chronic pancreatitis, WITHDRAWN without bacterial infections, and without pancreatic duct or parenchymal stones are more likely to attain successful islet cell yields. Additionally, patients with preoperative HgbA1c less than 5.6 were associated with higher outcomes of success. These factors should be considered when selecting patients for TP-IAT.

69 Genetic Counseling and Multigene Testing Experiences of Pancreatic Adenocarcinoma Patients

Diagnostic Yield of Micro Forceps Biopsies during Endoscopic Ultrasound Evaluation of Cathryn Koptiuch, MSc, Wendy Kohlmann, MS, Jingsong Zhao, Sara Johnson, Kimberly Kaphingst, PhD. Pancreatic Cystic Lesions Huntsman Cancer Institute, Salt Lake City, UT

Chetan Mittal, MD1, Joshua Obuch, MD, MSCS1, Sachin Wani, MD2, Steven Edmundowicz, MD2, Raj Introduction: Multigene testing is a routine tool for identifying patients with hereditary cancer con- Shah, MD2, Brian Brauer, MD2, Hazem Hammad, MD2, Augustin Attwell, MD3, Mihir S. Wagh, MD1. ditions. Research is needed to determine patient responses to complex results generated from this 1University of Colorado Anschutz Medical Campus, Denver, CO; 2University of Colorado Anschutz Medical approach. Increased risk for pancreatic adenocarcinoma (PAC) is associated with many hereditary can- Campus, Aurora, CO; 3University of Colorado, Denver, CO cer syndromes, but PAC patients may struggle to understand and share these results with relatives due to poor health and prognosis. This study examined cognitive and affective responses to genetic counseling Introduction: Evaluation of pancreatic cystic lesions (PCLs) can be challenging and requires a combi- (GC) and multigene testing among PAC patients. nation of radiologic imaging, morphologic characteristics on endoscopic ultrasound (EUS), cyst fluid Methods: Over 18 months beginning in December 2015, we attempted to contact all PAC patients pre- cytology, and cyst fluid analysis. Through-the-needle micro forceps (Moray micro forceps, US Endos- senting to Huntsman Cancer Institute for GC. A subset of patients that had GC completed questionnaires copy, Mentor, OH) are a recent addition to EUS armamentarium. These forceps allow tissue sampling about attitudes toward multigene testing pre-GC and 4 weeks post-results. 15-gene panel testing was from PCLs and may improve diagnostic yield. To date, scant data exists on the use of micro forceps completed in a CLIA lab. biopsy for these lesions. The main aim of this study was to assess diagnostic yield and safety of EUS- Results: Of 174 PAC patients, 121 were contacted. 59.5% agreed to GC; 83.1% of those pursued genetic guided micro forceps biopsy for PCLs. testing (Figure 1). Pathogenic variants were identified in 5 (8.5%) patients (1 had pathogenic variants Methods: Our electronic endoscopy database was queried to identify patients who underwent EUS- in BRCA2 and ATM). 13 patients had a variant of uncertain significance (VUS). Among the 18 patients FNA of PCLs and biopsies with the micro forceps, during the same procedure. Cysts were identified (Table 1) who completed questionnaires: 2 declined testing, 1 tested positive, 2 had VUS. All patients on EUS and punctured using 19-G FNA needle. All patients received peri-procedural antibiotics. completed the baseline, 6 completed the follow-up questionnaires. At Baseline, 78% thought genetics Cyst fluid was collected for cytologic analysis, CEA and amylase levels, and wall of the cyst was likely played a role in their diagnosis, 83% wanted to learn more about genetic testing and 72% had heard biopsied using micro forceps through the 19-G needle. Adverse events were recorded per published of genetic testing. None of the patients that completed follow-up had tested positive (Table 2). All patients ASGE criteria. reported having shared their result: 60% a healthcare provider, 100% a relative. Patients frequently shared Results: Twenty patients (mean age 64.2 years [range 34 -87]; 8/20 male [40%]) underwent EUS-FNA their results with sisters (80%), sons (60%), and daughters (40%) but also with brothers, mothers and and micro-forceps biopsy of PCLs from 02/2016 to 05/2017. The mean cyst size was 34.4 mm (range 13 others. -70 mm), with 10 cysts in the pancreatic head/uncinate and 10 in the body/tail region. Micro forceps Conclusion: 8.5% of PAC patients tested positive for conditions known to increase the risk for PAC and biopsies were technically successful in all cases, and provided a diagnosis of neoplasia in 11 cases (55%), other cancers. Most patients pursued testing to benefit their relatives. The majority of patients antici- benign tissue in 8 cases (40%) and insufficient quantity in 1 case. In 4 patients (20%), micro forceps pated that a hereditary condition would explain their diagnosis and pursued multigene testing. Patients biopsy results drastically changed patient management, providing diagnoses (1 Neuroendocrine tumor, who tested negative generally understood they did not have a hereditary cancer condition, but most still 1 mucinous cystic neoplasm, and 2 IPMNs) otherwise not suggested by cytology or cyst fluid analysis. believed their relatives should pursue genetic testing and did not understand their close relatives still had However, cytology provided a diagnosis of mucinous neoplasm in 3 cases not detected by micro forceps a familial risk for PAC. Generally, patients expressed relief and positive reactions and few expressed nega- biopsies. No adverse events were noted. tive reactions to the multigene testing process. Conclusion: Micro forceps biopsies were associated with high yield and excellent safety profile. Biopsies provided a diagnosis of neoplasia not detected by cytology in a significant number of patients but also did not identify mucinous neoplasms in a few cases as suggested by cyst fluid cytology. Hence micro forceps biopsies may be a useful adjunctive tool for EUS guided assessment of PCLs, complementing existing EUS-FNA sampling protocols.

[71A] .

[71B] .

[71] Follow-up Outcomes for Multigene Testing 72

Patient Reactions Percentage Prevalence and Risk Factors for Choledocholithiasis After Cholecystectomy

Patients who believed they are not at increased risk for cancer 80% Joseph Spataro, MD1, Mazen Tolaymat, MD2, Charles A. Kistler, MD1, Michael Jacobs, BS3, Jeffrey Fitch, BS4, Monjur Ahmed, MD1. 1Thomas Jefferson University Hospital, Philadelphia, PA;2 University of Mary- Patients who believed their relatives were at increased risk for cancer 20% land Medical Center, Baltimore, MD; 3Thomas Jefferson University, Philadelphia, PA; 4Thomas Jefferson Patients who did not believe their relatives were at increased risk for cancer 60% University School of Medicine, Philadelphia, PA

Patients who believed their relatives should still pursue genetic testing 60% Introduction: Clinical manifestations of choledocholithiasis include biliary colic, obstructive jaun- Patients who described genetic testing as at somewhat/very helpful for their family 60% dice, pancreatitis, and acute cholangitis. Prior studies have suggested that presentations of these conditions within 3 years of cholecystectomy (CCY) are more likely due to residual stones missed at Patients who reported never/rarely feeling uncertain about how their result affects their own or 100% time of surgery as opposed to formation of recurrent stones. Complications of retained common their relatives’ cancer risks bile duct (CBD) stones are well recognized, however there is limited data on pattern and time of Patients who reported never/rarely thinking about how their result affecting their home or work life 100% presentation. The aim of this study was to determine the pattern of presentation in patients with and without prior CCY. Patients who reported never/rarely feeling concerned about how their result affects their 100% Methods: A retrospective chart review was conducted of patients with biliary pancreatitis, acute cholan- insurance gitis, or symptomatic choledocholithiasis at a secondary and tertiary care center over a two-year period. Patients who reported never/rarely worrying about familial conflicts caused by genetic testing 100% The electronic medical record was reviewed to collect the following data: age, sex, race, body mass index, history of CCY and perioperative bile duct evaluation, presence of CBD stone or sludge, and endoscopic Patients who reported never/rarely feeling upset, anxious, guilty or feeling a loss of control 100% retrograde cholangiopancreatography (ERCP) results including maximum CBD diameter and presence because of their test result in the last week of periampullary duodenal diverticulum. Patients who reported never/rarely feeling sad because of their test result in the last week 80% Results: 358 patients with biliary pancreatitis, acute cholangitis, or biliary colic due to choledocholithiasis were assessed (Table 1). 100 (27.9%) patients were post-CCY prior to presentation. Demographic Patients who reported sometimes/often feeling relieved and/or happy because of their test result 60% and clinical data were reviewed (Table 2). Patients with or without a CCY presented with: 32 (12.4%) in the last week vs 6 (6.0%) for biliary pancreatitis (P= 0.078), 33 (12.8%) vs 32 (32.0%) for acute cholangitis (P< 0.001), and 196 (76.0%) vs 62 (62.0%) for symptomatic choledocholithiasis (P=0.008), respectively. Post-CCY Patients who reported sometimes/often feeling supported by their family (100%) and satisfied 100% patients with a known procedure date had a median recurrence of choledocholithiasis of 210 days with 24 with family communication (100%) regarding their result and testing process in the last week (80%) requiring an ERCP within 3 years.

The American Journal of GASTROENTEROLOGY VOLUME 112 | SUPPLEMENT 1 | OCTOBER 2017 www.nature.com/ajg Abstracts S33

Conclusion: Recurrent biliary obstruction due to post-CCY choledocholithiasis is not uncommon. The disproportionate prevalence of acute cholangitis in post-CCY patients is multifactorial. Given the [72_A] Comparison of Presentation of Choledocholithiasis Between Patients with and frequency of gallstone-related disease recurrence within 3 years after CCY, the role of pre-CCY chol- Without a CCY angiography should be re-evaluated in future studies. Specifically, the identification of risk factors that predispose patients to recurrent disease would assist in standardization of perioperative CBD evaluation, Clinical Presentation No CCY (n=261)* Post-CCY (n=100) P-value with a goal of preventing further episodes.

