FREE ORALS (FP01-FP31)

FP01 e Free papers 1 (mini oral) - Biliary: Introduction: The American Joint Committee on Cancer (AJCC), 8th edition, has revised the T system for distal Cholangiocarcinoma cholangiocarcinoma (DCC) from layer-based to depth- FP01-01 based approach. The aim of this study was to propose an NATIONWIDE TREATMENT, OUTCOME optimal T classification using a measured depth in resect- AND PREDICTORS FOR SURVIVAL IN able DCC. DISTAL CHOLANGIOCARCINOMA Method: Patients who underwent pancreatoduodenectomy for DCC at 32 hospitals between 2001 and 2010 were M. Strijker1, A. Belkouz1, L. G. van der Geest2, included. The distance between the basal lamina and the O. R. Busch1, T. M. van Gulik1, J. Verheij1, advancing cancer cells was measured as depth of invasion J. W. Wilmink1, B. Groot Koerkamp3, H. J. Klümpen1, (DOI). The invasive cancer foci was measured as invasive M. G. Besselink1 and Dutch Pancreatic Cancer Group tumor thickness (ITT). Log-rank c2 was used to determine 1Cancer Center Amsterdam, Academic Medical Center, the cutoff points, and C-index was used to assess the sur- University of Amsterdam, 2Netherlands Comprehensive vival discrimination of each T system. Cancer Organisation (IKNL), and 3Erasmus University Results: Some 404 patients were included. DOI and ITT Medical Center, The Netherlands were measurable in 182 and all patients, and the medians Background: Distal cholangiocarcinoma has a poor were 2.3 and 5.6 mm, respectively. They both showed a prognosis. Published cohorts focusing on distal chol- strong positive correlation (correlation coefficient, 0.854), angiocarcinoma, especially from Western countries, are and the cutoff points of ITT were 1, 5, and 10 mm. The lacking. This study investigated treatment, outcome and median survival time was significantly shortened with in- predictors for survival in a nationwide cohort of patients crease in ITT: 12.4 years for ITT< 1 mm, 5.2 years for with distal cholangiocarcinoma. 1ITT< 5 mm, 3.0 years for 5 ITT< 10 mm, and 1.5 Methods: A population-based cohort derived from the years for ITT10 mm . This 4-tier ITT classification Netherlands Cancer Registry (NCR) was studied. Patients exhibited the most favorable prognostic discrimination, fi with pathologically con rmed distal cholangiocarcinoma, compared to the T systems of AJCC 7th/8th editions (C- resected (2005-2015) and non-resected (2009-2015), were index: 0.6463, 0.6218, and 0.6243). included. Survival was analyzed using Kaplan Meier Conclusions: ITT is a versatile approach for depth methods and multivariable Cox analysis using backward assessment in DCC. The 4-tier ITT classification with selection. cutoff points of 1, 5, and 10 mm can be used as an optimal T fi Results: A total of 794 patients was identi ed; 513 resected system. patients and 321 non-resected patients. Of the resected patients (7.8%) received (neo-)adjuvant treatment. Of 321 non-resected patients, 147 (45.8%) presented with M1 disease and 63 (19.6%) received palliative chemotherapy. FP01-03 Median overall survival for resected, non-resected M0, and EVALUATION OF THE NEW AMERICAN non-resected M1 disease was 23 months (95% CI 21-26), 6 JOINT COMMITTEE ON CANCER months (95% CI 5-8) and 4 months (95% CI 4-5) (p< STAGING MANUAL 8TH EDITION FOR 0.001), respectively. In multivariable analysis, T3/T4 stage PERIHILAR CHOLANGIOCARCINOMA < (p=0.006), higher lymph node ratio (p 0.001), poor dif- 1 1 1 1 ferentiation (p=0.001) and R1 resection (p=0.002) were M. Gaspersz , S. Buettner , J. van Vugt , J. de Jonge , W. Polak1, M. Doukas2, J. Ijzermans1, negative prognostic factors in resected patients. In non- 1 3 resected patients, increasing age (p=0.007), lymph node B. Groot Koerkamp and F. Willemssen 1Surgery, Erasmus MC, University Medical Center metastases (p=0.013), distant metastases (p< 0.001), no Rotterdam, 2Pathology, Erasmus MC, University Medical surgical exploration (p=0.011) and no palliative chemo- Center Rotterdam, and 3Radiology and Nuclear Medicine, therapy (p< 0.001) predicted worse survival. Erasmus MC, University Medical Center Rotterdam, The Discussion: This nationwide, Western study includes the Netherlands largest population with all stages of distal chol- th angiocarcinoma. The study identified predictors for sur- Introduction: To evaluate the 8 edition of the American vival in both the resected and non-resected population Joint Committee on Cancer (AJCC) staging system for which can be useful to stratify future trials with (neo-) perihilar cholangiocarcinoma (PHC), and to compare the th th adjuvant or palliative treatment. prognostic value of the 7 and 8 editions of the AJCC staging system for PHC. Methods: All patients with PHC between 2002-2014 FP01-02 were included. Imaging at the time of presentation was reassessed and AJCC Tumor-Nodal-Metastasis (TNM) PROPOSAL OF AN OPTIMAL DEPTH- stage was determined according to the 7th and 8th edition BASED TUMOR CLASSIFICATION IN of the AJCC staging system. Overall survival (OS) was DISTAL CHOLANGIOCARCINOMA estimated using the Kaplan-Meier method. Comparison H. Aoyama, T. Ebata, Y. Yokoyama, T. Igami, T. Mizuno, of the prognostic accuracy of staging systems was J. Yamaguchi, S. Onoe and M. Nagino performed using the concordance index (c-index) and Nagoya University Graduate School of Medicine, Japan Brier-score.

HPB 2018, 20 (S2), S182eS294 Free Orals (FP01-FP31) S183

Results: A total of 248 PHC patients were included. were as good as the prognosis of ICC patients underwent Median OS of the entire cohort was 9.7 months (8.0-11.5). upfront R0 surgery. Prognostic accuracy was comparable between the 7th and Conclusion: If R0 resection is expected, aggressive sur- 8th AJCC staging systems (c-index 0.59 vs 0.58). Prog- gical resection with regional lymph nodes dissection is nostic accuracy of the 8th edition was higher in the sub- recommended in advanced ICC patients with LNM. group of resected patient (0.67), but expanding the staging However, downsizing chemotherapy is useful for patients system with sub-stages lowered the c-index to 0.62. The with initially unresectable locally advanced ICC for prognostic accuracy of the AJCC staging system in unre- improving prognosis. sectable patients was poor with a c-index of 0.57 in both the normal and expanded staging system. th th Conclusions: The 7 and 8 editions of the AJCC staging FP01-05 system for PHC have comparable prognostic accuracy. A COMPARISON OF TREATMENT AND Although developed to be used for resectable and unre- sectable patients, the discriminative performance in unre- OUTCOMES OF PERIHILAR sectable patients was poor. Therefore, the AJCC staging CHOLANGIOCARCINOMA IN AN system may need modifications to improve its prognostic EASTERN AND WESTERN CENTER accuracy in patients with unresectable PHC, compromising P. Olthof1,2, M. Miyasaka3, W. Jarnagin4, the majority of patients. T. van Gulik1 and T. Noij3 1Academic Medical Center Amsterdam, 2Syrgeru, Reinier de Graaf Gasthuis, The Netherlands, 3Hokkaido Univer- FP01-04 sity Graduate School of Medicine, Japan, and 4Memorial MULTIDISCIPLINARY TREATMENT Sloan Kettering Cancer Center, United States COMBINED WITH NEOADJUVANT Background: Perihilar cholangiocarcinoma (PHC) is a rare DOWNSIZING CHEMOTHERAPY AND tumor that requires multimodal multidisciplinary treatment AGGRESSIVE REGIONAL LYMPH and is associated with substantial morbidity and mortality. Treatment strategies and outcomes vary greatly between NODE DISSECTION TO ACHIEVE R0 the eastern and western parts of the world. This study aimed RESECTION IMPROVES PROGNOSIS IN to compare a major eastern and western PHC cohort in PATIENTS WITH ADVANCED terms of patient characteristics, treatment strategies and INTRAHEPATIC outcomes including a propensity score matched analysis. CHOLANGIOCARCINOMA Methods: All consecutive patients who underwent com- bined biliary and liver resection for PHC between 2005 and S. Kuboki, H. Yoshitomi, K. Furukawa, T. Takayashiki, 2016 at two western and one eastern center were included. S. Takano, M. Miyazaki and M. Ohtsuka A propensity score matched analysis based on baseline Department of General Surgery, Graduate School of characteristics to compare perioperative outcomes and a Medicine, Chiba University, Japan multivariable analysis for survival were performed. Background: R0 resection is the only chance for long-term Results: A total of 210 western patients were compared to survival in patients with intrahepatic cholangiocarcinoma 164 eastern patients. After propensity score matching (1:1), (ICC); however, the prognosis of ICC with lymph node the two cohorts both including 79 patients had similar rates metastasis (LNM) is poor and sometimes difficult to of liver failure and morbidity, however biliary leakage achieve R0 resection. (33% versus 17%, p=0.03) and morality rates (18% versus Aim: To evaluate the usefulness of neoadjuvant downsiz- 5%) were higher in the west. Despite matching based on ing chemotherapy and aggressive regional lymph node liver volume, portal vein embolization was more frequent dissection to achieve R0 resection for improving prognosis. in the east (58% versus 11%, p< 0.01). At multivariable Methods: 135 ICC patients underwent R0/R1 resection analysis, western patients had inferior survival compared to were enrolled. For control, 28 locally advanced ICC pa- the east (hazard-ratio 1.72 (1-23-2.40) P < 0.01) corrected tients receiving chemotherapy only (gemcitabine, cispla- for age, ASA score, tumor stage and margin. tine, or S-1) were selected. Discussion: There are major differences in patient charac- Results: Larger tumor size, positive vascular/lymphatic teristics, treatment strategies, perioperative outcomes and invasion, positive intrahepatic metastasis, positive LNM, survival between eastern and western PHC cohorts. These and R1 resection were significant risk factors for poorer results indicate that PHC research should not be translated prognosis. R1 resection was the strongest prognostic across the world before cross-cardinal external validation factor. When prognostic factors in R0 cases were further evaluated, positive LNM was an independent risk factor for poorer prognosis. However, even though LNM FP01-06 existed, aggressive R0 resection improved prognosis. To e achieve R0 resection, we aggressively performed COMPARISON ANALYSIS OF LEFT neoadjuvant downsizing chemotherapy for initially VERSUS RIGHT-SIDED RESECTION IN unresectable locally advanced ICC with massive LNM. BISMUTH TYPE III HILAR Of 41 patients with initially unresectable ICC, conver- CHOLANGIOCARCINOMA sion surgery was performed in 13 patients (31.7%) after Y. J. Lee, D. W. Choi, S. H. Choi, J. S. Heo, I. W. Han and chemotherapy (median: 4.4 months), and R0 resection S. Han was achieved in 11 of 13 cases (84.6%) underwent Department of Surgery, Samsung Medical Center, Re- conversion surgery, and the prognosis of these patients public of Korea

HPB 2018, 20 (S2), S182eS294 S184 Free Orals (FP01-FP31)

Introduction: Several studies reported worse prognosis cholangiocarcinoma (IHCC), and 34.7% (n=8) of patients a after left-sided liver resection than right in patients with perihilar cholangiocarcinoma (PHCC). Conversely, in the perihilar cholangiocarcinoma(CCC). We compared out- KRAS/TP53 group, the incidence of patients with GB, comes of left- and right-sided resections in Bismuth type III IHCC, and PHCC was 18.4% (n=9), 20.4% (n=10), and hilar cholangiocarcinoma and analyzed factors affecting 61.2% (n=30), respectively (p=0.003). Patients in PBRM1/ survival. IDH1/BAP1 group had a 5-year OS of 40% compared with Method: From May 1995 to December 2011, a total of 179 13% for KRAS/TP53 group (p=0.01). In the multivariable patients underwent surgery for type III hilar CCC in model, margin status (R1 vs. R0, HR 2.57, p=0.008), tumor Samsung Medical Center. Among them, 138 patients grade (poorly/undifferentiated vs. well/moderate, HR 2.54, received hepatectomies for adenocarcinoma with curative p=0.009), microvascular invasion (present vs. absent, HR intent. 103 patients underwent right-sided resections (IIIa 1.93, p=0.095), and lymph-node status (N1 vs. N0, HR group), whereas 35 patients underwent left-sided resections 2.29, p=0.027) were strongly associated with patents’ (IIIb group). Perioperative demographics, morbidity, mor- prognosis. Moreover, patients in KRAS/TP53 group had a tality, and overall and disease-free survival rates were 2.5-fold increased risk of death compared with patients in compared between the groups. BRM1/IDH1/BAP1 group (HR 2.56, p=0.010). The model Results: BMI was higher in IIIa group (24Æ2.6kg/m2 vs demonstrated a good to strong ability to predict patients’ 22.7Æ2.8kg/m2; p=0.012). Preoperative portal vein prognosis (c-index:0.743). embolization was done in 23.3% of patients in IIIa group, Conclusions: Pattern of gene mutations might increase the whereas none in IIIb group. The R0 rate was 82.5% in ability to predict the prognosis of patients undergoing the IIIa group and 85.7% in the IIIb group (p=0.796). 3a surgery for BDC. complication of Clavien-Dindo classification showed significant difference in between the two groups (10.7% in the IIIa group versus 23.3% in the IIIb group; p=0.002). The 5-year overall survival rate was 33% in the IIIa group and 35% in the IIIb group (p=0.983). The 5-year disease-free survival rate was 28% in the IIIa group and 29% in the IIIb group (p=0.706). Advanced T stage 3 and 4 and LN metastasis were independent prognostic factors for survival and recurrence by multi- variate analysis. Conclusions: There was no significant difference in out- comes according to lesion side in patients with Bismuth type III hilar CCC after curative surgery. Figure

FP01-07 PATTERNS OF GENE MUTATIONS IN FP01-08 BILE DUCT CANCERS: IS IT TIME TO ENBD IS ASSOCIATED WITH OVERCOME THE ANATOMICAL DECREASED TUMOR DISSEMINATION CLASSIFICATION? COMPARED TO PTBD IN PERIHILAR F. Bagante1,2, A. Ruzzenente1, S. Conci1, P. Capelli3, CHOLANGIOCARCINOMA T. Campagnaro1, T. Pawlik2, C. Luchini3, C. Iacono1, R. Higuchi1, T. Yazawa1, S. Uemura1, W. Izumo1, A. Scarpa3,4 and A. Guglielmi1 T. Furukawa2 and M. Yamamoto1 1Department of Surgery, University of Verona, Italy, 1Department of Surgery, Tokyo Woman’s Medical Uni- 2Department of Surgery, Ohio State University Wexner versity, Institute of Gastroenterology, and 2Histopatholo- Medical Center, United States, 3Department of Di- gy, Tohoku University Graduate School of Medicine, agnostics and Public Health, Section of Anatomic Pa- Japan thology, University of Verona, and 4ARC-Net Research Center, University of Verona, Italy Background: Little is known regarding the risk of tumor dissemination when percutaneous biliary drainage (PTBD) Background: Two studies based on multi-omics data is used before surgery of perihilar cholangiocarcinoma analyses have identified distinct subtypes of bile-duct (PHC). The aim of this study was to compare the incidence cancers (BDC) that might have important implications in of tumor dissemination following preoperative endoscopic terms of disease classification, patients’ selection and biliary drainage (ENBD) and PTBD for PHC. treatment. Methods: Data from 208 consecutive patients who un- Methods: Patients with mutations in KRAS, NRAS, TP53, derwent PHC resection between 2000 and 2013 were and ARID1A genes were in KRAS/TP53 group while pa- retrospectively analyzed. The influence of drainage type on tients with mutations in IDH1/2, BAP1, and PBRM1 were incidence of tumor dissemination was examined. Seventy- in BRM1/IDH1/BAP1 group. six patients underwent ENBD (37%), 87 underwent PTBD Results: Among the 72 patients included in the study, 23 (42%), and 45 underwent surgery without preoperative (31.9%) were in the BRM1/IDH1/BAP1 group and 49 biliary drainage (WD, 22%). (68.1%) in the KRAS/TP53 group. In the BRM1/IDH1/ Results: The estimated cumulative incidence rate of tumor BAP1 group, 4.3% (n=1) of patients had a gallbladder dissemination was significantly lower in the ENBD group cancer (GB), 60.9% (n=14) an intrahepatic than in the PTBD group (14.6% vs. 35.9% at 5 years, p =

HPB 2018, 20 (S2), S182eS294 Free Orals (FP01-FP31) S185

.024) and equivalent to that in the WD group (15.9% at 5 FP01-11 years, p = 0.84). Univariate and multivariate analyses of the risk factors for tumor dissemination showed that the PTBD SURGICAL PROCEDURE AND RESULTS was an independent risk factor for tumor dissemination OF CONCOMITANT VASCULAR (hazard ratio [HR] vs. ENBD, 2.80). The 2- and 5-year RESECTION FOR PERIHILAR AND disease-specific survival rates were higher in the ENBD DISTAL CHOLANGIOCARCINOMA group (67.6/47.3%) than in the PTBD group (56.6/27.8%, T. Noji, K. Okamura, K. Tanaka, Y. Nakanishi, T. Asano, p = 0.032), and equivalent to that in the WD group (64.9/ T. Nakamura, T. Tsuchikawa and S. Hirano 53.8%, p = NS). However, drainage type was not an in- Gastroenterological Surgery II, Hokakido University dependent risk factor in multivariate analysis of disease- Faculty of Medicine, Japan specific survival. Conclusion: The associated risk of postoperative tumor Background: Concomitant major vascular resection was dissemination for PHC in the ENBD group was lower required to achieve curative resection (R0 resection) for fi than in the PTBD group and equivalent to that in the WD cholangiocarcinoma; however, it is also dif cult to go group. together R0 resection and surgical safety. Purpose: To evaluate short- and long-term surgical results for patients with cholangiocarcinoma who underwent concomitant vascular resection (hepatic artery and portal FP01-09 vein). PREOPERATIVE BILIARY DRAINAGE Patients: From January 2000 to October 2017, we IN PERIHILAR performed 392 surgeries on patients with chol- CHOLANGIOCARCINOMA: RESULTS angiocarcinoma. Among these patients, 157 underwent OF AN ITALIAN MULTICENTER STUDY concomitant vascular resection and were evaluated. 1 2 1 3 Results: There were 112 male patients. The median age F. Ardito , L. Aldrighetti , F. Giuliante , A. Ferrero , was 68 years (range, 44-82 years). Hepatectomy (Hx), PD, A. Guglielmi4, A. D. Pinna5, L. De Carlis6, U. Cillo7, 8 9 and HPD were performed in 112, 18, and 27 patients, E. Jovine , N. Portolani and Italian Chapter of the respectively. Hepatic artery resection (AR), portal vein International Hepato-Pancreato-Biliary Association resection (PVR), and both portal vein and hepatic artery 1Catholic University, Rome, 2San Raffaele Hospital, 3 4 5 resection (APR) were performed in 19, 112, and 29 pa- Mauriziano Hospital, University of Verona, University tients, respectively. The operation time was 696 min (range, ’ 6 of Bologna Sant Orsola-Malpighi Hospital, Niguarda 385-1250 min). The volume of operative bleeding was Ca’ Granda Hospital, Milan, 7Policlinico Universitario, 8 9 1805 ml (range, 390-27,860 ml). The morbidity and mor- Padua, Maggiore Hospital, Bologna, and University of tality rates were 54% and 8.3%, respectively. We achieved Brescia, Italy 83.4% of R0 resection, and the 5-year survival rate was Introduction: Endoscopic biliary drainage (EBD) or 33.1%. The 5-year survival rate based on surgical proced- percutaneous transhepatic biliary drainage (PTBD) are ure did not show significant differences (p=0.199; Hx, recommended to improve liver function before major 37.7%; PD, 26.7%; HPD, 17.1%, respectively). The 5-year hepatectomy for perihilar cholangiocarcinoma (PHC). survival rate based on vascular resection site also did not However infectious drainage-related complications are show significant differences (p=0.158; PVR, 34.1%; AR, associated with both methods. Aim of this study was to 15%; APR, 37.4%, respectively). evaluate the operative risk associated with EBD and PTBD. Conclusion: Our surgical results of concomitant vascular Method: Retrospective multicenter study including 14 resection for cholangiocarcinoma under our guidelines Italian hepatobiliary centers, members of the Italian were thought to be acceptable. However, a new strategy for Chapter of the International Hepato-Pancreato-Biliary-As- decreasing morbidity and mortality is required. sociation. Patients undergoing associated major hepatec- tomy for PHC between 2000 and 2016 were included. Results: Data were available on 639 patients. Preoperative FP01-12 biliary drainage was performed in 69% of patients (range 38%-93%). PTBD was used in 51% of patients. Morbidity LYMPHOVASCULAR INVASION AND rate after biliary drainage was 35.8% and it was not SURGICAL COMPLICATIONS PREDICT significantly different between the two methods (27.1% CLINICAL OUTCOME IN PATIENTS after PTBD and 26.7% after EBD; p=ns). Infectious com- WITH PERIHILAR AND INTRAHEPATIC plications were the most frequent cause of morbidity fi CHOLANGIOCARCINOMA (64.5%). Morbidity was signi cantly higher in Bismuth 1 1 1 type 1-2 than in Bismuth type 3-4 (22.6% vs. 38.6%; J. Bednarsch , Z. Czigany , I. Amygdalos , D. Morales Santana1, F. Meister1, J. Böcker1, p=0.01). In Bismuth type 3-4, morbidity rate was not 1 2 1 fi T. F. Ulmer , C. H. C. Dejong , U. P. Neumann and signi cantly different according to the type of drainage 1 (37.3% after PTBD vs. 44.4% after EBD; p=ns). Unilateral G. Lurje 1Department of Surgery and Transplantation, University drainage of the future remnant liver was the preferred 2 method of drainage (78.7%) and its failure rate was 14.8%. Hospital Aachen, Germany, and Maastricht University Conclusions: Morbidity rate following biliary drainage Medical Center (MUMC), The Netherlands was not correlated with the type of drainage but with the Background: Cholangiocarcinoma (CCC) is a relatively complexity of biliary stricture. Unilateral biliary drainage rare malignancy that is typically diagnosed at an advanced of the future remnant liver was associated with low risk of disease stage. Major liver resection with portal vein failure. reconstruction has evolved as the mainstay of treatment for

HPB 2018, 20 (S2), S182eS294 S186 Free Orals (FP01-FP31) patients with perihilar (PHCC) and intrahepatic chol- success was achieved in 102 of 331 patients (30.8%). No angiocarcinoma (IHCC). Despite recent advancements, the resection was performed in 176 patients (53.2%), mainly overall- (OS) and recurrence-free survival (RFS) in CCC because of occult metastatic or locally advanced disease. remains lower than for most other solid tumors. Here we Of the 155 patients (46.8%) who underwent resection, 38 aimed to identify prognostic markers of clinical outcome in patients (24.5%) had a R1 resection. Of the remaining CCC-patients that underwent surgical resection in curative 117 patients (35.3%), 15 (12.8%) died within 90-days. intent. Independent poor prognostic factors for surgical success Methods: Between 2010 and 2016, 162 patients with CCC were higher age, preoperative cholangitis, involvement of (PHCC: n=91, IHCC; n=71) underwent surgery in curative the hepatic artery on imaging, suspicious lymph nodes, intent at our institution. Preoperative characteristics, peri- and Blumgart stage 3. A preoperative prognostic model operative data and oncological follow-up were obtained based on these factors was developed with a concor- from a prospectively managed institutional database. The dance-index of 0.71. External validation showed good associations of RFS and OS with clinico-pathological concordance (0.70). characteristics were assessed using univariate and multi- Conclusions: Surgical success (R0 resection without 90- variate survival analyses. day mortality) was achieved in only 30% of PHC patients Results: The median OS and RFS were 38 and 36 months undergoing exploratory laparotomy. The preoperative for PHCC and 25 and 13 months for IHCC, respectively. model for surgical success may be used in shared decision Lymphovascular invasion (LVI) as well as surgical com- making. plications as assessed by the comprehensive complication index were independently associated with OS for the PHCC (LVI; Exp(B)=2.28, p=0.042; CCI; Exp(B)=1.04, p< 0.001) and IHCC cohorts (LVI, Exp(B)=5.08, p=0.028; CCI, Exp(B)=1.04, p=0.002), respectively. No other clin- ical variable including R0-status and Bismuth classification was associated with OS. Conclusions: Surgical resections for CCC are safe in experienced high-volume liver centers. Tumor and patient characteristics were not associated with clinical outcome. In patients with PHCC and IHCC, LVI and CCI are asso- ciated with OS, suggesting a similar tumor biology. Figure [Preoperative prognostic model for surgical success]

FP01-13 PREOPERATIVE PROGNOSTIC MODEL TO PREDICT SURGICAL SUCCESS IN FP01-14 PATIENTS WITH PERIHILAR IS SEGMENTAL RESECTION CHOLANGIOCARCINOMA SUFFICIENT FOR THE TREATMENT OF EXTRAHEPATIC M. Gaspersz1, S. Buettner1, E. Roos2,M.D’Angelica3, R. DeMatteo4, F. Willemssen5, T. van Gulik2, CHOLANGIOCARCINOMA? W. Jarnagin3, J. Ijzermans1, B. Groot Koerkamp1 and COMPARISON WITH Perihilar Cholangiocarcinoma Study Group PANCREATICODUODENECTOMY 1Surgery, Erasmus MC, University Medical Center USING A PROPENSITY-SCORE 2 Rotterdam, Surgery, Academic Medical Center, Amster- MATCHING ANALYSIS dam, 3Surgery, Memorial Sloan Kettering Cancer Center, United States, 4Surgery, University of Pennsylvania, and S. Han, S. H. Choi, D. W. Choi, J. S. Heo, I. W. Han, 5Radiology and Nuclear Medicine, Erasmus MC - Uni- Y. H. You, D. Park and Y. Ryu versity Medical Center Rotterdam, The Netherlands Department of Surgery, Samsung Medical Center, Re- public of Korea Introduction: Patients with resectable PHC on imaging have a substantial risk of occult metastatic or locally advanced Background: The long-term outcomes following disease, incomplete (R1) resection, and 90-day mortality. We segmental resection (SR) and pancreaticoduodenectomy aimed to develop and validate a preoperative prognostic model (PD) for extrahepatic cholangiocarcinoma (EHCCC) have to predict surgical success, defined as complete (R0) resection shown discrepancies in previous studies. We conducted this without 90-day mortality, in patients with resectable perihilar study to compare the outcomes of EHCCC after curative cholangiocarcinoma (PHC) on imaging. resection using a propensity-score matching analysis. Methods: PHC patients who underwent exploratory lapa- Method: From November 1994 to December 2012, a total rotomy in three tertiary referral centers were identified. of 650 patients underwent SR or PD without liver resection Multivariable logistic regression analysis was performed to for EHCCC in our institution. Among them, 403 patients identify preoperative prognostic factors for surgical suc- achieved R0 resection and were divided into SR (n=115) cess. A prognostic model was developed using data from and PD (n=387) groups. One-to-one propensity score two European centers, and validated in one American matching was performed to minimize the effects of po- center. tential confounders. Results: In total, 671 PHC patients underwent explor- Results: 82 patients from each group were matched. Age, atory laparotomy. In the derivation cohort, surgical sex, ASA score, and pathologic confounding factors

HPB 2018, 20 (S2), S182eS294 Free Orals (FP01-FP31) S187 including T and N stages were balanced between the two FP01-16 groups. However, tumor size was larger (22.9Æ7.5mm versus 26.8Æ12.4mm; p=0.018) and number of harvested PROPOSAL ON THE MINIMAL NUMBER lymph nodes was higher (13.2Æ7.5 versus 21.9Æ11.3; p< OF RETRIEVED LYMPH NODES FOR 0.001) in the PD group. The 5-year overall survival rates of ACCURATE STAGING OF the SR and PD groups were 41.5% and 52.6%, respectively EXTRAHEPATIC (p=0.023) and the 5-year disease-free survival rates of the CHOLANGIOCARCINOMA AND SR and PD groups were 37.5% and 57.6% (p=0.047). The pattern of recurrence was not different while the number of CLINICAL VALIDATION OF AJCC 8TH recurred patients was higher in the SR group (p=0.042). N STAGE Lymph node metastasis was an independent prognostic J. S. Kang1, R. Higuchi2,J.He3, M. Yamamoto2, factor, whereas the operative procedure itself was not. C. L. Wolfgang3, J. L. Cameron3, Y. Han1, W. Kwon1, Conclusion: PD is recommended in EHCCC for better S. -W. Kim1 and J. -Y. Jang1 oncologic outcomes. And in patients underwent SR due 1Surgery and Cancer Research Institute, Seoul National to poor general performance, more aggressive nodal University College of Medicine, Republic of Korea, 2Sur- dissection is required for accurate staging and improved gery, Institute of Gastroenterology, Tokyo Women’s outcome. Medical University, Japan, and 3Surgery, Johns Hopkins Hospital, United States Background: The three-tier N staging system of the 8th FP01-15 edition of the American Joint Committee on Cancer EXTENDED LEFT VERSUS EXTENDED (AJCC) for extrahepatic bile duct (EBD) cancer is yet to be RIGHT HEPATECTOMY WITH HILAR validated. The minimal number of retrieved lymph nodes (MNRLN) for accurate staging remains unclear. The pre- EN-BLOC RESECTION IN PERIHILAR sent study aimed to propose the MNRLN, to validate the CHOLANGIOCARCINOMA three-tier N staging system, and compare the prognostic J. Bednarsch1, Z. Czigany1, I. Amygdalos1, predictability of 8th N staging with that of the 7th. D. M. Santana1, F. Meister1, J. Böcker1, T. F. Ulmer1, Methods: Between 1991 and 2015, patients with patho- C. H. C. Dejong2, U. P. Neumann1 and G. Lurje1 logically confirmed EBD adenocarcinoma who underwent 1Department of Surgery and Transplantation, University curative-intent pancreatoduodenectomy in Seoul National Hospital Aachen, Germany, and 2Department of Surgery, University Hospital, Tokyo Women’s Medical University, Maastricht University Medical Center (MUMC), The and Johns Hopkins Medical Institute were enrolled in this Netherlands study. The MNRLN was calculated via log-rank test based Background: Hilar en-bloc resection with portal vein on cut-off values. The concordance index (C-index) was resection (PVR) has emerged as the mainstay of treatment utilized to compare the discrimination capability of the two- for patients with perihilar cholangiocarcinoma (PHCC). and three-tier N staging. Whether liver resection should be carried out as extended Results: A total of 789 consecutive patients were enrolled. fi left- (LH) or right-sided hepatectomy (RH) is still subject of The survival rate was signi cantly different between node- ongoing debate. Here we evaluated perioperative compli- negative and -positive patients in whom at least 10 lymph fi cations and oncological outcome after RH or LH with hilar nodes were retrieved. The ve-year overall survival rates fi en-bloc resection and PVR in patients with PHCC. between each 8th N stage was signi cantly different (N0 vs < < Methods: Between 2010 and 2016, 91 patients with PHCC N1, P 0.037; N1 vs. N2, P 0.003). The C-index of the underwent surgery in curative intent at our institution. 8th N stage was higher than that of the 7th (0.590 vs. 0.575). fi Perioperative and survival data from all patients undergo- Conclusions: For accurate strati cation, at least 10 lymph ing surgical resection for PHCC were analyzed. PVR was nodes should be retrieved. The new three-tier system for N carried out in all cases as well as arterial reconstruction staging of EBD cancer is valid and has more accurate (n=5) if necessary. Patients undergoing hepatoduodenec- prognostic predictability than the 7th stage. tomy (n=8) or ALPPS (n=2) were excluded from the analysis. Results: Tumor grading, microvascular invasion, FP01-17 lymphovascular invasion, N-category, T-category, R-status R1 RESECTION IN HILAR and UICC tumor staging were equally distributed among CHOLANGIOCARCINOMA MIGHT NOT the LH (n=36) and RH (n=45) groups. Perioperative morbidity and mortality were higher after RH compared to TRANSLATE TO INFERIOR PATIENT LH (15.6% vs. 8.3%, p=0.003). While 3-year OS was SURVIVAL comparable between LH and RH (55% vs. 48%), we M. Passeri, W. Lyman, R. Kirks, A. Cochran, D. Iannitti, observed a non-significant difference in 5-year OS with J. Martinie, E. Baker and D. Vrochides 18% and 43% for LH and RH respectively (p=0.820, log Carolinas Medical Center, United States rank). Introduction: Hilar cholangiocarcinoma (HC) represents Conclusions: LH and RH hilar en-bloc resections an aggressive tumor embedded in an anatomically sensitive demonstrate comparable 3-year OS. While RH hilar en- area. Surgeons must balance the goal of obtaining negative bloc resection might result in better long-term 5-year sur- margins with the need to preserve enough proximal tissue vival, this may be at cost of an increase in perioperative to permit a feasible bilioenteric reconstruction. This study morbidity and mortality.

HPB 2018, 20 (S2), S182eS294 S188 Free Orals (FP01-FP31) aims to revisit the importance of obtaining an R0 resection suspended in perihilar tumor due to insufficient evidence. The when excising HC in the era of modern adjuvant therapy. aim of the present study was to investigate the prognostic Methods: A retrospective review of a prospectively impact of invasion depth in perihilar cholangiocarcinoma. maintained database was employed to compile a list of Methods: Patients with perihilar cholangiocarcinoma who patients with HC who were taken to the operating room underwent hepatectomy between 2001 and 2014 were with curative intent at our institution from 2008 to 2016. reviewed. Invasion depth was defined as a distance vertically Those cases that resulted in resection were subjected to from the top of the tumor to the deepest invasive cells. After further analysis, with a focus on oncologic outcomes. Pa- providing the optimal cutoff points by recursive partitioning, tients with both R0 and R1 margins underwent adjuvant the stratification ability of the classification by invasion depth chemoradiation. was compared with the T classification in the 8th edition of Results: 100 patients were included in the study. 70 re- the AJCC staging system using concordance index. sections were aborted after diagnostic laparoscopy and Results: The median invasion depth was 6.0 mm (range, liver/portal ultrasonography, either because of peritoneal 0-45 mm). Invasion depth was categorized as follows: D1, metastasis, or local disease extending beyond what was < 1 mm ( n=25, 5.7%); D2, 1to< 5 mm ( n=138, 31%); seen on preoperative imaging. 30 patients underwent and D3, 5 mm ( n=277, 63%). Each progressive category resection. Details of their cases and outcomes are included of invasion depth was associated with decreased survival in Figure 1. There were no cases of R2 resection. 2 year rate (Figure 1). The classification by invasion depth had a overall and progression free survival were similar between higher concordance index than the AJCC T classification R0 and R1 resection groups (p=0.86). system (0.582 vs 0.530). Conclusions: As with the resection of any malignancy, Conclusions: Invasion depth stratified survival of patients principles of oncologic resection should be adhered to with perihilar cholangiocarcinoma, suggesting invasion- when resecting cholangiocarcinoma. Striving for micro- depth based stratification would be a candidate of T scopically negative margins is one of those principles; classification. however, given the technical difficulty of obtaining a proximal margin, it should be considered that R1 resection does not preclude comparable outcomes.