Biliary Pancreatitis 32 (12.4%) 6 (6.0%) 0.078

Acute Cholangitis 33 (12.8%) 32 (32.0%) <0.001

Biliary Colic 196 (76.0%) 62 (62.0%) 0.008

*3 of the 258 patients presented with both biliary pancreatitis and acute cholangitis. 73

Cholecystectomy versus Wait and Watch Strategy After Endoscopic Sphincterotomy in Choledocholithiasis: A Systematic Review and Meta-Analysis

Chiranjeevi Gadiparthi, MD, MPH1, Muhammad Ali Khan, MD1, Khwaja F. Haq, MD2, Claudio R. Tombazzi1, C. Mel Wilcox, MD, MSPH, FACG3. 1University of Tennessee Health Science Center, Memphis, TN; [72_B] Demographic and Clinical Data 2New York Medical College at Westchester Medical Center, Valhalla, NY; 3University of Alabama at Birmingham, Birmingham, AL Patients No CCY (n=258) Post-CCY (n=100) P-value Introduction: Endoscopic retrograde cholangiography and sphincterotomy (ERC-S) with subsequent Male 150 (58.1%) 66 (66.0%) 0.174 cholecystectomy is the standard of care for the management of patients with choledocholithiasis. There Mean Age 62.4 64.6 0.352 is conflicting evidence in terms of mortality reduction, prevention of complications specifically biliary pancreatitis and cholangitis with the use of early cholecystectomy. We conducted this meta-analysis Caucasian 189 (73.3%) 76 (76.0%) 0.596 of randomized controlled trials (RCTs) to compare the early cholecystectomy versus wait and watch

2 strategy after ERC-S. Mean Body Mass Index (kg/m ) 28.8 29.6 0.299 Methods: We searched several databases for RCTs comparing the two strategies in the management of Presence of Periampullary Diverticulum 50 (19.38%) 26 (26.0%) 0.085 choledocholithiasis after ERC-S. Our primary outcome of interest was difference in mortality. We evaluated several secondary outcomes including difference in development of acute pancreatitis, biliary colic Timing of recurrence from cholecystectomy* and cholecystitis, cholangitis and recurrent jaundice, major adverse events and length of stay in hospital. <3 years 24 We conducted two predetermined subgroup analysis based on patient comorbidities (high risk vs low risk) and type of cholecystectomy (open versus laparoscopic). ≥3 years 6 Results: Seven studies with 916 patients (455 cholecystectomy group and 461 wait and watch group) were included in the meta-analysis. Rates of mortality for cholecystectomy group and wait and watch group Median Recurrence 210 days (range 6-5160 days) were 9.3% and 13% respectively. Pooled RR with 95% confidence interval (CI) for mortality was 1.51 2 Maximum CBD Diameter 13.5mm (range 4-25mm) (0.94, 2.44), I =13%. In high risk patient group pooled RR was 1.50(0.75, 2.98) and in low risk population pooled RR was 1.53(0.79, 2.96). Likewise, there was no difference in mortality based on laparoscopic *Only representative of patients with confirmed procedure dates. cholecystectomy RR 1.17 (0.71, 1.94) or open cholecystectomy RR 1.98 (0.97, 4.03). Pooled RR for acute pancreatitis was 1.64 (0.46, 5.81) with no heterogeneity. No difference in rate of acute pancreatitis patients based on high risk versus low risk patients. Pooled RR for occurrence of biliary colic and cholecystitis

[73] Forest plot to compare mortality.

[73] Outcomes and subgroup analyses evaluated 75

Outcome Subgroups Risk Ratios (95%CI) Heterogeneity Program for the Investigation of Premalignant Lesions and Early Pancreatic Cancer in a High- Risk Population: Report From our First Year of Experience Mortality High risk patients 1.50 (0.75, 2.98) 40% Hui Jer Hwang, MD1, Ignacio Málaga, MD1, Cecilia Curvale, MD1, Martin Guidi, MD2, Julio De Maria, Low risk patients 1.53 (0.79, 2.96) 0% MD2, Raul Matano, MD2, Sandra Basso, MD1. 1El Cruce Hospital, Buenos Aires, Argentina; 2El Cruce Hospital, La Plata, Buenos Aires, Argentina Open cholecystectomy 1.98 (0.97, 4.03) 20%

Laparoscopic cholecystectomy 1.17 (0.71, 1.94) 0% Introduction: Pancreatic cancer (PC) in early stages is potentially curable. High risk groups are st Acute pancreatitis High risk patients 2.22 (0.48, 10.23) 0% as follows: 2 o more relatives with PC, one 1 grade relative <50 years old with PC, Peutz-Jeghers syndrome (PJS), Lynch syndrome and relative with PC, being carrier of a mutation in the BRCA2 Low risk patients 0.89 (0.08, 8.07) 0% gene and relative with PC, hereditary pancreatitis and chronic pancreatitis. The methods recommended for the diagnosis are the endosonography and annual nuclear magnetic cholangioresonance Biliary colic and cholecystitis High risk patients 8.42 (3.02, 23.41) 0% (NMCR). Objective: to detect premalignant lesions and pancreatic tumors in high risk population for PC. Low risk patients 13.28 (3.17, 55.54) 0% Methods: Multicenter and prospective study. Patients with high risk for PC were enrolled. We made a Cholangitis and recurrent High risk patients 3.81 (1.24, 11.66) 0% detailed medical history with a genogram and performed endosonography and NMCR. jaundice Results: During 2016, 25 patients were included, average age: 47.6 years, 15 of which were men. Risk factor: 13 patients with chronic pancreatitis, 11 patients with PC family background and 1 patient with Low risk patients 1.65 (0.76, 3.56) 0% PJS. We performed endosonography and found the following: 13 chronic pancreatitis patients, 1 atrophic pancreas, 1 pancreatic mass with guided puncture negative results for neoplasm with good evolution Major adverse events High risk patients 0.73 (0.34, 1.53) 0% during the monitoring period, 1 intraductal papillary mucinous tumor (IPMN) and 10 normal studies. When NMCR was performed, the results were as follows: 7 patients with signs of chronic pancreatitis, 1 Low risk patients 0.51 (0.13, 1.99) 31% with increase of the pancreatic head, 1 with IPMN and 7 normal studies. The CA 19-9 was performed in Open cholecystectomy 0.73 (0.34, 1.53) 0 13 patients and only one of them had an altered level (x 2). Conclusion: During the first year of this program for the investigation of PC with endosonography and Laparoscopic cholecystectomy 0.51 (0.13, 1.99) 31% NMCR, no premalignant neoplasms were found; one premalignant lesion was diagnosed (IPMN).

Minor adverse events High risk patients 0.26 (0.04, 1.41) 40%

Low risk patients 0.88 (0.29, 2.07) 0% 76 Open cholecystectomy 0.26 (0.04, 1.41) 40%

Laparoscopic cholecystectomy 0.88 (0.29, 2.07) 0% The Diagnostic and Etiological Role of the Endosonography on the Idiopathic Pancreatitis: Prospective, Observational and Multicenter Study Length of stay in hospital High risk patients -3.34 (-6.38, -0.30) 66% Hui Jer Hwang, MD1, Ignacio Málaga, MD1, Cecilia Curvale, MD1, Martin Guidi, MD2, Julio De Maria, Low risk patients -2.21 (-4.89, 0.45) 72% MD2, Raul Matano, MD2, Sandra Basso, MD1. 1El Cruce Hospital, Buenos Aires, Argentina; 2El Cruce Hospital, La Plata, Buenos Aires, Argentina Open cholecystectomy -3.34 (-6.38, -0.30) 66%

Laparoscopic cholecystectomy -2.21 (-4.89, 0.45) 72% Introduction: 30% of the cases of acute pancreatitis are idiopathic (IAP). These patients require more complex studies. However, 15% of the cases will remain in the same condition. Objective: the aim is to determine the diagnostic and etiological capacity of the endosonography in patients with IAP. Methods: Prospective, observational and multicenter study. Endosonography was performed on con- during follow up was 9.82 (4.27, 22.59), I2=0%. Pooled RR for cholangitis and recurrent jaundice was 2 secutive patients with IAP. Enolic and metabolic etiologies were discarded, and the previous ultrasound 2.16(1.14, 4.07), I =0%. However, there was no difference in rate of cholangitis between the two groups scan and computerized axial tomography were negative. in low risk patients. Length of stay was shorter in wait and watch group pooled mean difference was 2 Results: During 24 months we included 32 consecutive patients of which 21 were women. Average age: -2.70(-4.71, -0.70), I =78%. 47.5 years. Episodes of acute pancreatitis: 3.1. Nuclear magnetic cholangioresonance was performed on 31 Conclusion: Although there is no difference in mortality between the two strategies after ERC-S, but patients (severe chronic pancreatitis: 2; pancreatic cyst: 1; pancreas divisum: 1; bile duct and wirsung duct laparoscopic cholecystectomy should be recommended as it is associated with lower rates of recurrent expansion: 1). Etiology could not be determined in 26 patients (84%). The diagnosis after endosonogra- cholecystitis, cholangitis and biliary colic down the road. phy was the following: undetermined chronic pancreatitis: 8 patients (25%); consistent chronic pancreatitis (calcifying): 6 (18.7%); microlithiasis: 7 (21.9%); Intraductal papillary mucinous tumor: 2 (6.2%); neoplasic periampullary tumor: 1 (3.1%); pancreas divisum: 1 (3.1%). Etiology could be determined due to the endosonography in 72% of the patients; 28% remained idiopathic. Limiting element: unavailability to perform Sphincter of Oddi manometry and genetic test. Conclusion: The cause of IAP could be determined in 75% of the patients, being the capacity of the endo- 74 sonography greater than that of the nuclear magnetic cholangioresonance. The knowledge of the etiology of the AP is essential for the treatment and prevention of its progression. Clinical Impact of the Endoscopic Ultrasound in Patients With Intermediate Probability of Choledocholithiasis: Prospective and Comparative Study of 2 Treatment Strategies Hui Jer Hwang, MD1, Ignacio Málaga, MD1, Cecilia Curvale, MD1, Martin Guidi, MD2, Julio De Maria, 77 MD2, Raul Matano, MD2. 1El Cruce Hospital, Buenos Aires, Argentina; 2El Cruce Hospital, La Plata, Buenos Aires, Argentina Is There a Weather Seasonal Effect on Acute Pancreatitis Etiology and Outcomes?

1 2 3 Introduction: Due to the complications with ERCP (5-10%), its indication in patients with inter- Palashkumar Jaiswal, MBBS , Yuchen Wang, MD , Mohamed Elkhouly, MD , C. Roberto Simons-Linares, 1 1 2 mediate pre-test of choledocholithiasis (CL) should be backed up by less invasive and more sensi- MD, MSc . John H. Stroger, Jr. Hospital of Cook County, Chicago, IL; Cook County Health and Hospitals 3 tive methods such as the EUS. In several published studies, the EUS showed to be very effective System, Chicago, IL; Cook County Health and Hospital Systems, Chicago, IL at avoiding the performance of ERCP (30-73.3%) in patients with intermediate risk of CL. This is the first study made in Argentina that assesses the clinical impact that the EUS has as a “filtering” Introduction: Approximately 20% of Acute pancreatitis (AP) cases are deemed to be idiopathic. method to avoid unnecessary ERCPs. Objective: to determine the clinical impact of the EUS in the Gallstone disease has been reported to have increased incidence during summer and this could treatment of patients with intermediate probability of CL and the amount of ERCPs that could be be due to variations in temperature during these months. The aim of this study was to identify the avoided. prevalence of different AP etiologies during the weather seasons and to investigate the association Methods: Prospective study of patients with intermediate risk of CL (according to the ASGE criteria) between temperature variation, seasonal effect and the onset of acute pancreatitis, its severity and in which two strategies are compared: a) Group with EUS: EUS was performed as a filter. ERCP was outcomes. performed immediately in the cases in which CL is detected, while in the negative cases there was a clini- Methods: We performed a retrospective cohort study of consecutive patients admitted with a very first cal control; b) Group without EUS: ERCP was performed directly. The group was selected according to episode of acute pancreatitis at a large public hospital in Chicago between 01/2013 and 12/2014. We the availability of the physician performing the study. We consecutively included patients derived to our identified acute pancreatitis diagnosis by ICD9 code and/or lipase ≥ 3 times the normal upper limit. Two hospital to perform ERCP with suspicion of CL from January 2016 to April 2017. physicians reviewed each case to include only first episodes of acute pancreatitis. We excluded patients Results: 44 patients were included in the EUS group (average age: 37, 37 females) and 27 in the group who were transferred to our center. We included a detailed weather season history and temperature in without EUS (average age: 38, 21 females). In the first group, the results were negative for CL in 29 Celsius degree on the day of admission for each AP case from the publically available website: weather. cases (66%); CL was detected in 15 patients (34%) to whom we performed ERCP and confirmed the com. We constructed multivariable logistic regression models. diagnosis in 12 of them (80%). There was only one complication: post-ERCP low-grade pancreatitis. Results: We analyzed 460 patients. Mean age was 48 years (range 17 to 89 years), 54% were males. In In the second group, the one without EUS, the ERCP was normal in 15 patients (56%), we found Univariate and Multivariate analysis, having an AP episode during winter was found to be protective presence of CL in 10 cases (37%), biliary stenosis in 1 patient (3.7%) and cannulation failure in 1 case against Acute Kidney Injury (AKI) (aOR: 0.4, P<0.005, CI 0.2-0.9). After adjusting for admission SIRS, (self-limited hemorrhage after the infundibulotomy). There was one case of post-ERCP moderate age and gender; Weather seasons were found to impact AP etiology nor be independent risk factors for pancreatitis. Persistent Systemic Inflammatory Response Syndrome (SIRS), AKI, ARDS, pancreatic necrosis, mortality, Conclusion: The EUS has shown to be very useful as a “filtering” method in the diagnostic algorithm of ICU, Length of Stay (LOS), in-hospital infections, BISAP score, nor recurrent AP. In addition, the mean patients with intermediate risk of CL. In the EUS group, ERCP could be avoided in 2/3 of the patients, temperature (T) of Chicago (15°C) was compare to lower temperatures as well as higher temperature reducing the number of normal ERCPs and the risk of complications. We recommend the indication of variations; Higher temperatures than the average Chicago T was found to be protective against Persistent EUS in the group of patients with intermediate suspicion of CL as previous instance for the performance SIRS at 48h (aOR 0.6, P<0.05, CI 0.3-0.9). No other significant impact on these temperatures variations of the ERCP. was found.

The American Journal of GASTROENTEROLOGY VOLUME 112 | SUPPLEMENT 1 | OCTOBER 2017 www.nature.com/ajg Abstracts S35

Conclusion: Weather Seasonal effect was not found to impact the etiology of acute pancreatitis. All study does not report the LOS. One study reported significantly lower mortality in plasmapheresis group etiologies were equally distributed among all seasons. However, AKI during winter was found to have a (9/41 vs 22/50, p =0.001). significantly lower incidence and Persistent SIRS had lower incidence during higher or warmer tempera- Conclusion: These 5 studies do not yield clinically significant inference. The existing literature does not tures. This is the largest study to date to examining weather seasonal effect in AP. match the baseline characteristics e.g. age, gender, TG levels, the severity of AP and the sample size is rather modest in all these studies. Most of them do not reflect on short-term clinically significant outcomes like mortality, complications like acute kidney injury, sepsis, respiratory failure or long-term outcomes like mortality and readmission. It is necessary to establish guidelines on a few issues like the indications of initiating TPE, indications of transitioning to insulin or stopping TPE before its role can be established in the management of HTP. Currently, the precise role of TPE in the management of HTP remains obscure. We hope that our review would lay the foundation for a well-designed randomized 78 control trial that can match the patient’s baseline characteristics and help determine long-term and short- term outcomes of TPE in HTP. The Role of Therapeutic Plasma Exchange in Management of Hypertriglyceridemic Pancreatitis: A Systematic Review

Palashkumar Jaiswal, MBBS1, Sachit Sharma, MD1, Radhika Jaiswal, MBBS2, Yuchen Wang, MD3, Munish Ashat, MD1, Gurpartap S. Sidhu, MBBS, MS3, Bashar M. Attar, MD, PhD4, Carlos Roberto Simons-Linares, MD1, Pranav D. Patel, MD5, Ashu Acharya, MD1, Edward Villa, MD1, Gijo Vettiankal, MD4. 1John H. Stroger, Jr. Hospital of Cook County, Chicago, IL; 2Forest Hills Hospital, Forest Hills, NY; 3Cook 79 County Health and Hospitals System, Chicago, IL; 4Cook County Health and Hospital Systems, Chicago, IL; 5Montefiore Medical Center, Bronx, NY Factors Affecting Opioid Use in Hospitalized Patients With Acute Pancreatitis

Nasim Parsa, MD1, Mahya Faghih2, Robert Moran, MD2, Ayesha Kamal, MD1, Niloofar Yahyapourjalaly, Introduction: Hypertriglyceridemia is the third most common cause of acute pancreatitis, behind gall- MD1, Venkata Akshintala, MD2, Haitham Al-Grain, MD2, Martin Makary, MD2, Mouen Khashab, MD2, stones and alcohol. It is usually treated with conservative management and lipid lowering pharmacologic Anthony N. Kalloo, MD, FACG2, Vikesh K. Singh, MD2. 1Johns Hopkins University School of Medicine, agents. Therapeutic plasma exchange (TPE) is a promising non-pharmacological modality for treatment Baltimore, MD; 2Johns Hopkins University Hospital, Baltimore, MD of severe hypertriglyceridemic pancreatitis (HTP) but the American Society of Apheresis has categorized TPE treatment as a category III strategy which implies “Optimum role of TPE is not established; the decision must be individualized”. The objective of this study is to take a closer inspection at the role Introduction: Opioid analgesics are commonly used to treat abdominal pain in patients hospitalized for of TPE in the treatment of HTP by systematically reviewing the studies which compare the efficacy of acute pancreatitis (AP). However, it is not known whether the quantity of opioid analgesia required is TPE vs. standard management. based on patient or disease specific characteristics in AP. The aim of this study is to quantify the amount Methods: A computer-assisted literature search of PubMed, Ovid and Google Scholar search engine was of and determine the factors associated with opioid analgesic in hospitalized patients with AP. To deter- conducted from 1946 – 2016 (Fig 1). We included all the human studies that enrolled patients with HTP, mine the factors associated with high opioid analgesic use. level of TG >500mg/dl and had at least 2 patients in TPE arm. We excluded all studies that did not have Methods: The records of all adult patients with the diagnosis of AP from 2006-2016 were reviewed a control group and studies that included patients with pancreatitis due to other etiologies like alcohol, and abstracted for demographic and clinical data. The revised Atlanta classification (RAC) was used to gallstones. define AP and severe AP. Exclusion criteria included hospital transfers, incomplete data, readmission Results: Only 5 studies met the required criteria and were included in our review (Table 1). 3/5 studies for AP within 2 weeks of discharge, underlying chronic pancreatitis (CP) as defined by the presence report a shorter length of stay (LOS) in plasmapheresis group, one study reported longer LOS and one

[78] Search strategy for our systematic review. Key words and MeSH terms included “hypertriglyceridemia”, “hyperlipemia”, “pancreatitis”, “hypertriglyceridemic pancreatitis”, “hyperlipemic pancreatitis”, “plasma apheresis”, “plasmapheresis”, “plasma exchange”, “apheresis” as well as combinations of these terms with Boolean operators (“AND” and “OR”). Irrelevant studies* are the ones which were based on completely unrelated subject e.g. pathophysiology of hypertriglyceridemic pancreatitis, use of apheresis in the management of other diseases like guillain-barre disease, mystenia gravis.

[78] Comparision of outcomes in Plamapheresis and non-plasmapheresis groups from the five studies included in systemic review

Study Intervention Control Other Outcomes

Type and sample Mean age Mean TG on Mean TG LoS Type and Mean age Mean TG on Mean TG LoS (days) size (years) admission at endpoint (days) sample size (years) admission at endpoint (mg/dl) (mg/dl) (mg/dl) (mg/dl)

Chen et al. Plasmapheresis NA 2019±780 691±331 after NA Standard NA NA NA NA No statistical difference in mortality, systemic (2004) (N=10) plasmapheresis therapy complications and local complications. sessions (N=19)

Afari et al. Plasmapheresis 41±1 6575±4214 NA, decreased 20.7±3.1 Insulin drip 40±6 5307±4932 NA, decreased 10.3±5.4 2 patients in plasmapheresis group had major (2015) (N=3) by 92.6% on (N=8) by 85.2% on complications of respiratory failure requiring discharge discharge intubation and tubular nephritis requiring hemodialysis.

Mahemuti Plasmapheresis 40 5137±3454 885±885 17±16 Medical NA 4867±3008 3097±1150 28±20 The mortality in the plasmapheresis group was et al (2015) (N=41) therapy upon when? significantly as compared to non-plasmapher- (N=50) esis group. (9 vs 22, P=0.001)

Chang et al. Double filtration 41 13576±1855 NA 5 (4-7) Standard 42 (35-51) 5648±829 NA 10 (7-13) Subgroup with STAP associated with (2016) plasma apheresis (38.0-44.0) therapy TG>5000 mg/gl had no complications on (N=12) (N=24) plasmapheresis.

Huang et al. Plasmapheresis 27.6±6.8 1803±656 463±318 after 17.3±6.7 Standard 31.6±3.6 2386±2132 NA 37.0±20.8 The prevalence of SIRS decreased from 100% (2016) (N=5) plasmapheresis therapy in plasmapheresis group post plasmapheresis sessions sessions to 28.6%.