Figure [Overall survival]

FP02 e Free papers 2 (long oral) e Liver: Metastases 1 Figure 1 FP02-02 ALPPS IMPROVES RESECTABILITY COMPARED TO CONVENTIONAL TWO- STAGE HEPATECTOMY IN PATIENTS FP01-18 WITH ADVANCED COLORECTAL INVASION DEPTH IS A CANDIDATE LIVER METASTASIS FOR THE T FACTOR IN PERIHILAR 1 2 3 4 CHOLANGIOCARCINOMA P. Sandström , B. Rösok , E. Sparrelid , G. Lindell , P. Nørgaard Larsen5, N. A. Schultz5, B. A. Björnbeth6, 1 2 1 1 K. Shinohara , Y. Shimoyama , T. Ebata , T. Mizuno and B. Isaksson7, M. Rizell8 and B. Björnsson9 1 M. Nagino 1Dept of Surgery, University Linkoping, Sweden, 2Dept of 1 Surgical Oncology, Nagoya University Graduate School Hepato-Pancreato-Biliary Surgery, Oslo University Hos- 2 of Medicine, and Department of Pathology and Clinical pital, Norway, 3Dept of Clinical Science, Intervention and Laboratories, Nagoya University Hospital, Japan Technology, Division of Surgery, Karolinska Institute, Introduction: The 8th edition of the American Joint Com- 4Dept of Surgery, Lund University, Sweden, 5Dept of mittee on Cancer (AJCC) staging system in distal chol- Surgical Gastroenterology and Transplantation, Univer- angiocarcinoma employed invasion depth for the T sity of Copenhagen, Denmark, 6Dept of Hepato-Pancrea- determinant, whereas the use of invasion depth was to-Biliary Surgery, University Hospital, Norway, 7Dept of

HPB 2018, 20 (S2), S182eS294 Free Orals (FP01-FP31) S189

Surgery, Akademiska University Hospital, 8Dept of database. Cox proportional hazards modeling was Transplantation and Liver Surgery, Sahlgrenska Academy, performed. Only first recurrences were analyzed. University of Gothenburg, and 9Depts of Surgery and Results: A total of 1614 patients were included Clinical and Experimental Medicine, Linköping Univer- (SYS+HAI=708,43.9%; SYS=906,56.1%). Median sity, Sweden follow-up was 106 months, and 1050 patients (65.1%) The aim was to evaluate ALPPS could increase resection developed a recurrence. Patients treated with SYS+HAI rates (RRs) compared with two stage hepatectomy (TSH) in had more advanced disease with a high clinical risk score in a randomized controlled trial (RCT). Radical liver metastasis 50.1% versus 37.1% (p< 0.001). Median OS was 80 resection offers the only chance of a cure for patients with months for SYS+HAI versus 56 months for SYS (p< metastatic colorectal cancer. Patients with colorectal liver 0.001). Five-year RFS for SYS+HAI was 30.9% versus metastasis (CRLM) and an insufficient future liver remnant 24.5% for SYS (p=0.07, Figure 1). A lower rate of initial (FLR) volume are traditionally treated with chemotherapy intrahepatic-only recurrence was found in the SYS+HAI with portal vein embolization or ligation followed by hep- group (17.1% versus 23.1%,p=0.003). No differences were atectomy (TSH). This treatment sometimes fails. A pro- found for simultaneous initial intra- and extrahepatic spective, multicenter RCT was conducted between June recurrence (10.0% versus 12.1%,p=0.18). In multivariable 2014 and August 2016. It included 97 patients with CRLM analysis SYS+HAI was an independent favorable prog- and a standardized FLR (sFLR) of < 30%. Primary nostic factor for OS (HR:0.70,95%CI:0.60-0.80,p< 0.001) outcome, resection rates (RRs) were measured as the per- and RFS (HR:0.76,95%CI:0.67-0.87,p< 0.001). centages of patients completing both stages of the treatment. Conclusion: Patients with resected CLM who received Secondary outcomes were complications, radicality and 90- perioperative SYS+HAI had more advanced disease and a day mortality measured from the final intervention. lower initial intrahepatic-only recurrence rate than patients Baseline characteristics, besides body mass index, did not receiving SYS only. SYS+HAI was an independent differ between the groups. The RR was 92% (C.I. 84- favorable factor for both OS and RFS. 100%)(44/48) in the ALPPS arm compared to 57% (43- 72%)(28/49) in the TSH arm (rate ratio 8.25 [95% CI 2.6- 26.6]; P< 0.0001). No differences in complications (Clavien- Dindo 3a) (43% (19/44) vs. 43% (12/28) (1.01 [95% CI 0.4- 2.6]; P=0.99), 90-day mortality (8.3% (4/48) vs. 6.1% (3/49) (1.39 [95% CI 0.3-6.6]; P=0.68) or R0 RRs (77% (34/44) vs. 57% (16/28) (2.55 [95% CI 0.9-7.1]; P=0.11)) were observed. Of the patients in the TSH arm that failed to reach a sFLR of 30%, 12 were successfully treated with ALPPS. ALPPS is superior to TSH in terms of RR, with comparable surgical margins, complications and short-term mortality. Figure 1. [Recurrence free survival]

FP02-03 FP02-04 RECURRENCE RATE AND PATTERN MORPHOLOGIC AND BIOLOGIC AFTER RESECTION OF COLORECTAL CHARACTERISTICS OF DISEASE ARE LIVER METASTASES WITH AND ASSOCIATED WITH PROGNOSIS OF WITHOUT HEPATIC ARTERIAL PATIENTS WITH COLO-RECTAL LIVER INFUSION CHEMOTHERAPY METASTASIS UNDERGOING F. E. Buisman1, E. P. van der Stok2, P. J. Allen3, NEOADJUVANT CHEMOTHERAPY A. Cercek4, W. R. Jarnagin3, N. E. Kemeny4, F. Bagante1,2, A. Ruzzenente2, K. Sasaki3, E. Beal1, T. P. Kingham3, B. Groot Koerkamp5 and K. Merath1, S. Conci2, A. Guglielmi2, M. Weiss3, M. I. D’Angelica3 C. Iacono2 and T. Pawlik1 1Surgery, Erasmus MC, Kanker Instituut locatie Danïel den 1Department of Surgery, Ohio State University Wexner Hoed, 2Surgery, IJsselland Ziekenhuis, The Netherlands, Medical Center, United States, 2Department of Surgery, 3Surgery, Memorial Sloan Kettering Cancer Center, 4Med- University of Verona, Italy, and 3Department of Surgery, ical Oncology, Memorial Sloan Kettering Cancer Center, Johns Hopkins Hospital, United States United States, and 5Surgery, Erasmus MC, The Netherlands Background: Similar to the “metro-ticket” tool utilized for Background: Perioperative systemic chemotherapy hepatocellular carcinoma, “tumor-burden score” (TBS) can (SYS) extends early recurrence-free survival (RFS) but account for the impact of tumor morphology on prognosis of not overall survival (OS) in patients with resected patients undergoing resection of colorectal liver metastases colorectal liver metastases (CLM). Studies have found (CRLM). We sought to assess the impact of tumor biology that the combination of systemic and hepatic arterial as determined by response to preoperative chemotherapy infusion chemotherapy (SYS+HAI) significantly pro- (pChemo) relative to tumor morphology defined by TBS. longs OS. The aim of this study was to compare the rate Methods: 669 patients with resectable CRLM who and pattern of recurrence in patients who underwent a received pChemo between 2000-2015 were identified. TBS resection of CLM with perioperative SYS+HAI versus was calculated based on tumor size and total tumor number. SYS alone. Overall survival was assessed relative to preoperative Methods: Consecutive patients (period:1992-2012) after tumor response by comparing pre- versus post-pChemo resection of CLM were included from a prospective TBS.

HPB 2018, 20 (S2), S182eS294 S190 Free Orals (FP01-FP31)

Results: PriortopChemomediantumorsizeand median number of tumors was 7 (2-33). One-hundred number of metastases was 2.7 cm (IQR, 1.8-4.1) and twenty-eight (65%) patients underwent portal vein 2(IQR, 1-4), respectively, for a pre-pChemo TBS of 3.8 embolization. More patients received chemotherapy prior (IQR,2.7-5.6). Following receipt of pChemo, median to the first stage compared with chemotherapy adminis- TBS was 4.1 (IQR,2.8-6.3). TBS increased in tration preceding the second stage (92% vs 60%). Median 188(28.1%) patients post-pChemo for a median increase overall survival was 50 months and median follow-up in TBS of +55.2%(IQR,+35.3-+85.8). In contrast, TBS was 28 months. A hepatic artery infusion pump was decreased or stayed the same in 463 (69.2%) patients placed in 64 (32%) patients with similar overall survival post-pChemo for a median TBS of 0% (IQR,-17.8- (OS) to those managed without a pump (OR 0.956; p = +3.6).Patients with an increase in TBS after pChemo 0.848). Postoperative morbidity following the second had a 5-year OS of 37.3% (IQR, 29.3-47.5) versus stage resection was 47%. Chemotherapy prior to the 52.9% (IQR,46.9-59.6) among patients with TBS second stage did not increase complications rate (OR decreasing after pChemo (p=0.004). In the multivariable 0.794, p=0.736). Readmission following the second stage analysis, while a high pre-pChemo TBS(HR 1.11, 95% was 10% and was associated with a decrease in disease CI:1.06-1.16) was associated with an increased risk of free survival. OS was significantly decreased by positive death, patients who had a decrease in TBS following resection margin and increased estimated blood loss pChemo had a markedly reduced risk of death(HR 0.65, (EBL) (p < 0.05). 95%CI:0.52-0.80)(both p< 0.001) Conclusion: This is the largest TSH series in the US and Conclusion: Both morphological characteristics (pre- demonstrates safety and feasibility of this approach to pChemo TBS) and biological behavior (decreased TBS bilateral CRLM. Chemotherapy prior to the second stage of after pChemo) were strongly associated with prognosis resection does not increase morbidity. Optimal outcomes among patients undergoing resection of CRLM. are influenced by margin status and operative EBL.

FP02-06 AUTOLOGOUS MESENCHYMAL STEM CELLS TRANSPLANTATION IS AN EFFECTIVE METHOD FOR INCREASE OF FUTURE LIVER REMNANT VOLUME IN PRIMARY NON- RESECTABLE COLORECTAL LIVER METASTASES V. Treska, T. Skalicky, V. Liska, J. Fichtl, J. Ludvik and J. Bruha Surgery, Charles University in Pilsen, Czech Republic FP02-04 Figure Introduction: Portal vein embolization (PVE) is not an optimal method for future liver remnant volume (FLRV) growth. The aim of study was to evaluate the effect of autologous mesenchymal stem cells (aMSC) trans- FP02-05 plantation with PVE for expansion of FLRV in patients TWO-STAGE HEPATECTOMY FOR with primary non- resectable colorectal liver metastases BILATERAL COLORECTAL LIVER (CLMs). Method: PVE with aMSC was used in 28 patients with METASTASES: A MULTI- primary non-resectable CLMs between 3/2011 and 12/ INSTITUTIONAL ANALYSIS 2017. Inclusion criteria included patients indicated for large M. I. Chavez1, C. G. Ethun2, G. A. Margonis3, hepatectomy, FLRV < 30% healthy or < 40% damaged S. Gholami4, B. J. Kim5, S. K. Maithel2, T. M. Pawlik6, liver parenchyma. Exclusion criteria were presence of M. I. D’Angelica4, T. A. Aloia5 and T. C. Gamblin1 extrahepatic metastases, pregnancy and liver insufficiency. 1Surgical Oncology, Medical College of Wisconsin, 2Surgical PVE was performed on the site of CLMs and aMSC were Oncology, Emory University, 3Johns Hopkins Hospital, applied to the contralateral liver lobe during one stage 4Surgical Oncology, Memorial Sloan Kettering Cancer procedure. FLRV growth was examined by computed to- Center, 5Surgical Oncology, M.D. Anderson Cancer Center, mography volumometry. Liver resection was performed as and 6Ohio State Medical Center, United States soon as FRLV was sufficient. Background: Two-stage hepatectomy (TSH) is an Results: FLRV was sufficient in all patients during 3 weeks important tool in the management of bilateral colorectal after procedure (p<0.0002). There was zero mortality and liver metastases (CRLM). This study sought to examine the 3.6% morbidity rate. R0 hepatectomy was performed in 24 presentation, management and outcomes of patients (85.7 %). One and three years overall survival (OS) was completing TSH. 88.1, resp. 68.7 %, progression free survival (PFS) 67.2, Methods: A retrospective review from 5 US liver centers resp.32.1% in patients after liver resection. Negative identified patients who underwent TSH for bilateral CRLM. prognostic factors for OS were CLMs volume (p< 0.01) Results: From December 2000 to March 2016, a total of and the right site localization of primary tumor 196 patients completed a TSH. The majority of proced- (p < 0.05).Growth of CLMs volume was found in 19 (82.6 ures (90%) were performed with an open technique. The %) patients ( p< 0.00009) and was significant for OS.

HPB 2018, 20 (S2), S182eS294 Free Orals (FP01-FP31) S191

Conclusion: aMSC transplantation together with PVE is a of OS, whereas first recurrence in multiple sites and RAS safe and promising method for FLRV growth in patients mutation were independent risk factors for OS (Table1). with primary non-resectable CLMs. Supported by Conclusion: In patients who have undergone TSH, recur- PROGRES research project. rence is frequent and should be treated with resection whenever possible. RAS wild-type patients fare particularly well with aggressive surgical management. FP02-07 SURGICAL RESECTION FOR RECURRENCE AFTER TWO-STAGE HEPATECTOMY FOR COLORECTAL LIVER METASTASES IS FEASIBLE, SAFE AND IMPROVES SURVIVAL H. Lillemoe1, Y. Kawaguchi1, G. Passot2, G. Karagkounis3, Y. -Q. You1, Y. S. Chun1, C. Conrad1, C. -W. Tzeng1, T. Aloia1 and J. -N. Vauthey1 1Department of Surgical Oncology, MD Anderson Cancer Center, United States, 2Department of Oncologic and General Surgery, Centre Hospitalier Lyon Sud, Hospices Civils de Lyon, France, and 3Department of Surgery, Figure 1 Cleveland Clinic, United States

FP02-07 Table 1Uni- and multi-variate analysis of OS N (%) 5-year Univariate Hazard 95% confidence Multivariable survival rate, % P value ratio interval P value All patients 83 42 Preoperative chemotherapy 55 (66) 47 0.798 Primary tumor location, rectum 30 (36) 37 0.264 Total number of tumors, >5 51 (61) 33 0.306 Positive surgical margin (either stage) 29 (35) 39 0.900 RAS mutation 36 (46) 19 0.006 2.25 1.16-4.50 0.016 Major complication (either stage, 19 (23) 34 0.287 Clavien-Dindo Grade  III) First recurrence at multiple sites 25 (30) 24 0.005 2.28 1.17-4.37 0.016 Resection for recurrence 31 (37) 67 <0.001 0.25 0.10-0.54 <0.001

Introduction: Recurrence rates are high after two-stage hepatectomy (TSH) for bilateral colorectal liver metastases FP02-08 (CLM), and there is no established treatment approach. RECURRENCE RATE AND PATTERNS This study aimed to evaluate the feasibility, safety, and prognostic impact of surgical resection for recurrence after WITH AND WITHOUT PERIOPERATIVE TSH and the prognostic role of RAS mutation in this cohort. SYSTEMIC CHEMOTHERAPY AFTER Methods: 137 patients planned to undergo TSH from RESECTION OF COLORECTAL LIVER 2003-2016. Clinicopathologic factors were compared using METASTASES univariate and multivariate analysis. F. E. Buisman1, E. P. van der Stok2, B. Galjart1, Results: 111 patients (81%) completed TSH. Patients P. J. Allen3, W. R. Jarnagin3, T. P. Kingham3, completing both stages had significantly higher overall D. J. Grünhagen1, C. Verhoef1,M.I.D’Angelica3 and survival (OS) compared to patients who did not (p< 0.001). B. Groot Koerkamp4 The median recurrence-free survival in patients completing 1Surgery, Erasmus MC, Kanker Instituut locatie Danïel TSH was 12 months. Of 83 patients who recurred, resection den Hoed, 2Surgery, IJsselland Ziekenhuis, 3Surgery, was performed in 31 patients (37%) with 11 patients un- Memorial Sloan Kettering Cancer Center, United States, dergoing multiple resections for recurrence. Of 48 total and 4Surgery, Erasmus MC, The Netherlands operations, there were 23 repeat hepatectomies (48%), 14 pulmonary resections (29%), 5 locoregional resections Background: The aim of this study is to compare the rate (10%), and 6 concurrent resections in multiple organ sites and patterns of recurrence after resection of colorectal liver (13%). Review of perioperative outcomes revealed an metastases (CLM) in patients treated with and without appropriate safety profile and major complication rate of perioperative systemic chemotherapy (CTx). 8%. The median OS for patients who underwent resection Methods: Consecutive patients with resected CLM were fi was 143 months vs. 49 months for those who did not included from two centers. Patients were strati ed by (Figure 1, p< 0.001). On multivariate analysis, resection Clinical Risk Score (CRS) in cox proportional hazards for recurrence was associated with significantly lower risk modeling. Only initial recurrences were analyzed.

HPB 2018, 20 (S2), S182eS294 S192 Free Orals (FP01-FP31)

Results: A total of 2115 patients were included; 1525 implication of lymph node dissection (LND) have been (72.1%) in the CTx group and 590 (27.9%) in the no-CTx debated. group. The median overall survival (OS) was similar in both Materials and methods: Three hundred and eighty-five ICC groups (CTx 57 months versus no-CTx 52 months, p=0.35). patients with radical resection between were retrospectively Median recurrence free survival (RFS) was also similar in analyzed by the Okayama Study group of Hepatobiliary and both groups (CTx 14 months versus no-CTx 13 months, Pancreatic Surgery, which consist of 16 HBP programs. p=0.75). However, 44.1% of patients in the CTx group had Results: LND were performed in 264 patients (69%), a high CRS versus 26.8% in the no-CTx group (p< 0.001). including 137 cases of extended LND beyond hepatoduo- In multivariable analysis stratified by CRS, CTx was an denal ligament. In LND cases, 107 (40%) were pathologi- independent favorable prognostic factor for OS in patients cally confirmed LN metastasis (pN+). MST in pN+ was with a high CRS (HR 0.76, 95%CI 0.60-0.97, p=0.03), and significantly lower than that in pN- (16 vs 52 months; p< not in patients with a low CRS (HR 0.99, 95%CI 0.83-1.18, .0001). Multivariate analysis revealed independent risk p=0.93). No differences were found considering RFS in factors of LN metastasis: located in hepatic hilum or pre- multivariable analysis. High CRS patients had fewer initial dominant in left lobe (OR 1.9), invasion of serosa (OR 2.0), simultaneous intra-and-extrahepatic recurrences with CTx mod/poorly differentiation (OR 3.6), intraductal infiltrating (16.4% versus 27.0%, p=0.003). No differences in recur- (OR 4.1), and high serum CA19-9 (p< .0001). There was rence rate and pattern were found for low CRS patients. no difference between normal LND and extended LND (33 Conclusion: In patients with resected CLM and high CRS vs 31 months; p=.92). However, 12 patients who survived who received perioperative CTx, OS was superior, and over five years despite pN+ were all performed extended initial simultaneous intra- and extrahepatic recurrences LND. On the other hand, of 121 patients without LND, were less frequent. only 7 patients showed LN-limited recurrence. Adjuvant

FP02-08 Table 1Recurrence rates and patterns Fong CRS** Site of recurrence CTx no-CTx p-value Low (0-2) Intrahepatic-only 171 (19.9%) 77 (18.6%) 0.58 Intra- and extrahepatic 91 (10.6%) 48 (11.6%) 0.59 Extrahepatic-only 233 (27.1%) 120 (28.9%) 0.48 Total 495 (57.5%) 245 (59.0%) 0.60 High (3-5) Intrahepatic-only 180 (31.8%) 35 (23.6%) 0.05 Intra- and extrahepatic 93 (16.4%) 40 (27.0%) 0.003* Extrahepatic-only 152 (26.9%) 40 (27.0%) 0.97 Total 425 (75.1%) 115 (77.7%) 0.51

** Fong Clinical risk score (0-5 points): Node-positive primary, disease-free interval <12 months, >1 CLM, size largest CLM >5cm, preoperative CEA > 200 ng/ml chemo (GEM or S1) were introduced to 145 patients (38%). FP03 - Free Papers 3 (long oral) - Liver: tumours In patients with pN+, adjuvant chemo improved survival FP03-01 outcome (Chemo+: 19 vs Chemo-: 12 months; p=.003). CLINICAL IMPLICATION OF LYMPH Conclusions: Routine LND may not be suitable for ICC. However, it would have therapeutic value in some limited NODE DISSECTION FOR cases and in identification of nodal status which could lead INTRAHEPATIC to prediction of adjuvant chemotherapy. CHOLANGIOCARCINOMA: A MULTICENTER ANALYSIS OF 385 RESECTED CASES FP03-02 PATHOLOGICAL CHARACTERISTICS T. Kojima1, Y. Umeda2, D. Satoh3, Y. Endo4, K. Sui5, M. Inagaki6, M. Oishi7, T. Ota8, T. Yagi2, AND EARLY POST-HEPATIC- T. Fujiwara2 and Okayama Study Group of Hepatobiliary RESECTION OUTCOME OF PATIENTS and Pancreatic Surgery WITH HEPATOCELLULAR 1Department of Surgery, Okayama Saiseikai General Hospi- CARCINOMA OCCURRED AFTER 2 tal, Department of Gastroenterological Surgery, Okayama HEPATITIS C TREATMENT WITH NEW University, 3Department of Surgery, Hiroshima City Hiroshima Citizens Hospital, 4Department of Surgery, DIRECT-ACTING ANTIVIRALS: A Himeji Japanese Red Cross Hospital, 5Department of MULTICENTER COHORT STUDY Gastroenterological Surgery, Kochi Health Sciences Center, A. Vitale1, F. P. Russo1, C. Sposito2, A. Cucchetti3, 6Department of Surgery, National Hospital Organization G. Levi Sandri4, S. Gruttadauria5, S. Di Sandro6, Fukuyama Medical Center, 7Department of Surgery, Tottori D. Ghinolfi7, D. Nicolini8, F. Trevisani3 and Special Municipal Hospital, and 8Department of Surgery, National Interest Group on Hepatocellular Carcinoma and New Hospital Organization Okayama Medical Center, Japan Anti-HCV Therapies of the Italian Association for the Introduction: Lymph node (LN) metastasis is well known Study of the Liver (AISF) 1 as a worse prognostic factor in intrahepatic chol- Department of Surgery, Oncology and Gastroenterology, 2 angiocarcinoma (ICC). However, clinical effectiveness and University of Padua, General Surgery and Liver

HPB 2018, 20 (S2), S182eS294 Free Orals (FP01-FP31) S193

Transplantation Unit, University of Milan and Istituto HCC, and prospectively validated by a cohort of 394 naïve Nazionale Tumori (National Cancer Institute), IRCCS, HCC patients. Initial and best response was defined as 3Department of Medical and Surgical Science, Semeiotica radiological response after first or at least 2nd session of Medica, University of Bologna, 4Division of General TACE, respectively. Surgery and Liver Transplantation, San Camillo Hospital, Results: Patients with SRD 6 months had longest median 5Mediterranean Institute for Transplantation and overall survival (OS), followed by patients with initial and Specialization Therapies (IRCCS-ISMETT), 6Division of best objective response, and then patients with persistent no General Surgery and Abdominal Transplantation, ASST response (67.7, 55.8, 53.2, 23.6 months, respectively, p< Grande Ospedale Metropolitano Niguarda, 7Hepatobili- 0.001). For patients with SRD 12 months, initial com- ary Surgery and Liver Transplantation, University of Pisa plete response (CR), subsequent CR, best partial response, Medical School Hospital, and 8Hepato-biliary and the median OS were 77.1, 68.9, 53.1, 47.0 months, Abdominal Transplantation Surgery, Polytechnic Univer- respectively (p< 0.001). SRD had the best ability to predict sity of Marche, A.O.U. ’Ospedali Riuniti’, Italy 5 year OS comparing to initial response, best response (The Introduction: There are no studies evaluating the patho- area under the receiver-operating characteristic curve were logical characteristics of recurrent or naïve hepatocellular 0.913, 0.840, 0.822, respectively, p< 0.001). SRD 6 carcinoma (HCC) occurring after anti-hepatitis-C (HCV) months (Hazard ratio (HR) 0.145, p< 0.001) was found to therapy with direct-acting antivirals (DAAs). Moreover, the be the independent prognostic factor for OS. Tumor early postoperative outcome of these HCC patients after number (Odds ratio (OR) 3.177, p=0.004), tumor size (OR hepatic resection is unknown. 1.687, p=0.001), capsule (OR 0.478, p< 0.001) were sig- Methods: Prospectively collected data from 420 consecu- nificant associated with SRD 6 months after TACE. The tive patients with HCC and HCV cirrhosis undergoing liver significance of SRD was confirmed in validation cohort. resection in 18 Italian hepato-biliary surgical units between Conclusion: SRD is the most significant prognostic factor in January 2014 and December 2016 were analysed. Seventy- patients with unresectable HCC after TACE. Achievement of seven patients (18.3%) who develop recurrent or de novo SRD 6 months is the robust predictor for a favorable OS. HCC after DAAs therapy represented the study group, while the remaining 343 HCC patients formed the control group. The aim of this study was to compare these two groups in FP03-05 terms of pathological characteristics (primary endpoint) and OUTCOME AFTER LIVER early postoperative outcome (secondary endpoint). TRANSPLANTATION, RESECTION AND Results: Primary endpoint (pathological characteristics): the study group showed significantly smaller tumors than ABLATION FOR HEPATOCELLULAR the control group (25 mm vs. 35 mm), while no significant CARCINOMA IN SWEDEN e A differences were found in terms of numbers of nodules, NATIONAL REGISTRY-BASED STUDY grading, vascular invasion, and satellitosis. Secondary 2008 e 2016 endpoint (early postoperative outcome): after inverse M. Sternby Eilard1,2, P. Naredi1, O. Hemmingsson3, probability of treatment weighting, the 2 groups became B. Isaksson4, G. Lindell5, P. Sandström6, C. Strömberg7, well-balanced for all baseline characteristics. Patients in the G. Söderdahl8 and M. Rizell1,2 study group showed a significantly lower incidence of 1Department of Surgery, Sahlgrenska Academy, University severe (Clavien score > 2) complications (3.4% vs. 9.3%) of Gothenburg, Institute of Clinical Sciences, 2Trans- and early (within 6 months) postoperative mortality (2.0% plantation Centre, Sahlgrenska University Hospital, vs 5.4%) than those in the control group. 3Department of Surgical and Perioperative Sciences, Conclusion: DAAs therapy doesn’tseemtomodifythebio- Umeå University, Division of Surgery, 4Department of logical aggressiveness of recurrent or de novo HCCs under- Surgical Sciences, Upper Abdominal Surgery, Uppsala going liver resection. Conversely, DAAs therapy significantly University, 5Department of Clinical Sciences, Division V, improves the early postoperative outcome of these patients. Surgery, Lund University, Faculty of Medicine, 6Depart- ment of Surgery, Linköping University, 7Department of Clinical Science, Intervention and Technology (CLIN- FP03-04 TEC), Karolinska Institute, and 8Department of Trans- SUSTAINED RESPONSE DURATION: A plantation, CLINTEC, Karolinska Institute and Karolinska ROBUST PREDICTOR FOR University Hospital, Sweden UNRESECTABLE HEPATOCELLULAR Introduction: Patients with limited hepatocellular carci- CARCINOMA AFTER noma (HCC) and well-preserved liver function can be CHEMOEMBOLIZATION treated by liver transplantation, resection or ablation. Transplantation offers good long-term results, but with a 1 2 3 Z. Peng , Y. Zhang and M. Zhang relatively high post-operative morbidity and long-term 1 Oncology, The First Affiliated Hospital, Sun Yat-sen Uni- complications. Liver resection is associated with a high risk 2 3 versity, Cancer Center, Sun Yat-sen University, and The of recurrent cancer. Liver ablation has increased lately, both First Affiliated Hospital, Sun Yat-sen University, China as a single curative option and while waiting for trans- Purpose: To explore the role of sustained response dura- plantation, as complication rates are favorable. Our objec- tion (SRD) by modified Response Evaluation Criteria in tives were to compare outcome after liver transplantation, Solid Tumors (mRECIST) after transarterial chemo- resection or ablation respectively, in relation to tumor stage, embolization (TACE) for hepatocellular carcinoma (HCC). liver function and comorbidities. Materials: Role of SRD after initial TACE was determined Methods: Prospectively collected data was retrieved from by mRECIST in 2853 consecutive patients with naïve national Swedish Quality Registry (SweLiv) for HCC-

HPB 2018, 20 (S2), S182eS294 S194 Free Orals (FP01-FP31) patients, treated 2008-2016 in Sweden. Overall survival years was 45.7%. There was no significant survival difference and cumulative incidence of tumor recurrences were between B3 and B4 groups (p = 0.404). TNM Stage (p < analyzed for each treatment group. 0.001) and presence of fibrosis/cirrhosis (p = 0.002) were Preliminary results: During 2008 - 2016, 3590 patients independent predictors of long-term survival. Both perform- were diagnosed with HCC; 31% by surveillance, while ing liver resection equal to or greater than hemihepatectomy 11% were incidental radiologic findings. Curatively aiming and combined bile duct resection (BDR) decreased risk of treatments were given in 1253 patients (35%); trans- both overall [hazard ratio, HR: 0.61 (0.38 - 0.99); p =0.044 plantation in 273, resection in 544 and ablation in 436 and HR: 0.51 (0.31 - 0.84); p = 0.008, respectively] and patients. Median time from diagnosis to treatment was 223, disease-free survival [HR: 0.59 (0.38 - 0.91); p =0.018and 83 and 107 days respectively. HR: 0.61 (0.42 - 0.89); p = 0.009, respectively] significantly. Five year overall survival was 52% for patients who had Clinical outcomes were mostly influenced by tumor resection, 76% for transplantation and 35% for ablation. stage and underlying liver function rather than presence of After 2 years, the probabilities for recurrence and for death BDTT. Therefore, aggressive surgical approach including without recurrence were 9% and 5% respectively after major liver resection combined with BDR to increase transplantation, 36% and 12% after resection and 44 and 14 chance of R0 resection is strongly recommended. % respectively after ablation. Conclusion: Recurrences and deaths without recurrence were more common after resection/ablation than after liver FP04 e Free Papers 4 (long oral) e Pancreas: transplantation. Adjusted subgroup analyses will be done. Neoplasms FP04-02 PLATELET RNA YIELDS A NEW DIAGNOSTIC TEST FOR PANCREATIC CANCER: A MULTICENTER COHORT T. Le Large1,2,3, L. L. Meijer1,3,S.G.J.G.in’t Veld3, M. Best3, N. Funel4, M. F. Bijlsma2,5, M. Besselink5,6, E. Giovannetti3,4, T. Wurdinger3 and G. Kazemier1,3 1Surgery, VU University Medical Center, 2LEXOR, Aca- demic Medical Center, 3Cancer Center Amsterdam, VU University Medical Center, The Netherlands, 4Pathology, University Hospital Pisa, Italy, 5Cancer Center Amster- 6 Figure [Recurrences and deaths after HCC-treatment] dam, Academisch Medisch Centrum (AMC), and Surgery, Academic Medical Center, The Netherlands Introduction: Pancreatic ductal adenocarcinoma (PDAC) FP03-06 is often diagnosed at advanced disease stages and invasive SURGICAL OUTCOMES OF procedures are needed to confirm diagnosis. Innovative HEPATOCELLULAR CARCINOMA blood-based approaches are warranted to enable easier and WITH BILE DUCT TUMOR THROMBUS: earlier diagnosis. Platelet-derived RNA recently emerged as a new diagnostic biomarker in patients with lung cancer. A KOREA-JAPAN MULTICENTER The aim of this study was to evaluate specific platelet RNA STUDY profiles for patients with PDAC. D. S. Kim1, M. Kaibori2, M. Kon2, M. Yamamoto3, Methods: Platelets of 135 PDAC patients (all disease stages), H. J. Wang4 and Korea-Japan HBP Collaborative Research 250 age-matched healthy controls (HC) and 72 patients with Group benign pancreatobiliary disease were collected at four centers 1Department of Surgery, Korea University College of in Italy and the Netherlands. Samples were allocated to a Medicine, Republic of Korea, 2Department of Surgery, training or validation cohort. After RNA isolation and ampli- Kansai Medical University, 3Department of Surgery, fication, Illumina sequencing was performed. Differential Tokyo Women’s Medical University, Japan, and expression was tested by ANOVA-test. Sequentially, we 4Department of Surgery, Ajou University College of performed swarm-enhanced classification to establish the most Medicine, Republic of Korea discriminating profiles between malignant and benign disease. fi Prognosis of patients with hepatocellular carcinoma (HCC) Results: Patients with PDAC had signi cantly different fi < accompanying bile duct tumor thrombus (BDTT) has been platelet RNA pro les compared to HC (p-value 0.0001). In fi known to be poor. There have been significant controversies the rst validation cohort, the algorithm predicted PDAC from regarding treatment and risk factors due to small number of control groups with an accuracy of over 90%. Moreover, we fi cases in previous reports. The aim of this study is to identify classi ed early stage disease (stage I-IIa) correctly as PDAC optimal surgical methods and risk factors for long-term sur- and distinguished them from advanced and benign disease. vival in patients with HCC accompanying macroscopic Conclusions: There is an urgent need for minimally-invasive BDTT. biomarkers to diagnose PDAC. Here we described our novel Records of 257 cases (B3 group = 151, B4 group = 106) method to diagnose PDAC from platelet-derived RNA. from 32 centers in Korea and Japan (1992 w 2014) were This method showed robust differences between PDAC and analyzed for overall and disease-free survival using Cox control groups, even at early disease stages. Evaluation of proportional hazard model. premalignant lesions and treated patients is ongoing to prove Curative surgery was performed in 244 (94.9%) patients its monitoring value and large-scale multicenter validation is with operative mortality of 5.1%. Overall survival rate at 5 warranted to facilitate implementation in the clinic.

HPB 2018, 20 (S2), S182eS294 Free Orals (FP01-FP31) S195

FP04-03 Methods: 3D pancreatic tumor organoids were established from pancreatic tumor tissue, obtained from pancreatic THE NATURAL HISTORY OF cancer patients (n=3) with cachexia. Conditioned medium INTRADUCTAL PAPILLARY was collected every 2-3 days. Simultaneously, human MUCINOUS NEOPLASM OF THE visceral SMCs were grown to confluency on basement PANCREAS: REAPPRAISAL OF THE membrane matrix coated surfaces and kept in a contractile INDOLENT PRECURSOR OF phenotype. After 6 days, SMCs were stimulated for 24h with 25% organoid conditioned medium. mRNA expres- PANCREATIC CANCER sion of contractile SMC markers (ACTA2, ACTG2, 1 2 2 G. Marchegiani , S. Andrianello , T. Pollini , TAGLN, SMTN) and extracellular matrix components 2 2 2 2 A. Caravati , M. Biancotto , G. Malleo , C. Bassi and (COL1A1, COL3A1, ELN) were analyzed by RT-qPCR. 2 R. Salvia Results: Conditioned medium from pancreatic tumor orga- 1 2 Surgery, University of Verona, and Verona University, Italy noids resulted in decreased expression levels of Acta2 (a- Aim: Aim of the present study is to describe the natural smooth muscle actin; 0.83-fold) and Actg2 (g-smooth muscle history of IPMNs observed at a high-volume center for actin; 0.52-fold), indicating loss of the mature contractile thirty years. phenotype. Concurrently, we observed increased expression Methods: Growth rates were analyzed through a linear- of elastin (ELN; 3.4-fold) indicating the contractile SMC mixed model. The development of worrisome features phenotype shifts towards a so-called synthetic phenotype. (WF), high-risk stigmata (HRS), pancreatic cancer (PC), Conclusion: These data suggest that pancreatic tumor cells survival and risk for surgery were analyzed. from cachectic patients secrete factors that are able to Results: From a total of 2189 observed patients, 1529 were modulate the contractile SMC phenotype which may be the included. The overall growth rate was 1mm/year. For about underlying cause of gastrointestinal motility problems. half of cases no dimensional change was documented. The presence of high risk stigmata (HRS), age < 75 years, and cyst size >30mm at diagnosis were associated to a faster growth FP04-05 rate. During follow-up, IPMNs developed WF in 6.5% of THE IMPACT OF FAILURE TO cases and HRS in 0.6%. Overall, 3.5% of patients developed ACHIEVE SYMPTOM CONTROL AFTER PC after a median time of 28 months. Of these PC patients, RESECTION OF FUNCTIONAL 72% previously developed HRS/WF. Of 1043 initially observed indolent branch duct (BD) IPMNs, 16 eventually NEUROENDOCRINE TUMORS: AN 8- developed PC with 10% occurring after 15 years of follow-up. INSTITUTION STUDY FROM THE US HRS and growth rate were independent predictors of PC. NEUROENDOCRINE TUMOR STUDY Growth rate was the only difference between IPMNs devel- GROUP oping PC and those remained stable after more than 5 years of M. Zaidi1, A. Lopez-Aguiar1, G. Poultsides2, M. Dillhoff3, follow-up (n=399). The mean estimated disease specificsur- F. Rocha4, K. Idrees5, H. Nathan6, E. Winslow7, vival (DSS) for the overall population exceeded 19 years. R. Fields8 and S. Maithel1 Conclusions: IPMN of the pancreas is the indolent pre- 1Surgical Oncology, Emory University, 2Surgery, Stanford cursor of PDAC that will not show a detectable growth University, 3Surgical Oncology, Ohio State University during follow-up in half of the cases. Those rapidly Comprehensive Cancer Center, 4Surgical Oncology, Vir- growing (>2.75 mm/year) will likely progress to pancreatic ginia Mason Medical Center, 5Surgical Oncology, cancer through the development of WF and HRS during the Vanderbilt University Medical Center, 6Surgical first year of follow-up. Oncology, University of Michigan, 7Surgical Oncology, University of Wisconsin School of Medicine and Public Health, and 8Surgical Oncology, Washington University FP04-04 School of Medicine, United States PANCREATIC TUMOR ORGANOID Introduction: The goals of resection for pts with functional CONDITIONED MEDIUM MODULATES neuroendocrine tumors(F-NETs) are two-fold: oncologic THE SMOOTH MUSCLE CELL benefit and symptom control. The interaction between the CONTRACTILE PHENOTYPE two is not well-understood. Method: All pts with F-NETs of the pancreas, liver, duo- R. Vaes1, M. Aberle2, D. van Dijk2, S. Olde Damink2 and 2 denum, and ampulla who underwent curative-intent resection S. Rensen fi 1 2 from 2000-2016 were identi ed. Cox regression analysis was Surgery, Maastricht University, and Maastricht utilized to determine factors associated with reduced recur- University rence-free survival(RFS). A multivariable model was created Introduction: Pancreatic cancer patients often suffer from by incorporating variables associated with RFS with p< 0.1. gastrointestinal-related symptoms which may be the Results: Of 260 pts with resected F-NETs, 230 underwent consequence of underlying gastrointestinal motility prob- curative-intent resection. 53% were insulinomas, 35% lems. Although muscle loss in cachectic pancreatic cancer gastrinomas, and 12% other. 21% had a known genetic patients is most obvious in skeletal muscle, these clinical syndrome(majority MEN-1), 23% had LN(+) disease, 80% symptoms suggest that cachexia may manifest itself also in underwent an R0 resection, and 14% had no postoperative smooth muscle, a tissue responsible for contraction of the symptom improvement(SI). Pts who did have SI were more gastrointestinal tract. We aimed to investigate whether the likely to have an insulinoma(p=0.01), no genetic syndro- smooth muscle cell (SMC) contractile phenotype is me(p=0.001), and R0 resection(p=0.007). LN(+) disease did affected in cachectic pancreatic cancer patients. not correlate with postoperative SI. Factors associated with

HPB 2018, 20 (S2), S182eS294 S196 Free Orals (FP01-FP31) reduced RFS included non-insulinoma histology, known to 12 (71%) from the 17 livers submitted to the combined genetic syndrome, LN(+) disease, R1 margin, and lack of SI. protocol (p=0.050). Non-viable livers had higher donor On MV analysis, only the failure to achieve SI following BMI (p=0.048) and liver weight (p=0.011) and were resection persisted as being associated with reduced discarded more often for steatosis (p=0.001). On multi- RFS(Table). Considering only pts with an R0 resection, variable analysis livers declined for steatosis (OR:50.6; failure to achieve SI was still associated with worse 3-yr RFS 95%CI:3.7-688.9; p=0.003), donor height (OR:1.1; 95% compared to pts with SI(36%vs80%;p=0.006). CI:1.1-1.2; p=0.039) and NMP alone (OR:10.3; 95% Conclusions: Failure to achieve symptomatic improve- CI:1.5-70.9; p=0.018) were independent risk factors for ment after resection of functional NETs is associated with non-achievement of the viability criteria. worse recurrence-free survival, even when accounting for Conclusion: Combination of HMP and NMP can increase histologic type, presence of genetic syndromes, R1 resec- the rescue of very-high risk extended criteria donor livers. tion, and LN involvement. These patients may benefit from However steatosis remains the major risk factor for non- short-interval periodic imaging postoperatively to assess viability achievement. New strategies to improve those for earlier radiographic recurrence of disease. organs still needed.