Abbreviations: LoS, length of Stay; N, Number; NA, not availbale; TG, triglyceride; SIRS, systemic inflammatory response syndrome; STAP, severe hypertriglyceridemia induced pancreatitis.Abbreviations: LoS, length of Stay; N, Number; NA, not availbale; TG, triglyceride; SIRS, systemic inflammatory response syndrome; STAP, severe hypertriglyceridemia induced pancreatitis.

[79A] .

[79C] .

BISAP≥3 defined to predict severity. The medication administration record was examined to determine the total quantity of opioid analgesics administered in the first 7 days of hospitalization and was con- verted to oral morphine equivalents (OME), using the CDC opioid conversion chart. The total OME was divided by the total number of days of opioid analgesics administration to obtain a mean OME per day(s) of treatment. Results: There were 879 patients with AP, of whom 267 were included, with 56.2% males, the mean age of 47(+/- 14) years, alcoholic etiology in 55.5 % and 39.9% with the first episode of AP. According to RAC 136(84%), 15(9.32%) and 10(6.21%) of patients had mild, moderate and severe disease respectively. The mean OME/day was 59mg. When OME/day was assumed as an indirect index of pain severity, it was found that age (β =-1.4, P=0.008), black race (β =-21, P= 0.004), first episode of AP β( =-1.09, P=0.049) and Hematocrit>44% (β =25, P<0.001) were significant predictors in multivariable analysis. Male gender, alcoholic etiology, history of smoking, BUN> 25 on day 1, disease severity per RAC were not predictors of OME/day in acute pancreatitis. Conclusion: Advanced age, black race, and first episode of AP are independently associated with less, but hemoconcentration is associated with increased opioid use in patients hospitalized with AP.

Evaluation of Biliary Bacterial Resistance in Patients With Frequent Biliary Instrumentation: One Size Does Not Fit All [79B] . Maen Masadeh, MD1, Subhash Chandra, MD2, Frederick Johlin, MD1, Daniel Livorsi, MD1, William Silverman, MD1. 1University of Iowa, Iowa City, IA; 2University of Iowa Hospitals and Clinics, Iowa City, IA

of calcification(s) and/or a dilated main pancreatic duct (≥ 5 mm) on past or current abdominal imag- Introduction: Bacteremia due to ascending cholangitis can be encountered as a complication of bil- ing, psychiatric comorbidities, intubation with mechanical ventilation, chronic opioid analgesic use, iary instrumentation. Patients who undergo frequent endoscopic retrograde cholangiopancreatography and illicit drug use. Co-morbidities were quantified using the Charlson Co-morbidity Index (CCI). The (ERCP) usually receive periprocedural prophylactic antibiotics.

The American Journal of GASTROENTEROLOGY VOLUME 112 | SUPPLEMENT 1 | OCTOBER 2017 www.nature.com/ajg Abstracts S37

[80] Demographics and clinical characteristics.

Methods: This is a retrospective review of patients with more than one ERCP, and bacteremia due to 81 ascending cholangitis. Conventional empiric antibiotics were defined as ciprofloxacin, penicillins alone or first or second-generation cephalosporin. Antibiotics with broader spectrum were defined as wide Initial Experience With Endoscopic Ultrasound-Guided Fine Needle Biopsy in a Safety Net spectrum. Patient Population Results: A total of 140 patients over a period of 6 years were reviewed. Sixty-two patients were excluded due to bacteremia secondary to other source. Patient characteristics are described in table 1. Over 50% of st nd Mohammad Shahshahan, MD, Ali Fakhreddine, MD, Tahmina Haq, MD, Daniel Shue, MD, Linda Hou, bacteria were not sensitive to conventional empiric antibiotics for biliary sepsis (ciprofloxacin or 1 or 2 MD, Anuj Datta, MD, Viktor Eysselein, MD, Sofiya Reicher, MD. Harbor UCLA Medical Center, Tor- generation cephalosporin). Forty-one patients received post procedural antibiotics for one week starting rance, CA from the day of the procedure, of those 58% grew bacteria resistant to the antibiotic used for prophylaxis and 26 patients (63%) required a wider-spectrum antibiotic for treatment. Number of ERCPs was not associated with resistance to prophylactic antibiotics (p 0.7103) or needing broader-spectrum antibiotics Introduction: EUS-guided fine needle biopsy (EUS-FNB) has emerged as a novel alternative to EUS- for treatment of ascending cholangitis associated bacteremia (p 0.1868). Routine use of antibiotic prophy- guided fine needle aspiration (EUS-FNA) by providing tissue amendable for histologic evaluation with laxis after ERCP was associated with trend towards need for wider-spectrum antibiotics for ascending preserved cell morphology and architecture. Recent technologies have focused on modified bevel design cholangitis associated bacteremia, chi-square 3.7, P= 0.0540. to improve EUS-FNB tissue acquisition. Our goal was to evaluate the performance of EUS-FNB with Conclusion: Biliary bacterial resistance to conventional empiric recommended antibiotics is an emerging new generation biopsy needles in a safety net patient population. problem, and consideration for wider spectrum antibiotics should be made in certain patients, particu- Methods: Electronic database of endoscopic procedures was retrospectively queried for all procedures larly in patients with prior antibiotic exposure. Blood cultures are needed to guide therapy. performed with EUS-FNB needles (Acquire, Boston Scientific or Shark Core, Medtronic) from 5/2016 to 5/2017. A standard technique was used with an average of 2-5 passes and 10 to-and-fro movements per pass; all procedure were performed by 3 experienced endoscopists. All specimens were collected, fixed in formalin, and were processed as histology. Data extraction was performed for patient demographics, target lesion, diagnostic adequacy, diagnostic accuracy, and complications. Diagnostic accuracy was assessed based on clinical follow-up in thirty-five cases, surgical follow-up in three cases and repeat biopsy in two cases.

Results: 40 biopsies were performed with the following target sites: pancreatic lesions (53%), liver paren- 84 chyma (15%), abdominal mass (7.5%), submucosal lesions (7.5%), lymph nodes (7%), gastric mass (5%), hilar (2.5%) and ampulla masses (2.5%). Of the pancreatic lesions, 33% displayed a cystic component. Race/Ethnicity-Specific Disparities in Gallbladder Cancer Stage at Diagnosis and Its Impact on Overall diagnostic adequacy of EUS-FNB was 32/40 (80%). Overall diagnostic accuracy was 29/38 (76%), Receipt of Curative Therapies with 2 patients pending follow-up at this time. Diagnostic adequacy and accuracy for pancreatic lesions was 15/21 (71%) and 13/20 (65%), respectively. Of non-cystic pancreatic lesions 12/14 (86%) were diag- Veeravich Jaruvongvanich, MD1, Buravej Assavapongpaiboon, MD2. 1University of Hawaii, Honolulu, HI; nostically adequate while 11/14 (79%) were diagnostically accurate. Diagnostic adequacy and accuracy 2Chulalongkorn University, Bangkok, Krung Thep, Thailand for liver parenchymal, abdominal and submucosal mass biopsy were 12/12 (100%). Diagnostic accuracy for adequate specimens was 31/32 (97%). There were no procedural complications. Conclusion: Our data suggests that EUS-FNB is a safe alternative to EUS-FNA with adequate and accu- Introduction: Gallbladder (GB) cancer is a rare malignancy in most countries. Overall survival has rate core tissue acquisition while used in a complex patient population with a variety of target lesions. improved in all races/ethnicities except for Hispanics and Blacks. There are few publications outlin- ing the race/ethnicity-specific disparities in stage of GB cancer at diagnosis, which reflect disparities in receiving potentially curative therapies. The aim of this study was to investigate race/ethnicity-specific disparities in GB cancer staging and types of treatment. 82 Methods: Using the 2000–2013 Surveillance, Epidemiology and End Results 18 registries in the United States. Race/ethnicity-specific cancer stage at diagnosis and treatment received among adults with GB The PLP Score: Predicting Mortality in Pancreatic Cancer With a Novel Nutritional and cancer were evaluated. Differences in gallbladder cancer stage at presentation, gallbladder cancer treat- Inflammatory Index ment methods and number of lymph nodes (LN) examined among race/ethnicity groups were evaluated using multivariate logistic regression models adjusting to age, sex and year of diagnosis. Early stage of GB Ammar Nassri, MD1, Mayssan Muftah, MD2, Zeeshan Ramzan, MD3. 1University of Florida Jacksonville, cancer was defined as patients with T1, T2 or T3 and M0 disease, which are amenable for curative surgery. Austin, TX; 2Emory University School of Medicine, Springhill, FL; 3Dallas VA Medical Center, Dallas, TX Optimal LN clearance was defined as more than 4 LN examined. Results: A total of 7,507 patients with GV cancer were included in this study. There was no racial disparity in stage of disease at presentation. With regards to disparity in treatment, Blacks were significantly Introduction: The aim of this study was to determine the utility of inflammatory and nutritional indices less likely to receive curative surgery compared to Whites in all patients with GB cancer (AOR 0.70, such as the Prognostic Nutritional Index (PNI), Lymphocyte to Monocyte Ratio (LMR), and Platelet to 95%CI 0.59-0.83; P<0.001) and early-stage GB cancer (AOR 0.65, 95%CI 0.53-0.79; P<0.001) (Table 1). Lymphocyte Ratio (PLR) individually, and as a combined score (PLP), for predicting survival in patients Asians were significantly less likely to receive radiation compared to Whites (AOR 1.25, 95%CI 1.01-1.56; diagnosed with pancreatic cancer (PC). P=0.04) (Table 1). There was no racial disparity for treatment in advanced-stage GB cancer. In patients Methods: The records of all patients diagnosed with PC at our institution were reviewed. The PNI, LMR who underwent surgery, Hispanics were significantly less likely to have optimal LN clearance compared and PLR at diagnosis were calculated (Table 1) and univariate Cox regression analysis was performed. to Whites (AOR 0.59, 95%CI 0.47-0.74; P<0.001) (Table 2). We used the median values of LMR (2.05), PLR (186) and a PNI of (35) to separate patients into low and Conclusion: Among US adults with GB cancer, no racial disparity in stage of disease at presentation was high index score groups. Kaplan-Meier analysis (KM) with log rank test was used to compare survival noted. However, Blacks, Asians and Hispanics were less likely to receive curative surgery, radiation and between groups. A combined score, the PLP, was created using the PNI, LMR and PLR (Table 1). Only optimal LN clearance, respectively. patients who had all three values available were included. Survival between low PLP (0-1) and high PLP (2-3) score was compared. Results: A total of 109 patients were identified (mean age 66 ± 0.9 years; 100% male; 67% white, 30% black). 89% of patients had Adenocarcinoma. Stage at presentation was I (1.8%), II (24.8%), III (13.8%) and IV (59.6%). On Kaplan-Meier analysis overall median survival was 154 ± 18 days. In univariate Cox [84_A] Race/ethnicity-specific variations in gallbladder cancer treatment by stage at regression analysis, a low PNI and LMR and a high PLR was associated with poorer overall survival diagnosis (P=0.003, P=0.015) and (P=0.002) respectively. On KM analysis, there was poorer survival in the low PNI group compared with the high PNI group [56 ± 37 vs 159 ± 24 days, P=0.002, 95% CI 120-188] as was the Race/ Total GB cancer Early-stage GB cancer Advanced-stage GB cancer survival of patients in the low-LMR group compared to the high- LMR group [93 ± 26 vs. 198 ± 28 days, ethnicity P=0.007, 95% CI 104-198]. Survival in the low-PLR group compared to the high-PLR group was higher [159 ± 45 vs. 138 ± 38 days; P=0.023, 95% CI 104-198]. Patients with a low PLP score had a significantly poorer survival than those with high PLP score [94 ± 45 vs. 162 ± 30 days, P=0.008]. AOR* 95%CI P AOR* 95%CI P AOR* 95%CI P Conclusion: A high PNI, LMR and low PLR score at the time of diagnosis of PC predicts better overall survival and the use of a combined score, the PLP, attenuates these findings. Noninvasive measures Surgery versus no surgery obtained at diagnosis such as a CBC with differential and serum albumin can be used to obtain the PLP White 1 Reference 1 Reference 1 Reference score and can provide important prognostic information for patients with PC and their clinicians. Black 0.70 0.59-0.83 <0.001 0.65 0.53-0.79 < 0.001 0.78 0.53-1.14 0.20

Asian 0.93 0.77-1.12 0.43 0.94 0.75-1.18 0.60 0.76 0.50-1.15 0.19

[82] Formulas Hispanic 1.06 0.92-1.21 0.44 1.10 0.93-1.30 0.27 0.90 0.68-1.21 0.50

Radiation versus no radiation PNI 10 × serum albumin (g/dL)+0.005 × total lymphocyte count (per mm3) White 1 Reference 1 Reference 1 Reference LMR Absolute Lymphocyte Count (ALC)/ Absolute Monocyte Count (AMC) Black 1.04 0.85-1.28 0.70 1.08 0.86-1.35 0.51 0.81 0.43-1.53 0.51 PLR Platelet count (k/mcl)/Absolute Lymphocyte Count (ALC) Asian 1.25 1.01-1.56 0.04 1.22 0.97-1.55 0.10 1.41 0.76-2.59 0.27 PLP One point each for high PNI (>35), high LMR (>2.05) and low PLR (<186) Hispanic 0.90 0.76-1.06 0.22 0.90 0.75-1.08 0.26 0.90 0.56-1.45 0.67 PNI: Prognostic Nutritional Index; LMR: Lymphocyte to Monocyte Ratio; PLR:Platelet to Lymphocyte Ratio. *Adjusted for age, sex and year of diagnosis.

83 [84_B] Race/ethnicity-specific variation in number of lymph node examined in patients Alteration of Pancreatic Cystic Neoplasm Glycopatterns Between MCN and SCN with gallbladder cancer

1 2 3 4 3 Ying Wang, MD , Ming Yang, MS , Jia Feng, MD , Yufa Sun, MD , Ningli Chai, MD, PhD , Fuquan Yang, ≥3 lymph node assessed vs. no lymph node assessed or AOR* 95%CI 2 3 1 P MD, PhD , Enqiang Linghu, MD, PhD . Department of Gastroenterology, Chinese PLA General Hospital, unknown Beijing, China (People's Republic); 2Laboratory of Proteomics, Institute of Biophysics, Chinese Academy 3 of Sciences, Beijing, China (People's Republic); Chinese PLA General Hospital, Beijing, China (People's White 1 Reference Republic); 4Department of Health Care, Central Guard Bureau, Beijing, China (People's Republic) Black 0.84 0.64-1.11 0.22

Introduction: The pathophysiology of malignant pancreatic cystic including Mucinous cystic neoplasm Asian 0.78 0.57-1.07 0.12 (MCN)，Intraductal papillary mucinous neoplasm (IPMN), Mucinous cystic adenocarcinoma (MCA) are not fully understood and there are no diagnostic or predictive biomarkers. Glycosylation modified as Hispanic 0.59 0.47-0.74 < 0.001 many as 70% of all human proteins can sensitively reflect various pathological changes. However, little is known about the alterations of glycosylation and glycoproteins in MCN. *Adjusted for age, sex and year of diagnosis. Methods: This study is based on the results of surgical pathology and cystic fluid cytology in 35 cases from 120 cases diagnosed by CT or MRI in patients with pancreatic cystic tumor samples. Of the 35 cases, 17 are from the MCN group and 18 are from the serous cystic adenoma (SCN) group. The liquid samples are gained through fine needle biopsy under endoscopic ultrasonography (endoscopic ultrasonography- guided fine needle aspiration, EUS- FNA). In former data shownlectin microarrayswas used to character the altered glycosylation between MCN and SCN. The T test results show that 6 lectins(STL, WGA, BPL, DBA, PTL–I and MAL–I) showed different binding signals between two groups P( <0. 05). So this study 85 is based on the results of exhibited increased binding signals in the MCN cyst,wheat germ agglutinin (WGA) and solanum tuberosum (potato) lectinbinding glycoproteins (STL), magnetic particle conjuga- Common Bile Duct Stone Size Is an Independent Predictor of Stone Detection by Magnetic teassisted LC-MS/MS analysesand bioinformatics analysis. Resonance Imaging Results: The lectin binding glycoproteins among these STL, WGA exhibited character the altered glycopatterns between MCN and SCN, the protein was evaluated in three repetitions.Through bioinformatics Sehem Ghazala, MD1, Bhupesh Pokhrel, MD2, Manish P. Shrestha, MD2, Shilpa Junna, MD3, John Cun- analysis and StringProtein interaction diagrams,We concluded that the glycoprotein LAMC1; YWHAG; ningham, MD1, Hemanth Gavini, MD1. 1University of Arizona, Tucson, AZ; 2University of Arizona College YWHAQ; YWHAZ; S100A11; HSPA5; LDHA; ACTB; KRT18 were participated in the development of of Medicine, Tucson, AZ; 3University of Arizona Medical Center, Tucson, AZ pancreatic cystic tumor,which indicated the abundance in MCN group. Conclusion: The glycoprotein of LAMC1; YWHAG; YWHAQ; YWHAZ; S100A11; HSPA5; LDHA; ACTB; KRT18 may correlated with a significant increase in the mucin secretion of cancer epithelial cells. Introduction: Endoscopic retrograde pancreatography (ERCP) is now primarily considered a therapeutic modality for management of common bile duct (CBD) stones. Endoscopic ultrasound (EUS) and

The American Journal of GASTROENTEROLOGY VOLUME 112 | SUPPLEMENT 1 | OCTOBER 2017 www.nature.com/ajg Abstracts S39

[85_A] Study population characteristics based on CBD stone size (univariate analysis) 86 Non-alcoholic Fatty Pancreas Disease Is Associated With Prediabetes in Hispanics Population charac- Total Cohort Small stone* Large stone* P=value*** teristics (n=110) (n=72,65.5%) (n=38,34.5%) Javier Jose Hernandez Buen Abad, MD, Maria Cristina Moreno del Castillo, MD, Alain Sanchez Rodriguez, MD, Maria Emilia Mendizabal Rodriguez, MD, Jonathan Salazar Segovia, MD, Ricardo Garcia-Buen- Age, median (IQR#), 53 (27-68) 46 (23.5-65) 64 (35-75) 0.005 Abad, MD, Jose Ramon Mena Ramirez, MD, Natalie Atkinson, MD, Ignacio Garcia Juarez, MD. American years British Cowdray, Medical Center, Mexico City, Distrito Federal, Mexico Gender, No. (%) 0.15

Male 42 (38.2) 24 (33.3) 18 (47.4) Introduction: Non-alcoholic fatty pancreas disease (NAFPD) has been associated with the development of type 2 diabetes mellitus (DM2), increased acute pancreatitis severity and the development of Female 68 (61.8) 48 (66.7) 20 (52.6) pancreatic cancer. The aim of our study is to describe the association of NAFPD and prediabetes (PDM) in asymptomatic patients. Body mass index, 28.5 (25.1-33.2) 28.9 (25.2-33.8) 28.1 (24.6-32.3) 0.67 Methods: Cross-sectional observational study from a single center in Mexico City. We defined NAFPD median (IQR), kg/m2 based on ultrasound features and alcohol consumption of less than 20 gr/day in men or 10 g/day in women. PDM was defined by impaired fasting glucose (IFG) after a 12 hour fast (with a glucose value of Diabetes, No. (%) 16 (14.4) 11 (15.3) 5 (13.2) 0.99 100-125 mg/dL) and/or a glycated hemoglobin (A1C) test between 5.7 and 6.4 percent. Hypertension, No. (%) 41 (37.3) 25 (34.7) 16 (42.1) 0.53 Results: We included data from 458 asymptomatic ambulatory patients who underwent health check-up. The prevalence of NAFPD detected by ultrasound was 17.3% (95% CI, 13-20), and 37% of patients in this Hyperlipidemia, No. (%) 22 (19.8) 16 (22.2) 6 (15.8) 0.46 group had associated non-alcoholic fatty liver disease (NAFLD). The global prevalence of PDM was 26.6 % and 10/122 (91.8%) were newly detected cases, 62.7% were male, mean BMI was 26.02 (SD +/- 4.0) kg/m2. Cholecystectomy, 0.01 Patients with PDM had associated NAFPD [OR=2.8 (95% CI 1.025 – 7.8), P=0.038]. No. (%) Conclusion: We found an association between NAFPD and PDM in asymptomatic outpatients. To our knowledge there have not been any such descriptions in our population. Prospective studies are required Pre-ERCP 43 (39.1) 27 (37.5) 16 (42.1) to determine the importance of pancreatic fat deposition and its consequences in Hispanics. Post-ERCP 52 (47.3) 40 (55.6) 12 (31.6)