FP04-05 Table 1 Variable Univariable Multivariable Hazard Ratio 95% CI p-value Hazard Ratio 95% CI p-value Type of Functional Tumor Insulinoma Ref ––Ref –– Gastrinoma 2.8 (1.3-6.1) 0.006 1.1 (0.6-2.0) 0.75 Other (glucagonoma, somatostatinoma, VIPoma) 2.7 (1.0-7.2) 0.042 ––– Known Genetic Syndrome 1.8 (0.9-3.5) 0.077 0.68 (0.2-2.0) 0.49 Lymph Node Positive 1.8 (0.9-3.6) 0.080 1.6 (0.6-4.6) 0.35 R1 Resection Margin 2 (1.0-3.9) 0.052 0.45 (0.1-1.8) 0.25 Failure of Symptom Improvement 3.1 (1.3-7.2) 0.008 4.7 (1.3-16.6) 0.016

FP05-02 FP05 - Free Papers 5 (long oral) - Transplant 1 EX VIVO SUBNORMOTHERMIC FP05-01 REGULATED HEPATIC REPERFUSION FACTORS PREDICTING VIABILITY IS SUPERIOR TO NORMOTHERMIC ACHIEVEMENT ON DISCARDED MACHINE PERFUSION: A NOVEL DONOR LIVERS SUBMITTED TO CONCEPT OF LIVER RESUSCITATION EXTRA-CORPOREAL MACHINE AFTER CARDIAC DEATH IN LARGE PERFUSION ANIMAL MODEL Y. Longatto Boteon1,2, R. Laing1,2, A. Schlegel1, J. Kim1, W. Y. Shin1, M. A. Zimmerman1, A. Martin1, J. Attard1,2, L. Wallace2, R. Bhogal1, D. Neil1, D. Mirza1, J. Yee1, C. Hagen2, J. McGraw3, A. Camara4, H. Mergental1 and S. Afford2 D. Stowe4 and J. C. Hong1 1Liver Unit, Queen Elizabeth Hospital Birmingham, and 1Transplant Surgery, Medical College of Wisconsin, 2Pa- 2National Institute for Health Research, Birmingham Liver thology, Medical College of Wisconsin, 3Concordia Uni- Biomedical Research Centre, University of Birmingham, versity, and 4Anesthesiology, Medical College of United Kingdom Wisconsin, United States Introduction: Ex-vivo machine perfusion of livers (MP) Introduction: Livers procured after cardiac death suffer may offer superior preservation for marginal donor organs. extreme degree of ischemia - reperfusion injury (IRI) and at During normothermic machine perfusion (NMP) is increased risk of graft failure after liver transplantation (LT). possible to assess liver viability before transplantation. A Regulated hepatic reperfusion (RHR) is a novel organ resus- period of hypothermic machine perfusion (HMP) prior to citation therapy that facilitates hepatic mitochondrial and NMP may restore the liver metabolism. The aim of this microcirculatory recovery from IRI using an energy substrate- study was to compare viability criteria achievement in enriched, leukocyte-depleted, oxygen-saturated perfusate livers subjected to NMP-only or a combination of HMP delivered in a pressure, and temperature -controlled milieu. followed by NMP; and, analyse factors associated with the Methods: Using a swine donation after cardiac (DCD) positive outcome. model, all livers were subjected to 1 hour of in situ warm Methods: We analysed 46 discarded human livers sub- ischemia and 2 hours of cold static (CS) preservation. All mitted to MP including NMP alone for 6 hours or a com- livers were subsequently reperfused and monitored for 2 bined protocol of HMP for 2 hours followed by NMP for hours in the ex vivo perfusion machine. Four methods of further 4 hours. Viability was assessed at the end of the reperfusion were compared. 5 animals in each treatment NMP in both groups using the lactate clearance criteria. group. Group I: Sanguineous-Subnormothermic (SS); Results: Steatosis was the main for organ discard (46%). Group II: RHR-Subnormothermic (RHR-S); Group III: 24/46 (52%) organs achieved viability criteria. From 29 Sanguineous-Normothermic (SN); Group IV: RHR- livers that had NMP-only 12 (41%) met viability compared

HPB 2018, 20 (S2), S182eS294 Free Orals (FP01-FP31) S197

Normothermic (RHR-N). We analyzed the effects of RHR Introduction: Although there is data on steroid-free on post-reperfusion biochemical, metabolic, immunolog- immunosuppression following deceased donor liver ical and histological profiles and mitochondrial function. transplantation, robust evidence is lacking in LDLT. Results: Compared to Groups I, III, and IV, Group II Our aim was to evaluate the efficacy of steroid-free (RHR-S) significantly mitigated IRI: AST (p=0.04), total immunosuppression(basiliximab induction) following bilirubin (p=0.03), TNF-alpha (p< 0.01), IL-10 (p< 0.01); LDLT. and IL-6 (p< 0.01). Mitochondrial calcium retention ca- Method: Out of 95 LDLTs performed in 1 year, after pacity and respiration confirmed preservation of mito- excluding 11 patients(renal dysfunction-3, APOLT-2, chondrial function in Group II. Histopathological Multiorgan transplant-1, ABO incompatable-2, Retrans- evaluation of hepatocyte viability showed minimal plant-1, Massive bleed-2), 84 patients were randomized ischemic changes in Group II; however, a significant degree either to SF-arm(Basiliximab + Tacrolimus + Azathio- of hepatocyte necrosis was observed the other groups. prine, n=42) or S-Arm(Steroid + Tacrolimus + Azathio- Conclusion: Ex Vivo subnormothermic regulated hepatic prine, n=42). Primary objective was to compare metabolic reperfusion mitigates IRI, preserves mitochondrial function complications. and facilitates liver function recovery after donation after Results: Baseline characteristics like age, etiology, cardiac death. This novel strategy has potential applica- comorbidities, MELD score, duration of surgery, blood bility to human liver transplantation. loss, cold ischemia and GRWR were comparable in both groups. The incidence of new onset diabetes after transplantation(NODAT) was significantly higher in S- FP05-04 arm at 3 months (68.2% vs 21.7%, p-0.002) and at 6 RANDOMISED STUDY ON STEROID months(54.5% Vs 13%, p-0.003). Likewise incidence of new onset hypertension(32.1% Vs 6.1%, p-0.021), FREE IMMUNOSUPPRESSION WITH hypertriglyceridemia(27.8% Vs 7.5%, p-0.041) at six BASILIXIMAB INDUCTION VS months were significantly higher in S-arm. However STANDARD TRIPLE DRUG incidence of biopsy proved rejection (16.7% Vs 21.4 %, IMMUNOSUPPRESSION IN ADULT LIVE p-0.57), time to first rejection (56 vs 44 days, p-0.080), DONOR LIVER TRANSPLANTATION and graft survival(549 Vs 499 days, p- 0.329)(median follow up of 10.7 months), were not different between (LDLT). CTRI/2017/08/009508 two groups (SF-arm Vs. S-arm). Interestingly ICU stay M. Kathirvel, S. Mallick, M. S. Durairaj, K. Nair, was significantly less in SF-arm compared to S-arm(8.8 J. Shaji Mathew, S. Binoj, R. N. Menon, D. Balakrishnan, Vs 10.7 days, p-0.042). There were no differences in U. Gopalakrishnan and S. Sudhindran renal functions, readmission rates, hospital stay, infec- GI Surgery & Solid Organ Transplantation, Amrita tious, biliary and vascular complications. Institute of Medical Sciences & Research, India

FP05-04 Table [Study End Points] Variable SF-arm S-arm P-value NODAT at 3 months 5 (21.7%) 15 (68.2%) 0.002 NODAT at 6 months (6M) 3 (13%) 12 (54.5%) 0.008 NOSHT at 3 months 2(6.1%) 9(32.1%) 0.021 NOSHT at 6 months 2(6.1%) 9(32.1%) 0.021 Hypercholesterolemia at 6 months 8 (20%) 13(36.1%) 0.117 Hypertriglyceridemia at 6 months 3 (7.5%) 10 (27.8%) 0.041 Mean Total cholesterol at 6M 171.39 ±35.33 187.76 ±56.23 0.139 Mean Triglycerides at 6M 101.5 ±45.64 124.88 ±62.98 0.066 Biopsy proved rejection (within 6M) 7(16.7%) 9(21.4%) 0.578 Time to rejection 56.8 ± 15.47 44 ±50.18 0.080 Corticosteroid-resistant rejection 2 (4.8%) 1 (2.4%) 1.000 Severity of rejection (RAI) 5.86 ± 1.06 5.44 ± 0.88 0.495 Graft & Patient survival (days) 549 ± 23.8 499.5 ±31.64 0.329 Overall mortality 4 (9.5%) 7(16.7%) 0.519 90 - day mortality 2 (4.8%) 6(14.3%) 0.293 1-year survival 88.5% 83.3% Mean eGFR at 3rd month (mL/min) 87.25 ±32.31 94.20 ±31.72 0.399 Mean eGIFR at 6th month (mL/min) 93.96 ± 37.11 103.28 ±35.04 0.124 Renal dysfunction 24 (60%) 18(50%) 0.381 (eGFR < 90mL/min) at 3 months Renal dysfunction 24 (60%) 13 (36.1%) 0.037 (eGFR < 90mL’min) at 6 months

(continued on next page)

HPB 2018, 20 (S2), S182eS294 S198 Free Orals (FP01-FP31)

(continued) Variable SF-arm S-arm P-value Overall Infections 12 (28.6%) 14(33.3%) 0.637 Blood borne sepsis 8(19%) 7(16.7%) 0.776 Urinary sepsis 2 (4.8%) 6(14.3%) 0.265 Abdominal sepsis 6(14.3%) 7(16.7%) 0.763 ICU stay 8.86 ± 5.78 10.76 ±8.64 0.042 Total Hospital stay 20.31 ± 10.77 20.14 ±9.46 0.545 Hepatic artery thrombosis 0 (0%) 4 (9.5%) 0.124 Middle hepatic vein thrombosis 3(7.1%) 8(19%) 0.196 Portal vein thrombosis 0 (0%) 1(2.4%) 1.000 Bile leak 3(7.1%) 8(19%) 0.196 Biliary stricture 0 (0%) 5(11.9%) 0.065 Re-admission 12 (28.6%) 17(40.5%) 0.251 Re-exploration 2 (4.8%) 5(11.9%) 0.430

Conclusions: Steroid-free immunosuppression with basi- Conclusions: The combined use of initial perfusion with liximab induction is as effective as the steroid based HTK followed by UW in liver transplantation is safe, re- triple drug regimen with significantly lower metabolic duces LIR and improves early liver function, in relation to complications. isolated perfusion with HTK. The potential benefit on graft preservation should be studied through long-term survival analysis. FP05-05 COMBINED FLUSH WITH HISTIDINE- TRYPTOPHAN-KETOGLUTARATE AND FP06 - Free Papers 6 (long oral) - Biliary: UNIVERSITY OF WISCONSIN Neoplasms SOLUTION IN LIVER FP06-01 RE-EVALUATION OF TRANSPLANTATION: RANDOMIZED CLINICOPATHOLOGICAL CLINICAL TRIAL CHARACTERISTICS OF INTRADUCTAL F. J. León Díaz, J. L. Fernandez Aguilar, B. Sanchez Perez, C. Montiel Casado, J. A. Perez Daga, PAPILLARY NEOPLASM OF THE BILE J. M. Aranda Narvaez, M. A. Suarez Muñoz and DUCT (IPNB): THE MULTI- J. Santoyo Santoyo INSTITUTIONAL STUDY OF THE JAPAN Liver Transplantation Unit, University Regional Hospital, BILIARY ASSOCIATION FOR JAPAN- Spain KOREA COLLABORATIVE STUDY Introduction: Ischemia-reperfusion injury (IRI) is the K. Kubota1, Y. Nakanuma2, N. Fukushima3, main cause of early graft dysfunction (EAD) and subse- T. Furukawa4, Y. Zen5, J. -Y. Jang6, T. Matsumoto1, quent primary graft failure (PAF). The combination of the K. Inui7 and M. Unno8 advantages of a low viscosity solution (Histidine Trypto- 1Second Department of Surgery, Dokkyo Medical Uni- phan Ketoglutarate, HTK) with a higher viscosity solution versity, 2Department of Pathology, Saiseikai Hospital, (University of Wisconsin, UW) could be an alternative to 3Department of Pathology, Jichi Medical University, reduce the LIR. 4Department of Pathology, Tohoku University Graduate Methods: A randomized trial was performed to compare School of Medicine, 5Department of Diagnostic Pathology, outcomes in liver recipients who underwent transplantation Kobe University Graduate School of Medicine, 6Depart- surgery in the University Regional Hospital of Malaga, ment of Surgery, Seol National University Hospital, Re- Spain. Seventy patients were randomized to two groups, public of Korea, 7Department of Gastroenterology, Second perfused with HTK and UW vs. HTK alone. Primary Military Medical University Teaching Hospital Fujita endpoints included IRI, EAD, PAF, reintervention, acute Health University, and 8Department of Surgery, Tohoku cellular rejection (ACR), retransplantation, arterial com- University Graduate School of Medicine, Japan plications (AC), and biliary complications (BC) at 90 Introduction: To clarify clinicopathological characteristics postoperative days. Results: Postoperative GOT (1623.15 Æ 1214.66 UI/L vs of IPNB and re-evaluate pathological diagnostic criteria, 989.13 Æ 822.11 UI/L, p=0.04) and GPT (1211.23 Æ Japan Biliary Association (JBA) and Korean Hepato- 1057.23 U/L vs 750.43 Æ 740.09 U/L, p=0.03) were Biliary-Pancreatic Surgery Association are performing a significantly higher in patients perfused with HTK alone. A collaborative study. Herein, Japan-side data are reported. Methods: clear tendency was observed in recipients perfused with A multi-institutional, retrospective study on pa- HTK alone to present moderate to severe IRI (8 in HTK/ tients with IPNB or papillary cholangiocarcinoma was UW vs 16 in HTK, p=0.08), EAD (10 in HTK/UW vs 25 performed by JBA. Pathological diagnostic criteria were re- HTK, p=0.15), PAF (1 HTK/UW vs 1 in HTK, p=0.22). evaluated by pathologists of Japan and Korea, and new

HPB 2018, 20 (S2), S182eS294 Free Orals (FP01-FP31) S199 diagnostic criteria were proposed: Type-1 IPNB (classical cancer specific survival (CSS) was analyzed. A 1:1 pro- IPNB) is histologically similar to IPMN of the pancreas. pensity score matching was performed to achieve compa- Type-2 IPNB (so-called papillary cholangiocarcinoma) has rable baseline characteristics between groups. more complex histological architecture with irregular papil- Results: The study included 238 patients: 190 (80%) ORR lary branching or with foci of solid-tubular components. and 48 (20%) LRR. Sixteen (33%) LRR patients underwent Results: 469 of 497 cases presented from 44 institutes conversion. Positive liver margin at time of index chole- could be pathologically evaluated: 193 cases were diag- cystectomy and interval between surgeries >60-days were nosed as Type-1 IPNB, 112 cases as Type-2 and 163 cases predictive for conversion. Thirty-five patients in each group were given other diagnoses. Type-1 lesions were observed were eligible for 1:1 matching. Blood loss, operative time, in the liver and perihilar area with a significant difference, number and positive lymph nodes, R1, residual cancer, while Type-2 lesions were observed in the distal bile duct. overall morbidity, Clavien grade III, 90-days mortality and Mucobilia was observed significantly higher in Type-1 le- recurrence patterns were similar. Median hospital stay for sions than in Type-2 lesions. The values of AST, ALT, LRR was significant shorter (4 vs 6 days, p=0.032). CSS in ALP and total bilirubin were significantly higher in Type-2 LRR vs. ORR was 78% vs. 66% at 3 years (p=0.42). In lesions than in Type-1 lesions. Otherwise, there were no multivariate analysis, predictive factors for poor CSS were significant differences. The incidences of IPNB with an residual cancer (HR 5.6; p < 0.001) and blood transfusion associated invasive component and papillary adenocarci- (HR 1.0; p=0.002). noma were significantly higher in Type-2 than in Type-1. 5- Conclusion: LRR for selected patients with IGBC is safe year OS rates of type-1 and Type-2 were 85.3% and 71.6%, and oncologically effective. Dissemination of advanced respectively (p< 0.001). laparoscopic skills and timely referral of IGBC patients to Conclusion: Type-1 and Type-2 IPNB had different clin- specialized centers within 60-days may allow more patients icopathological characteristics with different overall sur- to benefit from this demanding operation. vival rates. It is reasonable to classify IPNB into two types.

FP06-03 BISMUTH-CORLETTE TYPE 4 PERIHILAR CHOLANGIOCARCINOMA IS NO LONGER A CONTRAINDICATION TO SURGICAL RESECTION T. Campagnaro, A. Ruzzenente, S. Conci, G. Di Filippo, S. Manfreda, S. Valcanover, R. Ziello, C. Iacono and A. Guglielmi Surgery, University of Verona, General and Hepatobiliary Figure [OS curve] Surgery Unit, Italy Introduction: Surgical indication for Bismuth-Corlette (BC) type 4 perihilar cholangiocarcinoma (PHC) is still under debate for some authors. We aimed to compare FP06-02 surgical results of PHC type 4 with those of the other types (1-2-3a-3b) after resection with radical intent. SAFE LAPAROSCOPIC ONCOLOGIC Methods: One-hundred-eleven patients submitted to sur- RE-RESECTION OF INCIDENTALLY gery with curative intent for PHC between January 1990 DISCOVERED GALLBLADDER and December 2016 were evaluated. Patients were divided CANCER: A PROPENSITY SCORE in two groups according to tumour extension along intra- fi MATCHING ANALYSIS hepatic bile ducts, as de ned by Bismuth-Corlette classi- fication: group 1 (BC type 1-2-3a-3b) (n=64) and group 2 E. A. Vega1,2, X. De Arextabala3, F. Castillo3, 2,4 3 3 1 (BC type 4) (n=47). Short-term and long-term surgical re- M. Sanhueza , G. Rencoret , M. Vivanco , J. E. Lee , sults were evaluated. J. -N. Vauthey1, E. Vinuela2,4 and C. Conrad1 1 Results: The two groups of patients had similar clinical and Sugical Oncology, UT MD Anderson Cancer Center, pathological characteristics. Major liver resection and United States, 2Digestive Surgery, Hepato-Bilio-Pancre- 3 caudate lobe resection were performed more frequently in atic Surgery Unit, Hospital Sotero del Rio, Digestive group 2 than in the other: 91.5% vs 64.1% (p=0.001) and Surgery, Hepato-Bilio-Pancreatic Surgery Unit, Clinica 4 85.1% vs 62.5% (p=0.007), respectively. Lymph-node Alemana, and Cirugia General, Hepato-Bilio-Pancreato dissection (96.9% vs 95.7%; p=0.56), radical resection rate fi Unit, Ponti cia Universidad Católica de Chile, Chile (69.2% vs 59.6%; p=0.50), incidence of positive lymph Introduction: Evidence regarding the safety of laparo- nodes (46% vs 54%; p=0.255), median postoperative hos- scopic re-resection (LRR) of incidentally discovered gall- pital stay (16 vs 17 days; p=0.7) and incidence of global bladder cancer (IGBC) is limited. Concerns exist regarding postoperative complications (40.7% vs 48.9%; p=0.27) did inadequacy of extended of oncologic resection and safety. not significantly differ between group 1 and group 2. This study aims at a comparative analysis of LRR vs. open Overall survival was not statistically different between re-resection (ORR) with consideration for selection bias. group 1 and group 2 (26 vs 30 months; p=0.47) as well as Methods: Patients with IGBC who underwent re-resection recurrence free survival (21 vs 24 months; p=0.48). with curative intent at 3 US and Chilean centers between Conclusion: Resection of perihilar cholangiocarcinoma BC 1999-2017 were analyzed. In this intention to treat analysis type 4 has similar short-term and long-term surgical results

HPB 2018, 20 (S2), S182eS294 S200 Free Orals (FP01-FP31) when compared to the other BC types, though an accurate 6Ospedale San Raffaele, Italy, 7Emory University Hospi- preoperative evaluation and extended resection is mandatory. tal, United States, 8Beaujon Hospital, France, and 9Erasmus University Medical Centre, The Netherlands Background: Intrahepatic cholangiocarcinoma (ICC) is FP06-04 morphologically classified as mass forming (MF), peri- OUTCOME OF PERIOPERATIVE CARE ductal infiltrating (PI) or intraductal growth (IG) type. We OF PERIHILAR sought to investigate the characteristics and long-term outcomes of IG patients compared with MF and PI ICC. CHOLANGIOCARCINOMA IN THE Methods: 1,083 patients who underwent curative-intent NEDERLANDS liver resection for ICC between 1990-2015 were identified. E. Roos1, R. J. Coelen1, M. G. W. Dijkgraaf2, Data on clinico-pathological characteristics, operative de- T. M. van Gulik1 and DRAINAGE Trial Study Group tails, and morphological status were analyzed. 1Surgery, Academic Medical Center Amsterdam, and Results: ICC morphology consisted of MF (n=911, 84%), 2Clinical Research Unit, Academic Medical Center PI (n=142, 13%), or IG (n=30, 3%) subtypes. Among pa- Amsterdam, The Netherlands tients with IG ICC, 63% has a T1a/T1b tumor and the Introduction: Liver resection for perihilar chol- majority of patients (67%) were node negative. Most pa- angiocarcinoma (PHC) is associated with high periopera- tients with MF (52%) or PI (78%) had more advanced T2/ tive morbidity and mortality. This is mainly due to the T3/T4 disease; the incidence of lymph-node metastasis > biliary obstruction caused by the tumor and subsequent among MF (34%) or PI (59%) was comparable (p 0.05). fi cholestasis of the liver. We recently finished the On nal pathology, the incidence of R0 margins was also DRAINAGE trial: a randomized controlled trial on the similar among IG (93%) and MF (89%)(p=0.46), yet lower optimal method of preoperative biliary drainage (PBD). We among patients with PI (77%)(p=0.04). Lympho-vascular observed an uncommonly high perioperative mortality rate invasion (IG, 13% vs. MF, 29% vs. PI, 46%) was lower < in this study. Therefore we decided to analyze perioperative among patients with IG (p 0.001). In addition, the inci- outcome of patients that were not included in the trial. dence of perineural invasion was also lower among IG Methods: Patients with a potentially resectable PHC that patients (IG, 17% vs. MF, 18% vs. PI, 38%; p=0.03). While were treated between June 2013 and June 2016 were 5-year overall survival was comparable among patients selected. Demographics, morbidity and mortality were with IG (41%) and MF (42%), long-term outcome was calculated. These results were compared to those of the lower among patients with PI (25%)(p=0.04). trial. Results: We selected 100 patients. Of these 67 underwent preoperative biliary drainage. The majority of these patients (44/67) had an endoscopic biliary stent. Of all patients, 87 underwent a resection, 41% of which were extended re- sections. Vascular reconstruction was performed in 26%. Of all patients 50% had Clavien-Dindo grade 3 or higher postlaparotomy complications. Biliary leakage and intra- abdominal abscess were the most frequent complications. Postlaparotomy 90-day mortality was 10%. Discussion: Postoperative mortality from the DRAINAGE trial was unexpectedly high: 26%. This was especially increased in the PTBD group. High mortality has been attributed to bile loss and subsequent liver impairment by external bile drainage through percutaneous drains. Endo- Figure scopic drainage was more frequently used in the national, not-included, cohort. The outcome of our national cohort is Conclusion: Patients with IG tended to present with earlier comparable to other Wester series in terms of morbidity and T-stage disease, yet had the same overall incidence of nodal mortality. metastasis. Long-term outcomes were comparable to the most common MF subtype, yet was better than PI. FP06-05 LONG-TERM OUTCOMES OF PATIENTS WITH INTRADUCTAL GROWTH SUB- FP06-06 TYPE OF INTRAHEPATIC CIRCULATING MICRORNAS CHOLANGIOCARCINOMA DISCRIMINATE F. Bagante1,2, M. Weiss3, S. Alexandrescu4, H. Marques5, CHOLANGIOCARCINOMA FROM L. Aldrighetti6, S. Maithel7, O. Soubrane8, PANCREATIC CANCER 9 2 1 B. G. Koerkamp , A. Guglielmi , T. Pawlik and L. L. Meijer1, J. R. Puik1, T. Y. S. Le Large1,2, M. Heger3, Intrahepatic Cholangiocarcinoma Study Group F. Dijk4, N. Funel5, I. Garajová6, E. Giovannetti2,5 and 1 Department of Surgery, Ohio State University Wexner G. Kazemier1 2 Medical Center, United States, Department of Surgery, 1Cancer Center Amsterdam, Department of Surgery, VU 3 University of Verona, Italy, Department of Surgery, Johns University Medical Center, 2Cancer Center Amsterdam, 4 Hopkins Hospital, United States, Fundeni Clinical Insti- Departement of Medical Oncology, VU University Medical 5 tute, Romania, Curry Cabral Hospital, Portugal, Center, 3Experimental Surgery, Academic Medical Center

HPB 2018, 20 (S2), S182eS294 Free Orals (FP01-FP31) S201

Amsterdam, 4Pathology, Academic Medical Center fistula. Three-dimensional(3D) vision may especially Amsterdam, 5Cancer Pharmacology Lab, AIRC-Start-Up, improve surgical performance during construction of the Department of Translational Research and The New pancreaticojejunostomy(PJ) and hepaticojejunostomy(HJ). Technologies in Medicine and Surgery, University of Pisa, Yet, data on the added value of 3D vision are lacking. and 6Medical Oncology, University of Bologna, Italy Methods: We performed an experimental randomized Introduction: Accurate diagnosis of distal chol- controlled cross-over trial including 20 expert laparoscopic angiocarcinoma (CCA) and pancreatic ductal adenocarci- surgeons and 20 surgical residents capable of laparoscopic noma (PDAC) is a challenge with clinical consequences. suturing. Participants performed a PJ and a HJ twice, using Both are lethal malignancies with distinct therapeutic op- 3D- and 2D-laparoscopy on artificial organ models(LifeLike tions. This study aimed to identify a circulating microRNA BioTissueÒ) according to the Pittsburgh method. Partici- (miRNA) signature to diagnose and discriminate distal pants were randomized for the sequence of laparoscopy, i.e., CCA from PDAC. 3D or 2D first. Primary endpoint was the time needed to Methods: In the discovery phase, microarray profiling of complete both procedures. Secondary endpoints were sur- 752 miRNAs was performed on plasma samples of seven gical performance, assessed using a modified Birkmeyer patients with distal CCA and seven age- and sex-matched scoring platform, and side-effects of 3D-laparoscopy. healthy controls. Significant candidate miRNAs were Results: A total of 71 HJs and 70 PJs were completed. The selected for validation based on predefined selection criteria. use of 3D-laparoscopy reduced the time to complete the PJ In the validation phase, these miRNAs were analyzed by RT- and HJ with 16.9%, from 84.0 to 68.3 minutes (median 15.1 qPCR in an independent cohort of healthy controls (n=32), minutes, CI -6.5- -24.3 min, P=0.002). Performance scoring benign periampullary disease (n=15), distal CCA (n=22), increased with 3D-laparoscopy (mean 0.78 increase in 5- PDAC (n=30), and other hepatobiliary malignancies (n=38). point Likert scale summary score, P=0.001). This effect was Data were normalized to a combination of two reference also seen in the subgroup of experts, including an increase in genes. The optimal diagnostic combination of miRNAs was summary score. Of all participants, 97% stated to prefer 3D- determined by logistic regression. Sensitivity and specificity laparoscopy, whereas 24.3% reported minor side effects, were evaluated by ROC curves and AUC values. and 5.4% severe side effects (e.g. eye strain). Results: In the discovery phase, 19 miRNAs were signif- Conclusion: 3D-laparoscopy, as compared to 2D, icantly deregulated in patients with distal CCA compared to demonstrated substantial reduction in procedure time of healthy controls. In the validation phase, 13 candidate completing both PJ and HJ in bio-tissue, and improved miRNAs were considered. A three-miRNA panel was results in surgical performance grades for both experts and constructed with the most robust signature to discriminate trained residents. malignant from benign disease (AUC=0.881). The combi- nation of two of these miRNAs was able accurately distinguish distal CCA from PDAC, with an AUC of 0.814. FP07-02 Conclusion: This is the first study to identify a combined ESTABLISHING ERAS IN panel of plasma miRNAs which shows promising diag- PANCREATICODUODENECTOMY IN A nostic capability to serve as distal CCA signature with the potential to discriminate distal CCA from PDAC. TERTIARY CENTRE: A 9-YEAR RETROSPECTIVE ANALYSIS S. Koek1,2 and M. Ballal1,2 FP07 - Free Papers 7 (long oral) - HPB 1 1Fiona Stanley Hospital,, and 2The University of Western FP07-01 Australia, Australia ADDITIONAL VALUE OF 3D-VISION Introduction: Recent studies have demonstrated that DURING LAPAROSCOPIC enhanced recovery after surgery (ERAS) protocols in pancreaticoduodenectomy (PD) may decrease morbidity PANCREATODUODENECTOMY BIO and length of stay. This retrospective study critically TISSUE DRILLS (LAELAPS 3D2D): A assessed the implementation of ERAS in patients who have RANDOMIZED CONTROLLED CROSS- undergone PD since 2011. OVER TRIAL Methods: A retrospective review of all patients who un- derwent a PD from 2009 to 2017 at Fremantle Hospital and M. J. W. Zwart1, J. van Hilst2, I. P. Fuente3, T. de Rooij2, Fiona Stanley Hospital in Western Australia was assessed. S. van Dieren2, L. B. van Rijssen2, M. P. Schijven1, The implementation of ERAS since 2011 was critically O. R. C. Busch1, M. D. Luyer4, D. J. Lips5,6, assessed based on outcome measures of length of stay, S. Festen7 and M. G. H. Besselink1 morbidity and mortality during the pre-ERAS, transition, 1Surgery, University of Amsterdam, Cancer Center early ERAS and established ERAS period. Amsterdam, Academic Medical Center, 2University of Results: One hundred and ninety-six patients underwent a Amsterdam, Cancer Center Amsterdam, Academic Medi- PD between 2009-2017 (pre-ERAS, n=29; transition, cal Center, The Netherlands, 3Hospital Italiano de Buenos n=14; early ERAS, n=53; established ERAS, n=73) with a Aires, Argentina, 4Surgery, Catharina Hospital, 5Surgery, mean age (Æ SD) of 64 Æ 11.3 years. ERAS significantly Jeroen Bosch Hospital, 6Surgery, Medisch Spectrum increased the proportion of patients reaching the target Twente, and 7Surgery, Onze Lieve Vrouwen Gasthuis, The length of stay and reduced mortality (p < .05). Delayed Netherlands gastric emptying (DGE) significantly reduced from 82.8% Background: Laparoscopic pancreatoduodenecto- pre-ERAS to 45.1% in the early ERAS phase (p < .013). my(LPD) is a technically challenging procedure with There was no significant impact on pancreatic fistula, ileus, concerns about higher rates of postoperative pancreatic infection or haemorrhage (p > .05).

HPB 2018, 20 (S2), S182eS294 S202 Free Orals (FP01-FP31)

Conclusion: ERAS was safe with success in reducing rates Mauriziano Umberto I, 4Ospedale San Camillo Forlanini, of DGE in the early phase, increasing the odds to reach 5Ospedale Universitario GB Rossi, 6Policlinico Gemelli, target LOS and reducing mortality. Our findings support and 7Ospedale Pinetagrande, Italy the continued development of ERAS in PD as an oppor- Background: The I Go MILS prospective registry was tunity to standardise care, reduce morbidity and the po- established in 2014 with the goals to create a national tential to reduce health care costs. hub for data and projects and to promote the diffusion and implementation of MILS programs. The present study gives a snapshot of the real diffusion and out- FP07-03 comes of MILS in Italy, while analyzing the role of the A COMPLEXITY-BASED AND STEPWISE registry in implementation of MILS programs LEARNING CURVE IN 607 nationwide. Material and methods: The I Go MILS Registry is a LAPAROSCOPIC LIVER RESECTIONS: prospective and intention-to-treat registry opened to any A SINGLE SURGEON ANALYSIS TO Italian center performing MILS, without restriction criteria. DEFINE A STANDARD FOR TECHNICAL The Registry is based on 34 clinical variables, regarding IMPROVEMENT indication, intra- and postoperative course. Results: Between November 2014 and December 2017, L. Aldrighetti, F. Ratti, F. Cipriani and G. Fiorentini data from 2081 MILS performed in 50 centers have been Hepatobiliary Surgery Division, IRCCS San Raffaele collected. 22% of procedures were performed for benign Hospital, Italy and 78% for malignant disease (HCC constituted 48% and Importance: The standard educational path to complete the CRLM 32.6% of malignant tumors). Major liver resections learning curve in laparoscopic liver resections (LLR) should be were 11% of the series. Mean blood loss was 200 Æ 230 fi de ned taking into account the degree of surgical complexity, mL Morbidity rate was 20.4% and mortality was 0.3%. fi to maintain an adequate pro le of safety until its completion. 9.6% of cases were converted to open approach. Median fi Objective: To de ne a single surgeon learning curve in a length of stay was 5 days. MILS/total resections ratio in 13 series of procedures grouped according to their complexity experienced centers increased from 14 to 30% after Reg- fi fi calculated by dif culty index to de ne a standard for istry establishment. technical improvement. Conclusion: MILS programs are well established in Design: 607 LLR performed by a single surgeon (2005- Italy, with progressive increase in number of cases and in fi fi fi 2017) were strati ed by dif culty scores: Low-Dif culty numerosity of centers. The I Go MILS Registry is fi (LD, n=227); Intermediate-Dif culty (ID, n=197) and playing a crucial role in monitoring the development of fi High-Dif culty (HD, n=183). The learning curve effect MILS in the real world on a national basis while giving a was analyzed in the three Groups using the Cumulative significant contribution to the implementation of MILS Sum (CUSUM) method taking into consideration the ex- programs. pected risk of conversion. Results: Along with increasing numerosity and complexity of cases, the ratio laparoscopic/total liver resections increased from 5.8% (2005) to 72.8% (2017). The CUSUM analysis per FP07-06 group showed that the average value of the conversion rate AUTOLOGOUS VEIN GRAFT FOR was reached at the 50th case in the LD Group and at the 15th VASCULAR RECONSTRUCTION IN in the ID and HD groups. The evolution from LD to ID and PANCREATIC SURGERY: STRATEGY HD procedures occurred only when learning curve in LD TO REDUCE COMPLICATION AND fl resections was concluded. Re ecting different degree of PORTAL VEIN CLAMP TIME complexity, procedures showed significantly different blood loss, morbidity and conversions among groups. A. Maki Conclusion: A standard educational model - stepwise and Hepato-Pancteato-Biliary Surgery, Saitama Medical Uni- progressive - is mandatory to allow surgeons to define the versity Medical Center, Japan technical and technological background to deal with a specific Aim: Feasibility of autologous vein graft (AVG) for degree of difficulty, providing an help in the definition of vascular reconstruction is widely accepted in pancreatic indications to laparoscopic approach in each phase of training. surgery. However, portal vein (PV) reconstruction using AVG had brought longer PV clamp time and larger blood FP07-04 loss (EBL). We here report our series of PV reconstruction using AVG and our strategy for reducing PV clamp time DIFFUSION, OUTCOMES AND and EBL. IMPLEMENTATION OF MINIMALLY Patients: 298 patients underwent pancreatic surgery since INVASIVE LIVER SURGERY: A Jan. 2013 to Nov. 2017 in our facility. 70 patients had PV SNAPSHOT FROM THE I GO MILS reconstruction. AVG was applied for PV reconstruction in (ITALIAN GROUP OF MINIMALLY 17 patients (AVG), 53 patients were directly anastomosed (DA). Operative time (OR time), EBL and PV clamp time INVASIVE LIVER SURGERY) REGISTRY were retrospectively reviewed. 1 1 2 3 L. Aldrighetti , F. Ratti , U. Cillo , A. Ferrero , Results: 12 left renal vein, 3 right iliac vein, 1 splenic 4 5 6 7 G. M. Ettorre , A. Guglielmi , F. Giuliante , F. Calise and vein and 1 superficial femoral vein (SFV) were utilized I Go MILS as AVG. Mean OR time and EBL were not different 1 Hepatobiliary Surgery Division, IRCCS San Raffaele between AVG and DA. PV clamp time was significantly 2 3 Hospital, Policlinico Universitario di Padova, Ospedale longer in AVG. PV obstruction was observed in 14 out

HPB 2018, 20 (S2), S182eS294 Free Orals (FP01-FP31) S203 of 53 patient from DA, whereas 3 out of 17 patients from 75.5%, and 42.6%, and 77.0%, 56.4%, and 28.4%, both P AVG. Temporary renal insufficiency was observed in 2 < 0.001). Multivariable analyses revealed that preopera- out of 12 patients who procured left renal vein. AVG tive prealbumin level, but not albumin level, was an in- anastomosis to distal SMV before dissecting PV collat- dependent predictor of OS (HR, 1.789; 95% CI: 1.544 eral flow to prepare porto-caval shunt was devised to -2.072, P < 0.001) and RFS (HR, 1.420; 95% CI: 232- reduce intestinal congestion. 1.636, P < 0.001). Conclusion: PV reconstruction using AVG was acceptable Conclusions: Preoperative prealbumin level is useful for in terms of vascular patency. However, appropriate graft predicting long-term prognosis in patients undergoing liver selection should be considered to avoid complications. resection. Prealbumin may be suitable to displace albumin, AVG anastomosis to distal SMV prior to collateral vein yielding to an updated Child-Pugh grade for accessing liver dissection may be acceptable strategy to reduce PV clamp function. time and EBL.

FP08 - Free Papers 8 (mini oral) - Liver: Surgical Outcomes 1 FP08-01 PREOPERATIVE PREALBUMIN LEVEL AS AN INDEPENDENT PREDICTOR OF LONG-TERM PROGNOSIS AFTER CURATIVE LIVER RESECTION OF HEPATOCELLULAR CARCINOMA: A MULTICENTER STUDY OF 1,483 PATIENTS J. D. Li1,21, J. Han11,H.Wu11, Y. H. Zhou31,W.M.Gu41, H. Wang51, T. H. Chen61, Y. Y. Zeng71, F. Shen11 and T. Yang11 1Department of Hepatobiliary Surgery, Eastern Hepato- biliary Surgery Hospital, Second Military Medical Uni- versity, 2Department of Clinical Medicine, Second Military Medical University, 3Department of Hepatobili- ary Surgery, Pu’er People’sHospital,4First Department of General Surgery, Fourth Hospital of Harbin, Figure 5Department of General Surgery, Liuyang People’s Hospital, 6Department of General Surgery, Ziyang First People’s Hospital, and 7Department of Hepatobiliary Surgery, Mengchao Hepatobiliary Hospital, Fujian FP08-02 Medical University, China IS PREOPERATIVE RED BLOOD Introduction: Serum prealbumin is more sensitive to profile nutritional status and liver function than albumin, CELL DISTRIBUTION WIDTH (RDW) which could hardly be affected by infusion supplement. A PROGNOSTIC FACTOR FOR This study aims to identify the relationship between pre- SURVIVAL AFTER RESECTION OF operative prealbumin level and the long-term prognosis HEPATOCELLULAR CARCINOMA? after curative resection of hepatocellular carcinoma T. Campagnaro, A. Ruzzenente, S. Conci, G. Lazzari, (HCC). M. De Angelis, E. Lombardo, A. Dorna, A. Guglielmi and Methods: Patients undergone HCC curative resection be- C. Iacono tween 2001 and 2014 at six institutions in China were Surgery, University of Verona, General and Hepatobiliary enrolled. By using 170 mg/dl as cut-off value of serum Surgery Unit, Italy prealbumin level, these patients were divided into the low and normal preoperative prealbumin groups. The overall Introduction: Red blood cell distribution width (RDW) is fl fi survival (OS) and recurrence-free survival (RFS) were an in ammation biomarker with prognostic signi cance for analyzed and compared. Univariable and multivariable survival in different types of tumor. We sought to evaluate Cox-regression analyses were performed to identify pre- the preoperative value of RDW as a prognostic factor for dictive factors of OS and RFS. survival after radical resection for hepatocellular carcinoma Results: Among 1,483 patients, 437 (29.5%) had a low (HCC). prealbumin level within a week before surgery. The 1-, 3-, Methods: A retrospective analysis of 223 patients sub- and 5-year OS and RFS rates of patients in the low mitted to curative resection for HCC between 2006 and prealbumin group were 83.8%, 57.0%, and 31.1%, and 2016 was performed to highlight the relationship between 67.0%, 39.8%, and19.9%, respectively, which was preoperative serum RDW and survival. significantly poorer than those in the normal group (93.0%, Results: Patients were divided in two groups according to preoperative RDW value: 121 with RDW 14.5% and 102 with RDW >14.5%. The two groups were similar All authors contributed equally to this work. with regard to preoperative clinical (aetiology of

HPB 2018, 20 (S2), S182eS294 S204 Free Orals (FP01-FP31) cirrhosis, Child-Pugh stage, BCLC stage, alfa-fetopro- postoperative complications and ascites in patients tein) and pathological characteristics (tumour diameter, with CRLM, which was also significant in multivariate vascular invasion, grading) (p>0.05). However, in pa- analysis. Pathological examination revealed chemo- tients with high RDW the incidence of multifocal HCC therapy related liver damage in high ALICE grade was higher (23.7% vs 9.3%; p 0.007). Median overall CRLM patients. survival was significantly longer in patients with low Conclusions: This new grading system is a simple method RDW (75 vs 42 moths, p=0.008). RDW (p=0.008), for prediction of the postoperative short-term outcomes in aspartate aminotransferase (AST) (p=0.001), gamma both HCC and CRLM patients. glutamiltransferase (GGT) (p=0.001), platelets (PTLs) (p=0.006), albumin serum level (p=0.001), BCLC stage (p=0.006) and multifocal tumor (p=0.004) resulted as FP08-04 prognostic factors for survival at univariate analysis. THE 3e60 CRITERIA CHALLENGE RDW (HR=1.78, p=0.03), AST (HR=2.36, p=0.003), PTLs (HR=0.33, p=0.02) and multifocality (HR=2.43, ESTABLISHED PREDICTORS OF p=0.02) were confirmed as prognostic factors for survival POSTOPERATIVE MORTALITY AND at multivariate analysis. ENABLE TIMELY THERAPEUTIC Conclusion: Preoperative high level of RDW is an inde- INTERVENTION AFTER LIVER pendent negative prognostic factor for survival after sur- RESECTION gical resection for HCC. RDW could be a novel prognostic G. Gyoeri1, D. Pereyra1, F. Offensperger1, factor for a better preoperative selection of the patients to 2 1 1 submit to surgical treatment. F. Klingelmüller , R. Baumgartner , S. Holzer , M. Gnant1, T. Grünberger3 and P. Starlinger1 1Dept. of Surgery, Medical University of Vienna, 2Dept. for Statistics, Informatics, Intelligent Systems, Medical FP08-03 University of Vienna, and 3Dept. of Surgery, KH Rudolf- ASSESSMENT OF PREOPERATIVE stiftung, Austria LIVER FUNCTION IN PATIENTS Background: To date, definitions of liver dysfunction UNDERGOING LIVER RESECTION e (LD) after hepatic resection rely on late postoperative time THE ALICE GRADE points. Further, the used parameters are markedly influ- enced by perioperative management. Thus, we aimed to T. Kokudo, C. Shirata, T. Toda, Y. Nishioka, evaluate a very early postoperative score for prediction of T. Yamaguchi, J. Arita, J. Kaneko, N. Akamatsu, postoperative mortality. Y. Sakamoto and K. Hasegawa Methods: Liver related parameters were evaluated after Hepato-Biliary-Pancreatic Surgery Division and Artificial liver resection in a retrospective evaluation cohort of 228 Organ and Transplantation Division, and Department of colorectal cancer patients with liver metastasis (mCRC) and Surgery, Graduate School of Medicine, University of subsequent validation in a prospective set of 177 consec- Tokyo, Japan utive patients undergoing hepatic resection was performed. ’ Objective: Precise preoperative evaluation of the patient s Results: C-reactive protein (CRP,AUC=0.739,P< 0.001) liver function is essential for surgical decision making in and antithrombinIII-activiy. patients with hepatocellular carcinoma (HCC) and colo- (ATIII,AUC=0.844,P< 0.001) on the first post- rectal liver metastasis (CRLM). The aim of the study is to operative day (POD) were found to significantly predict validate the usefulness of the new preoperative liver func- postoperative LD. Cut-off values for CRP at 3 mg/dL tion evaluation system, Albumin-Indocyanine Green and for ATIII at 60% were found to significantly identify Evaluation (ALICE) grading system, for predicting the high-risk patients for postoperative LD and mortality (P< postoperative outcomes after liver resection in both HCC 0.001) and thus defined the 3-60 criteria on POD1. The and CRLM patients. 3-60 criteria showed superior sensitivity and specificity Methods: The ALICE grading system, incorporating only compared to established criteria for LD (3-60 criteria: serum albumin level and the indocyanine green retention total positive patients: 26 patients [70% mortality detec- rate at 15 minutes (ICG R15), was developed based on the ted], odds ratio (OR):48.8; ISGLS: total positive patients: overall survival of 1868 HCC patients: linear predictor = 43 [70% mortality detected], OR:23.3; Peak 7: total   0.663 log10ICG R15 (%) - 0.0718 albumin (g/L) (cut- positive patients: 9, [30% mortality detected], OR:27.8; off value: -2.20 and -1.39). We analyzed the predictive 50-50: total positive patients: 9 [30% mortality detected], power for the postoperative short-term outcome after liver OR:27.8; P< 0.001, respectively). Further, only the 3-60 resection in 1025 patients with HCC and 348 patients with criteria (OR:50.4) and peak 7 (OR:10.3) remained inde- CRLM. pendent predictors of postoperative mortality upon Results: Determination of the ALICE grade allowed multivariable analysis. better prediction of the risk of postoperative liver fail- Conclusions: The 3-60 criteria on POD1 predict post- ure than the previously reported model based on the operative LD and mortality after liver resection with a presence/absence of portal hypertension (PH). A com- comparable or better accuracy than established criteria, bination of ALICE grade and PH further predicted the allowing for early identification of high-risk patients risk of postoperative liver failure. Determination of the already on POD 1. ALICE grade also allowed prediction of the risk of

HPB 2018, 20 (S2), S182eS294 Free Orals (FP01-FP31) S205

FP08-05 in 3,399 (6.7%) patients. Propensity matching produced 2,336 well-matched pairs. ROLE OF OCTREOTIDE IN PORTAL Results: After matching, the overall incidence of major INFLOW MODULATION AND complication was comparable among patients underling PROPHYLAXIS OF LR+C versus LR (2.2% vs. 2.3%, p=0.64). However, the POSTHEPATECTOMY LIVER FAILURE. incidence of certain complications was higher among pa- PROSPECTIVE CLINICAL STUDY tients who underwent LR+C vs. LR including sepsis(2.1% vs. 0.5%), surgical site infection (5.5% vs. 10.4%,) and R. Alikhanov, M. Efanov, P. Kim, I. Kazakov, renal failure(5% vs. 3.2%)(all p< 0.001). In turn, overall A. Vankovich, S. Iskhagi and V. Subbotin mortality was slightly higher among patients who under- fi HPB Surgery Department, Moscow Clinical Scienti c went LR+C versus LR(2.3% vs. 1.4%, p< 0.001). Hospital Center, Russian Federation length-of-stay (LOS) was also longer among LR+C patients Background: High portal vein pressure (PVP) after major by an average of 2 days (p< 0.001). The incremental cost hepatic resection(MHR) may increase risk of post- associated with a concomitant colon resection was $+6,335 hepatectomy liver failure (PHLF).Whether octreotide may (95% Confidence interval(CI), $4,644 to $8,026). The decrease inflow PVP and risk of PHLF remains unclear and additional costs were partially attributable to additional investigated prospectively in this study. length-of-stay (average cost of each additional LOS day: Methods: 90 patients with hepatic and bile ducts malig- $+1,821) and (p< 0.001). In addition, among patients who nancies after MHR were prospectively included in this experienced at least one major complication, the incre- study and divided in two branches: first (from 2016 to 2017 mental cost attributable to colon resection was $+11,537 year) - comparative analysis of 40 patients who received (95% CI $5,821 to $17,253, p< 0.001). octreotide starting intraoperatively and continuous post- Conclusions: Concomitant colon resection with hepatec- operatively 3-5 days in dose 800-1000 microgram per day tomy was associated with a slightly higher incidence of and control group who do not received octreotide. The certain complications including infection, as well as overall dynamics of ALT, AST, bilirubin, albumin, INR, serum mortality. The costs associated with LR+C versus LR were lactate and type of PHLF (ISGLS) were assessed. Second modestly higher, yet increased significantly in the setting of (from 2017-2018) -30 patients after MHR. The inflow PVP a complication. were assessed just after hepatectomy and through one hour after applied octreotide in dose 300 microgram bolus intravenously. FP08-07 Results: There were no difference postoperatively in dy- namics of laboratory data and PHLF in both groups. Mor- LYMPHOID INFILTRATE PREDICTS tality was higher in group that do not received octreotide: PROGNOSIS OF MASS-FORMING 5% and 0% (p< 0,05). The inflow PVP decreased average INTRAHEPATIC in 2 mmHg (range from 0-7) after octreotide infusion. A CHOLANGIOCARCINOMA significant reduction of inflow PVP was marked if the UNDERGOING COMPLETE LIVER level exceeded 16 mmHg. One patient with PVP 23 mmHg RESECTION was resistant to reduction of portal flow after octreotide 1 1 2 3 applying and died on 20th postoperative day due to type C L. Viganò , C. Soldani , A. Lleo , L. Di Tommaso , 1 1 1 PHLF. B. Franceschini , M. Cimino , M. Donadon and 1 Conclusion: Octreotide could be considered as an efficient G. Torzilli 1 2 pharmacological portal vein modulation and prophylaxis of Hepatobiliary and General Surgery, Internal Medicine 3 PHLF. Resistance to reduction of portal vein pressure by and Hepatology, and Pathology, Humanitas University, octreotide may be indication for surgical modulation of Humanitas Clinical and Research Center, Milan, Italy portal flow. Introduction: The lymphoid infiltrate has shown a prog- nostic impact in several tumors. No study focused on intrahepatic mass-forming cholangiocarcinoma (MFCCC). FP08-06 Methods: All consecutive patients undergoing surgery for INCREMENTAL COSTS AND MFCCC between 2005 and 2015 were considered. The inclusion criteria were complete resection and follow-up OUTCOMES OF HEPATECTOMY WITH 12 months. Patients with operative mortality were VS WITHOUT COLON RESECTION FOR excluded. Tissue sections from MFCCC were immuno- METASTATIC CANCER stained for CD3+, CD4+, CD8+, Foxp3+ and CD68+. J. Idrees, F. Bagante, Q. Chen, K. Merath, O. Akgul, Results: Overall, 53 patients were analyzed. MFCCC were > J. Cloyd, M. Dillhoff, C. Schmidt and T. Pawlik multiple in 10 patients, 50 mm in 26, and N+ in 12. At fi Ohio State University, Wexner Medical Center, United univariate analysis, the following lymphoid in ltrate values > States had a prognostic impact: CD3+ 0.10% (overall survival, OS p< 0.001, recurrence-free survival, RFS p< 0.001); Introduction: We sought to determine incremental costs CD8+ >0.10% (OS p=0.044, RFS p=0.001), CD4+ and compare perioperative outcomes of LR(Liver-Resec- >0.30% (OS p=0.094, RFS p=0.009), and Foxp3+ present tion) versus LR+C(Liver+Colon-Resection) among pa- (OS p=0.097). At multivariable analysis, CD3+ value was a tients with colorectal liver metastasis(CRLM). prognostic factor of OS and RFS [>0.10%, 5-year OS Method: The Nationwide Inpatient Sample(NIS) was used 66.9% vs. 18.2% if 0.10%, HR=0.287, p=0.049; 3-year to identify 50,397 patients who underwent a hepatectomy RFS 48.1% vs. 9.1%, HR=0.232, p=0.001] and Foxp3+ between 2002 and 2011 for CRLM. LR+C was performed was a prognostic factor of OS [present, 5-year OS 21.4%

HPB 2018, 20 (S2), S182eS294 S206 Free Orals (FP01-FP31) vs. 61.9% if absent, HR=2.924, p=0.044].CD3+ values collections) and renal impairment, were also significant. stratified prognosis in T1 patients (5-year OS 73.9%/ Statins remained as a significant(p< 0.01) variable on 14.3%, p< 0.001; 3-year RFS 60.8%/14.3%, p< 0.001), in multivariate analysis. N+ patients (OS 71.4%/0%, p=0.028; RFS 42.9%/0%, Conclusion: Peri-operative use of statins were shown to p=0.011) and in patients without lymph-node metastases have a protective effect on the incidence of PHLI in this (RFS 49.7%/20.0%, p=0.062). observational study. Its role as a modifiable factor of PHLI Conclusions: The lymphoid infiltrate impacts prognosis of needs further investigation. MFCCC after complete surgery. CD3+ infiltrate is associ- ated with higher survival and lower recurrence risk, while Foxp3+ is associated with worse prognosis. CD3+ infiltrate FP08-09 fi allows to re ne prognosis in early tumors and across THE USE OF NEOADJUVANT different N stages. RADIOEMBOLIZATION PRIOR TO FP08-08 RESECTION OF HEPATIC IS PREOPERATIVE STATIN THERAPY MALIGNANCIES L. Tsamalaidze1,2, J. Stauffer1, J. LeGout3, K. Croome4, ASSOCIATED WITH REDUCED POST- 3 3 1 3 HEPATECTOMY LIVER G. Frey , D. Sella , H. Asbun , R. Paz-Fumagalli , J. Burns4 and B. Toskich3 INSUFFICIENCY? 1Surgery, Mayo Clinic, United States, 2Tbilisi State Med- B. V. Dasari, A. Pathanki, R. Marudanayagam, ical University, Georgia, 3Radiology, Mayo Clinic, and M. Abradelo, K. Roberts, D. Mirza, J. Isaac, 4Transplant Surgery, Mayo Clinic, United States R. Sutcliffe and P. Muiesan Introduction: Arterial radioembolization with yttrium-90 Queen Elizabeth Hospital Birmingham, United Kingdom (Y-90) has demonstrated efficacy for hepatic malignancies. Background: Post-hepatectomy liver insufficiency (PHLI) Little is known about use of Y-90 in a neoadjuvant fashion. after elective liver surgery is primarily influenced by the We describe our use of unilobar Y-90 therapy for liver extent of resection in addition to several other modifiable directed treatment and hepatic conditioning in preparation and non-modifiable factors. Preoperative use of statins is for resection. known to improve the postoperative outcomes following Methods: 14 patients undergoing Y-90 therapy with cardiac, and non-cardiovascular surgery. There is evidence subsequent surgical resection from 2007 to 2017 were that statins may increase hepatic angiogenesis and increased reviewed. Patients underwent mapping angiography with liver regeneration response. This study investigated if the angiosomal analysis via cone beam CT and hepatic arte- use of statins would have reduced incidence of PHLI. rial Tc-MAA SPECT CT. Dosimetry was accomplished Methods: Patients who underwent hepatectomy between using MIRD and BSA for glass and resin spheres January 2013 and January 2017 were included. ISGLF respectively. definition was used to identify patients with PHLI. The Results: Patient descriptions are given in Table 1. The extent of liver resection was categorised as minor, major mean age was 62.7 and included hepatocellular carcinoma, (right and left hepatectomy) and extra major hepatectomy. cholangiocarcinoma, and liver metastasis in 5 (36%), 3 Univariate and multivariate (Binary logistic regression (21%), and 6 (43%) respectively. All patients were non- model) analyses were performed to identify predictors of surgical candidates due to insufficient contralateral volume PHLI. All analyses were performed using IBM SPSS 22 or high risk biology prior to treatment. Mean contralateral (IBM Corp. Armonk, NY), with p< 0.05 being statistically hypertrophy with neoadjuvant Y-90 alone resulted in a 549 significant. ml to 789 ml (30% to 45%) in 9 patients within a median Results: A total of 890 patients underwent liver resection 202 days. Y-90 with PVE in 5 patients resulted in 531 ml to during the study period. Of these, 18% patients were on 710 mL (38% to 54%) hypertrophy within a median 190 statins in the peri-operative period. PHLF (Grades A,B,C) days. Liver resection included right hepatectomy, left was encountered in 19% of the patients. On univariate hepatectomy, and trisegmentectomy in 9 (64%), 2 (14%) analysis, use of statins(p< 0.01) had a protective effect on and 3 (22%) respectively with a mean length of stay of 8.5 the incidence of PHLI. Extent of resection(p< 0.001), post- days and major morbidity of 14%. No perioperative mor- operative septic complications (bile leak, intra-abdominal tality occurred.

FP08-09 Table Characteristics and post hepatectomy outcomes in patients with previous Y-90 for liver malignancy. Pt Age/Sex Diagnosis Tumor PVE Y-90 Y-90 – FLRV Type of LOS 90-Day size Dosage surgery Pre /post Y-90 Surgery Complications (cm)/N (Gy) (days) (ml) (%) (CD-grade) 1 73/M HCC 10.0/1 Yes 101 190 514(19%)/786(31%) Open TRS 6 Pleural effusion, (Cl–I) 2 59/M HCC 6.3/1 No 118 210 517(30%)/875(49%) HA RH 22 Bile leak, (Cl–IIIb) 3 67/F MCA 3.2/multiple No 105 217 421(33%)/627(56%) Lap. RH 7 Subcortical infarct (Cl-II) 4 60/M HCC 11.0/1 No 122.7 202 625(25%)/789(36%) Open RH 6 Anemia, (Cl-II) 5 66/F HCC 13.0/1 No 106 122 447(35%)/631(48%) Open TRS 6 None 6 68/M CCA 4.0/1 No 116.7 167 310(16%)/535(35%) Open TRS 20 DVT, PE (Cl-IIIb) 7 27/F MMA 6.0/1 Yes 110 126 530(34%)/527(43%) Lap. RH 4 None

HPB 2018, 20 (S2), S182eS294 Free Orals (FP01-FP31) S207

FP08-09 Table (continued) Pt Age/Sex Diagnosis Tumor PVE Y-90 Y-90 – FLRV Type of LOS 90-Day size Dosage surgery Pre /post Y-90 Surgery Complications (cm)/N (Gy) (days) (ml) (%) (CD-grade) 8 72/M MCA 2.8/multiple Yes 117 153 411(43%)/372(53%) Open RH 6 None 9 70/F MCA 3.8/multiple Yes 113 481 733(58%)/854(68%) Open LH 6 Anemia, (Cl-II) 10 61/F CCA 11.0/1 No 369 141 598(31%)/1046(47%) Open RH 6 None 11 63/M MCA 2.0/multiple Yes 104 459 469(34%)/1012(73%) Open RH 8 Ileus, (Cl-II) 12 44/M CCA 8.8/multiple No 225 243 1008(43%)/934(47%) Lap. LH 8 Anemia, (Cl-II) 13 61/M MCA 2.3/1 No 231 182 469(34%)/808(49%) Open RH 5 None 14 87/M HCC 9.8/1 No 412 259 543(27%)/855(41%) HA RH 8 N/A

Abbreviations: HCC, hepatocellular carcinoma; MCA, metastatic carcinoma; MCA, metastatic melanoma; TRS, trisegmentectomy; HA, hand assisted; RH, right hepatectomy; LH, left hepatectomy; Cl, Clavien-Dindo classification; LOS, length of stay; DVT, deep vein thrombosis; PE, pulmonary embolism; PVE, portal vein embolization; CD, Clavien-Dindo-Score.

Conclusion: Neoadjuvant radioembolization contributes 2012RAC patients. In MSAP-2012RAC patients, however, to contralateral hypertrophy and tumor control during sPCD was associated with an increased OR (P = 0.009) and surviellance in patients with aggressive biology. Surgical IR (P = 0.040). intervention after Y-90 is feasible without adding Conclusions: In virtue of 2012RAC, we can conclude that morbidity. early-stage sPCD improve the prognosis in SAP-2012RAC patients with drainable AIPFCs by reducing the OR, IR, and mortality, but did not do so in MSAP-2012RAC FP09 - Free Papers 9 (mini oral) - Pancreas: patients. Pancreatitis 1 FP09-01 EFFICACY OF STERILE PERCUTANEOUS CATHETER DRAINAGE IN ACUTE PANCREATITIS OF VARIOUS SEVERITIES ASSOCIATED WITH ACUTE INFLAMMATORY PANCREATIC FLUID COLLECTION Y. Zhang and T. -B. Liang The Second Affiliated Hospital of Medical College, Zhejiang University, China Introduction: To evaluate the efficacy of early-stage sterile percutaneous catheter drainage (sPCD) of drain- able acute inflammatory pancreatic fluid collections (AIPFCs) in acute pancreatitis (AP) of different sever- ities. A retrospective analysis was performed based on the presence of drainable AIPFCs and different AP severity classifications. Figure [sPCD for AP with AIPFCs] Methods: 162 patients with drainable AIPFCs were clas- sified as severe acute pancreatitis (SAP) per the 1992 Atlanta classification (SAP-1992AC). Of these, 94 patients with persistent organ failure were reclassified as SAP ac- cording to the 2012 Revised Atlanta classification (SAP- FP09-02 2012RAC); whereas, 68 patients with transient organ fail- THE PRESENCE OF ACUTE ure and/or local complications were reclassified as having CHOLANGITIS DEFINED BY 2013 moderate severity acute pancreatitis (MSAP-2012RAC). TOKYO GUIDELINES (TG13) IN ACUTE Early-stage sPCD was performed among 64 SAP- NECROTISING PANCREATITIS IS 2012RAC and 41 MSAP-2012RAC. Outcome variables ASSOCIATED WITH POOR OUTCOMES including operation rate (OR), pancreatic infection rate (IR), and mortality was compared between patients with E. Pando, P. Alberti, L. Vidal, J. N. Hidalgo, C. Dopazo, and without sPCD. C. Gomez, M. Caralt, L. Blanco, J. Balsells and R. Charco Results: 2012RAC divides the SAP-1992AC patients into Hepato-Pancreato-Biliary and Transplant Surgery, Hos- ’ two groups: SAP-2012RAC and MSAP-2012RAC pa- pital Universitario Vall d Hebron, Universidad Autónoma tients. The Early-stage sPCD caused no change in outcome de Barcelona, Spain variables among SAP-1992AC-classified patients with Background: Is difficult to establish whether acute AIPFCs, but cause obvious decreases in the OR (P = cholangitis (AC) co-exists as an independent entity with 0.006), IR (P = 0.020), and mortality (P = 0.009) in SAP- acute pancreatitis (AP) owing to overlaping systemic

HPB 2018, 20 (S2), S182eS294 S208 Free Orals (FP01-FP31) inflammatory response between entities. Also, the admitted with incident acute pancreatitis. A diagnosis of ascending route of the bile duct has been reported as a alcoholism (OR: 0.3; 95% CI: 0.2-0.5) or concurrent biliary source of infection in AP. The aim of our study was to disease (OR: 0.2; 95% CI: 0.1-0.3) were associated with a determine whether the presence of AC according to TG13 low risk of underlying pancreatic cancer. is associated with poor outcomes in AP. Conclusions: Our findings suggest that, in patients Methods: Data of 270 consecutive patients with AP from admitted with acute pancreatitis, age 65-70 years, chronic our prospective database were analysed. Outcomes pancreatitis and pancreatic cyst/pseudocyst are predictors analysed were: persistent organ failure (POF), persisten of underlying pancreatic cancer, whereas alcoholism or multi-organ failure (PMOF), mortality, duration of hospital concurrent biliary disease were predictors of no underlying stay, use of antibiotics, common bile duct stones (CBDS) pancreatic cancer. requiring ERCP, intensive care unit (ICU) admission, infected pancreatic necrosis (IPN) with positive culture after necrosectomy, and need for an interventional pro- FP09-04 fi cedure against necrosis. AC was de ned according to TG13 COMPLICATIONS OF PERCUTANEOUS in all patients between admission and the first week post AP onset. DRAINAGE IN STEP-UP APPROACH Result: A positive moderate or severe TG13 AC was found FOR MANAGEMENT OF PANCREATIC in 64(23.7%) patients, and was associated with POF NECROSIS: EXPERIENCE OF TEN (58.8% vs. 11.1%, p=0.000), PMOF (29.4% vs. 7.1%, YEARS FROM A TERTIARY CARE p=0.001), ICU admission (23.5% vs. 5.1%, p=0.003), CENTRE hospital stay > 20 days (41.2% vs 17.4%,p=0.015), use of R. Gupta1, A. Kulkarni1, G. Sharma2, P. Vaswani3, antibiotics (94.1% vs 33.2%,p=0.000), blood positive cul- 2 1 1 1 tures (29.4% vs 7.5%, p=0.002), and CBDS requiring R. Nimje , R. Gupta , S. Shenvi , R. Y. Babu , M. Kang3 and S. Rana4 ERCP (64.7% vs. 13%,p=0.000). No association was found 1 2 regarding IPN, need for an interventional procedure and Division of Surgical Gastroenterology, Department of General Surgery, 3Department of Radiology, and mortality. 4 Conclusion: In our series, AP with acute cholangitis Department of Gastroenterology, PGIMER Chandigarh, defined by TG13 is associated with poor outcomes, India including necrosis infection. Large series are needed to Introduction: Use of percutaneous drainage as initial confirm these findings. treatment in necrotising pancreatitis led to improved out- comes and obviated need for surgery in a significant pro- portion. However, percutaneous drainage (PCD) has its FP09-03 own complications including injury to visceral or vascular structures. In this study, we reviewed our experience with PREDICTORS OF UNDERLYING PCD-related complications. PANCREATIC CANCER IN PATIENTS Method: Retrospective analysis of prospectively main- WITH ACUTE PANCREATITIS tained database of patients with moderately severe and J. Kirkegård1, F. V. Mortensen1, U. Heide-Jørgensen2 and severe acute pancreatitis (revised Atlanta) who were treated D. Cronin-Fenton2 by step-up approach in our unit was performed. All patients 1Department of Surgery, and 2Department of Clinical who underwent percutaneous catheter drainage between Epidemiology, Aarhus University Hospital, Denmark April 2008 to December 2018 were included. PCD-related complications (ECF, bleeding) were reviewed in detail. Introduction: Acute pancreatitis may be the initial symp- Result: Total of 707 PCD catheters were utilized in 314 tom of pancreatic cancer in some patients. However, patients(median 2,range 1-6). Total number of inter- identification of patients with an underlying pancreatic ventions(insertion, repositioning, upsizing) was cancer is challenging. 1194(median 4,range 1-11). Most commonly used PCD Methods: Using the Danish National Patient Registry and size was 10Fr and most commonly used imaging modality Danish Cancer Registry, we conducted a population-based was ultrasound. Most common region of PCD placement cohort study of all patients diagnosed with acute pancre- was peripancreatic region in 44.6%. atitis in Denmark, 1980-2012. We defined pancreatic Enteric communication was seen in 8.9%(28/314) of cancers diagnosed within one year after acute pancreatitis patients and 3.9%(28/707) of all PCD’s. Fistula was as an underlying pancreatic cancer. We a priori selected communicating with colon in 71.4%(20/28), duodenum in variables that could potentially predict pancreatic cancer in 17.8%(5/28) and jejunum in 10.7%(3/28) patients. Fistulae patients with acute pancreatitis, and included these in a were conservatively managed in 71.4% (20/28). Operative multivariate logistic regression model to estimate the odds management was required in 30% of colonic and 40% of ratio (OR) of pancreatic cancer associated with each duodenal fistulae. variable. PCD-related bleeding complications were seen in Results: We included 43,972 patients with incident acute 7.3%(23/314) patients. 30.4% were managed conserva- pancreatitis, the median age was 56 years and 55% were tively and 30.4% required angiography and embolisation of men. 353 (0.8%) patients had an underlying pancreatic pseudoaneurysms. Surgery was needed in 39.1% (9/23). cancer at the time of acute pancreatitis diagnosis. Age 65- Mortality in patients who bled was 30% (7/23). 70 years (OR: 3.6; 95% CI: 2.7-4.9), a previous diagnosis Conclusion: Majority of PCD-related ECF’s can be of chronic pancreatitis (OR: 2.5; 95% CI: 1.6-4.0), or managed conservatively with success. Bleeding complica- pancreatic cyst/pseudocyst (OR: 1.8; 95% CI: 0.8-4.1) were tions had highest rate of mortality. predictors of underlying pancreatic cancer in patients