No Cholecystectomy 15 (13.6) 5 (6.9) 10 (26.3) 87 Dilated Common bile 61 (55.4) 36 (50) 25 (65.8) 0.16 duct, No. (%) Xanthogranulomatous Cholecystitis: The Needle in the Haystack *Small stone defined as stone size£5 mm based on ERCP findings. **Large stone defined as stone size>5 mm based on ERCP findings. Shraddha Shetty, MBBS, MS, MRCS, Aashish R. Shah, MBBS, MS, FRCS, Kapali Neelamegam, MBBS, MS, FRCS. Fortis Hospital, Bangalore, Karnataka, India ***P-values cited compare patients with small and large stones. Dilated common bile duct defined as common bile duct diameter >8 mm. #IQR Interquartile range. Introduction: The rise of laparoscopic cholecystectomy (LC) as the gold standard for management of gallstone disease has exponentially increased the number of sacrificed gallbladders. In this gargan- tuan pool of specimens, several rare variants of inflammation, infection, premalignant and malignant transformation have become manifest. One of these is the xanthogranulomatous variety of cholecystitis (XGC), an aggressive inflammatory condition known for throwing the surgeon off course from a simple LC to aggressive liver resection by virtue of its malignant mimicry. Methods: We retrospectively analysed 1000 consecutive LCs performed in a tertiary care hospital from [85_B] Multivariate Logistic Regression Results for the CBD stone detection by MRCP January 2010 to December 2016. In the cohort found to have XGC on histopathology, we analysed the gender predilection, age preponderance, associated comorbidities, clinical presentation, radiological fea- Variable Adjusted Odds Ratio (95% confidence interval); tures, intraoperative findings, complications and postoperative course. We then compared this with the P-value rest of the source population. Results: Of the 1000 patients that constituted the study population, 35 patients were found to have XGC. Age 0.99 (0.96-1.02); 0.76 We found a male predilection of 5:4, with age ranging from 32 to 76 years, in opposition to the study population that varied from 2 to 85 years. The cohort had a 20% incidence of diabetes mellitus, compa- Male gender 2.43 (0.82-7.21); 0.11 rable to the source population. No difference was found in either duration or severity of symptoms in the BMI 0.94 (0.88-1.00); 0.07 two groups. Preoperative radiological diagnosis of XGC was not done in any of the cases. Intraoperatively, 14.3% of the patients were found to have Mirizzi's syndrome, which formed the majority of the Mirizzi's Hypertension 2.46 (0.62-9.83); 0.20 in the source population. Empyema of the gall bladder was found in 5 of the patients, with incidence similar to the source population. Operative time in the XGC cohort was significantly higher (80±23 vs Diabetes 0.56 (0.12-2.56); 0.45 43±7 minutes) which also formed a sizable fraction of the subtotal excisions performed (26.7%). While the incidence of complications was similar in both groups, postoperative stay was significantly longer in Hyperlipidemia 0.36 (0.08-1.48); 0.15 the XGC cohort (2.7±1.3 days vs 1.1±0.3 days). Only 3 cases of gall bladder malignancy were found in the Cholecystectomy entire study population, none of whom exhibited features of XGC. Conclusion: XGC continues to be the bane of the surgeon, confounding findings and causing even Pre-ERCP cholecystectomy 1(Reference) seasoned professionals to second guess their clinical decisions. Albeit rare, a working knowledge of the condition is of paramount importance in order to provide the patient the best and most appropriate Post-ERCP cholecystectomy 1.49 (0.54-4.08); 0.43 care.

No cholecystectomy 2.74 (0.37-20.40); 0.32

Stone size>5 mm 5.57 (1.66-18.70); 0.005 88 CBD diameter>8 mm 0.81 (0.32-2.01); 0.14 Multivariate Predictors of In-hospital Mortality With Acute Pancreatitis Admissions in the United States

Shreyans Doshi, MD1, Yash Shah, MD2, Jiten Desai, MD3, Rajkumar P. Doshi, MD, MPH4, Christopher L. Magnetic resonance imaging (MRI) are less invasive modalities used in the diagnosis of CBD stones Bray, MD, PhD5. 1University of Central Florida College of Medicine, Gainesville, FL; 2VA Hospital Philadel- prior to ERCP. The aim of the study is to detect if there is an additional diagnostic yield of EUS over MRI phia, Philadelphia, PA; 3Nassau University Medical Center, East Meadow, NY; 4North Shore University Hos- in detecting small common bile stones. pital, Jersey City, NJ; 5UCF COM/HCA GME Consortium Internal Medicine Gainesville, Gainesville, FL Methods: A retrospective cross-sectional study was performed in patients who underwent ERCP between July 2011 and March 2017 for suspected CBD stone. The study included patients with CBD stone confirmed on ERCP and had a MRCP and an EUS within 72 hours from ERCP. Patient’s demographic data, Introduction: Acute Pancreatitis (AP) is one of the common gastrointestinal disease associated with medical/surgical history, and imaging findings were collected from electronic health record (EHR). Imag- long term hospitalizations and higher in-mortality rate. Most AP presentations are self-limited but ing findings included CBD stone size on ERCP (categorized as small ≤5 mm vs. large >5 mm), CBD ranges from mild to life threatening attack and mortality associated with AP can be up to 30% which diameter (categorized as dilated if CBD diameter >8 mm), and presence or absence of CBD stone on depends on the area of sterile versus infected necrosis. Timely identification of severity of the disease EUS and MRCP. is important which can help alleviate further complications. Although all scoring systems have been Results: A total of 110 patients were identified to have CBD stone on ERCP. In the final analysis, there shown to correlate with mortality, it remains difficult to accurately identify severity of AP patient on were 110 patients with MRCP detecting stone in 66.4% patients. Seventy-two (65.5%) patients had small admission. We have analyzed multivariate predictors of in-hospital mortality with AP admissions in CBD stone and 38 (34.5%) had large CBD stone. Of this study population, the median age was 53 years, the US. and 61.8% were female. The median BMI was 28.5. A total of 41 (37.3%) had hypertension; 22 (19.8%) Methods: It is a retrospective, observational study which analyzes various factors that estimate in-hospital had hyperlipidemia and 16 (14.4%) had diabetes. Most patients had cholecystectomy (47.3%) after ERCP mortality in patient with AP using mixed-effect multivariate predictor model. The sample for this study and 39.1% before ERCP while the rest still have their gallbladder. Sixty-one (55.4%) had dilated CBD was obtained from the National Inpatient Sample (NIS) for the years 2010-2014 using ICD-9 CM diag- (Table 1). The multivariate logistic regression analysis revealed that larger stone size was associated with nostic code of 577.0 which is specific for AP. Analysis were performed in SAS 9.4 (SAS institute Inc., a higher likelihood of stone detection by MRCP (adjusted odds ratio: 5.57; 95% confidence interval [CI] Cary, NC). 1.66 -18.7; P=0.005) (Table 2). Results: A total of 449,357 patients were identified with diagnosis of AP. Higher age is associated with Conclusion: MRCP has lower detectability rate for small (≤ 5 mm) common bile duct stone compared to higher mortality. Black patients have higher odds of in-hospital mortality compared to white. Because of EUS and ERCP. Sensitivity of MRCP for stones larger than 5 mm was satisfactory. Patients with a low or more high-risk patients admitting to urban hospitals compared to rural, these hospitals have higher odds intermediate clinical index of suspicion for choledocholithiasis, further investigation with EUS is recom- of in-hospital mortality. Also, presence of liver disease, electrolyte disorder, neurological disorder, chronic mended despite negative MRCP. pulmonary disease, congestive heart failure, renal failure, and cancers were associated with greater risk of mortality. In contrast, females, private insurance as a primary payer, diabetes, hypertension, anemia and depression are associated with lower risk of mortality. Conclusion: Patients presenting with positive predictors identified our study requires close monitoring and extra-care to further avoid in-hospital mortality. Black patients with higher age and multiple

[88] Multivariate Predictors of In-hospital Mortality with Acute Pancreatitis.

comorbidities should be well evaluated as they have higher risk of in-mortality. Further prediction models thrombosis (DVT) or pulmonary embolism (PE) is a proven life-threatening complication of AP. Mor- require addition of biomarkers, inflammatory mediators, and genetic factors to better predict the severity tality related to VTE did not decrease even after advancements of treatment and preventive measures. In with AP admissions. recent years, deaths related to PE in acute pancreatitis also reported in case reports. We sought to analyze trends of DVT and PE in AP Admissions in US. Methods: Our study cohorts were identified from Nationwide inpatient sample between 2010-2014 using appropriate diagnostic codes for VTE and AP. Jonckheere Terpstra trend test was performed and trend were analyzed for hospitalization, in-hospital mortality, and length of hospitalization stay. Analysis was 89 performed using SAS 9.4 (SAS institute Inc., Cary, NC). Results: A total of 449,357 cohorts were identified having AP, out of which 1.9% cohorts had DVT and Trends and In-hospital Mortality of Venous Thromboembolism in Patients With Acute 0.5% cohorts had PE. Occurrence of DVT (Ptrend =0.91) and PE (Ptrend=0.18) over the study period is Pancreatitis consistent. In-hospital mortality in those without DVT or PE is decreasing over the course of the study (Ptrend=<0.001). In-hospital mortality associated with DVT increased significantly over the course of the Shreyans Doshi, MD1, Yash Shah, MD2, Jiten Desai, MD3, Rajkumar P. Doshi, MD, MPH4, Christopher L. study period (Ptrend =0.02). However, Length of stay in admission with DVT or PE (Ptrend=<0.01) with Bray, MD, PhD5. 1University of Central Florida College of Medicine, Gainesville, FL; 2VA Hospital Philadel- AP reduced over the course of the study period. phia, Philadelphia, PA; 3Nassau University Medical Center, East Meadow, NY; 4North Shore University Hos- Conclusion: VTE remains a significant factor for morbidity and mortality in admission with AP. Further pital, Jersey City, NJ; 5UCF COM/HCA GME Consortium Internal Medicine Gainesville, Gainesville, FL research warrants to identify high risk population for VTE specifically in acute pancreatitis patients. These patients can be sent for close monitoring of DVT and PE which are preventable.

Introduction: Acute Pancreatitis (AP) is one of the common gastrointestinal disease associated with long term hospitalizations. Inflammatory process in AP can trigger local as well as systemic inflammatory process which can eventually cause Venous Thromboembolism (VTE). Although rare, Deep vein

The American Journal of GASTROENTEROLOGY VOLUME 112 | SUPPLEMENT 1 | OCTOBER 2017 www.nature.com/ajg Abstracts S41

[89A] Prevalence of DVT and PE in Patients with Pancreatitis.