HPB 2018, 20 (S2), S182eS294 Free Orals (FP01-FP31) S209 fl FP09-05 explored the effect of dexamethasone and in iximab in L- arginine induced experimental SAP in mice. QUANTUM: PATIENT REPORTED Methods: Ninety-four, male, Swiss albino mice were OUTCOMES (PROS) AFTER MINIMAL randomly divided into 9 groups. Severe AP was induced by ACCESS RETROPERITONEAL intraperitoneal injections of L-arginine (4.5 g/Kg, two PANCREATIC NECROSECTOMY TO doses, one hour apart). The four treatment groups were TREAT ACUTE PANCREATITIS e AN given either dexamethasone (10 mg/Kg intravenous) or infliximab (16 mg/KG intravenous) either prophylactically EXPLORATORY STUDY (30 minutes prior to L-arginine injection) or therapeutically 1 2 1 3 1 P. Zelga , J. Rees , E. Iaculli , C. Johnson and A. Jah (24 hours after L-arginine injection). Animals were eutha- 1 HPB and Transplant Surgery Unit, Cambridge University nized at 72 hours, histopathological evaluation and 2 Hospitals NHS Foundation Trust, Bristol Centre for biochemical analyses performed on the harvested blood, Surgical Research, School of Social and Community lungs and pancreas. 3 Medicine, University of Bristol, and University Surgical Results: The histopathological scores were not signifi- Unit, Southampton University Hospital, University of cantly different between the SAP and the SAP with treat- Southampton, United Kingdom ment with dexamethasone. Pancreatic histopathology Introduction: A minimally invasive approach to pancre- didn’t significantly improve with infliximab. Therapeutic atic necrosectomy (MIRPN) decreases morbidity and infliximab decreased lung histopathology score (p = 0.005) mortality from necrotising acute pancreatitis. However, the but lung myeloperoxidase remained unchanged in the drug impact of MIRPN on patient reported outcomes is unclear. treated groups (p = ns). Plasma amylase was not signifi- Methods: Patients with acute pancreatitis who underwent cantly different in the SAP, or treatment with dexametha- MIRPN between January 2010 and December 2016 were sone (median 5809.9 vs 3697.6 vs 3327.5 U/L, p = ns) or identified and invited to participate completing the EORTC infliximab (median 5809.9 vs 4830.7 vs 4566.8 U/L, p = QLQ-C30, & EORTC PAN-28(CP) questionnaires. PROs ns). No significant difference in the serum levels of cyto- were compared using Pearson and Anova (Significance p< kines TNF-alpha, IL-6 and IL-10 was noted. 0.01) with an age- and sex-matched normative population Conclusions: Dexamethasone and infliximab did not of Western Europe and stratified by age and also the decrease the severity of L-arginine-induced murine AP. duration after MIRPN (< 1y,1-3y,>3y). A >15 point change in PRO score was clinically significant. Results: 52 eligible patients were identified and 46 FP09-07 participated (88%). 74% (n=34) returned questionnaires. NUTRITIONAL SUPPLEMENTATION IN After MIRPN all patients reported poorer physical and social functioning scores with a clinically significant PEOPLE WITH ACUTE PANCREATITIS: (>15point) reduction in those 50 years old; whilst fa- A SYSTEMATIC REVIEW AND tigue, pain and insomnia symptom scales had the highest NETWORK META-ANALYSIS < scores(42,26 and 30 points respectively). Patients 1 year M. Di Martino1,2, L. H. Eusebi3 and K. S. Gurusamy4 after MIRPN, had similar scores for all functional scales 1Department of Surgery, University Hospital La Princesa, and global HRQL(health related quality of life) when Spain, 2Department of Surgery, Royal Free London, compared to the normative data. However, with time United Kingdom, 3Department of Surgical and Medical global HRQOL deteriorated and by three years scores for Sciences, University of Bologna, Italy, and 4Department of < functioning and symptoms were statistically(p 0.0001) Surgery, University College London, United Kingdom and clinically(>15 points) worse than age matched Introduction: Nutritional deficiencies in patients with controls. acute pancreatitis (AP) may increase the complications and Conclusion: One-year after MIRPN patients report similar/ mortality. There is significant uncertainty on whether improved HRQL but more fatigue when compared to the nutritional supplementation of enteral or parenteral nutri- population data. However, patients should be informed that tion with glutamine, or polyunsaturated fatty acids (PUFA) HRQL deteriorates with time often due to chronic pancre- are beneficial in AP. atitis and pancreatic insufficiency. This study supports the Methods: We searched the Cochrane register, MEDLINE, use of MIRPN and prospective HRQL studies in this setting Embase, Science Citation Index Expanded, and trial reg- isters until March 2017 to identify randomized controlled trials. Two review authors independently identified trials FP09-06 and extracted data. We performed a network meta-analysis EFFECT OF DEXAMETHASONE AND using Bayesian methods, calculating odds ratio (OR), rate INFLIXIMAB ON THE SEVERITY OF L- ratio (RaR), or mean differences (MD) with 95% credible ARGININE INDUCED MURINE ACUTE intervals (95%CrI). Results: We included 23 trials (1072 participants). The PANCREATITIS network meta-analysis showed that glutamine reduced 1 2 3 1 S. H. S. Reddy , T. G. Jacob , P. K. Garg and P. Sahni short-term mortality (OR 0.30, 95%CrI 0.13-0.65), mor- 1 2 GI Surgery and Liver Transplantation, Anatomy, and tality at maximal follow-up (OR 0.33, 95%CrI 0.14-0.68), 3 Gastroenterology and Human Nutrition, AIIMS, India serious adverse event rate (RaR 0.61 95%CrI 0.43-0.84), Introduction: Systemic inflammation in severe acute total adverse event rate (RaR 0.64, 95%CrI 0.48-0.83), pancreatitis (SAP) is due to excessive release of pro-in- sepsis (OR 0.11, 95%CrI 0.02-0.45), and intensive therapy flammatory cytokines, which may be decreased by gluco- unit (ITU) stay (MD -31.18 days 95%CrI -5.71,-0.72) when corticoids or specific monoclonal antibodies. Here, we compared with control. PUFA reduced severe adverse

HPB 2018, 20 (S2), S182eS294 S210 Free Orals (FP01-FP31) event rate (OR 0.60, 95%CrI 0.37-0.95) and hospital stay the maximum IAP during the treatment reached 18.9Æ1.6 (MD -7.50 days, 95%CrI -14.40,-0.78). There was no ev- vs. 21.8Æ1.7, p=0.047 in the Early and Late groups idence of differences in other outcomes (serious adverse accordingly. Early CVVH resulted in a shorter ICU (14 event proportion, total adverse event proportion, require- Æ3.8 vs. 18Æ2.4, p=0.069) and hospital stay (21Æ7.2 vs. ment for endoscopic or radiological drainage of abdominal 28Æ6.8, p=0.091). Mortality rate was similar (9.8% vs. collection and costs) in the trials that reported these. 11.1%, p=0.89). Conclusion: Based on low-quality evidence, glutamine Conclusions: Early CVVH is a rational treatment strategy supplementation decreases mortality, serious and total for the management of SAP and the prevention of IAH and adverse event rate, sepsis, and ITU stay, while PUFA can be successfully implemented in the setting of the uni- supplementation decreases severe adverse event rate and versity hospital. hospital stay when compared to no supplementation.

FP09-09 FP09-08 MANAGEMENT OF CHRONIC EARLY CONTINUOUS VENO-VENOUS PANCREATITIS e STEP UP? HEMOFILTRATION MAY PREVENT P. Varshney, R. K. Singh, S. Sharma, A. Kumar, DEVELOPMENT OF INTRA- A. Prakash, A. Behari, A. K. Gupta, V. K. Kapoor and ABDOMINAL HYPERTENSION IN R. Saxena PATIENTS WITH SEVERE ACUTE Surgical Gastroenterology, Sanjay Gandhi Post Graduate PANCREATITIS Institute of Medical Sciences, India Introduction: Endoscopic treatment is used in several K. Zeiza1,2, M. Pavlovics1, R. Starinskis2, K. Atstupens1, units prior to surgical treatment for pain in chronic V. Fokin1,2, H. Plaudis1,2 and G. Pupelis1,2 pancreatitis. There is limited information on patients un- 1Department of General and Emergency Surgery, Riga dergoing surgical ’salvage’ after endoscopic failure. We East University Hospital, and 2Rigas Stradins University, conducted a comparative study between patients who un- Latvia derwent surgery after prior non-surgical intervention and Introduction: Early recognition, prevention and treatment upfront surgery. of intra-abdominal hypertension (IAH) are crucial in pa- Methods: Patients who underwent surgical drainage in our tients with severe acute pancreatitis (SAP). institution over the last 6 years were reviewed, in two fi Aim: To determine the incidence of IAH and the signi - groups- Group A (n=29) - surgery with prior non-surgical cance of an early application of CVVH in the treatment interventions (ESWL and/or endoscopy); Group B (n=79) - protocol of SAP. upfront surgery. Pain scores and QOL scores were Methods: Analysis of prospectively collected ICU treat- measured prospectively. ment protocols of patients who developed IAH (intra- Results: The two groups were comparable for baseline > abdominal pressure (IAP) 12mmHg) during their clinical characteristics except that group A patients had more fi course of SAP de ned by the criteria of the Revised Atlanta strictures/ stones in body/tail region (p < 0.05). Short term fi fi classi cation. Patients were strati ed in two groups ac- morbidity was more in Group A vs Group B (65% vs 26 %, fi cording to whether CVVH was started within the rst 48 p < 0.01) - wound infection (45 % vs 10 %, p < 0.01; hours after admission (Early group) or later (Late group). Gastroparesis (10% vs nil, p < 0.01). On long term The main variables were the dynamics of IAP and the main followup complete pain relief was worse in Group A pa- outcomes. tients (37% vs 68%, p=0.05). Quality of life scores Results: During the period from January 2000 to February (WHOQOL- BREF) in social domain was significantly 2017, 174 patients with SAP developed IAH and 125 of better in Group B. Improvement in exocrine and endocrine them underwent CVVH. Of all, CVVH was performed insufficiency was similar in both groups. within 48 hours in 71, and in 54 later than 48 hours after Conclusion: Patients with chronic pancreatitis who un- admission. The mean IAP before the commencement of dergo ’salvage’ surgery after non-surgica interventions are Æ CVVH was 15.1 4 mmHg vs. 12.8 mmHg, p=0.003, and at increased risk of postoperative morbidity, lower quality

FP09-09 Table Group A (n [ 24) Group B (n[ 54) p value Follow UP duration (months) * range 58± 22 (19-92) * 66 ± 20 (17-92) * 0.170 IFC pain score at last follow-up 3.7 ± 2.4 2.67 ± 2.1 0.104 Complete pain relief, n (%) 9 (37) 37 (68) 0.05 WHOQOL- BREF Environmental 61±12 65±10 0.115 Physical 52 ± 7 52± 10 0.937 Psychological 55±10 56 ±10 0.731 Social 59±16 69±14 0.012 Endocrine function, n Insufficiency resolved/improved 3/8 5/16 0.218 Exocrine function, n Insufficiency resolved/improved 2/8 4/16 0.217

[Long term outcomes] Long term outc...

HPB 2018, 20 (S2), S182eS294 Free Orals (FP01-FP31) S211 of life and poor pain control as compared to those under- General, Hepato-Bilio-Pancreato Unit, Pontificia Univer- gone upfront surgery. Patients with pancreatic body/tail sidad Católica de Chile, Chile strictures/stones have poor outcomes with non-surgical Introduction: While it has been shown that the presence of interventions and may be considered for upfront surgery. residual cancer (RC) at oncologic extended resection (OER) in patients with incidental gallbladder cancer (IGBC) is associated with poor disease-specific survival FP10 - Free Papers 10 (mini oral) - Biliary: (DSS), prognostic factors at time of index cholecystectomy Gallbladder Cancer are poorly understood FP10-01 Methods: This retrospective study included patients with ONCOLOGIC OUTCOME OF T2 IGBC who underwent OER with curative intent at 2 centers (USA and Chile) between 1999-2015 and had no evidence GALLBLADDER CANCER AND THE of RC. Pathologic features were examined, and predictors OPTIMAL SURGICAL TREATMENT: of DSS were analyzed. KOREAN MULTI-INSTITUTIONAL Results: The study included 114 patients. Thirteen patients ANALYSIS (11.4%) had a positive cystic duct margin at the index cholecystectomy. Twenty-one patients (18%) underwent W. Kwon1, J. -Y. Jang1, Y. Han1, J. Kim1, J. Heo2, resection of the common bile duct. DSS was significantly S. -J. Park3 and S. -W. Kim1 worse in the patients with a positive than in those with a 1Department of Surgery, Seoul National University Col- negative cystic duct margin at index cholecystectomy (DSS lege of Medicine, 2Department of Surgery, Sungkyunkwan rates at 3 and 5 years, 56% vs 85% and 45% vs 78%, University School of Medicine, and 3Department of Sur- respectively; p< 0.014). In multivariate analysis, only a gery, National Cancer Center, Republic of Korea positive cystic duct margin was predictive of worse DSS Background: Extended cholecystectomy (EC) is generally (hazard ratio, 4.1; p=0.015). recommended in T2 gallbladder cancer (GBC). Recently, Conclusions: In patients with incidental gallbladder some advocated simple cholecystectomy (SC) for T2 GBC cancer, a positive cystic duct margin at index cholecys- on serosal side. The aim was to investigate the oncologic tectomy is a strong and independent predictor of worse outcome of T2 GBC and its optimal surgery extent. DSS even if no further RC is found OER. A positive cystic Methods: Data on 577 T2 GBC patients from 3 high- duct stump may prompt consideration of additional multi- volume centers were collected. Pathology data and CT were modal therapy to achieve optimal oncologic outcome. reviewed. Survival analyses were performed. Results: The mean age was 64 years with female domi- nance. 168 patients received SC and 409 EC. Tumor was located on liver and serosal side in 396 and 155 patients, FP10-03 respectively. The LN metastasis rate was 38.3%. The 5- GEMCITABINE-OXALIPLATIN FIRST- YSR was 59.4%. The independent prognostic factors were LINE CHEMOTHERAPY IN LOCALLY CEA, CA19-9, location, operation type, R status, LN ADVANCED AND METASTATIC metastasis. A subgroup of 500 curatively treated patients GALLBLADDER CANCER - RESULTS was analyzed to compare SC and EC. EC was then matched FROM A TERTIARY CENTRE IN NORTH to SC with propensity score for age and ASA score (1:2). Finally, 118 SC and 239 EC patients were analyzed. EC INDIA showed better survival than SC (p< 0.001). EC was A. Kulkarni1, T. D. Yadav1, I. Santosh2, P. Kumar S2, significantly better with tumor on liver side (p=0.001). R. Kumar3, V. Bansal2 and S. K. Vadi3 Even when the tumor was located on serosal side, EC 1Division of Surgical Gastroenterology, 2Department of demonstrated superior outcome over SC (p=0.015). General Surgery, and 3Department of Nuclear Medicine Conclusion: T2 GBC had fair survival outcome, but still Introduction: In view of encouraging results of ABC-02 unsatisfactory compared to other gastrointestinal cancers. trial in advanced biliary tract cancer, we aimed to evaluate Extended cholecystectomy should be performed for T2 efficacy of gemcitabine-oxaliplatin chemotherapy in our GBC whenever the patients’ physical condition permits. patients with locally advanced/ metastatic gallbladder cancer (GBC). FP10-02 Methods: Retrospective analysis of prospectively main- POSITIVE CYSTIC DUCT AT INDEX- tained database of patients with locally advanced and CHOLECYSTECTOMY IS AN metastatic GBC(July 2015- December 2017) was IMPORTANT NEGATIVE performed. Inclusion criteria - histologically proven locally advanced GBC(T3 or N2 disease) or metastatic GBC, no PROGNOSTICATOR IN INCIDENTAL prior chemotherapy, ECOG status2, normal hematologic, GALLBLADDER CANCER EVEN IF NO renal and hepatic functions. Chemotherapy regimen RESIDUAL CANCER IS FOUND AT included injection Gemcitabine 1250mg/m2 infusion on ONCOLOGIC EXTENDED RESECTION days 1 and 8 and oxaliplatin 100mg/m2 on day 1 every 3 weeks. Primary endpoints were overall survival(OS), pro- E. A. Vega1, E. Vinuela2,3, M. Sanhueza2,3, C. Diaz2,3, gression-free survival(PFS) and R0 resection rate in those M. Okuno1, K. Joechle1, E. Simoneau1, J. E. Lee1, given preoperative chemotherapy. Secondary endpoints J. -N. Vauthey1 and C. Conrad1 were response rate and toxicity of chemotherapy regimen. 1Sugical Oncology, UT MD Anderson Cancer Center, Results: Total of 52 patients received chemotherapy(mean United States, 2Digestive Surgery, Hepato-Bilio-Pancre- age: 52 years, 73% females) of which preoperative atic Surgery Unit, Hospital Sotero del Rio, and 3Cirugia

HPB 2018, 20 (S2), S182eS294 S212 Free Orals (FP01-FP31) chemotherapy was given in 38.5 %(n=20) and palliative ICPN and multiple bilIN may suggest a bad prognosis. chemotherapy in 61.5%(n=32). Median OS was 8.3 months Continued follow up and research on ICPN should be needed. and median PFS was 4.7 months. Complete response was seen in 1.9%, partial response in 11.5%, stable disease in 21.2% and progression in 65.4%.In locally advanced FP10-05 group, 30%(n= 6/20) underwent R0 resection. R0 resection VALIDATION OF AJCC 8TH EDITION was possible in 85.7% (6/7) patients who showed response to chemotherapy. Patients who underwent surgery had STAGE FOR GALL BLADDER CANCER significantly better OS(p< 0.01) and PFS(p=0.01). D. J. Park, J. S. Heo, S. H. Choi, I. W. Han and Conclusion: Combination chemotherapy with gemcitabine- D. W. Choi oxaliplatin showed good response rate in a select subgroup Samsung Medical Center, Sungkyunkwan University of patients. Curative surgery was feasible in majority of School of Medicine, Republic of Korea those who had good response. Patients who underwent Purpose: This study evaluated discriminatory value of curative resection had significant improvement in survival. newly proposed AJCC 8th staging system compared to the AJCC 7th staging system for gallbladder cancer. Methods: We retrospectively reviewed database of 522 patients who underwent curtive intent surgery for gall- bladder cancer at a single institute from January 2006 through June 2016. Clinicopathologic characteristics and survival rates were analyzed based on 7th and 8th edition of AJCC staging system, respectively. Results: There was significant different in survival rates between T2a and T2b(P< 0.001). Among pathologically- stage patients, P values for pairwise comparisons among different 8th AJCC stage were significant(P< 0.05) for Figure [Pre and post chemotherapy images] stage IIa vs IIb and IIIa vs IIIb. Notably, the new edition improved the power of discrimination slightly in overall survival and disease-free survival (c-indices: 0.711, 0.723) FP10-04 compared with the 7th edition (c-indices:0.698, 0.714). SINGLE CENTER EXPERIENCE IN There were no significant difference between N1 vs N2 in PATIENTS WITH INTRACHOLECYSTIC both stage group(P=0.578, 0.283). Conclusions: Compared to the 7th system, the 8th system PAPILLARY-TUBULAR NEOPLASM bring significant prognostic improvement for stage IIa vs (ICPN) IIb. However, lymph node staging does not reflect well the S.-R. Kim, Y.-D. Yu, Y.-I. Yoon and D.-S. Kim prognosis in both classification by number or location Surgery, Korea University Anam Hospital, Republic of based staging system. Korea Introduction: Intracholecystic papillary-tubular neoplasm (ICPN) has recently been reported as a new concept in the FP10-06 classification of gallbladder tumors. ICPN is defined as FLT-PET CT SCAN CAN AVOID mass-forming preinvasive neoplasm (>1cm) of the gall RADICAL SURGERY IN bladder and shows many similarities to intraductal papillary MASQUERADING neoplasm of the bile duct (IPNB) and intraductal papillary mucinous neoplasm (IPMN). We analyzed the cases of PANCREATOBILIARY LESIONS ICPN and compared with gall bladder cancer (non-ICPN) G. Krishnamurthy, V. Kumar, G. R. Verma, B. Mittal, in our center. K. Agarwal, R. Kochhar, V. Singh, L. Kaman and Method: We retrospectively reviewed the medical records R. Singh of patients who diagnosed ICPN and/or Gall bladder cancer PGIMER Chandigarh, India in our center from January 2010 to October 2017. Patient Background: The purpose of this study was to evaluate the data and clinical outcomes were collected and analyzed. efficacy of this marker (FLT-PET CT scan) and to compare Results: There were 11 patients who diagnosed ICPN and it with FDG-PET scan in differentiating benign from ma- 144 patients who diagnosed gall bladder cancer (non- lignant diseases in suspected bilio-pancreatic tumors. ICPN) in our center. There was no significant difference in Methodology: It is a prospective observational study demographics between two groups. Two year survival rate from July 2013 to April 2017. Suspected Pancreatico- was 83.3% in ICPN group, 73.2% in non-ICPN group. biliary malignancies were included. FDG-PET/CT and (p=0.11) Among the 11 patients in ICPN group, Three FLT-PET/CT was done. Histopathology of the resected patients diagnosed ICPN with high grade dysplasia, 8 pa- specimen or FNAC was considered the gold standard for tients diagnosed ICPN with adenocarcinoma (Invasive diagnosis. ICPN). Two patients died due to the disease in ICPN group. Results: Fifty-six patients harboring 58 suspected lesions of Among the two patients, one patient who diagnosed ICPN pancreatobiliary malignancy were included in this study with high grade dysplasia and multiple BilIN recurred early Two patients had synchronous lesions in gall bladder and with peritoneal seeding. pancreas. Majority of patients(75.8%,44/58) had gall Conclusions: ICPN group shows better survival rate than non- bladder carcinoma. Thirty-five patients underwent laparot- ICPN group without statistical significance. Concomitant omy, thirty of them were resorted to radical surgery.

HPB 2018, 20 (S2), S182eS294 Free Orals (FP01-FP31) S213 fi Eighteen lesions(31%) were con rmed as FP10-08 benign(XGC+Chr. Cholecystitis=14, Ch.Pancreatitis+- Pancreatic lipodystrophy=4) on histological examination. MUC2, MUC4 AND MUC5AC The result of FLT-PET were better than FDG-PET. The EXPRESSION IN CARCINOMA sensitivity, specificity, PPV and NPV & accuracy of FLT- GALLBLADDER PET was 92.5%, 100%, 100%, 84.2% and 94.6% respec- P. Puneet1, P. Shukla1, S. C. U. Patne2 and A. Khanna1 fi tively. The sensitivity, speci city, PPV and NPV & accuracy 1Department of Surgery, and 2Department of Pathology, of FDG-PET was 97.5%, 31.2%, 78%, 83.3% and 78.5% Institute of Medical Sciences, Banaras Hindu University, respectively. Adding CA19-9 for detecting malignancy ac- India curacy increases to 98.2%. On analyzing with ROC, at Aim: To study the expression of MUC2, MUC5AC and SUVmax = 2 on FLT-PET, the sensitivity and specificity for MUC4 mucin in gallbladder cancer (GBC) and its corre- detecting malignancy in all pancreaticobiliary lesions was lation with clinicopathological characteristics. 92.5% and 100% respectively(AUC = 0.91). AUC for FDG- Material and methods: Fifty histopathologically proven PET CT in pancreaticobiliary lesions was 0.624. GBC (cases) and 10 patients undergoing surgery for Conclusion: FLT-PET is better than FDG-PET scan in chronic cholecystitis (CC) (control) were included in the differentiating benign pancreaticobiliary lesions. By study. The immunohistochemical expression of MUC2, employing FLT-PET CT as the diagnostic tool, it is MUC5AC and MUC4 were studied in formalin-fixed, predicted that 43.3% of radical surgery done for benign paraffin-embedded specimens from patients with GBC and bilio-pancreatic lesions can be avoided. CC patients and correlated with clinicopathological characteristics. Results: The mean age of GBC group was 52.24Æ11.39 FP10-07 years (range 30-70 years) and in CC group was 44.40Æ OUTCOME OF NEOADJUVANT 18.60 years (range 28-65 years). Majority of patients were CHEMOTHERAPY IN ‘LOCALLY females. No significant correlation of MUC2 expression ADVANCED/BORDERLINE was found in GBC and CC. MUC2 expression was ’ significantly correlate with lower T stage (T1+T2) as RESECTABLE GALLBLADDER compared with higher T stage (T3+T4) (p=0.001). In GBC CANCER: THE NEED TO DEFINE cases MUC5AC was positive in 26(54%) patients than CC INDICATIONS group (p=0.015) and expression of MUC5AC was signifi- S. Patkar1, V. Chaudhari1, V. Ostwal2, A. Ramaswamy2, cantly positive with higher T stage (T3/T4) in comparison S. Shrikhande1 and M. Goel1 to CC group (p=0.002). In GBC cases MUC4 was positive 1Surgical Oncology, and 2Medical Oncology, Tata Me- in 24(48%) patients than CC group (p=0.026) and expres- morial Hospital, India sion of MUC4 was significantly associated with poor dif- Introduction: Studies evaluating neo-adjuvant chemo- ferentiation, higher T stage and perineural invasion. therapy (NACT) exclusively in gallbladder cancer (GBC) Conclusion: MUC4 and MUC5AC expression is related to fi are few and there are no randomised trials on the subject. higher stage (T3/T4 stage), MUC4 is signi cantly associ- Locally advanced disease is not yet clearly defined and there ated with poor differentiation of tumour, perineural inva- fi are no objectively defined indications for NACT in GBC. sion in GBC group. MUC2 expression was signi cantly Material and methods: We analysed a prospectively associated with lower stage (T1/T2 stage) in GBC. maintained database of 160 GBC patients who received NACT at Tata Memorial Hospital from January 2010 to February 2016. Decision to offer NACT was based on FP10-09 clinico-radiologic criteria suggesting high risk disease. PROGNOSTIC PERFORMANCE OF (Tata Memorial Hospital criteria for borderline resectable VARIOUS NODAL STAGING SYSTEMS /locally advanced GBC - TMH Criteria). FOR RESECTED GALLBLADDER Results: Contiguous hepatic infiltration for > 2 cm in 109 (68.7%) patients and radiologically significant N1 group CANCER USING A RESTRICTED CUBIC lymphadenopathy in 105 (65%) patients were the SPLINE MODEL: THE RESULTS OF commonest indications for NACT. 66 (41.2%) patients MULTICENTER STUDY could be offered curative intent resection. 27 patients were W. Lee, C. -Y. Jeong, J. Y. Jang, S. -C. Hong and Busan- deemed inoperable intra-operatively (17%). With a median Ulsan-Kyeungnam HBP Surgery Study Group follow-up of 33 months, the median OS and EFS of the Gyeongsang National University College of Medicine, entire cohort were 13 months (95% CI: 8.7 to 17.2 months) Republic of Korea and 8 months (95% CI: 7 to 8.9 months) respectively. Pa- Background: Current nodal staging system is insufficient tient undergoing curative surgery had a statistically supe- to determine the prognoses in gallbladder cancer (GBC) rior OS (49 vs. 7 months; p=0.0001) and EFS (25 months patients with lymph node (LN) metastasis. This study was vs. 5 months; p= 0.0001) compared to those who did not. aimed to identify the best prognostic nodal staging system. Conclusion: The use of NACT in locally advanced/ Methods: We reviewed clinical data for 398 patients with borderline resectable GBC, led to either curative surgical resected GBC and compared nodal staging systems based resection or disease stabilisation in significant proportion of on the number of positive LNs (PLN), positive/retrieved patients. This resulted in improved survival in this select LN (RLN) ratio (LNR), and log odds of positive LNs group of ’high risk’ GBC patients who otherwise would be (LODDS). Restricted cubic spline (RCS) model was used treated with palliative intent chemotherapy. to determine the cut-off values for subgroups. Prognostic

HPB 2018, 20 (S2), S182eS294 S214 Free Orals (FP01-FP31) performance was evaluated with the C-index, Akaike’s between groups. The incidence risks of major complications information criterion (AIC), and the Bayesian information were 8.0% (16/199) in the drainage group and 2.5% (5/201) in criterion (BIC). the no-drainage group, respectively. The risk difference was Results: Subgroups were divided based on RCS plot -5.6% (95% confidence interval: -9.9%, -1.2%) and the p accordingly: PLN 3 (PLN = 0, 1-2,  3), PLN 4 (PLN = 0, values for non-inferiority and for superiority of the no- 1-3,  4), LNR (LNR = 0, 0-0.269,  0.27), and LODDS drainage to the drainage group were < 0.001 and 0.011, (LODDS < À0.8, À0.8-0,  0). Oncologic outcome was respectively. No differences were found in the incidence of different significantly among subgroups in each system. surgical site infection. Bile leakage did not occur in the no- PLN 4 (C-index: 0.730; AIC: 523.784; BIC: 531.685) and drainage group, while 18 patients (9.0%) had bile leakage in PLN 3 systems (C-index: 0.734; AIC: 980.483; BIC: the drainage group. The length of postoperative hospital stay 988.363) were the best prognostic discriminators of sur- was similar between the two groups (median, 8 days). vival and recurrence, respectively. Internal validation Conclusions: Abdominal drain should not be placed showed good discrimination power using 1000 times routinely because drainage can increase the incidence of bootstrapped calibration test. When RLN  5, LODDS major complications. showed the best prognostic performance for survival (C- index: 0.741; AIC: 208.704; BIC: 211.060) and recurrence (C-index: 0.703; AIC: 430.38; BIC: 434.122). FP11-03 Conclusion: The PLN system showed the best prognostic DOES KETOROLAC HAVE AN OPIOID performance in all GBC patients. However, LODDS was the best prognostic discriminator for prognosis in GBC SPARING EFFECT AFTER OPEN LIVER patients with RLN  5. RESECTION? DATA FROM THE OSLO- COMET TRIAL Å. A. Fretland1,2, J. Hausken3, B. Edwin1,2, FP11 - Free Papers 11 (long oral) - Liver: M. H. Andersen4, V. J. Dagenborg5, R. Kristiansen6, Miscellaneous K. Røysland7, G. Kvarstein8 and T. I. Tønnessen2,3 FP11-02 1Department of HPB Surgery and the Intervention Centre, 2 THE SAFETY OF LIVER SURGERY Oslo University Hospital, University of Oslo, Institute of Clinical Medicine, 3Division of Emergencies and Critical WITH NO-DRAIN POLICY: A Care, Oslo University Hospital, 4Department of Trans- MULTICENTER RANDOMIZED plant Medicine, Norway, 5Department of Gastrointestinal CONTROLLED TRIAL Surgery, 6The Intervention Centre, Oslo University Hos- 7 Y. Mise1, J. Arita2, K. Sakamaki3, T. Yamanaka4, pital, Institute of Basic Medical Sciences, University of 8 A. Saiura1, Y. Sakamoto2, M. Konishi5, M. Hashimoto6, Oslo, and University of Tromsø, Norway T. Sano7 and K. Uesaka8 Introduction: The “Opioid Crisis” has led to an increased 1Department of Hepatobiliary Pancreatic Surgery, Cancer focus on the use of opioids for postoperative analgesia. Institute Hospital, 2Hepato-Biliary-Pancreatic Surgery Thoracic epidural analgesia is widely used and can reduce Division, Department of Surgery, University of Tokyo, the need for intravenous and oral opioids. There is, however, 3Department of Biostatistics and Bioinformatics, Univer- a systemic opioid absorption also from the epidural space. sity of Tokyo, 4Department of Biostatistics, Yokohama City Objective: We conducted a randomized, controlled trial to University School of Medicine, 5Department of Hepato- investigate if intravenous multimodal, patient controlled biliary and Pancreatic Surgery, National Cancer Center analgesia (IV-PCA) was non-inferior to thoracic epidural Hospital East, 6Hepatobiliary-Pancreatic Surgery Divi- (TEDA) in patients undergoing open liver surgery. The sion, Department of Gastroenterological Surgery, Tora- multimodal analgesia was enhanced with the non-steroidal nomon Hospital, 7Department of Gastroenterological anti-inflammatory drug ketorolac, in order to reduce the Surgery, Aichi Medical University Hospital, and need for systemic opioids. 8Department of Hepatobiliary and Pancreatic Surgery, Methods: Patients assigned to open liver surgery in OSLO- Shizuoka Cancer Center, Japan COMET (NCT01516710) were randomly assigned (1:1) to Background: Despite the conclusions of previous ran- receive either IV-PCA (n=66), or TEDA (n=77). All pa- domized-controlled trials (RCTs), no-drain policy has not tients were included in an enhanced recovery after surgery been standardized in daily practice of liver surgery. The aim (ERAS) protocol. of this multicenter RCT is to assess the safety no-drain Results: The primary noninferiority endpoint, mean NRS policy after hepatectomy utilizing endpoint which is clini- pain score, was 1.6 in the IV-PCA group and 1.85 in the cally relevant. EDA group, establishing non-inferiority. The total use of Methods: Patients undergoing elective hepatectomy were intravenous morphine equivalents on postoperative day 0-2 randomly assigned to drainage and no-drainage groups. was 96 in the IV-PCA group and 109 in the TEDA group The primary endpoint was the incidence of major compli- (p=0.024), indicating an opioid sparing effect of ketorolac. cations after hepatectomy (Clavien-Dindo grade 3 or There was no difference in postoperative morbidity be- higher). Calculated sample size was 200 patients per group tween the groups (Comprehensive Complication Index: 8.2 when the study had an 80% power to prove non-inferiority vs 10.3, p=0.43), while hospital stay was 3.1 days in the of no-drainage group with a non-inferiority margin of 4%. PCA group and 4.3 days in the TEDA group (p=0.001). Results: Between May 2015 and July 2017, 405 patients Conclusion: IV-PCA with ketorolac was non-inferior to (drainage: 203, no-drainage: 202) were enrolled from 7 hos- TEDA for the treatment of postoperative pain in patients un- pitals. In 400 patients (drainage: 199, no-drainage: 201) of the dergoing open, parenchyma sparing liver resection. IV-PCA per protocol set, demographic factors were comparable with ketorolac may be an opioid sparing alternative to TEDA.