[89B] In-Hospital Mortality Associated with Acute Pancreatitis- Stratified by Presence of DVT or PE.

[89C] Trends of Length of Stay in Acute Pancreatitis- Stratified by DVT or PE.

90 evidence of post ERCP pancreatitis (PEP). Patients with a diagnosis of PEP were excluded. Patients with NPPEP were compared with procedural complexity-matched controls without post ERCP abdominal pain in 1:2 ratio for demographic and procedure related factors. Descriptive analysis as well as paired t In-hospital Outcomes After Concomitant Occurrence of Atrial Fibrillation With Acute test and chi-square test were performed for analysis. Pancreatitis: Insights From National Inpatient Sample Results: A total of 174 ERCPs performed in patients with divisum were reviewed. Out of all procedures, the incidence of NPPEP was 14%. (24/174) which was significantly higher than observed incidence of Yash Shah, MD1, Shreyans Doshi, MD2, Jiten Desai, MD3, Rajkumar P. Doshi, MD, MPH4, Arthur C. Lee, 5 1 2 PEP of 5% (9/174) (P – 0.006). In patients with NPPEP, the average age was 34.6 years, 60% were male, MD . VA Hospital Philadelphia, Philadelphia, PA; University of Central Florida College of Medicine, 70% were Caucasian. 77% had complete divisum. 37% procedures were performed on native papilla. 80% Gainesville, FL; 3Nassau University Medical Center, East Meadow, NY; 4North Shore University Hospital, 5 procedures were ASGE complexity grade 3. The patients with NPPEP were admitted for an average of Jersey City, NJ; The Cardiac and Vascular Institute, Gainesville, FL 3.3 days. 40% received rectal Indomethacin during the procedure. The average procedure time was 71.5 mins. 40% of patients were active smokers and 40% drank alcohol. 50% of patients were on home opioids. Introduction: Acute pancreatitis is a potentially fatal disease with an overall mortality of 2.1%–7.8% Compared to an ASGE complexity-matched control group, patients in the NPPEP group were younger with about half of the deaths occurring within the first 1–2 weeks. Patients with AP have an acute rise (35 vs. 55 years, P=0.02). ROC curves showed age <49 years had sensitivity and specificity of of 64% and in inflammatory markers such as C-reactive protein (CRP) and these markers track the severity of AP. 63% respectively. There were no other differences in demographics and procedure related interventions Recent literature also suggests that an elevated level of CRP is associated with an enhanced risk of atrial between the two groups. fibrillation (AF). In addition, the duration of AF is also correlated to the levels of circulating CRP. Con- Conclusion: The incidence of NPPEP was 17%, significantly higher than the observed incidence of PEP comitant presence of both AP and AF can lead to increased length of hospital stay, more likely require in this cohort. NPPEP was more common in younger patients (<49 years). Neither Indomethacin nor skilled nursing facility placement and increased mortality rates. While the existing literature suggests an minor papilla stent placement impacted the risk of NPPEP. Attention to the rate of NPPEP following association between AP and AF, there is scarcity of data on outcomes of patients admitted for AP and minor papilla intervention may help allay unnecessary testing in patients undergoing ERCP with minor have concomitant AF. Our aim of the study was to assess the outcome of patients hospitalized with an papilla therapy. In the study institution, many patients have a scheduled admission following the index episode of AP who have concomitant AF. ERCP for pancreas divisum for close observation and pain management. Methods: Our study cohorts were identified from nationwide inpatient sample between 2010-2014 using appropriate ICD-9 diagnostic codes for AF and AP. Propensity score matched analysis (1:1) was performed between those 2 groups. Outcomes of interest were in-hospital mortality, length of hospitalization stay and disposition to home. 92 Results: AP patients with and without AF were 27,779 and 421,578 respectively (Table 1). Patients with concomitant AP and AF had increased rate of in hospital mortality (6.4% vs. 4.2%, P<0.001), increased Long-term Risk of Surgery and Cancer in Patients Meeting AGA 2015 and Fukuoka 2012 duration of hospitalization (8.5 vs. 6.4 days, P<0.001) and less patients going home (50.7% vs 61.3%, Management Criteria for Pancreatic Cystic Lesions P<0.001) as compared to patients without concomitant AF. Also, concomitant occurrence of AF in hospitalized AP patients is increasing during our study period. Mohammad Al-Haddad, MD1, Sara Jackson, PhD2, Nicole Toney, MPH2, Christina Narick, MD2, Sydney Conclusion: Patients with AP who develop AF during their hospitalization have an increased rate of Finkelstein, MD2, Nadim G. Haddad, MD3. 1Indiana University, Indianapolis, IN; 2Interpace Diagnostics morbidity, in hospital mortality and decreased rate of discharge to home as compared to patients with AP Corporation, Pittsburgh, PA; 3MedStar Georgetown University Hospital, Washington, DC without AF. Close monitoring with additional care is warranted if occurrence of AF is noticed with AP. Introduction: We examined the real-world risk of necessary and unnecessary surgery in patients tri- aged using Fukuoka (2012) and AGA (2015) management guidelines. Patients who met each criteria were reclassified by integrated molecular pathology (IMP) to determine if molecular results altered 91 risk of surgery and provided a net health benefit to the outcomes of patients in which management was altered. Non-pancreatitis-Related Abdominal Pain Following ERCP in Patients With Pancreas Divisum Methods: This is analysis of data from a national pancreatic cyst registry cohort.1 Patients with any worrisome features (WF) were considered Fukuoka positive (FP) (Table 1); Patients lacking all WF were Rushikesh Shah, MD, Kara L. Raphael, MD, Parit Mekaroonkamol, MD, Qiang Cai, MD, PhD, FACG, considered Fukuoka negative (FN). AGA positive (AP) criteria included: severe cytological atypia and/or Steven A. Keilin, MD, FASGE, Field F. Willingham, MD, MPH. Emory University School of Medicine, the presence of at least two positive imaging features. Patients lacking such criteria were considered AGA Atlanta, GA negative (AN). IMP “Benign” and “SI” were considered low risk IMP results; “SHR” and “Aggressive” high risk. Probability of surgery-free and malignancy-free survival for up to 7.7yrs was determined using by Kaplan Meier analysis. Hazard ratio was used to assess relative risk of surgery and cancer. Introduction: Non-pancreatitis Post ERCP abdominal pain (NPPEP) is not uncommon following ERCP Results: Patients who met FN criteria (274/491) were at comparably lower risk of undergoing surgery with minor papilla therapy. ERCP for pancreas divisum is technically demanding procedure requir- and lower risk of cancer than all patients, while patients who met AN criteria (469/491) did not have ing direct pancreatic duct manipulation. Non-Pancreatitis related pain following ERCP in patients with statistically different risk (Table 2). Low risk IMP results decreased risk of surgery and cancer in AN divisum has not been well evaluated. patients but not in FN patients. High risk IMP increased risk of surgery and cancer in both FN patients Methods: A retrospective case-control study in patients with divisum undergoing ERCP was con- and AN patients. Patients who met FP criteria (217/491) and those who met AP criteria (22/491) were at ducted. NPPEP was defined as post ERCP abdominal pain without clinical, biochemical and radiological

The American Journal of GASTROENTEROLOGY VOLUME 112 | SUPPLEMENT 1 | OCTOBER 2017 www.nature.com/ajg Abstracts S43

[90A] In-Hospital Outcomes in Patients Admitted with Acute Pancreatitis- Stratified by Presence or Absence of Atrial Fibrillation After Propensity Score Matched Analysis.

[90B] Trend of Atrial Fibrillation in Patients Admitted with Acute Pancreatitis.

comparably higher risk of undergoing surgery and higher risk of cancer than all patients. Low risk IMP infrequent, the presence of AP criteria increases risk of necessary surgeries. However, the absence results decreased risk of surgery and cancer in FP patients but not in AP patients. High risk IMP increased of AP criteria does not alter risk of surgery or cancer. Low risk molecular results reduce the risk of risk of surgery and cancer in FP patients but not in AP patients. unnecessary surgery when WF are present; high risk results increase the risk of necessary surgery in Conclusion: The absence of all WF lowers risk of unnecessary surgery. High risk molecular informa- such patients. tion increases necessary surgeries in the few patients who lack such features but have cancer. Although References: 1. Loren D, et al. Diagnostic Pathology 2016; 11:5.

93 [91] Comparison between patients with non-pancreatitis post-ERCP pain (NPPEP) with patients without any post-procedure complications Predictors of Readmission for Patients Admitted With Acute Pancreatitis

Factors Non pancreatitis Post No post ERCP P Value 2017 Presidential Poster Award ERCP abdominal pain complication (NPPEP) (n- 24) (n-48) Abhinav Goyal, MD1, Kshitij Chatterjee, MD2, Sujani Yadlapati, MD3, Kamolyut Lapumnuaypol, MD3, Shailender Singh, MD4. 1Albert Einstein College of Medicine, Philadelphia, PA; 2University of Arkansas Age 35.42 54.68 0.02 for Medical Sciences, Little Rock, AR; 3Albert Einstein Medical Center, Philadelphia, PA; 4University of Nebraska Medical Center, Omaha, NE Gender (M:F) 2 1.6 0.6