HPB 2018, 20 (S2), S182eS294 Free Orals (FP01-FP31) S215

4 FP11-04 Hospital of Lille, and University Hospital of Paul Brousse, France HEALTH RELATED QUALITY OF LIFE Background: Mortality after hepatectomy has not AFTER LAPAROSCOPIC AND OPEN decreased over the past decade at specialized HPB-cen- LIVER RESECTION. DATA FROM THE ters (4.7%) or nationally (5.8%). Prognostic models OSLO-COMET RANDOMIZED inaccurately predict mortality. Sensitizing HPB-centers CONTROLLED TRIAL to their outcome also fails to improve the postoperative course. A root-cause analysis, a method employed Å. A. Fretland1,2, V. J. Dagenborg2,3, G. M. Bjørnelv4, outside medicine to retrospectively explore and over- D. Aghayan4, R. Kristiansen4, M. W. Fagerland5, come the causes of adverse events, of postoperative L. Barkhatov4, B. A. Bjørnbeth1 and B. Edwin1,2 mortality was performed. 1Department of HPB Surgery and the Intervention Centre, Study design: The study population was a prospective Oslo University Hospital, 2University of Oslo, Institute of cohort of hepatectomies performed at 13 HPB-centers be- Clinical Medicine, 3Oslo University Hospital, 4The Inter- tween 10/2012 and 12/2014. Of the 1906 patients, 90 pa- vention Centre, and 5Oslo Centre for Biostatistics and tients died within 90-days (4.7%). The root-cause analysis Epidemiology, Oslo University Hospital, Norway was performed independently by a senior HPB-surgeon and Introduction: Most surgical treatments will cause a tem- a HPB fellow. The aims were to assess if: porary decline in Health-Related Quality of Life (HRQoL). (1) the cause of death had been identified by the Laparoscopic surgery has been shown to reduce this attending surgeons; decline. Evidence on HRQoL after laparoscopic liver (2) the patient had been managed according to Interna- resection (LLR) is however limited. The aim of this study tional guidelines; was to compare HRQoL after open versus laparoscopic (3) the intra and post-operative management had been liver resection. inappropriate; fi Methods: This was a prede ned sub-study of the OSLO- (4) the death was preventable; COMET trial (ClinicalTrials.gov NCT01516710). A total (5) a typical root-cause of postoperative death could be of 280 patients with colorectal liver metastases (CLM) identified. were randomly assigned to open liver resection (OLR) Results: The cause of death was identified by the index (n=147) or LLR (n=133). HRQoL was assessed with the surgeon and by the root-cause in 84% and 88% of the pa- Short Form 36 (SF-36) at baseline, 1 month and 4 months tients. The root-cause was intra-operative in 33% (technical after surgery. error, 24%; inappropriate judgement, 22%) or related to Results: A total of 272 patients underwent open (n=143) delayed postoperative management (23%). Guidelines were and laparoscopic (n=129) surgery. A total of 264 patients not followed in 57%. Overall, 47% of deaths were pre- completed at least 2 questionnaires, and 671 (82%) ques- ventable. The typical cause of death was a patient operated tionnaires were available for analysis. The decline in for a malignancy, with insufficient evaluation (tumor stage HRQoL was compared between the groups. The decline or progression), in whom a larger than expected procedure fi from baseline was signi cantly smaller after LLR for was performed. Bodily Pain (p=0.001), Role Physical (p=0.003), Vitality Conclusion: Compliance with guidelines and team (p=0.023) and Social Functioning (p=0.026) 1 month after communication if an unplanned operative finding occurs surgery, and for Role Physical (p=0.019) at 4 months after should be strictly implemented. surgery. The time from operation to initiation of adjuvant chemotherapy did not differ between the groups (46 vs 43 days, p=0.39), and the median number of courses given was 8 in both groups. FP11-07 Conclusion: Our results indicate that postoperative BENCHMARKS VALUE FOR HRQoL is better after LLR compared to OLR. This dif- INCIDENCE OF POST- ference lasts for up to four months. There was no difference HEPATECTOMY LIVER FAILURE in time to chemotherapy or number of courses given. AFTER MAJOR LIVER SURGERY: A VALIDATION AND INTEGRATION ANALYSIS FP11-05 1,2 2 1 2 HALF OF POSTOPERATIVE DEATHS F. Bagante , A. Ruzzenente , E. Beal , T. Campagnaro , K. Merath1, S. Conci2, O. Akgül1, C. Iacono2, AFTER HEPATECTOMY ARE A. Guglielmi2 and T. Pawlik1 PREVENTABLE: RESULTS OF THE 1Department of Surgery, Ohio State University Wexner ROOT-CAUSE ANALYSIS OF A Medical Center, United States, and 2Department of Sur- PROSPECTIVE MULTICENTER gery, University of Verona, Italy COHORT Background: Post-hepatectomy liver failure (PHLF) re- mains one of the most dreaded complications following I. Khaoudy1, O. Farges2, E. Boleslawski3, E. Vibert4, hepatectomy. Recently, an incidence of 5% was proposed O. Soubrane2 and J. -M. Regimbeau1 as the “benchmark” value of PHLF to define best practices. 1Digestive and Metabolic Surgery, Universitary Hospital We sought to validate the previously proposed benchmark of Amiens, 2University Hospital of Beaujon, 3University

HPB 2018, 20 (S2), S182eS294 S216 Free Orals (FP01-FP31) fi value, assess the incidence, and de ne factors associated FP11-08 with PHLF following hepatectomy. Methods: The National Surgery Quality Improvement CYTOSOLIC NUCLEIC ACID SENSORS Program (NSQIP) database was used to develop OF THE INNATE IMMUNE SYSTEM Bayesian models to estimate the probability of devel- PROMOTE LIVER REGENERATION oping PHLF based on patient characteristics. The AFTER PARTIAL HEPATECTOMY benchmark values were calculated as the 75th percen- D. Hartmann, S. Schulze, C. Stöß, M. Lu, B. Wang, tile of the distribution resulting from the Bayesian G. O. Ceyhan, H. Friess, B. Holzmann and N. Hüser models. Department of Surgery, Klinikum rechts der Isar, Tech- Results: A total of 6,918 patients underwent minor nical University of Munich, Germany (64.5%), right (RH:18.2%), left (LH:9.0%) or extended (EH:8.3%) hepatectomy. A small subset of patients Introduction: DNA sensors localized in the cytosol require had preoperative biliary stenting (5.3%) and 30.9% the adapter protein STING for the generation of immune received neoadjuvant therapy; 18.9% of patients had responses, while cytosolic RNA sensors use MAVS for cirrhosis. The indication for surgery included second- signal transduction. Although stimulation of cytosolic ary malignancy (49.3%), primary hepatobiliary cancer nucleic acid sensors by pathogen-derived nucleic acids are (29.0%), or benign disease (21.8%). Among patients important for initiating anti-microbial defense, inadvertent with benign disease undergoing LH, RH, and EH, stimulation through self-derived nucleic acids may fl PHLF benchmark values were 1.1%, 4.3%, and 4.1%, contribute to autoin ammation and cancer. fi respectively. In contrast, among patients with a ma- Methods: Mice with a combined homozygous de ciency lignant indication for surgery, PHLF benchmark values of MAVS and STING underwent partial hepatectomy at the ranged widely from 2.2% for patients undergoing LH age of 10 to 12 weeks. After surgery, we determined liver- for secondary malignancies to 28.3% for patients un- to-body-weight ratio and performed BrdU labeling as well dergoing RH for primary hepatobiliary cancers after as immunohistochemistry. In addition, we performed biliary stenting. Western blotting for p-STAT3, p-RB, p-CDK2, cyclin E1, cyclin D1, and CDK1 as well as quantitative RT-PCR for p21 and SOCS3. Results: Our analysis directly demonstrated that defective cytosolic nucleic acid sensing impairs hepatocyte prolifer- ation and delays the recovery of organ mass. Whereas combined MAVS and STING deficiency did not influence upregulation of the G1 phase cyclins D1 and E1, it sub- stantially reduced the hyperphosphorylation of RB protein, attenuated the activating phosphorylation of CDK2, and impaired both CDK1 upregulation and hepatocyte BrdU incorporation. Importantly, expression of the cell cycle inhibitor p21 was markedly elevated in mutant as compared with wildtype mice. Conclusions: Here, we report on a previously unrecog- nized role of cytosolic nucleic acid sensors for organ regeneration. Our findings suggest that deficiency of cytosolic nucleic acid sensing pathways enhances expres- sion of p21 and impairs hepatocyte cell cycle progression beyond the G1/S checkpoint leading to a marked impair- ment of liver regeneration after partial hepatectomy.

FP11-09 SEQUENTIAL TRANSARTERIAL CHEMOEMBOLIZATION AND PORTAL VEIN EMBOLIZATION VERSUS PORTAL VEIN EMBOLIZATION ALONE BEFORE MAJOR HEPATECTOMY FOR Figure PATIENTS WITH LARGE HEPATOCELLULAR CARCINOMA M. Terasawa, M. -A. Allard, N. Golse, O. Ciacio, Conclusion: While the previously proposed incidence G. Pittau, A. Sa Cunha, D. Castaing, D. Cherqui, “ ” of 5% might be the best achievable incidence of R. Adam and E. Vibert PHLF among healthy patients undergoing major hepa- Paul Brousse Hospital, France tectomy, this PHLF benchmark value may not be Introduction: The value of sequential transarterial applicable to certain patients including those with ma- chemoembolization (TACE) and portal vein embolization lignant indications, neoadjuvant chemotherapy, and (PVE) compared to PVE alone before major hepatectomy biliary stenting. for large hepatocellular carcinoma (HCC) remains little

HPB 2018, 20 (S2), S182eS294 Free Orals (FP01-FP31) S217 known. We aimed at comparing early and long-term out- recovery was 9 days (95%CI 6-12) after LPD versus 8 days comes of these two strategies. (95%CI 6-10) after OPD (P=0.90). The conversion rate of Methods: We included all consecutive patients with large LPD was 20%. Operative blood loss was 300 versus 400 HCC (> 50 mm) scheduled treatment for TACE-PVE or mL (P=0.20) and operative time 391 versus 235 minutes PVE alone before major hepatectomy from January 2005 to (P=0.001), for LPD and OPD respectively. Clavien-Dindo December 2015 in a single institution. The Hypertrophy of 3 complications (25(50%) versus 20(41%), P=0.40), the future remnant liver, morbidity, mortality and survival pancreatic fistula (B/C) (15(30%) versus 14(29%), P=0.90), (Overall survival (OS) and Recurrence-free survival (RFS)) hepaticojejunostomy leakage (B/C) (5(10%) versus were retrospectively compared. A Cox multivariate model 6(12%), P=0.2), and hospital stay (11 versus 10 days, was used for adjusting on confounders. P=0.60) were comparable for LPD and OPD. Results: A total of 56 patients were included as follow: Conclusion: This early terminated randomized multicenter TACE-PVE (n=27) and PVE alone (n=28). Baseline patient trial showed comparable time to functional recovery, and and tumor characteristics were similar in the two groups. morbidity after LPD versus OPD. The high mortality rate After TACE, no disease progression has been observed. after LPD is worrisome and does not align with the prior Tumor downstaging lead to change the strategy as follow: training program. limited resection (n=3) and liver transplantation (n=1). Hypertrophy of the future remnant liver after PVE was significantly higher after TACE (P=0.03). This effect of FP12-02 fi TACE remained signi cant after adjusting on the non-tu- DEVELOPMENT OF AN ENTIRELY moral parenchyma. Resection was finally achieved in 24 patients (88.9%) of the TACE-PVE group versus 19 pa- PREOPERATIVE RISK tients (67.9%) of the PVE alone group. The TACE-PVE STRATIFICATION SCHEME FOR group experienced significantly improved OS compared to CLINICALLY RELEVANT PANCREATIC PVE alone (P=0.03). FISTULA FOLLOWING Conclusion: Our results suggest that TACE before PVE PANCREATICODUODENECTOMY increases the degree of hypertrophy of the future remnant R. J. Ellis1,2, D. B. Hewitt2, R. P. Merkow2, J. B. Liu1, liver after PVE and yields improved oncological outcomes 1,2 2 2 in patients with large HCC planned for major hepatectomy. K. Y. Bilimoria , D. J. Bentrem and A. D. Yang 1American College of Surgeons, and 2Department of Surgery, Northwestern University, United States FP12 - Free Papers 12 (long oral) - Pancreas: Introduction: Clinically relevant postoperative pancreatic fi Surgical Outcomes 1 stula (CR-POPF) rates remain high despite decreasing mortality after pancreaticoduodenectomy. Previously FP12-01 developed fistula risk scores require intraoperative data LAPAROSCOPIC VERSUS OPEN which limits their utility in the preoperative setting. Our PANCREATODUODENECTOMY objective was to develop a CR-POPF risk score using only (LEOPARD-2): A MULTICENTER preoperatively ascertainable variables. PATIENT-BLINDED, RANDOMIZED Methods: American College of Surgeons National Surgical Quality Improvement Program (ACS-NSQIP) targeted CONTROLLED TRIAL pancreatectomy variables were used to identify 10,809 1 1 2 J. van Hilst , T. de Rooij , M. F. Gerhards , pancreaticoduodenectomies. The cohort was randomly 3 2 4 3 I. H. de Hingh , T. M. Karsten , D. J. Lips , M. D. Luyer , divided into two groups for risk score model development 1 2 1 O. R. C. Busch , S. Festen , M. Besselink and for the and internal validation. Models included only routine pre- LEOPARD-2 investigators and the Dutch Pancreatic operative variables known prior to surgery. Cancer Group Results: The overall rate of CR-POPF was 17.7%. Three 1 2 3 Academic Medical Center, OLVG, Catharina Hospital, factors increased risk of CR-POPF: male (OR 1.33 [95%CI 4 and Jeroen Bosch Hospital, The Netherlands 1.18-1.51]), BMI>30 (OR 2.08 [95%CI 1.78-2.45]), and Introduction: Laparoscopic pancreatoduodenectomy normal pancreatic duct diameter (OR 1.82 [95%CI 1.51- (LPD) may improve postoperative recovery but concerns 2.20]). Four factors decreased risk of CR-POPF: insulin- exist regarding increased pancreatic fistula rates and, in dependent diabetes (OR 0.48, [95%CI 0.38-0.61]), receipt low-volume centers, increased mortality. In the of neoadjuvant therapy (OR 0.59 [95%CI 0.45-0.78]), Netherlands, LPD was introduced within the LAELAPS-2 elevated bilirubin (OR 0.71 [95%CI 0.58-0.86]), and training program (114 LPDs, 3.5% 90-day mortality), pancreatic adenocarcinoma pathology (OR 0.68 [95%CI followed by a randomized controlled trial (RCT). 0.56-0.81]). Patients were grouped into low, moderate, Method: This multicenter patient-blinded RCT was high, and very high risk groups by risk score with CR- performed in 4 centers that each perform 20 pancreato- POPF rates of 6.7%, 12.3%, 19.2%, and 33.0%, respec- duodenectomies annually (median 37 (range 23-77)), tively. Stratification was similar in the validation cohort completed the LPD training program, and had performed (Figure). 20 LPDs (range 23-34). Patients with a neoplasm without Conclusions: This post-pancreaticoduodenectomy fistula signs of vascular involvement were included. Primary risk score is the first derived using entirely preoperative outcome was time (days) to functional recovery. variables and is a useful tool for risk stratification, patient Results: The LEOPARD-2 trial was stopped early after 99 counseling, and perioperative care planning. Preoperative of the projected 136 patients were included, because of a estimation of CR-POPF risk can inform decision-making difference in 90-day complication-related mortality (LPD regarding early somatostatin analogue therapy, enhanced 5/50(10%) vs. OPD 1/49 (2%), P=0.20). Time to functional recovery protocols, and drain management.

HPB 2018, 20 (S2), S182eS294 S218 Free Orals (FP01-FP31)

P>0.729), mortality (4.0% vs 2.9%, P=0.314), R1-re- sections (20 vs 23%, P=0.269), and hospital stay (mean 17 vs 17 days, P>0.99) were similar, POPF occurred more frequently after MIPD (23% vs 14%, P< 0.001). The association between MIPD and major morbidity (OR 0.95, P=0.686), mortality (OR 1.57, P=0.121), and POPF (OR 1.84, P< 0.001) remained similar after excluding, respectively, lower volume (10-20 MIPD/year) centers, first 10 MIPD cases, hybrid cases, hybrid and laparo- scopic cases. Conclusions: In this largest propensity-score-matched study on MIPD vs OPD to date, we found no signif- Figure [Clinically Relevant Fistula Rates by Risk Group] icant differences in major morbidity, mortality, R1- resection rate, and hospital stay, but more POPF after MIPD. MIPD can safely be performed in high-volume centers, but the increased rate of POPF should be FP12-03 addressed. MINIMALLY-INVASIVE VERSUS OPEN PANCREATODUODENECTOMY: A PAN- EUROPEAN PROPENSITY-SCORE FP12-04 MATCHED ANALYSIS IN HIGH- MINIMALLY INVASIVE VOLUME CENTERS PANCREATICODUODENECTOMY: IS S. Klompmaker1, J. van Hilst1, U. Wellner2, I. Khatov3, THERE A LEARNING CURVE M. Abu Hilal4, D. Fuks5, I. Poves6, T. Keck2, U. Boggi7, ASSOCIATED WITH MORTALITY? M. G. Besselink1 and The European consortium on R. Torphy, C. Friedman, A. Halpern, S. Ahrendt, Minimally Invasive Pancreatic Surgery (E-MIPS), the M. McCarter, R. Schulick and A. Gleisner Dutch Pancreatic Cancer Group (DPCG), and the Department of General Surgery, University of Colordao, Deutschen Gesellschaft für Allgemein- und United States Viszeralchirurgie (DGAV) Minimally invasive pancreatidoduodenectomy (MIPD) 1 Department of Surgery, Cancer Center Amsterdam, Ac- is becoming more prevalent at both high and low volume ademic Medical Center, University of Amsterdam, The centers, but the presence of a learning curve has not been 2 Netherlands, Department of Surgery, University Medical evaluated nationally. The objective of this study was to 3 Center Schleswig-Holstein, Germany, Department of assess the presence of a MIPD learning curve for mor- Surgery, Moscow Clinical Scientific Center n.a. A.S. tality at an institutional level using the National Cancer 4 Loginov, Russian Federation, Department of Surgery, Database (NCDB). Southampton University Hospital NHS Foundation Trust, 2,867 patients with pancreatic cancer undergoing cura- 5 United Kingdom, Department of HPB Surgery & Liver tive intent MIPD (laparoscopic or robotic) from 2010-2014 Transplantation, Beaujon Hospital, Assistance Publique were included in the study. Annual cumulative MIPD Hôpitaux de Paris, University Paris VII, France, (cMIPD) volume was calculated for each institution and 6 Department of Surgery, Hospital del Mar, Spain, and used as a surrogate for experience. Average annual volume 7 Department of General and Transplant Surgery, Cisa- of pancreaticoduodectomy (MIPD and open) was also nello University Hospital, University of Pisa, Italy calculated. A natural log transformation was performed for Background: We assessed short-term outcomes after cMIPD volume to better model its non-linear relationship minimally invasive (laparoscopic, robot-assisted, hybrid) with the probability of death. A random effects logistic pancreatoduodenectomy (MIPD) versus open pancreato- regression model was used to examine the adjusted asso- duodenectomy (OPD) among European high-volume (>10 ciation between 90-day mortality and cMIPD volume. MIPDs and >20 PDs overall per year) centers. cMIPD volume ranged from 1 to 246 cases (median 13, Methods: Multicenter propensity-score-matched (1:1) IQR 5-33). 90-day mortality was 5.02% (n=144) among all retrospective study on MIPD vs OPD for (pre-)malignant patients. Unadjusted 90-day mortality decreased as cMIPD tumors or cysts in 14 European MIPD centers (2012-2017) volume increased (Quartiles: 1-4 cases=7.99%; 5-12 and OPD data from Dutch and German pancreatic surgery cases=5.72%; 13-32 cases=3.76%; 33 cases=2.12%, p< registries (2014-2017). Propensity scores were based on 0.001). On multivariable analysis, increasing cMIPD age, sex, BMI, ASA, comorbidities, ECOG, tumor location, volume was associated with a decreased OR for 90-day suspected cancer, organ involvement, and venous resec- mortality (OR=0.72, p=0.002). There was no interaction tion. Primary outcome: 30-day major morbidity (Clavien- between cMIPD volume and annual pancreaticoduode- Dindo 3a-5). Secondary outcomes: grade-B/C pancreatic nectomy volume indicating the effect of cMIPD volume is fistula (POPF), R1-resection (< 1 mm) margin, hospital independent of hospital volume. stay and 30-day mortality. Nationally, there is a learning associated with 90-day Results: Of 4220 included patients, all 730 MIPD (413 mortality as institutions introduce MIPD. Predicted 90-day laparoscopic, 184 robot-assisted, 130 hybrid) were mortality after MIPD is significantly < 5% after 17 cu- matched to 730 OPD. Major morbidity (28% vs 29%, mulative cases (Figure 1).

HPB 2018, 20 (S2), S182eS294 Free Orals (FP01-FP31) S219

groups, incidence of CR-POPF was 2.3%, 8%, 14.2% and 17.5% in the no risk group, low risk group, moderate risk group and high risk group(p< 0.001). By comparison, ac- cording to a-FRS, CR-POPF occurred in 6.7%, 13.4% and 21.6% of patients in the low risk group, intermediate risk group and high risk group(p< 0.001). However, discrimi- nation with area under curve(AUC) was only 0.629 in o- FRS and 0.622 in a-FRS.

Figure 1. [MIPD Learning Curve]

FP12-05 VALIDATION OF ORIGINAL FISTULA Figure [ROC curve of o-FRS and a-FRS] RISK SCORE AND ALTERNATIVE Conclusion: o-FRS and a-FRS could reflect the incidence FISTULA RISK SCORE IN of CR-POPF to some extent, but some risk factors were POSTOPERATIVE PANCREATIC considered to have no or low statistical significance. These FISTULA measures are also of low value as predictive models, and Y. Ryu, I. W. Han, D. W. Choi, S. H. Choi, J. S. Heo, further research is needed to modify the FRS. Y. H. You, S. Han and D. J. Park Department of Surgery, Samsung Medical Center, Sung- FP12-06 kyunkwan University School of Medicine, Republic of HOW TO PREDICT CLINICALLY Korea RELEVANT POSTOPERATIVE fi Introduction: Postoperative pancreatic stula (POPF) is PANCREATIC FISTULA (CR-POPF, the most serious surgical complication after pancreatico- duodenectomy(PD). In 2013, Fistula Risk Score (FRS) was GRADES B AND C) FOLLOWING developed to assess the risk of clinically relevant post- PANCREATODUODENECTOMY (PD)? operative pancreatic fistula(CR-POPF). In 2017, the alter- N. Tabchouri1, H. Hermand1, O. Benoit1, J. -C. Loiseau2, native Fistula Risk Score (a-FRS) was proposed. In this S. Dokmak1, B. Aussilhou1, S. Gaujoux1, study, we validate the o-FRS and a-FRS to assess statistical O. Soubrane1 and A. Sauvanet1 significances following PD. 1Beaujon Hospital, Assistance Publique Hôpitaux de Method: From January 2007 to December 2016, a Paris, University Paris VII, France, and 2Ecole Nationale total of 1771 patients underwent PD for primary peri- Supérieure d’Arts et Métiers, France fi fi ampullary cancers. POPF was de ned and classi ed Introduction: CR-POPF remains the most dreaded according to the International Study Group for Pancre- complication following PD, occurring in 10-20% of cases. atic Fistula(ISGPF), 2016. All data were reviewed This study aimed to determine a predictive risk score for retrospectively. CR-POPF following PD, and compare it to existing risk Results: Pathologic diagnosis other than ductal adenocar- scores. cinoma, pancreas duct diameter and BMI were independent Methods: Data from all patients who underwent open PD risk factors of CR-POPF. However, pancreas texture and procedure between 2012 and 2017 were collected from a intraoperative blood loss volume were not associated with prospectively maintained database. Identified CR-POPF CR-POPF. risk factors were used to build a risk score, which was According to o-FRS, the CR-POPF incidence increased compared to currently used risk scores such as FRS (Call- < as the o-FRS score increased(p 0.001). Within o-FRS ery; JACS:2013) and NSQIP-FRS (Kantor; JACS:2017).

FP12-05 Table [Risk factor analysis for CR-POPF] Risk factor Univariate Multivariate HR 95% CI P HR 95% CI P Gland texture(soft) 1.515 1.143 - 2.009 0.004 1.099 0.816 - 1.480 0.534 Pathology (not PDAC or pancreatitis) 3.142 2.129 - 4.635 <0.001 2.718 1.833 - 4.030 <0.001 P-duct size, per mm increase 0.831 0.764 - 0.903 <0.001 0.866 0.798 - 0.939 0.001 Estimated volume loss 1.000 1.000 - 1.000 0.827 BMI 1.110 1.065 - 1.157 <0.001 1.093 1.047 - 1.140 <0.001

HPB 2018, 20 (S2), S182eS294 S220 Free Orals (FP01-FP31)

Results: 448 patients were included (56% men, 24.3kg/m2 5.0)mm, and a majority (53.1%) had hard gland. median BMI and 54% had pancreatic adenocarcinoma). 164(11.7%) patients developed grade B/C fistula. Soft pancreatic texture was noted in 46% of patients, Upon external validation an AUC of 0.82(95%CI:0.79- median main pancreatic duct size was 4mm and median 0.86) was observed for a-FRS, as compared to AUC of intraoperative blood loss was 300cc. 90-day postoperative 0.81(95%CI:0.78-0.84) for FRS (p=0.447)(Figure-1). The mortality rate was 1.8%, CR-POPF rate was 22%, post- mean observed risk for low-, intermediate-, and high-risk operative hemorrhage was 9% and Clavien-Dindo grades 3- groups as defined by a-FRS was 3.5%, 10.9%, 28.4% 4 postoperative complications reached a 14% rate. Inde- respectively (p< 0.001). pendent CR-POPF risk factors were: male gender, Conclusion: In this external validation, the a-FRS BMI>25 kg/m2, absence of preoperative radiotherapy and performed well and was comparable to the FRS. Further- soft pancreatic texture. Within CR-POPF sub-groups, more, based on the suggested cut-off, a significant and intraoperative blood loss was an independent predictive clinically relevant risk stratification was achieved. factor of grade C POPF. FRS and NSQIP-FRS AUCs reached 0.736 and 0.751 respectively. With presently identified risk factors, risk score reached an AUC of 0.792. FP12-08 fi Based on this modi ed score, negligible, low, intermediate DEFINING THE PRACTICE OF DISTAL and high risk patients presented with 4%, 8%, 19% and 66% CR-POPF respectively. PANCREATECTOMY AROUND THE Conclusions: Modern CR-POPF risk scores should include WORLD factors associated with present therapeutic management, L. Maggino1,2, G. Malleo2, C. Bassi2 and C. Vollmer1 including preoperative radiotherapy. Intraoperative 1Surgery, University of Pennsylvania, United States, and bleeding no longer represents an overall CR-POPF risk 2Surgery, University of Verona, Italy factor but remains associated with grade C POPF. Introduction: Despite considerable efforts, best manage- ment practices for distal pancreatectomy (DP) have not been conclusively defined. Surgeons’ choices are therefore FP12-07 often based on their backgrounds or surgical dogma, VALIDATION OF THE ALTERNATIVE possibly leading to substantial variation in the practice of FISTULA RISK SCORE FOR DP. PANCREATODUODENECTOMY BY A Methods: A survey assessing experience and management approaches for DP was distributed worldwide, in 8 native- SINGLE HIGH VOLUME CENTER IN language translations, through 56 surgical societies THE UNITED STATES (including the IHPBA). To evaluate global variance, re- A. Javed, D. Ding, M. Wright, J. Wang, I. C. Ye, gions were clustered: North America, South/Central J. Cameron, R. Burkhart, J. He, C. Wolfgang and America, Asia/Australia, and Europe/Africa. M. Weiss Results: 797 surgeons from 68 nations responded (median Johns Hopkins Hospital, United States age=47; years of experience=14; fellowship trained=62%). Introduction: Postoperative pancreatic fistula (POPF) is Most are HPB surgeons (61%) - greatest in Asia/Australia < the most common cause of postoperative morbidity after (76%, p 0.001) - with only 7% practicing pancreas pancreatoduodenectomy (PD). The alternative fistula risk surgery exclusively. Their median annual and career vol- score (a-FRS), a modified version of the fistula risk score umes are 6 and 46 DPs, respectively, with major regional < (FRS), has been proposed to predict the risk of POPF variations (p 0.001). Experience with minimally invasive after PD. The aFRS eliminates requirement of estimated (MI) techniques is also diverse-highest in North America < blood loss for prediction. The aim of this study was to (p 0.001). Laparoscopy is the most common MI validate the performance of the a-FRS and compare it to approach (85%), while robotics is rarely performed that of FRS. outside the US/Canada. The preferred means of pancreatic Methods: Patients undergoing PD at Johns Hopkins Hos- remnant closure is via stapler (65%) - more commonly pital between 2010 and 2015 were identified and clinico- applied in North America than in Europe/Africa (81% vs < pathological data were extracted from institutional 53%, p 0.001). Management techniques for the remnant fi fi database. Missing data were estimated using multiple im- and other stula mitigation strategies display signi cant putations. Performance of a-FRS and FRS was assessed by regional variability (Table). Also the use of drains is evaluating area under receiver-operating curve (AUC), and diverse, with the biggest disparity between North Amer- DeLong test was used to compare difference between ican and Asian surgeons (selective and routine drainers, AUCs. Observed mean risk was calculated for risk groups. respectively). Results: A total of 1406 patients were included and the Conclusions: Globally, there is wide regional variability in mean age was 62.9Æ11.7 years. A majority was the practice of DP. Many of these choices are not evidence- female(53.5%), and white (79.2%). The median BMI was based, precluding equipoise in management and, possibly, 25.1(IQR:22.3-28.5), median duct size was 3.4 (IQR:1.9- optimized outcomes.

HPB 2018, 20 (S2), S182eS294 Free Orals (FP01-FP31) S221

FP12-08 Table Variable North South/ Europe/ Africa/ Asia/ Oceania Overall p-value America Central Middle East America Surgeons, n (%) 182 (25.2) 109 (15.1) 195 (27.0) 235 (32.6) 721 (100) . Primary hand-sewn closure of a non- <0.001 stapled.cd transection line Never 56 (32.2) 29 (30.2) 40 (21.4) 54 (23.7) 179 (26.1) Occasionally (1-25%) 80 (46.0) 36 (37.5) 62 (33.2) 78 (34.2) 256 (37.4) Sometimes (26-7$%) 19 (10.9) 6 (6.3) 27 (14.4) 22 (9.6) 74 (10.8) Frequently (76-99%) 12 (6.9) 15 (15.6) 32 (17.1) 29 (12.7) 88 (12.8) Always 7 (4.0) 10 (10.4) 26 (13.9) 45 (19.7) 88 (12.8) Selective suture ligation of the main <0.001 pancreatic duct Never 51 (29.5) 31 (32.6) 36 (19.0) 58 (25.3) 176 (25.7) Occasionally (1 -25%) S3 (48.0) 23 (24.2) 59 (31.2) 64 (27.9) 229 (33.4) Sometimes (26-75%) 18 (10.4) 13 (13.7) 31 (16.4) 19 (8.3) 81 (11.8) Frequently (76-99%) 16 (9.2) 17 (17.9) 39 (20.6) 36 (15.7) 108 (15.7) Always 5 (2.9) 11 (11.6) 24 (12.7) 52 (22.7) 92 (13.4) Over-sew a staple-line transection with 0.024 additional suture Never 63 (35.6) 24 (23.8) 65 (35.3) 90 (40.5) 242 (35.4) Occasionally (1-25%) 68 (38.4) 35 (34.7) 59 (32.1) 80 (36.0) 242 (35.4) Sometimes (26-75%) 23 (13.0) 20 (19.8) 26 (14.1) 27 (12.2) 96 (14.0) Frequently (76-99%) 19 (10,7) 11 (10.9) 24 (13.0) 15 (6.8) 69 (10.1) Always 4 (2.3) 11 (10.9) 10 (5.4) 10 (4.5) 35 (5.1) Synthetic Mesh on the remnant <0.001 Never 139 (79.9) 81 (88.0) 159 (88.3) 141 (64.7) 520 (78.3) Occasionally (1-25%) 9 (5 2) 4 (4.3) 14 (7.8) 36 (16.5) 63 (9.5) Sometimes (26-75%) 7 (4 0) 4 (4.3) 5 (2.8) 12 (5.5) 28 (4.2) Frequently (76-99%) 16 (9.2) 2 (2.2) 2 (1-1) 12 (5.5) 32 (4.8) Always 3 (1-7) 1 (1.1) 0 17 (7.8) 21 (3.2) Biological Sealant on the remnant <1.001 Never 110 (62.5) 67 (72.0) 104 (57.1) 116 (52.7) 397 (59.2) Occasionally (1-25%) 32 (18.2) 17 (18.3) 47 (25.8) 41 (18.6) 137 (20.4) Sometimes (26-75%) 17 (9.7) 4 (4.3) 14 (7.7) 23 (10.5) 58 (8.6) Frequently (76-99%) 13 (7.4) 4 (43) 15 (82) 18 (8.2) 50 (7.5) Always 4 (2.3) 1 (1.1) 2 (1.1) 22 (10.0) 29 (4.3) Autologous tissue patch on the 0.003 remnant Never 89 (51.1) 67 (73.6) 130 (71.8) 145 (65.6) 431 (64.6) Occasionally (1-25%) 37 (21.3) 14 (15.4) 32 (17.7) 39 (17.6) 122 (18.3) Sometimes (26-75%) 20 (11.5) 3 (3.3) 11 (6.1) 19 (8.6) 53 (7.9) Frequenüy (76-99%) 20 (11.5) 4 (4.4) 6 (3.3) 11 (5.0) 41 (6.1) Always 8 (4.6) 3 (3.3) 2 (1.1) 7 (3.2) 20 (3.0) Pancreatico jejueal anastomosis/ 0.007 pancreaticogastrostomy Never 148 (84.6) 80 (88.8) 150 (83.8) 177 (81.2) 555 (83.8) Occasionally (1-25%) 24 (13.7) 3 (3.3) 24 (13.4) 37 (17.0) 88 (13.3) Sometimes (26-75%) 2 (1.1) 7 (7.8) 4 (2.2) 4 (1.8) 17 (2.6) Frequently (76-99%) 1 (0.6) 0 1 (0.6) 0 2(0.3) Always 0 0 0 0 0 Prophylactic trans-papillary pancreatic 0.027

(continued on next page)

HPB 2018, 20 (S2), S182eS294 S222 Free Orals (FP01-FP31)

(continued) Variable North South/ Europe/ Africa/ Asia/ Oceania Overall p-value America Central Middle East America stent placement Never 153 (85.0) 85 (85.9) 157 (84.4) 188 (83.2) 583 (84.4) Occasionally (1-25%) 26 (14.4) 7 (7.1) 26 (14.0) 33 (14.6) 92 (13.3) Sometimes (26-75%) 1 (0.6) 5 (5.1) 3 (1.6) 1 (0.4) 10 (1.4) Frequently (76-99%) 0 2 (2.0) 0 3 (1.3) 5 (0.7) Always 0 0 0 1 (0.4) 1 (0.1) Intraoperatively placed, externalized <0.001 drains Never 9 (4.9) 2 (1.8) 5 (2.6) 2 (0.9) 18 (2.5) Occasionally (1 -25%) 17 (9.3) 4 (3.7) 5 (2.6) 6 (2.6) 32 (4.4) Sometimes (26-75%) 13 (7.1) 3 (2.8) 5 (2.6) 10 (4.3) 31 (4.3) Frequently (76-99%) 42 (23.1) 15 (13.8) 31 (15.9) 33 (14.0) 121 (16.8) Always 101 (55.5) 85 (78.0) 149 (76.4) 184 (78.3) 519 (72.0) Prophylactic somatostatin analogues <0.001 Never 121 (67.6) 69 (65.7) 84 (44.0) 141 (61.0) 415 (58.8) Occasionally (1-25%) 26 (14.5) 15 (14.3) 26 (13.6) 37 (16.0) 104 (14.7) Sometimes (26–75%) 17 (9.5) 9 (8.6) 25 (13.1) 21 (9.1) 72 (10.2) Frequently (76–99%) 9 (5.0) 7 (6.7) 26 (13.6) 16 (6.9) 58 (8.2) Always 6 (3.4) 5 (4.8) 30 (15.7) 16 (6.9) 57 (8.1) Postoperative antibiotics specifically <0.001 for fistula prophylaxis/mitigation Never 153 (85.0) 52 (49.1) 130 (68.4) 96 (41.6) 431 (61.0) Occasionally (1-25%) 14 (7.8) 17 (16.0) 23 (12.1) 27 (11.7) 81 (11.5) Sometimes (26-75%) 6 (3.3) 8 (7.5) 17 (8.9) 23 (10.0) 54 (7.6) Frequently (76-99%) 4 (2.2) 16 (15.1) 10 (5.3) 23 (10.0) 53 (7.5) Always 3 (1.7) 13 (12.3) 10 (5.3) 62 (26.8) 88 (12.4)

*76 surgeons did not specify their region of practice

Methods: Among 80 randomly selected patients liver me- tastases, 70% and 30% were recognized as microsatellite FP13 - Free Papers 13 (mini oral) - Liver: stable and unstable, respectively. Samples of 2 ml of fresh Metastases 2 blood were collected preoperatively. The cell fraction pos- FP13-01 itive to CD309 (EPCS) was enumerated by flow cytometer EXPRESSION OF ENDOTHELIAL and expressed as the percentage of WBC. Serum samples for evaluation of the vascular endothelial growth factor (VEGF) PROGENITOR CELLS AND VASCULAR concentration were collected simultaneously for ELISA. A ENDOTHELIAL GROWTH FACTOR P-value < 0.05 was adopted as statistically significant in ACCORDING TO GENOMIC INSTABILITY multiparametric analysis to reveal the association of IN COLORECTAL LIVER METASTASES genomic instability and of angiogenesis. A control group of 36 volunteers (18 F, 18 M, ma 38) was also measured. W. Otto1, F. Macrae2, J. Sierdzinski3, M. Król4 and Results: The expression of EPCs and VEGF were signifi- E. Wolinska 5 cantly higher in MSI-H than MSS patients (0.01% versus 1General, Transplant & Liver Surgery, Medical University of 0.004% for EPCs, P< 0.008; 420.9 versus 300.2 pg/ml for Warsaw, Poland, 2Dept. Gastroenterology, Melbourne Royal VEGF). Multiparametric analysis showed angiogenesis Hospital, University of Melbourne, Australia, 3Medical nine times more active in MSI-H patients (OR=9.1, P< Informatics & Telemedicine, 4Oncology, Hematology & In- 0.01). They were characterized by more frequent metasta- ternal Medicine, Medical University of Warsaw, and 5Pa- ses in both liver lobes (OR=32.8 P< 0.001) and metastases thology, Medical University of Warsaw, Poland outside the liver (OR=8.3 P< 0.01). Introduction: Increased numbers of endothelial progenitor Conclusions: Patients with CRC metastases with MSI-H cells and high level of vascular endothelial growth factor have high circulating numbers of EPCs and serum levels of may indicate on higher tumor propensity to angiogenesis. VEGF. This may facilitate they regrowth. The data does The study aimed to evaluate the possible role of genomic point to the potential utility of anti-angiogenesis therapy in instability in stimulation of angiogenesis in liver metastases MSI-H metastatic disease. of colorectal carcinoma.

HPB 2018, 20 (S2), S182eS294 Free Orals (FP01-FP31) S223

FP13-02 Background: Aim of this study was to compare patients undergoing combined colorectal and hepatic surgery with RESECTION OF COLORECTAL LIVER and without neoadjuvant chemotherapy to clarify the METASTASES AFTER EARLY FAILURE prognostic advantage of preoperative oncological treatment OF ADJUVANT SYSTEMIC to identify factors predictive of good prognosis in a selected CHEMOTHERAPY FOR THE PRIMARY population of SCRLM. TUMOR: IS THERE A ROLE FOR Methods: 73 patients who underwent upfront elective combined surgery without preoperative CT for SCRLM in ADDITIONAL TREATMENT two European referral centers constituted the study group 1 1 1 2 T. Boerner , C. Zambirinis , J. Gagniere , J. Chou , (NoNACT Group). The NoNACT group was matched in a 2 1 1 1 M. Gonen , P. Kingham , P. Allen , J. Drebin , ratio of 1:1 with patients who were operated after chemo- 1 1 W. Jarnagin and M. D`Angelica therapy with neoadjuvant intent (NACT group). 1 2 Surgery, and Epidemiology-Biostatistics, Memorial Results: Preoperative characteristics of patients and dis- Sloan Kettering Cancer Center, United States ease were comparable between NoNACT and NACT Introduction: The utility of adjuvant systemic chemo- Group. In particular, median Fong score was 2 (1-3) both therapy after resection of colorectal liver metastasis in NoNACT and NACT group. The NoNACT Group, as (CRLM), especially after recent exposure to adjuvant compared to the NACT Group, had lower blood loss (450 treatment for the primary tumor, is controversial. This study mL versus 600 mL) with p=0.04. Postoperative stay and compared adjuvant therapy with hepatic artery infusion morbidity were comparable. 1 and 3-years disease free (HAI) and systemic chemotherapy (HAI+Sys) and Sys survival were respectively 83.6% and 69.9% in the alone to no treatment after resection of CRLM after recent NoNACT and 84.9% and 71.2% in the NACT group. At failure of adjuvant chemotherapy for the primary tumor. multivariable analysis, predictors of recurrence were: right Methods: Early CRLM was defined as the occurrence of colonic neoplasms, positive nodal status of the primary CRLM within 12 months of finishing adjuvant chemotherapy tumor, number of liver lesions, RAS mutational status, for their primary tumor. 240 patients who underwent a com- Fong score>3. plete resection of early CRLM for stage III CRC between Conclusion: Preoperative neoadjuvant chemotherapy in 1992 and 2014 were included from a single-center prospec- patients with colorectal cancer and synchronous resectable tively maintained database. Recurrence-free survival (RFS) liver metastases does not influence the risk of recurrence in and overall survival (OS) were estimated using the Kaplan- patients with favorable tumor biology (Fong < 3), while it Meier methods. A Cox proportional hazards model was used was associated with increased intraoperative blood loss. to evaluate the independent association between treatment There is no strong evidence, presently, to strongly recom- groups after liver resection and outcomes. mend upfront chemotherapy in the absence of negative Results: 79 pts (32.9%) received adjuvant HAI+Sys, 77 prognostic factors. (32.1%) adjuvant Sys alone and 84 (35%) no adjuvant therapy. Pts received HAI+Sys had a prolonged median RFS (HAI+Sys:1.8yr vs. Sys alone:1.2yr vs. Surveillan- FP13-04 ce:1.4yr,p=0.02) and OS (HAI+Sys:6.2yr vs. Sys HISTOPATHOLOGICAL GROWTH alone:4.0yr vs. Surveillance:4.1yr,p< 0.01). In multivari- able model, the risk of death or a recurrence was nearly PATTERNS AND POSITIVE MARGINS halved for pts treated with adjuvant HAI+Sys compared to AFTER RESECTION OF COLORECTAL pts with no adjuvant therapy (HR:0.6 [95%CI:0.43- LIVER METASTASES 0.93],p=0.02]. Other independent factors associated with P. M. H. Nierop1, E. P. van der Stok1, < < improved OS were largest tumor size 5cm (p= 0.01) B. Groot Koerkamp2, P. J. Allen3, W. R. Jarnagin3, < and left-sided colon tumors (p= 0.01). N. E. Kemeny4, T. P. Kingham3, D. J. Grünhagen1, Conclusion: Combined HAI and systemic chemotherapy is C. Verhoef1 and M. I. D’Angelica3 effective as adjuvant therapy after resection of early 1Surgical Oncology, 2Surgery, Erasmus University Medi- metachronous CRLM. cal Center, The Netherlands, 3Surgery, and 4Medical Oncology, Memorial Sloan Kettering Cancer Center, United States FP13-03 Introduction: It has been argued that positive resection TIMING OF PERIOPERATIVE margins after resection of colorectal liver metastases CHEMOTHERAPY DOES NOT (CRLM) are a reflection of underlying tumour biology. INFLUENCE LONG-TERM OUTCOME Histopathological growth patterns (HGPs) of CRLM are believed to be an expression of biological tumour behav- OF PATIENTS UNDERGOING iour. Three HGPs are distinguished; replacement, desmo- COMBINED COLORECTAL AND LIVER plastic and pushing HGP. The current study aimed to RESECTION IN BIOLOGICALLY investigate in a large multicentre cohort whether HGPs are FAVORABLE TUMORS associated to positive resection margins after resection of F. Ratti1, D. Fuks2, F. Cipriani1, B. Gayet2 and CRLM. Secondary objective was to assess prognostic value L. Aldrighetti1 of positive resection margins, while correcting for potential 1Hepatobiliary Surgery Division, IRCCS San Raffaele confounders including HGPs. Hospital, Italy, and 2Department of Digestive Disease, Methods: All consecutive patients treated surgically for Institut Mutualiste Montsouris, France CRLM between January 2000 and March 2016 at the Erasmus MC Cancer Institute or between January 2000 and

HPB 2018, 20 (S2), S182eS294 S224 Free Orals (FP01-FP31)

December 2012 at the Memorial Sloan Kettering Cancer Conclusion: Early liver-only and systemic recurrence after Center with known HGPs were considered for inclusion. LR for CRLM significantly decreased overall patient sur- Positive resection margin was defined as tumour cells at the vival and were associated with LNS. Early liver-only resection margin. recurrence was associated with R1 resection and patients Results: Of all patients (n=1302) included for analysis 13% benefit from AC. Early-systemic recurrence was associated (n=170) had positive resection margins. Replacement HGP with high hepatic tumour burden. These findings could was associated with a higher chance of a positive resection guide post-operative imaging protocols to detect early- margin (adjusted odds ratio (OR): 1.71, p=0.025). Higher disease recurrence. number of CRLM was also correlated with a greater chance of a positive resection margin (adjusted OR: 1.22, p< 0.001). FP13-06 After correction for several risk factors, including HGPs, IS R1 HEPATECOMY STILL RELEVANT positive resection margins remained significantly associ- ated with an impaired overall survival (adjusted hazard IN THE CONTEMPORARY ratio: 1.40, p=0.004). MANAGEMENT OF COLORECTAL Conclusion: The current study demonstrates that replace- LIVER METASTASES? LONG-TERM ment growth is associated with positive resection margins. OUTCOMES FROM A MULTI- This suggests that positive resection margins are at least INSTITUTIONAL BI-NATIONAL partially an expression of underlying tumour biology. ANALYSIS J. Hallet1, R. Memeo2, P. Karanicolas3, D. Goéré4, 5 6 7 FP13-05 E. Dixon , R. Hernandez-Alejandro , G. Martel , F. Navarro8, A. Sa Cunha9, P. Pessaux2 and French PREDICTORS OF EARLY LIVER-ONLY Colorectal Liver Metastases Working Group, Association AND SYSTEMIC RECURRENCE Française de Chirurgie (AFC) FOLLOWING LIVER RESECTION FOR 1Surgery, University of Toronto, Canada, 2Nouvel Hôpital COLORECTAL METASTASES Civil, France, 3University of Toronto, Canada, 4Institut Gustave Roussy, France, 5University of Calgary, 6Western Y. Longatto Boteon1,2, University, 7University of Ottawa, Canada, 8CHU Mont- A. Pinter Carvalheiro da Silva Boteon1, R. Laing1,2, pellier, and 9Hôpital Paul Brousse, France R. Haminder Bhogal1, R. Sutcliffe1, K. Roberts1, fi R. Marudanayagam1, J. Isaac1, P. Muiesan1 and D. Mirza1 Background: Bene ts of narrow margins below 1mm for 1Liver Unit, Queen Elizabeth Hospital Birmingham, and CRLMs resection have been suggested in historical co- 2National Institute for Health Research, Birmingham Liver horts, but remain controversial. We compared the out- £ > Biomedical Research Centre, University of Birmingham, comes of R1 (margin 1mm) to R0 (margin 1mm) United Kingdom hepatectomy with contemporary multi-modal management of CRLMs. Introduction: Early recurrence remains a major problem Methods: We performed a multi-institutional analysis of following liver resection (LR) for colorectal liver hepatectomy for CRLMs in France and Canada (2006- metastasis (CRLM). Understanding the factors that are 2013). Primary outcome of interest was overall survival associated with liver-only and/or systemic recurrence (OS). Multivariable regression analyses assessed the as- will assist in developing more effective post-operative sociation between R1 and OS, and factors associated surveillance protocols. The aim of this study was to with R1. investigate risk factors for early liver-only and systemic Results: Of 2439 hepatectomies, 985 (40.4%) had R1 recurrence. resections. 5-year OS was lower for R1 with 57.0% Methods: Retrospective review of a prospectively main- (95%CI 54.4-59.6%) compared to 70.4% (95%CI 68.6- tained database of patients undergoing first LR between 72.2) for R0 (p< 0.0001). After adjusting for country, 2005 and 2015 for CRLM. Early recurrence was defined as age, locally advanced primary, node positive primary, disease recurrence within 6 months following LR. A synchronous CRLM, more than 3 metastases, largest multivariable logistic regression model identified inde- metastasis over 5 cm, extra-hepatic disease, pre-hepa- pendent predictors of early-recurrence. tectomy chemotherapy, major liver resection, operative Results: 682 patients were included in the study. 48 pa- time, and perioperative transfusion, R1 was indepen- tients (7%) had liver-only early disease recurrence and 40 dently associated with increased risk of mortality (6%) had systemic recurrence. Median overall survival (in (HR 1.47, 95%CI 1.16-1.86). Factors independently months) was lower for the recurrence groups (liver-only associated with R1 included node positive primary 20 [Interquartile range:17-23] and systemic 19 [14-24] vs. (RR 1.26, 95%CI 1.03-1.55), more than 3 metastases no-recurrence 43 [37-45], both p< 0.001). Multivariable (RR 2.08, 95%CI 1.65-2.63), largest metastasis over analyses showed lymph node status of the primary tumour 5cm (RR 1.74, 95%CI 1.35-2.24), and receipt of (LNS) (p=0.003) and positive microscopic resection mar- pre-hepatectomy chemotherapy (RR 1.27, 95%CI gins (R1) (p=0.010) were independent risk factors for 1.03-1.58). liver-only recurrence, whereas LNS (p=0.002), major LR Conclusion: In a contemporary multi-institutional cohort (3 Couinaud segments) (p=0.038) and summative his- of resected CRLMs, R1 was independently associated with topathological tumour diameter (p=0.005) were indepen- inferior long-term outcomes than R0 resection. Factors dent risk factors for systemic recurrence. Adjuvant associated with R1 can help identify higher risk patients chemotherapy (AC) significantly reduced liver-only pre-operatively. R1 yielded OS superior to known results of recurrence (p=0.029).

HPB 2018, 20 (S2), S182eS294 Free Orals (FP01-FP31) S225 systemic therapy. While the need for narrower margins FP13-08 should not preclude hepatectomy, clinical discussion and THE DESMOPLASTIC GROWTH counseling should be informed by this data. PATTERN PREDICTS IMPROVED SURVIVAL AFTER RESECTION OF FP13-07 COLORECTAL LIVER METASTASES ONCOLOGICAL IMPACT OF P. M. H. Nierop1, B. Galjart1, E. P. van der Stok1, R. R. J. Coebergh van den Braak2, S. Daelemans3,4, POSTOPERATIVE COMPLICATIONS 5 1 4,5 AFTER RESECTION OF L. Y. Dirix , C. Verhoef , P. B. Vermeulen and D. J. Grünhagen1 COLORECTAL LIVER METASTASES: 1Surgical Oncology, 2Surgery, Erasmus University Medi- SYSTEMATIC REVIEW AND META- cal Center, The Netherlands, 3Medical Biochemistry, ANALYSIS University of Antwerp, 4HistoGeneX, and 5Translational D. Dorcaratto1, G. Mazzinari2, M. Garces1, E. Muñoz1, Cancer Research Unit, University of Antwerp, Belgium M. Fernandez1, L. Sabater1 and J. Ortega1 Background and aims: Recent studies suggest that pa- 1Hospital Clinico Universitario de Valencia, and 2Grupo tients with colorectal liver metastases (CRLM) displaying a de Investigación en Medicina Perioperatoria, Hospital desmoplastic growth pattern (dHGP) have favourable Universitario La Fe, Spain prognosis when compared to patients with non-desmo- Introduction: The aim of this systematic review and meta- plastic (non-dHGP) CRLM. The current study aimed to analysis was to study the influence of postoperative com- determine the prognostic impact of dHGP in a large ho- plications (POC) on overall (OS) and disease free survival mogeneous cohort of chemonaive patients with long-term (DFS) after surgical resection of colorectal cancer liver follow-up after resection of CRLM. Secondary objectives metastases (CRLM). were to determine the prevalence of dHGP after pre-treat- Methods: Systematic review of the literature using vali- ment with chemotherapy and to assess the prognostic dated methods of the Cochrane Collaboration and preferred impact of dHGP amongst pre-treated patients. reporting items for systematic reviews and meta-analyses Methods: A cohort study was conducted, for which all (PRISMA) was performed in the following databases: consecutive patients undergoing complete surgical treat- PubMed/MEDLINE, EMBASE Databases, Web of Sci- ment of CRLM between January 2000 and March 2015 at a ence, the Cochrane library, and Google Scholar. Articles tertiary referral centre were considered for inclusion. The showing POC, OS and DFS rates after CRLM resection growth pattern was assessed according to international were included. The influence of POC on OS/DFS was consensus guidelines. calculated using Forrest plots with Mantel-Haenszel Results: In total 732 patients were included. Amongst method and random effect model. I2 was used to test the chemonaive patients (n=367), 68 patients (19%) with heterogeneity between the included studies. Newcastle- dHGP were observed. The presence of dHGP was associ- Ottawa scale (NOS) was used to assess the quality of the ated with improved progression free survival (PFS) included studies. (Hazard Ratio (HR): 0.54, p=0.002) and overall survival Results: 38 articles including 11741 patients fulfilled (OS) (HR: 0.40, p=0.001). fi the inclusion criteria. POC significantly decreased 3 After preoperative chemotherapy (n=365), signi cantly years OS (O.R. 1.70, C.I. 1.45-2.00), 5 years OS (O.R. more patients with dHGP were seen (n=109, 30%) (odds < 1.68, C.I. 1.46-1.93), 3 years DFS (O.R. 1.48, C.I. 1.24- ratio: 2.71, p 0.001). In this cohort of patients, no sig- fi 1.77) and 5 years DFS (O.R. 1.64, C.I. 1.34-2.00). ni cant effect of dHGP on PFS (adjusted HR: 0.76, Heterogeneity was law in all comparisons (I2< 30%). p=0.069) or OS (adjusted HR: 0.92, p=0.627) was found. None of the included studies reached the maximum Conclusion: The current study demonstrates that dHGP is NOS punctuation. associated with superior survival outcomes after resection Conclusion: This is the largest systematic review studying of CRLM. Preoperative chemotherapy is associated with the oncologic impact of POC after resection of CRLM different proportions of HGPs and the prognostic impact of published to date. POC negatively influence oncologic HGPs is reduced in pre-treated patients. outcomes. POC rates should be included in recurrence risk scores and considered as a confounding factor in studies designed to calculate the risk of death and recurrence in FP13-09 patients with CRLM. COMPARISON OF HIGH VERSUS LOW DOSE STEREOTACTIC BODY RADIATION THERAPY (SBRT) FOR HEPATIC METASTASES E. N. D. Kok1, E. P. M. Jansen2, M. E. Nowee2, D. Evers1, N. F. M. Kok1, T. J. M. Ruers1 and K. F. D. Kuhlmann1 1Surgical Oncology, and 2Radiation Oncology, Netherlands Cancer Institute, The Netherlands Background: SBRT is an alternative treatment for liver metastases when surgical resection is not feasible. Previous research showed better local control with high dose SBRT, FP13-07 Figure [Forrest plot comparing five years overall survival] which led to an adjustment of our hospital protocol in 2013.

HPB 2018, 20 (S2), S182eS294 S226 Free Orals (FP01-FP31)

This retrospective study evaluates high versus low dose bilobar), the median number of CTX-cycles was 3(1-6). SBRT of liver metastases. Ninety-day-mortality was 3.5%, median survival 54 Methods: Between 2009 and 2017, 90 patients with 97 months (95%CI:38.0-69.7). ETS and MC were signifi- liver metastases (77% colorectal origin) received SBRT. cantly associated with survival (ETS present/absent: 60.9 Patients were divided in a low dose group of 45Gy (group months (95%CI:49.3-72.5) vs. 33.7 months (95%CI:20.4- 1; n=33) and a high dose group of 50Gy (group 2; n=57). 47.1);p=0.026); MC response/no response: 85.5 months The median dose in group 1 and 2 was 37.5Gy (range: 36- (95%CI:41.6-129.3) vs. 45.3 months (95%CI:28.9- 45Gy) and 60Gy (range: 51-60Gy) in 3-5 fractions. Pri- 61.7);p=0.033); RECIST Response/no response: 66.4 mary outcome was local control and toxicity. Secondary months (95%-CI:51.0-81.7) vs. 38.7 months (95%- outcomes were overall survival and progression-free CI:32.1-45.3);p=0.059). ETS was especially useful for survival. patients with no morphological change of lesions. Results: The median follow-up through CT/MRI/PET-CT Conclusion: Early assessment of response by ETS signif- was 15 months. One year local control rate in group 1 was icantly predicts postoperative survival. ETS and MC both 64% (CI: 45-82%) and in group 2 95% (CI: 90-100%) (p = have individual strengths and limitations to be considered 0.004). Two year local control rate was 64% (CI: 45-82%) by radiologists and surgeons likewise. and 85% (CI: 70-99%), respectively. Grade 3 toxicity (requiring hospitalization) occurred in 3% in both groups. Grade 4-5 toxicity was not observed. The 2-year survival in FP13-11 group 1 and 2 was 53% (CI: 35-71%) and 75% (CI: 61- IMPAIRED KINETIC GROWTH RATE 90%) (P=0.06). The median progression-free survival was 5 and 8 months (P=0.04), respectively. AND INCREASED POSTHEPATECTOMY Conclusion: High dose SBRT provides significantly better LIVER FAILURE RATES AFTER ALPPS local control than low dose SBRT for liver metastases FOR COLORECTAL LIVER without increased toxicity. When surgical resection is not METASTASIS feasible, high dose SBRT provides an effective and safe M. Teutsch1,2, B. Andreas3, X. Jin4, L. Fischer2, treatment for liver metastases. B. Nashan2 and J. Li2 1General Surgery, Schoen Clinic Hamburg-Eilbek, 2Department of Hepatobiliary and Transplant Surgery, FP13-10 3Department of Hematology and Oncology, PROGNOSTIC VALUE OF EARLY University Clinic Hamburg-Eppendorf, Germany, and TUMOUR SHRINKAGE DURING 4Center of Evidence-Based Medicine, Fudan University, NEOADJUVANT CHEMOTHERAPY FOR China LONG-TERM SURVIVAL AFTER Introduction: Associating Liver Partition and Portal Vein RESECTION OF COLORECTAL LIVER Ligation for Staged Hepatectomy (ALPPS) is a fairly new operative technique, which pushes the limits of resectability METASTASIS of liver tumors. It has especially become a valid treatment N. Fadinger1, E. Braunwarth1, S. Biggel2, G. Göbel3, option for patients with advanced colorectal liver metastasis B. Cardini1, E. Gassner2, S. Schneeberger1, D. Öfner1, (CRLM). It has the potential to facilitate much faster hy- S. Stättner1 and F. Primavesi1 pertrophy of the future liver remnant (FLR), thereby 1Department of Visceral, Transplantation and Thoracic increasing resectability of previously unresectable CRLM, Surgery, 2Department of Radiology, and 3Department of yet posthepatectomy liver failure (PHLF) as well as Medical Statistics, Informatics and Health Economics, impaired hypertrophy of the FLR may still pose a problem Medical University Innsbruck, Austria in this setting. Introduction: Response evaluation after chemotherapy Methods: To further investigate and understand these (CTX) for colorectal liver metastasis (CRLM) is crucial to issues, we retrospectively analyzed all consecutive patients select patients, which benefit from surgery. Previously, undergoing ALPPS for resection of CRLM between RECIST criteria were used, but studies showed that non- January 2013 and March 2017 in search for potential risk size-based criteria like morphological-criteria (MC) based factors. Cases of posthepatectomy liver failure (PHLF) on lesions’ appearance on imaging may better predict sur- were graded according to the consensus of the International vival. Recently, early restaging within 6-12 weeks after Study Group of Liver Surgery (ISGLS). The kinetic growth initiation of CTX with evaluation of early tumour shrinkage rate (KGR, ml/d) was calculated as mean volume increase (ETS) was suggested to improve prognostic accuracy in per day between the two stages of the ALPPS procedure. oncological studies. However, ETS has not been evaluated Results: A total of 22 patients were included in the anal- in surgical cohorts. ysis. Seven patients (31.8%) developed PHLF (all grade B). Methods: Patients undergoing resection after preoperative Multivariate regression analysis showed a statistically sig- CTX for newly diagnosed CRLM between 2005-2016 were nificant association between preoperative chemotherapy analysed. Baseline imaging and first restaging were involving EGFR inhibitors (cetuximab, panitumumab) and compared regarding RECIST, MC and ETS (size reduction grade B PHLF (p< 0.01) as well as a reduced KGR (p< 20% 12 weeks). Prognostic values of response criteria 0.05). were statistically evaluated. Conclusions: Impaired liver regeneration following EGFR Results: Ninety-four of 137 patients received CTX before signaling inhibition has been shown in rodents, but not in resection. Of these, 58 underwent restaging 12 weeks and the clinical setting as of yet. Pending further investigation were evaluated for response. The median size of the largest of these phenomena, we suggest caution when considering lesion was 20mm, with a median of 2 lesions (1-8; 41% ALPPS in this subgroup of patients.

HPB 2018, 20 (S2), S182eS294 Free Orals (FP01-FP31) S227  FP13-12 Background: A 1-mm margin is standard for resection of colorectal liver metastases (CLM). However, R1 resection RIGHT PORTAL VEIN VERSUS RIGHT is not rare (10%-30%), and chemotherapy could mitigate its PORTAL VEIN WITH SEGMENT 4 impact. The possibility of detaching CLM from vessels (R1 EMBOLIZATION: ANALYSIS OF vascular margin) has been described. A reappraisal of R1 IMPACT ON FUTURE LIVER REMNANT resection is needed. VOLUME Methods: A 19-question survey regarding R1 resection for CLM was sent to hepatobiliary surgeons worldwide. Seven S. Ali1, H. Haq1, J. P. A. Lodge2, G. J. Toogood2, 3 3 clinical cases were included. J. Patel and C. Hammond Results: In total, 276 surgeons from 52 countries 1University of Leeds, 2HPB and Transplant Unit, and 3 completed the survey. Ninety percent reported a negative Department of Vascular Radiology, Leeds Teaching impact of R1 resection (74% local recurrence, 31% hepatic Hospitals NHS Trust, United Kingdom recurrence, and 36% survival), but 50% considered it Introduction: Right portal vein embolization [RPVE] may sometimes required for resectability. Ninety-one percent of be used to increase the volume of the future liver remnant responders suggested that the impact of R1 resection is [FRL] before major hepatic resection. It is not established modulated by the response to chemotherapy and/or CLM whether embolization of segment 4 in addition RPVE characteristics. Half considered the risk of R1 resection to [RPVE+4] induces greater hypertrophy of the FRL. Limi- be an indication for preoperative chemotherapy in patients tations of prior studies include heterogenous populations who otherwise underwent upfront resection, and 40% and absence of correction for patient size. modified the chemotherapy regimen when the tumor Methods: From 2010 to 2015, consecutive patients un- response did not guarantee R0 resection. Nevertheless, 80% dergoing RPVE or RPVE+4 for colorectal liver metastases scheduled R1 resection for multiple bilobar CLM that [CRLM], who had not undergone prior major hepatic responded to chemotherapy. Forty-five percent considered resection, were included in this retrospective study. Hos- the vascular margin equivalent to R0 resection. However, pital electronic records [HER] were used to abstract base- for lesions in contact with the right hepatic vein, right line data. Volumetric assessments of segments 2-3 were hepatectomy remained the standard. Detachment from the made on cross-sectional imaging before and after emboli- vein was rarely considered (10%), but 27% considered sation and corrected for body surface area. Survival was detachment in the presence of multiple bilobar CLM. assessed from the HER. Conclusions: A negative margin is still standard for CLM, Results: Of 105 patients undergoing PVE, 60 met the in- but R1 resection is no longer just a technical error. R1 clusion criteria. 38 underwent RPVE and 22 underwent resection should be part of the modern multidisciplinary, RPVE+4. 45 patients had undergone a mean of 5.5 cycles aggressive approach to CLM. of prior chemotherapy. 13 patients had FRL metastases at PVE and 14 had already undergone subsegmental meta- stasectomy from the FRL. Assessments of hypertrophy FP13-14 were made at 43.2 Æ 28.0 days (mean Æ SD) after PVE. RPVE+4 resulted in a significantly greater increase in MISSING COLORECTAL LIVER corrected FRL than RPVE alone (97.0 cm2/m2 vs 62.9 cm2/ METASTASES: IS IT TIME FOR A m2; p=0.008). Multivariate analysis indicated that only PARADIGM SHIFT? RPVE+4 and the presence of left lobe metastases were R. García-Pérez1, E. Ramirez2, P. Guarner2, M. Pavel2, fi signi cantly associated with increased FRL post emboli- V. Molina2, F. Al Shwely2, J. Ferrer2, J. Fuster2, sation. Median survival post PVE was 2.4 years. J. C. García-Valdecasas2 and C. Fondevila2 Conclusion: RPVE+4 results in greater FRL hypertrophy 1HPB and Transplant Surgery, and 2Hospital Clinic than RPVE alone in patients with CLRM. Barcelona, Spain The effectiveness of modern chemotherapy regimens means that colorectal liver metastases(CRLM) may no FP13-13 longer be detected on post-therapy cross-sectional imaging. R1 RESECTION FOR COLORECTAL These missing metastases(MM) represent a clinical chal- METASTASES: A SURVEY lenge for surgeons. While microscopic tumor rests might QUESTIONING SURGEONS ABOUT ITS still be present and progress if not removed, there are le- sions that present complete pathological response, upon INCIDENCE, CLINICAL IMPACT, AND which any further treatment could only result in morbidity. MANAGEMENT Methods: We prospectively reviewed liver resections G. Costa1, L. Vigano1, M. Cimino1, F. Procopio1, performed for CRLM at our center between 01/2012 and M. Donadon1, J. Belghiti2, N. Kokudo3, M. Makuuchi4, 12/2016, evaluating the rate of MM as well as clinical, J. -N. Vauthey5 and G. Torzilli1 radiological, and pathological features. 1Hepatobiliary and General Surgery, Humanitas Univer- Results: Liver resections were performed in 264 patients, sity, Humanitas Clinical and Research Center, Milan, among whom 16 presented MM, demonstrating complete Italy, 2Beaujon Hospital, France, 3University of Tokyo, pathological tumor response 12 cases(75%). Primary Japan, 4Japanese Red Cross Medical Center, Tokyo, tumor was located more frequently in colon, being all Japan, and 5The University of Texas MD Anderson Cancer cases synchronous metastases. Eighty-four percent of pa- Center, Houston, United States tients presented T3-T4 as primary tumor with 67% of

HPB 2018, 20 (S2), S182eS294 S228 Free Orals (FP01-FP31) positive nodes. Most of the patients(57%) were treated FP13-16 with FOLFOX scheme. The mean of follow-up was 52 months.The total number of MM identified on baseline INFLUENCE OF PRIMARY TUMOR imaging was 64.At surgical pathology, 52 of 64 lesions LOCATION ON RESECTABILITY OF resected demonstrated non-viable-tumor. Recurrence SYNCHRONOUS COLORECTAL LIVER occurred in 44% of patients with a mean follow-up of 10 METASTASES AND ONCOLOGIC months +/- 2.4 months of which 18%(3 patients) relapsed OUTCOMES AFTER LIVER RESECTION in situ. 1 1 1 Conclusion: In our series, the pathological response rate S. T. Alexandrescu , D. Anastase , C. Zlate , A. Diaconescu1, Z. Ionel1, R. Grigorie1, D. Hrehoret1, after chemotherapy in MM, arise up to 80%, which brings 1 2 1 us to put into question the current treatment paradigm. We V. Brasoveanu , G. Droc and I. Popescu 1Dan Setlacec Centre of General Surgery and Liver believe it is necessary to restate the surgical treatment in 2 MM when it implies extra risk for the patient, opening the Transplantation, and Center of Anesthesiology and door to a new era where observation can be a reasonable Intensivce Care, Fundeni Clinical Institute, Romania alternative. Introduction: Recent studies revealed that prognosis of patients with metastatic colorectal cancer is poorer in right colon (RC) primary tumors than in patients with left colo- FP13-15 rectal primaries (LC). However, it is not clear if this RAS observation is still valid in patients who underwent cura- MUTATION IS ASSOCIATED WITH tive-intent surgery, especially those with synchronous UNSALVAGEABLE RECURRENCE liver-only metastases (SCLMs). FOLLOWING HEPATECTOMY FOR Methods: Between January 2006 and December 2015, 872 COLORECTAL CANCER LIVER patients with SCLMs underwent surgery in our institution. METASTASES Of these, 237 presented initially resectable SCLMs, un- dergoing curative-intent surgery (simultaneous or staged E. Simoneau, M. Okuno, C. Goumard, S. Kopetz, E. Vega, hepatectomy). We compared the resectability rates of RC K. Joechle, T. Mizuno, C. -W. Tzeng, J. -N. Vauthey and vs. LC patients and also the oncologic outcomes after liver C. Conrad resection in RC vs. LC patients (cut-off was spleen flexure). MD Anderson Cancer Center, United States Results: Among the 872 patients with SCLMs, 178 presented Background: RAS mutation status predicts survival after RC tumors. Resectability rate of SCLMs was significantly hepatectomy for colorectal liver metastases (CRLM) and lower (p = 0.0375) in RC patients (37/178 - 20.78%) than in survival after repeat hepatectomy for intrahepatic recur- LC patients (200/694 - 28.81%). In patients who underwent rence. This study aimed at determining the impact of RAS curative surgery, the characteristics of RC group were similar mutation on amenability of recurrence to local therapy and to those of LC group, except for the significant lower number on post-recurrence survival following hepatectomy. of patients who underwent neo-adjuvant chemotherapy in the Methods: CRLM patients with recurrence at any location first group (10.81% vs. 33.16%, p = 0.0056) (Table). The after curative intent hepatectomy during 2006-2015 were median DFS and OS were not significantly different between retrospectively analyzed. Factors associated with recur- the RC and LC group (13 vs. 11 months - p = 0.946, and 41 vs. rence not amenable to local therapy and with post-recur- 36 months - p = 0.666, respectively). rence survival were evaluated. Conclusion: In patients with SCLMs, right-sided primary Results: Of 566 patients with recurrence, 309 (54.6%) tumors are associated with significantly lower resectability underwent chemotherapy only, 189 (33.4%) underwent rates. In patients who underwent liver resection, the loca- surgical resection, 47 (8.3%) underwent ablation, and 21 tion of the primary did not significantly influence the (3.7%) underwent radiation therapy. Median post-recur- oncologic outcomes. rence survival was significantly longer in patients with local Table [Characteristics of RC group vs. LC group] therapy than in those with chemotherapy only (65.1 months vs. 26.5 months, p< 0.0001). RAS mutation (p=0.01), Right-sided Left-sided p value presence of extrahepatic metastases (p=0.0006), and posi- primary primary (n[37) (n[200) tive surgical margin at prior hepatectomy (p=0.01) were associated with recurrence not amenable to local therapy. Male/Female 24/13 121/79 0.7143 RAS mutation (HR: 1.49, p=0.0012), disease-free interval Tis-T2/T3/T4 0/32/2 9/164/17 0.7083 < 12 months (HR: 1.76, p< 0.0001), multi-site recurrence N0/N1/N2 11/12/10 57/60/67 0.7934 < (HR: 1.71, p 0.0001), and recurrence not amenable to <5cm/>=5cm 24/13 146/54 0.3264 local therapy (HR: 4.11, p< 0.0001) were independent risk SCLMs factors for shorter post-recurrence survival. RAS mutation Single/Multiple 21/16 119/81 0.7201 was associated with poor post-recurrence survival in pa- SCLMs tients who received local therapy and who received Unilobar/Bilobar 25/12 149/51 0.4203 chemotherapy only. Minor/Major 31/6 169/31 1 Conclusions: RAS mutation predicts recurrence not hepatectomy amenable to local therapy and shorter post-recurrence sur- Simultaneous 31/6 152/48 0.3950 vival after hepatectomy for CRLM. In the event that local resections/Staged treatment is feasible in this patient cohort, ablation, radia- resections tion and resection of the recurrence lead to similar Neoadjuvant 4/33 67/133 0.0056 outcomes. chemo/No neoadjuvant chemo

HPB 2018, 20 (S2), S182eS294 Free Orals (FP01-FP31) S229

Methods: We categorized liver resections by the distance FP13-17 from bilio-vascular structures to bepreserved:Type I: A NEW STRATEGY BASED ON FDG-PET 10mm (R0 easy to obtain), Type II: 9-3mm , type III: < fi FOR MANAGING LIVER METASTASIS 3mm.We analyzed the prevalence of rst-shot pathological (fs-p) R0 (recuts excluded) and localrecurrences, and tested FROM COLORECTAL CANCER for differences the two sites of a joint University (U) and A. Watanabe1, K. Araki1, N. Harimoto1, T. Yoshizumi2, Regional Hospital (R) program. K. Arima3, Y. Yamashita3, H. Baba3, T. Higuchi1, Results: Between 2012-2015, 154 hepatectomies for H. Kuwano1 and K. Shirabe1 CRLM were performed (10450 at the UR). Fs-p R0 re- 1Gunma University School of Medecine, 2Kyushu Uni- sections in group I were 86/97 (88%), in group II 13 /29 versity, and 3Kumamoto University, Japan (44%), in group III 7/28 (25%). Local recurrences were 5/ Background: Effective prognostic models are needed to 97 in group I, 8/29 in group II, and 8/28 in group III (p< manage colorectal liver metastasis (CRLM). It is unsolved 0.01). In Group I, the rates of R0 resections, and of tech- problerm which is appropriate of surgery first and chemo- nical faults responsible for R1 resections, were similar at therapy first. Thus, we developed an algorithm to facilitate the U and R hospitals (50 of 56 - 89% at U vs 36 of 41 - treatment based on the standardized uptake value (SUV) 88% at R, p=0.8; and 3/56 at UH and 3/41 at RH, p=0.7). from fluorodeoxyglucose-positron emission tomography Conclusions: The categories of resections were associated (FDG-PET). to the probability of fs-p R0 and local recurrences. The rate Methods: We retrospectively evaluated 154 patients who of fs-p R0 in type I resections was similar in the two hos- underwent surgery for CRLM, including 112 cases that pitals, suggesting an equivalent surgical quality, and that involved primary surgery and 42 cases that involved pre- the fs-R0 rate could be a useful additional benchmarking operative chemotherapy before conversion surgery. We parameter for liver surgery. evaluated the relationship between the perioperative SUV and postoperative prognosis in the primary surgery cases, as well as the relationship between the SUV change rate FP14 - Free Papers 14 (long oral) - Pancreas: (post-chemotherapy SUV / pre-chemotherapy SUV) and Pancreatitis 2 prognosis after conversion surgery. FP14-01 Results: In the primary surgery group, recurrence-free EARLY CHOLECYSTECTOMY IN survival (RFS) was independently predicted by an SUV of 6.04 (P = 0.042) and 4 liver metastases (P = 0.003). In PATIENTS WITH MILD GALLSTONE addition, the combination of an SUV of 6.04 and 4 liver ACUTE PANCREATITIS: A metastase was strongly associated with poor RFS (p < RANDOMIZED PROSPECTIVE STUDY. 0.001). In the conversion surgery group, the SUV change CLINICALTRIALS.GOV. NCT02590978 rate was associated with tumour size change, CA19-9 F. Riquelme1,2, B. Marinkovic1,2, M. Salazar2, change, and pathological response. An SUV change rate of W. Martinez2, F. Catan2, S. Uribe-Echevarría2, 0.293 was associated with a shorter RFS (P = 0.006) and F. Puelma2, J. Muñoz2, G. Díaz1 and M. Uribe1,2 also independently predicted RFS (P = 0.026). 1Cirugía Oriente, Universidad de Chile, and 2Cirugía Conclusion: We established a therapeutic algorithm for Hepatobiliopancreática, Hospital del Salvador, Chile managing CRLM based on these results (Figure). FDG- PET may be a useful modality for predicting recurrence and Introduction: There are two strategies for laparoscopic prognosis in cases of CRLM, and our algorithm may be cholecystectomy (LC) in mild gallstone pancreatitis useful for managing multiple CRLMs. (MGP): delayed-LC (D-LC) or early-LC (E-LC, within 48- 72 h. of admission). Aim: To demonstrate that E-LC would result in shorter FP13-18 hospital stay for MGP. STANDARDIZING THE DIFFICULTY OF Methods: Single-centre prospective randomized clinical trial. Patients were enrolled from January 2015 to December 2017. OBTAINING OBJECTIVE FIRST-SHOT Inclusion criteria: age 18-70y, MGP by Atlanta criteria12. PATHOLOGICAL R0 MARGINS. AN Exclusion criteria: acute colangitis/cholecystitis, gastric by ADDITIONAL TOOL FOR pass, supraumbilical abdominal surgery, alcohol consump- BENCHMARKING THE QUALITY OF tion, chronic hepatic/pancreatic disease. Arms: E-LC (within LIVER RESECTIONS FOR 72 h. of admission) or D-LC. Primary outcome: hospital stay. COLORECTAL METASTASES Secondary outcomes: complications 30 d., ERCP/chol- edocholithiasis, rate of convertion and readmission. 1 2 3 3 R. Balzarotti , G. C. Vitali , A. Andres , I. Fournier , Results: 137 patients were admitted with acute pancreatitis. 3 1 1 3 C. Toso , V. Bianchi , R. Rosso , P. Morel , 26 patients where assigned to E-LC and 26 to D-LC. There 3 3 L. Rubbia-Brandt and P. Majno-Hurst were no significant differences in demographic variables, 1 2 Hospital of Lugano, visceral and HPB surgery, Hospital comorbidities, blood tests and common bile duct diameter 3 of Lugano, and University of Geneva, Switzerland between groups. On the E-LC group the surgery was Objectives: To use the rate of R0 resections as a quality performed in a mean time of 31 hours upon admission parameter for the surgery of colorectal liver metastases versus 125 hours on the D-LC group. Mean hospital stay (CRLM). was less in the E-LC group (2,2Æ1,1 d.) than D-LC group Background data: Quality assessment of hepatectomies is (6,2 Æ 2,3 d., P0,001). 7 patients in the E-LC group (7/26, generally based on morbidity, while this is not the main aim 26,9%) underwent an ERCP procedure versus 6 patients in of surgery. the D-LC group (6/26, 23,1%, P0,33). No patients from the