Race (%Caucasian) 78% 78% 1 Introduction: Acute pancreatitis is commonly seen in clinical practice. Thirty-day readmission is an important quality indicator for Center for Medicare and Medicaid Services (CMS). Hospitals are often Smoking 37.5% 40% 0.81 penalized for readmissions. We attempt to identify 30-day readmission rate and risk factors for readmis- Alcohol 42% 40% 0.9 sion for acute pancreatitis in the United States. Methods: We used the Nationwide Readmissions Database for the year 2013. ICD-9 code 577.0 was Opioid use 37.5% 40.4% 0.81 used to identify index admissions for acute pancreatitis. We used the criteria for Index admissions and 30-day readmissions as defined by CMS. Patients who died during index admission, those who were non- Divisum (Complete: Incomplete) 2.1 5.4 0.1 residents of the state they were hospitalized in, or those with missing key variables were excluded as they were not at risk of readmission. Index discharges during December 2013 were excluded due to lack of Native Papilla (%) 33% 31% 0.8 sufficient time for a 30-day readmission. Manufacturer provided sampling weights were used to produce # of prior divisum intervention (Average) 2.09 2.24 0.73 national level estimates. Proportion of all cause 30-day readmissions for hospitalizations was determined. Logistic regression model was used to find the risk factors for 30-day readmissions (adjusted for patient Use of Indomethacin (%) 37.5% 20.8% 0.13 and hospital related factors). Results: A total of 216,395 eligible index hospitalizations were identified for acute pancreatitis during Procedure Time (Minute) 68 63.8 0.5 2013.14.8% were found to have one or more 30-day readmissions. Age and gender distribution was similar in the two groups (index admissions with and without readmissions). Alcohol abuse (32.3% vs 30.8%) # Prior hospital admissions due to pancreatitis 5.3 3.3 0.2 and discharge to a facility were more prevalent in the group with readmissions, and median length of stay (4 vs 3 days, P<0.001) was longer as well. Length of stay (LOS) was divided into quartiles for the purpose Divisum stent placement through minor papilla 87% 74.4% 0.2 of logistic regression analysis. Important risk factors for readmissions included LOS of 6 days or more (OR=1.52, P<0.001), discharge to a facility (OR=1.7, P<0.001), being a Medicare (OR=1.65, P<0.001) or Medicaid beneficiary (OR=1.69,P <0.001), higher comorbidity burden as per Grouped Charlson index (OR= 1.68, P<0.001), admission to a teaching hospital (OR=1.09, P=0.001), and drug abuse (add OR=1.31, P<0.001).39.1% of all 30-day readmissions were due to another episode of acute pancreatitis. [92_A] Fukuoka (2012) and AGA (2015) Management Guideline Criteria Mean cost of a 30-day readmission was $11824. Conclusion: All cause 30-day readmission rate for acute pancreatitis is 14.8%. Longer LOS, facility dis- Fukuoka Positive (FP) criteria for Surgery Any one positive feature: charge, Medicare or Medicaid insurance, high comorbidity burden, and drug abuse were important risk factors for readmissions. Jaundice

Main duct dilation ≥1cm Solid component 94 MCN Lung Metastasis in Pancreatic Cancer: Should Staging Chest CT be Routinely Performed? Size ≥3cm Shailendra Singh, MD, Bharat Rao, MD, Shifa Umar, MD, Manish Dhawan, MD, Abhijit Kulkarni, MD, Abrupt change in duct caliber Marcia Mitre, MD, Suzanne Morrissey, MD, FACG, Katie Farah, MD, Shyam Thakkar, MD. Allegheny Health Network, Pittsburgh, PA Pancreatitis

Suspicious cytology Introduction: National Comprehensive Cancer Network (NCCN) guidelines recommend chest x-ray AGA Positive (AP) criteria for Surgery Any one positive feature: or chest computed tomography (CT) for the staging of potential resectable pancreatic adenocarcinoma (PDA). However, there is limited data supporting these guidelines and the prevalence of lung metastasis At least two of the following: main duct dilation, Size ≥3cm or is not well defined on staging CT scans. We report our findings of patients with lung metastasis during solid component initial staging and follow-up of patients with PDA. Methods: Data was prospectively collected from May 2013 to September 2016 for PDA patients who Suspicious cytology were presented at a multidisciplinary pancreas conference (MDPC) at a large tertiary care center. All

[92_B] Risk of Surgery and Malignancy in Cysts Reclassified by Management Guidelines and IMP

Risk of Surgery Risk of Cancer

N % Surgery Events Probability of Surgery- Relative Hazard Ratio % Malignant Outcome Probability of Cancer-free Relative Hazard Ratio free Survival (95% CI) P-value Events survival (95% CI) P-value

All patients 491 42% 58% N/A 13% 85% N/A

Fukuoka Negative (FN)1 274 25% 75% 0.5 (0.4-0.7) P<0.001 4% 96% 0.2 (0.1-0.5) P<0.001

IMP Low2 257 21% 79% 0.8 (0.6-1.2) P=0.293 1% 99% 0.3 (0.1-1.1) P=0.077

IMP High2 17 82% 18% 5.6 (3.1-10.0) P<0.001 41% 42% 18.8 (7.0-50.2) P<0.001

AGA Negative (AN)1 469 39% 61% 9.0 (0.7-1.1) P=0.373 10% 89% 0.7 (0.5-1.0) P=0.074

IMP Low3 395 31% 69% 0.7 (0.6-0.9) P=0.008 3% 97% 0.3 (0.1-0.5) P<0.001

IMP High3 74 85% 15% 3.3 (2.5-4.4) P<0.001 46% 36% 7.4 (4.7-11.6) P<0.001

Fukuoka Positive (FP)1 217 63% 37% 1.8 (1.5-2.3) P<0.001 25% 68% 2.2 (1.6-3.2) P<0.001

IMP Low4 140 49% 51% 0.7 (0.5-0.9) P=0.006 6% 92% 0.2 (0.1-0.4) P<0.001

IMP High4 77 90% 10% 2.0 (1.5-2.7) P<0.001 61% 23% 3.3 (2.2-4.9) P<0.001

AGA Positive (AP)1 22 95% 5% 6.0 (3.8-9.4) P<0.001 91% 6% 15.0 (9.0-25.1) P<0.001

IMP Low5 2 50% 50% 0.2 (0.03-1.6) P=0.137 0% 100% 0.0 (0.0-1.5) P=0.997

I MP High5 20 100% 0% 1.2 (0.7-2.3) P=0.526 100% 0% 1.3 (0.7-2.3) P=0.470

1 Hazard Ratio compared to all patients. 2 Hazard Ratio compared to all FN patients. 3 Hazard Ratio compared to all AN patients. 4 Hazard Ratio compared to all FP patients. 5 Hazard Ratio compared to all AP patients. IMP= integrated molecular pathology. N/A=not applicable.

The American Journal of GASTROENTEROLOGY VOLUME 112 | SUPPLEMENT 1 | OCTOBER 2017 www.nature.com/ajg Abstracts S45

[94] Flow diagram showing prevalence of lung metastasis in pancreatic cancer.

Am J Gastroenterol 2017; 112:S45–S106; doi:10.1038/ajg.2017.296 [94] Comparison of patient and tumor characteristics

Characteristics Patients without Lung Patients with Lung P value metastasis N=265 Metastasis N=13

Age (yrs), mean (S.D) 68.6 64.8 0.22 ACCEPTED: COLON

Male (%) 48.4 69.2 0.14

Race, Caucasian (%) 90.2 100 0.36

Mass size (mm), mean (S.D) 26.9 31.1 0.16 95 Mass Location Risk of Metachronous High-Risk Adenomas and Large (>/= 1 cm) Serrated Polyps in Individuals Head (%) 76.7 46.2 0.01 With Serrated Polyps and Low-Risk Adenomas on Index Colonoscopy: Longitudinal Data From the New Hampshire Colonoscopy Registry Body/Tail (%) 23.3 53.8

CA 19-9, mean (S.D) 899 (1528) 961 (482) 0.90 2017 Category Award (Colon) Joseph C. Anderson, MD, MHCDS1, Amitabh Srivastava, MD2, Christina Robinson, MS3, Julie Weiss, MS1, Christopher Amos, PhD1, Lynn Butterly, MD3. 1Dartmouth College Geisel School of Medicine, Hanover, NH; 2Brigham & Women's Hospital, Boston, MA; 3Dartmouth-Hitchcock Medical Center, Lebanon, NH patients were staged with CT pancreatic protocol, CT chest and Endoscopic Ultrasound. Patients with findings of lung lesions on initial staging chest CT were followed prospectively. Metastatic lung lesions were determined based on definite imaging characteristics with clinical consensus or lung biopsy results. Introduction: There are limited longitudinal data for the risk of metachronous high risk adenomas Results: A total 278 PDA patients referred to MDPC were staged with CT chest (Table 1). Out of these, (HRA) in adults with clinically significant SPs on index colonoscopy. Clinically significant SPs may 36 (12.6%) patients were found to have either malignant (N= 6) or indeterminate (N=30) lung lesions on include large (>/= 1 cm) SPs, sessile serrated polyps (SSP) or traditional serrated adenomas (TSA). initial staging CT chest (Figure 1). Out of the 6 malignant lung lesions, 5 (1.8%) patients had metastatic One important clinical question is whether the presence of SPs in adults with LRA increase the risk PDA lesions and 1 (0.35%) patient had incidental primary lung cancer. On a follow-up of 30 patients for metachronous HRA versus having LRA only. Existing data are limited by sample size. We used the with indeterminate lung lesions, 8 patients (26.7 %) were later determined to be lung metastasis. The population based New Hampshire Colonoscopy Registry (NHCR) to examine the risk of metachronous overall prevalence of definite lung metastasis was at least 4.8% (13/278). The prevalence of lung metas- large (>/= 1 cm) SPs and HRA associated with serrated polyps (SPs) on index colonoscopy as character- tasis in pancreatic head cancer was 3.0 %, while body and tail masses was 10.5 %. Lung metastasis was ized by SP size or histology. almost 4 times more likely in the body and tail masses (OR=3.83, CI 1.2-11.8, P=0.02) compared to Methods: We used 2 separate schemes to stratify adults by index SPs: 1) SIZE (SPs >/= 1cm vs<1cm) head. Overall CT chest resulted in change in management plan in 9 (2.9%) patients due to change in AND 2) HISTOLOGY (SSP or TSA vs hyperplastic polyps (HP)). We further stratified SP groups by stage to metastatic (8) and diagnosis primary lung cancer (1). Staging with CT chest changed otherwise presence or absence of synchronous HRA (adenoma >/= 1cm, villous, high grade dysplasia, multiple (> resectable disease to unresectable/metastatic in 5 patients (1.8%) and borderline resectable to metastatic 2), cancer). Reference group had normal exams. We used logistic regression to adjust for risk factors (age, disease in 2 (0.7 %) patients. Prevalence of isolated PDA lung metastasis without any other metastasis sex, BMI, smoking, follow up time) and calculate 2 metachronous risks: 1) HRA; 2) Large SPs >/=1cm). was 2.8 % (8/278). Results: We excluded adults with index HRA leaving 4,616 adults (median age 61 yrs; 49.7% men) with 2 Conclusion: Our study showed that the prevalence of pulmonary metastasis in PDA was clinically rel- colonoscopies (median time to surveillance: 4.9 yrs)(6/2004-6/2015). Overall, the risk for metachronous evant to mandate routine staging with CT chest. Prevalence was significantly higher for pancreatic body HRA was 6.3% and large SP was 1.2%. We adjusted for patient age, sex, smoking, BMI and time between and tail cancer compared to the head. Staging CT chest resulted in change in the staging of PDA and the 2 exams. Index LRA were associated with an increased metachronous HRA risk (see Table). There was management decisions.