HPB 2018, 20 (S2), S182eS294 S230 Free Orals (FP01-FP31)

E-LC group presented complications versus 2 patients from Introduction: Most patients with chronic pancreatitis/CP the D-LC group (Clavien-Dindo< II). There were no suffer from strong, intractable pain. Recently, we could clearly conversions to open surgery in any of the groups. show that mast cells/MC were strongly increased around Conclusions: E-LC in patients with MGP significantly intrapancreatic nerves of patients with painful CP compared to reduces hospital stay with no impact in perioperative patient with CP without pain whereas other immune cells complications, technical difficulty or rates of chol- remain constant or were even strongly decreased. Therefore, edocholithiasis/ERCP. we considered that the inhibition of MC using MC-stabilizers as single or in combination with conventional analgesics may FP14-02 be a new therapeutic target against pain in CP. EARLY SURGERY VERSUS CURRENT Methods: Using repetitive intraperitoneal injections of STEP-UP PRACTICE FOR CHRONIC cerulean, acute pancreatitis/AP and CP were generated in C57BL/6-mice. Mice were randomized in six different PANCREATITIS: A MULTICENTER groups receiving vehicle, Ketotifen, Cromoglycate, Keto- RANDOMIZED CONTROLLED TRIAL tifen+Metamizol, Cromoglycate+Metamizol or Metamizol Y. Issa1, R. Kempeneers1, M. Bruno2, P. Fockens3, alone. von-Frey-Filament- and Open-Field-tests were J. W. Poley2, U. Ahmed Ali1, M. Besselink1, regularly raised to perform pain scores. H. van Santvoort4,5, M. Dijkgraaf6, M. Boermeester1 and Results: Dose-findings studies were performed to estimate the Dutch Pancreatitis Study Group doses of each drug (Metamizol 500mg/kg KG; Ketotifen 10mg/ 1Surgery, Academic Medical Center Amsterdam, kg KG; Cromoglycate 500mg/kg KG). Strikingly, whereas 2Gastroenterology, Erasmus MC, University Medical MC-stabilizers and Metamizol only showed a low potent and Center Rotterdam, 3Gastroenterology, Academic Medical moderate analgesic effect as single drugs, respectively, the Center Amsterdam, 4Surgery, St Antonius Hospital, 5Sur- combination of Ketotifen+Metamziol and Cromoglycate+Me- gery, University Medical Center Utrecht, and 6Clinical tamizol strongly decreased pain scores of von-Frey-Filament- Epidemiology, Academic Medical Center Amsterdam, The tests and leads to a nearly complete pain relief in mice with in AP Netherlands and CP, which was confirmed by the Open-Field-Test. Background: Surgical intervention for patients with Conclusion: Combination of MC-stabilizers and Meta- chronic pancreatitis is currently used as last resort treatment mizol has a high potent analgesic effect in Cerulean- when both medical and endoscopic treatment have failed. induced AP and CP. However, further prospective, ran- We conducted a randomized controlled trial to compare domized controlled trials with patients with CP were ur- early surgery with the current step-up approach. gently needed to check a possible translation of this effect Methods: We included patients with obstructive chronic from murine to human CP. pancreatitis and severe pain, who recently started using opi- oids. Patients with strong opioids for >2monthsorweak opioids for >6 months were excluded. Patients were ran- FP15 - Free Papers 15 (mini oral) - Pancreas: domized to early surgery (i.e. within 6 weeks) or to the step-up Surgical Outcomes 2 approach (medical treatment, endoscopy if needed, surgery if FP15-01 needed). Primary endpoint was the mean Izbicki pain score. POSTERIOR (SMA-FIRST) APPROACH Results: Eighty-eight patients were randomized, 44 to early surgery (41 underwent surgery) and 44 to the step-up VS. STANDARD approach. Patients in the early surgery group had a lower PANCREATODUODENECTOMY IN Izbicki pain score during follow-up compared to patients in PATIENTS WITH PERIAMPULLARY the step-up approach (35 vs. 48, P= 0.018). Pain relief CANCER: A RANDOMIZED TRIAL during follow-up was achieved in 54% of patients in early A. N. Singh, H. Jain, S. Pal, P. Das, R. Yadav, G. Sahota, surgery and in 33% of patients in the step-up approach (P< N. Dash, S. Gupta and P. Sahni 0.001). Fewer interventions were performed in the early All India Institute of Medical Sciences, India surgery group (median 1 vs. 3, P< 0.001). Complications, fi mortality (0%), hospital admissions, pancreatic function Introduction: The SMA- rst approach of pancreatoduo- and quality of life were comparable among groups. denectomy(PD) is postulated to improve oncological out- Conclusions: The preferred treatment strategy for patients comes by better clearance of the retroperitoneal margin. We fi with chronic pancreatitis and a dilated pancreatic duct is compared the posterior SMA- rst approach(SMA group) early surgery within the first months of opioid use, because with conventional PD(Classical group) in a randomized this provides better pain control with less interventions than trial in terms of long-term survival. the current step-up approach. Methods: All consenting patients undergoing PD for periampullary cancer at the department of GI Surgery, FP14-03 AIIMS were included. Standardized ’Leeds Histopatho- ’ INHIBITION OF MAST CELLS logical protocol was followed. Primary outcomes were the overall and recurrence free survival. The secondary DIMINISHES PAIN SENSATIONS IN outcome measures were intraoperative variables(operating MURINE ACUTE AND CHRONIC time, blood loss), postoperative complications and histo- PANCREATITIS pathology(R0 rate, number of lymph nodes harvested). H. Steenfadt, S. Klauss, S. Schorn, E. I. Demir, Results: Of 97 patients assessed, 77 were randomized and H. Friess and G. O. Ceyhan 70 analyzed(7excluded after randomization, 34 patients- fi Department of Surgery, Klinikum rechts der Isar, Tech- SMA, 36 patients-classical). The demographic pro le and nical University of Munich, Germany preoperative laboratory parameters were comparable. The mean operating time, blood loss and transfusion

HPB 2018, 20 (S2), S182eS294 Free Orals (FP01-FP31) S231 requirements were comparable in both the groups. ISGPS Conclusions: Perioperative BIVA assessment predicts the Grade B/C pancreatojejunostomy leak, post-pancreatec- onset of major and medical postoperative complications in tomy hemorrhage, delayed gastric emptying and Clavien- pancreatic cancer patients. Dindo grade III to V (major) complications were similar and occurred in 11(30.55%), 2(5.6%), 18(50%) and 12(33.3%) patients in the SMA group and in 14(41.17 %), 7(20.6%), FP15-04 20(58.8%) and 16(47.1%) patients in the classical group, LAPAROSCOPIC respectively. The total number of lymph nodes harvested was similar in the two groups (median: 20 in SMA group vs. PANCREATODUODENECTOMY FOR median: 21.5 in the classical group; p=0.80). R0 resections PATIENTS WITH BORDERLINE were higher in the SMA group but statistically insignificant RESECTABLE PANCREATIC DUCTAL [28(77.8%)vs.22(64.7%); p=0.22]. Both groups were com- ADENOCARCINOMA: FEASIBILITY parable in terms of median recurrence-free (27vs28 months) AND OUTCOMES and overall survival (29vs32 months). Conclusion: ’SMA- first’ and conventional approach Maeda S., M. J. Truty, R. L. Smoot, S. P. Cleary, pancreatoduodenectomy have similar long term overall and D. M. Nagorney, T. E. Grotz and M. L. Kendrick recurrence-free survival. Mayo Clinic Rochester, United States Introduction: Outcomes of laparoscopic pancreatoduode- nectomy(LPD) for borderline resectable (BR) pancreatic FP15-03 ductal adenocarcinoma (PDAC) have not been reported. Our aim is to evaluate the feasibility, safety, outcomes of CAN BIOIMPEDANCE VECTOR LPD for BR PDAC. ANALYSIS (BIVA) PREDICT THE Methods: Retrospective review of patients undergoing SURGICAL RISK IN CANCER PATIENTS LPD for BR and resectable PDAC at our institution (2010- UNDERGOING PANCREATIC 2017). NCCN definition used for BR. SURGERY? Results: 55 patients with BR and 112 with resectable PDAC fi < 1 2 1,3 1 were identi ed. Tumor size (3.0 vs. 2.5cm,p 0.001), and M. Sandini , S. Paiella , M. Cereda , M. Angrisani , receipt of neoadjuvant therapy (62% vs. 14%, p< 0.001) was G. Capretti3, L. Roccamatisi1, F. Casciani2, A. Zerbi3, 2 1 increased in the BR group. Operative time (444 vs. 350min, C. Bassi and L. Gianotti < 1 2 p 0.001), estimated blood loss (500 vs. 300mL, p=0.002), Surgery, University of Milano-Bicocca, Pancreas Insti- < 3 mesoportal vein resection (66% vs 13%, p 0.001) and tute, University of Verona, and Surgery, Humanitas conversion laparotomy (18% vs 4%, p=0.002) was greater in Research Hospital, Italy the BR group. Lymph nodes retrieved and R0 resection rates Introduction: Bioimpedance vector analysis (BIVA) is a were similar as were major complications, length of hospital validated tool for the assessment of body composition and stay, and 90-day mortality. hydration index. Several perioperative conditions, such as Of patients with BR disease, 34 patients received sarcopenia, sarcopenic obesity and hyperhydration, have neoadjuvant chemotherapy, 21 did not. Of patients with been associated with adverse postoperative outcomes after neoadjuvant treatment, tumor size and CA 19-9 were pancreatic surgery. Aim of our study was to evaluate the greater compared to those with upfront surgery (3.5 vs usefulness of BIVA in the assessment of perioperative risk 2.6cm, p=0.003, 288 vs 98U/mL, p=0.012, respectively). in cancer patients undergoing pancreatic resection. Mesoportal vein resection, major complications, median Methods: We performed a multicentric prospective length of hospital stay, and 90-day mortality did not differ observational study, including consecutive adult patients between groups. N0 (77% vs. 5%, p< 0.001) and R0 rates undergoing pancreatic surgery at 3 Italian Academic (97% vs. 76%, p=0.026) were significantly higher in the Medical Centers. Exclusion criteria were chronic kidney neoadjuvant group. failure, ASA score >3 and diagnosis of compartmentalized Conclusions: LPD for patients with BR PDAC is feasible fluid collection. BIVA was measured prior to surgery and and safe demonstrating outcomes similar to those reported on postoperative day 1. for open approaches. Neoadjuvant therapy is associated Results: Among104patients,64(61.5%)hadany-grade with improved nodal and margin negative rates. postoperative complications and 10 (9.7%) experienced major morbidity. The preoperative value of the standardized fi phase angle (SPA) was signi cantly lower in patients who FP15-05 experienced severe postoperative morbidity than who did not (- 0.60Æ1.13 vs. 0.56Æ1.64, p = 0.043). The predictive OUTCOME OF THREE DIFFERENT ability of SPA was investigated by the c-statistic (AUC = THROMBOPROPHYLACTIC SCHEMES 0.728) and the optimal cut-point value was assessed at -0.55. AFTER PANCREATIC SURGERY fi fi The values of speci c resistance (Rz/H), speci c reac- R. Hanna-Sawires1, J. Groen1, W. Mesker1, tance (Xc/H) and the Hydration Index (HI) on post- A. Vahrmeijer1, R. Tollenaar1, R. -J. Swijnenburg1, operative day 1 were significantly different in patients who 2 1 1 U E. Klok , B. Bonsing and S. Mieog either experienced medical complications, or not [Rz/H ( / 1Department of Surgery, Leiden University Medical Æ Æ m): 279.52.5 53.58 vs. 249.89 39.85, p = 0.041; Xc/H Center, The Netherlands, and 2Department of Thrombosis U Æ Æ ( /m): 23.61 9.21 vs. 18.33 5.06, p = 0.027 and HI: and Hemostasis, Leiden University Medical Center, The 78.2% (73.9-81.6) vs. 82.7% (77.7-87.4), p = 0.034]. Netherlands

HPB 2018, 20 (S2), S182eS294 S232 Free Orals (FP01-FP31)

Introduction: Post-Pancreatectomy Hemorrhage (PPH) Pancreatic Surgery, Oslo University Hospital, Institute of and Venous Thromboembolism (VTE) are serious com- Clinical Medicine, The Arctic University of Norway, plications following pancreatic surgery. Thromboprophy- Norway laxis by administration of Low Molecular Weight Heparin Introduction: Distal pancreatectomy (DP) is increasingly (LMWH) is the standard of care for the prevention of VTE. performed by laparoscopy yet outcomes reported from However, it remains a challenge to maintain a balance unselected routine practice are scarce. We describe out- between the risk of PPH and VTE. The aims of this study comes of laparoscopic and open DP in a complete cohort. were to investigate the incidence, timing and predictive Methods: A national cohort of all patients undergoing DP factors of PPH and VTE in three LMWH dosage groups: in a universal health care system (2012-2016) from the single-dose (2850IU), double-dose (5700IU) and split-dose Norwegian Patient Register. Short-term outcomes for (twice-daily 2850IU) after pancreatic surgery. evaluation include aggregated length of stay (aLOS) and Method: From November 2013 through September 2017, rates of reoperation, readmissions and mortality (90 days). pancreatic resections at the Leiden University Medical Risk is reported as odds ratio with 95% confidence interval Center were reviewed. PPH was classified according the (OR, 95% CI). International Study Group of Pancreatic Surgery. Time-to- Results: Of 546 procedures, 323 (59%) were laparoscopic. event (clinically relevant (CR)-PPH or VTE) analysis and DP rate increased during period. Median age was 66 years uni- and multivariable analysis were performed. (IQR 55-72), with 52% being female (Table 1). Women Results: In total, 240 patients were included. CR-PPH (OR 1.43, 95% c.i. 1.02-2.02; P=0.039) and patients with developed in 25.0%, 12.5%, and 13.8% of patients in the higher Charlson comorbidity score (OR 0.56, 95% c.i. double-dose, single-dose, and the split dose-group respec- 0.32-0.99; P=0.045) were significantly more often selected tively (P=0.069). There was no significant difference in to open DP. The majority of DPs were done for pancreatic VTE between the groups. Time-to-event (Figure 1) analysis tumours (n=258, 47%), of which 166 were done laparo- showed 70.7% of CR-PPH and 36.4% of VTE occurred scopic (64%). A splenectomy was included in 416 (76%) of within 10 postoperative days (PODs). Lastly, patients with all DPs, with no difference between open or laparoscopy CR-Postoperative Pancreatic Fistula (POPF) were at higher groups. Reoperation rate at 30 days was 5.7% (n=31), with risk for an event (HR=3.16, 95%CI=1.70-5.86, P< 0.001). 4.3% for laparoscopic (n=14) and 7.6% for open DP Conclusions: CR-PPH appeared borderline significantly (n=17). The median aLOS was 5.5 days shorter in laparo- more frequently in the double dose group compared to the scopic to open DP (median 8.6 days vs 14.1; P< 0.001). two other dosage groups. The incidence of VTE was not Readmission rate was 20%, 90-day mortality was 1.1%. significantly different between the dosage groups. Most Conclusion: Laparoscopy is becoming the preferred access CR-PPH occurred within 10 PODs. Based on these results, in a routine nationwide practice with reduction in length of a single dose of LMWH (2850 IU) can be considered. stay but otherwise no difference in short-term outcomes. Readmission rate is high after DPs.

FP15-06 Table 1 Variable Total Open DP Lap. DP P OR (95% CI) N[546 N[223 (41%) N[323 (59%) Female, n (%) Male, n (%) 284(52%) 264(48%) 127 (57%) 96 (43%) 155 (48%) 168 (52%) .039 1.43 (1.02-2.02) Age <65 yrs (n, %) 255 (47%) 104 (47%) 151 (47%) .979 0.99 (0.71-1.40) Comorb. CCI >1 (n, %) 52 (10%) 28 (13%) 24 (7%) .045 0.56 (0.32-0.99) Splenectomy, n (%) 416 (76%) 177 (79%) 239 (74%) .147 0.74 (0.49-1.11 Outcomes aLoS (median, iqr) 9 (5-14) 12 (8-19) 7 (5-11) <.001 n.a. Reoperations (n, %) 31 (6%) 17 (8%) 14 (4%) .103 0.55 (0.27-1.14) Readmissions (n, %) 107 (20%) 42 (19%) 65 (20%) .709 1.09 (0.71-1.67) Mortality 90-d, n (%) 6 (1.1%) 2 (0.9%) 4 (1.2%) .707 1.39 (0.25-7.63)

FP15-06 SHORT-TERM OUTCOMES IN OPEN AND LAPAROSCOPIC DISTAL FP15-07 PANCREATECTOMY IN A NATIONAL COMPARISON OF END TO SIDE COHORT STUDY VERSUS SIDE TO SIDE GASTRIC 1,2 3 4 5 K. Soreide , F. Olsen , L. S. Nymo , D. Kleive and ANASTOMOSIS AFTER PANCREATICO- K. Lassen6 1Clinical Surgery, Royal Infirmary of Edinburgh and DUODENECTOMY IN PREVENTING University of Edinburgh, United Kingdom, 2Department of DELAYED GASTRIC EMPTYING: A Gastrointestinal Surgery, Stavanger University Hospital, RANDOMIZED CONTROL TRIAL 3 4 UNN, Department of Gastrointestinal Surgery, Univer- V. Gupta1, N. Naru1, T. Yadav1, S. Sinha2 and sity Hospital og Northern Norway, Institute of Clinical R. Kochhar2 5 Medicine, The Arctic University of Norway, Department 1General Surgery, and 2Gastroenterology of Hepatobiliary and Pancreatic Surgery, Oslo University Aim: To compare end to side (ES) versus side to side (SS) Hospital, and 6Department of Hepatobiliary and gastric anastomosis in the occurrence of Delayed Gastric

HPB 2018, 20 (S2), S182eS294 Free Orals (FP01-FP31) S233

Emptying (DGE) following Pylorus resecting Pancreati- Background: Perioperative Packed Red Blood Cell coduodenectomy (PrPD). (PRBC) transfusion has been associated with worse sur- Methods: Of the 120 patients enrolled, 83 patients were vival in patients with pancreatic exocrine tumors. The effect randomised to end to side (n=43) and side to side(40) of transfusion in patients with pancreatic neuroendocrine gastric-jejunal anastomosis. Delayed Gastric Emptying tumors (PNETs) has not been examined. (DGE) was defined as per ISPGS guidelines. Operative Methods: A retrospective cohort study of patients with time, removal of nasogastric tube, resumption of oral feed PNETs was performed using the U.S. Neuroendocrine Study and hospital stay was compared. Group database (2000-2016). Demographic and clinical fac- Results: Both the groups were comparable in terms of tors were compared. Kaplan-Meier and log-rank analysis were demographic and operative data. The overall incidence of used for survival analysis. Logistic regression was used to DGE in this study was 59% with 60.5% in ES and 57.5% in assess factors associated with transfusion, overall (OS), EE group. Incidence of clinically relevant DGE was lesser recurrence free (RFS) and progression free survival (PFS). in the SS group (22.5% vs 48.8%, p =0.013). Patients with Results: Of the 1,169 with surgically resected PNETs, 85% ES and SS ananstomosis had similar incidence of primary underwent curative operations and 15% underwent non- DGE (30% vs 27.5%, p=0.787), however, the incidence of curative operations. 156 patients (14%) received perioper- clinically relevant primary DGE was lower in SS group ative transfusion. Transfused patients had a higher ASA (25.6% vs 5%, p= 0.0117). The incidence of secondary Class (p< 0.01), lower preoperative hemoglobin (Hgb; p< DGE (30% and 30%, p=0.984) and clinically relevant 0.01), larger tumors (4.6 v 3.0cm, p< 0.01), more nodal secondary DGE (23% vs 17.5%, p=0.515) was similar involvement (41 v 25%, p< 0.01) and increased major between the groups. In SS group, there was a trend towards complications (42 v 16%, p< 0.01). On multivariate anal- early removal of nasogastric tube (5.87+5.17 vs 7.21+6.44 ysis, ASA Class (p=0.01), Hgb (p< 0.01), and estimated days), early resumption of solid oral feeding (8.21+5.64 vs blood loss (p< 0.001) were associated with transfusion. 10.20+6.41 days) and early discharge (14.73+8.45 vs Transfused patients had worse 5-year OS (77 v 88%, p< 17.61+12.04 days). Other post-operative complications 0.01) in all patients. Transfusion was also associated with were comparable. worse RFS (59 v 75%, p< 0.01) in curatively resected and Conclusions: Side to side gastric anastomosis is more worse PFS for non-curatively resected patients (45 v 73%, effective in preventing primary delayed gastric emptying p=0.05)(Table 1). On multivariate analysis, transfusion was and this translates into the early resumption of oral feeding associated with worse OS (HR 2.27, < 0.001) and RFS and early discharge from the hospital. (HR 1.54, p=0.02) when controlling for TNM stage. Conclusion: Perioperative transfusion is associated with worse survival in patients with surgically resected PNETs.

FP15-08 Table [5- and 10-year Survival, Transfused vs. Non-transf] Number Patients Non-Transfused Transfused p-value Included (n) Patients Patients 5-year OS (all patients) 1,169 88% 77% p<0.0001 10-year OS (all patients) 75% 47% 5-year RFS (curative intent operations, R0 or R1 margins) 997 75% 59% p=0.009 10-year RFS (curative intent operations, R0 or R1 margins) 59% 39% 5-year PFS (non-curative intent operations, R2 margins) 172 73% 45% p=0.047 10-year PFS (non-curative intent operations, R2 margins) 28% 14%

FP15-08 EFFECT OF BLOOD TRANSFUSION ON FP15-09 SURVIVAL FOR RESECTED COMPARITIVE STUDY OF OPEN VS PANCREATIC NEUROENDOCRINE LAPAROSCOPIC SURGERY FOR PAIN TUMORS: ANALYSIS OF THE U.S. RELIEF IN CHRONIC CALCIFIC NEUROENDOCRINE STUDY GROUP PANCREATITIS P. Marincola Smith1, M. Dillhoff2, G. Poultsides3, S. Srivatsan Gurumurthy, V. Annamaneni, S. Sabnis, F. Rocha4, C. Cho5, E. Winslow6, R. Fields7, S. Maithel8, E. Senthil Anand, V. P. Nalankilli, A. V. Natesan, K. Idrees1 and U.S. Neuroendocrine Study Group P. Senthilnathan and C. Palanivelu 1Division of Surgical Oncology, Vanderbilt University Dept. of HPB, Minimally Invasive Surgery and Liver Medical Center, 2The Ohio State University Comprehen- Transplant, Gem Hospital & Research Centre, India sive Cancer Center, Ohio State Medical Center, 3Division of Surgical Oncology, Stanford University Medical Center, Introduction: Chronic pancreatitis is a painful, progres- fl 4Department of Surgery, Virginia Mason Medical Center, sive, in ammatory condition of pancreas. Various surgical 5Division of Hepatopancreatobiliary and Advanced strategies for the treatment of pain relief in chronic Gastrointestinal Surgery, University of Michigan, pancreatitis have been described, although the application 6Department of Surgery, University of Wisconsin School of of minimally invasive surgery is less frequently reported. Medicine and Public Health, 7Department of Surgery, This study is an attempt to examine the outcome of lapa- Washington University in St. Louis, and 8Winship Cancer roscopic procedures in the management of chronic Institute, Emory University, United States pancreatitis in comparison to open surgery from a tertiary

HPB 2018, 20 (S2), S182eS294 S234 Free Orals (FP01-FP31) care centre located in the more disease prevalent southern Introduction: There is a lack of information on the part of India. impact that serum alpha-fetoprotein(AFP)dynamics may Methods: This is a prospective comaparitve study of pa- have on hepatocellular carcinoma(HCC)recurrence and tients undergoing surgery for pain relief in chronic calcific survival after liver transplantation(LT).We aimed to pancreatitis, either by laparoscopy or open method, be- assess if this could serve as a surrogate of tumor tween 2015-2017. The primary endpoint is postoperative biology. pain relief at 3 months, 6 months and 1 year. Methods: We assessed all patients who underwent LT Results: During the study period, 24 patients underwent for HCC between January 2004 and December laparoscopic modified frey’s procedure and 26 patients 2014.Only patients with a measurement of serum AFP at underwent open procedure. There was a statistically sig- listing and at LT were included.AFP ratio was defined as nificant difference favourable to the open group regarding AFP at LT/AFP at listing.Patients with more than 2- shorter operative times (185 min vs 247.5 min; P = 0.002), times increase in AFP were compared to those with less however laparoscopic arm had lesser blood loss (120 mL vs than 2-times increase and those with a decrease or no 240 mL; P = 0.002), a shorter hospital stay (5 d vs 7d;P = change in AFP.The primary outcomes were overall 0.004) and comparable pain relief score (Izbicki’s score) at survival(OS)and disease-free survival(DFS).Multivariate 3 months, 6 months and 1 year follow up. Post-operative Cox regression was applied to identify predictors of weight gain and endocrine insufficiency rates were com- post-LT outcomes. parable in both arms. Results: 506 patients were transplanted for HCC;442 pa- Conclusion: Laparoscopic surgery is safe, effective and tients were included in the study.There were 292 pa- feasible technique for chronic pancreatitis in selected pa- tients(66.1%)with a decrease or no change in tients in the presence of adequately dilated pancreatic duct AFP(Group1),86(19.4%)with an increase< double(- and has a comparable outcome in short term follow-up Group2)and 64(14.5%)with a double increase(Group3). when compared to open surgery. These 3 groups were comparable.The 5-years DFS was 71.3%for Group1 vs 66.4%for Group2 and 52.2%for Group3,p=0.018.The 5-years OS was 75.7% for Group1 vs FP16 - Free Papers 16 (mini oral) - Transplant 2 72.9%for Group2 vs. 54.3% for Group3,p=0.03. The sole FP16-01 risk factors for recurrence after adjusting for confounders were double increase in the AFP ratio[HR2.4(1.4- ALPHA-FETOPROTEIN DYNAMICS IN 4.1),p=0.001]and maximum tumor size at listing THE WAITING LIST AS A BIOMARKER [HR1.12(1.06-1.19),p< 0.001];however the only risk OF HEPATOCELLULAR CARCINOMA factor for death was a double increase in AFP RECURRENCE AND MORTALITY [HR1.60(1.04-2.4),p=0.001]after adjusting for AFTER LIVER TRANSPLANT confounders. 1 2 1 1 Conclusion: Double increase in the serum AFP while pa- W. Zhang , A. Gorgen , A. Ghanekar , N. Selzner , tients are waiting for LT is a strong predictor of outcomes. L. Lilly1, I. McGilvray1, M. Bhat1, P. Greig1, 1 1 This dynamic serum AFP biomarker may serve to expand D. Grant and G. Sapisochin “ ” 1 the size and number criteria without impairing the results Multi-Organ Transplant. Division of General Suregery, of LT. and 2Multi-Organ Transplant Program, Division of Gen- eral Suregery, University of Toronto, Canada FP16-01 Table [Patient characteristics] AFP Ratio N (442) Group 1 Group 2 Group 3 p value £ (292, 66.1%) 1-2 (86, 19.4%) ‡ (64, 14.5%) Age, y* 58 (53-62) 58 (53-62) 58 (52-62) 57 (50-61) 0.661 Male, % 369 (83.5) 243 (83.2) 75 (87.2) 51 (79.7) 0.461 Bridging therapy, % 329 (74.4) 227 (77.7) 58 (67.4) 44 (68.8) 0.083 Radiofrequency ablation, % 208 (63.2) 150 (66.1) 31 (53.4) 27 (61.4) 0.197 Transarterial 79 (24.0) 51 (22.5) 15 (25.9) 23 (52.3) 0.001 chemoembolization, % MELD at listing* 10 (8-14) 10 (8-13) 10 (8-13) 11 (9-14) 0.654 Waiting time (months)* 6 (2-11) 6 (2-11) 6 (2-11) 6 (3-12) 0.651 Listing max. tumor size, cm* 2.5 (1.5-3.9) 2.5 (1.5-3.9) 2.5 (1.4-3.9) 2.5 (1.5-3.9) 0.835 PreLT max. tumor size, cm* 2.0 (0.0-3.2) 1.8 (0-3.0) 1.8 (0-3.3) 2.4 (1.3-3.4) 0.106 Listing tumor number* 1 (1-2) 1 (1-2) 1 (1-2) 1 (1-3) 0.292 PreLT tumor number* 1 (0-2) 1 (0-2) 1 (0-2) 2 (1-3) 0.028 Listing AFP* 13 (5-47) 11 (5-47) 18 (7-50) 21 (9-63) 0.196 PreLT AFP* 11 (5-44) 7 (5-16) 20 (9-63) 76 (27-522) <0.001 AFP difference* 0 (-5-3) -2 (-14-0) 4 (2-13) 59 (18-426) <0.001

*median (IQR) Abbreviation: BMI: Body mass index; MELD: Model for End-Stage Liver Disease; AFP: alpha-fetoprotein:

HPB 2018, 20 (S2), S182eS294 Free Orals (FP01-FP31) S235

FP16-02 Background: Alveolar echinococcosis is a severe zoonotic disease caused by Echinococcocus multilocularis infection SALVAGE LIVER TRANSPLANTATION that is in part dependent on local innate and adaptive OR REPEAT HEPATECTOMY FOR immune responses. In this study the effect of PD-L1 and RECURRENT HEPATOCELLULAR PD-1 inhibition on outcome of alveolar echinococcosis was CARCINOMA: AN INTENT-TO-TREAT tested in different mouse models. ANALYSIS Methods: Wild type (WT) and PD-1 knock out (KO)mice were infected with EM and received anti-PD-L1 mono- C. Lim1, H. Shinkawa2, K. Hasegawa2, P. Bhangui3, 1 4 2 clonal antibody during one month. Lesions count and C. Salloum , C. Gomez-Gavara , Y. Sakamoto , weight in the infected liver was measured at day 30 post P. Compagnon5, N. Kokudo6 and D. Azoulay7 1 2 infection. Immune responses were measured by assessing Henri-Mondor Hospital, France, Hepato-Biliary-Pancre- serum cytokines and intrahepatic and systemic immune atic Surgery Division, Graduate School of Medicine Uni- responses by flow cytometry. versity of Tokyo, Japan, 3Liver Transplantation, Medanta- 4 Results: After oral as well as after intraperitoneal infection The Medicity, Gurgaon, India, HPB and Liver Trans- the extent of AE was significantly reduced after pharma- ’ plantation, Vall d Hebron Institut of Oncology, Hospital cological inhibition or genetic deletion of PD-L1 or PD-1. Universitario Vall d’Hebron, Universidad Autónoma de 5 Mechanistically, we observed in mice infected with AE Barcelona, Spain, HPB and Liver Transplantation, General reduction of protective cytokines such as IL-22 secretion by University Hospital, Switzerland, 6National Center for 7 innate lymphoid cells. Adaptive immune cells such as CD4 Global Health and Medicine, Japan, and HPB and Liver and CD8 T cells are significantly elevated in response to Transplantation, Henri-Mondor Hospital, France infection. The salvage liver transplantation (LT) strategy was Injection of anti-PD-L1 restored the innate and adaptive conceived for initially resectable and transplantable hepato- immunity system balance with elevation of IL-22 secretion cellular carcinoma (HCC) to obviate upfront transplantation, by group 1 innate lymphoid cells. Frequencies of CD4 and with salvage LT in the case of recurrence. The aim of this study CD8 T cells decreased, while the frequency of CD4Treg was to perform an intention-to-treat analysis of overall survival increased after PD-1 blockade confirming its potential to (OS) comparing these 2 strategies for initially resectable and maintain the immune balance. transplantable recurrent HCC. From 1994 to 2011, 391 pa- Conclusion: Inhibition of both, PD-1 and PD-L1 reduces tients with HCC who underwent salvage LT (n = 77) or a the extent of AE infection via modulation of hepatic innate second resection (n = 314) were analyzed. Of 77 patients in the and adaptive immune cell responses. These findings reveal salvage LT group, 21 presented with resectable and trans- that the use of checkpoint inhibitors potentially may alter plantable recurrent HCC and 18 underwent transplantation. Of the clinical course of AE infection. 314 patients in the second resection group, 81 presented with resectable and transplantable recurrent HCC and 81 under- went a second resection. The 5-year intention-to-treat OS FP16-04 rates, calculated from the time of primary hepatectomy, were comparable between the 2 strategies (72% for salvage trans- IMPACT OF DONOR HEPATECTOMY plantation versus 77% for second resection; P = 0.57). In pa- TIME ON OUTCOMES FOLLOWING tients who completed the salvage LT or second resection DCD LIVER TRANSPLANTATION: THE procedure, the 5-year OS rates, calculated from the time of the UK EXPERIENCE second surgery, were comparable between the 2 strategies S. Farid, M. Attia, V. Dhakshina, V. Upasani, S. Willis, (71% versus 71%; P = 0.99). The 5-year disease-free survival E. Hidalgo and N. Ahmad (DFS) rates were 72% following transplantation and 18% St James University Hospital, United Kingdom following the second resection (P < 0.001). Similar results were observed after propensity score matching. In conclusion, Little data exists to evaluate the impact of hepatectomy although the 5-year OS rates were similar in the salvage LT time (HT) during donation after cardiac death (DCD) pro- and second resection groups, the salvage LT strategy still curement on outcomes following liver transplantation (LT). achieves better DFS. Second resection for recurrent HCC In this study we analyse the impact of the time from aortic might be considered to be the best alternative option to LT in perfusion to end of hepatectomy on outcomes across all UK the current organ shortage. transplant centers. 1112 adult patients receiving a first LT in the UK be- FP16-03 tween 1 January 2001 and 31 August 2015 from a DCD donor. Primary end points were PNF and all cause graft CHECKPOINT BLOCKADE IMPROVES survival. OUTCOME IN A MOUSE MODEL OF Incidence of PNF was 40 (4%) and in multivariate HEPATIC ALVEOLAR analysis only CIT >8 hrs. (HZ 2.186 (1.113-4.294, ECHINOCOCCOSIS INFECTION VIA p=0.023) and HT > 60 mins (HZ 3.669 (1.363-9.873, MAINTENANCE OF INNATE AND p=0.01) were correlated with PNF. Overall 90 day, 1 yr., 3 ADAPTIVE IMMUNE BALANCE yr. and 5 yr. graft survival in DCD LT was 91.2%, 86.5%, 80.9% and 77.7% (compared to a DBD cohort in the same 1 2 1 2 F. Jebbawi , J. Wang , D. Candinas , B. Gottestein and period (n=7221) 94%, 91%, 86.6%, and 82.6% respectively 1 G. Beldi (p< 0.001)). In multivariate analysis the factors associated 1 Visceral and Transplantation Surgery, University of with poorer graft survival were HT >60 mins (or more 2 Bern/DBMR, and University of Bern/ Infectious Diseases specifically, 53mins on a continuous spectrum), donor and Pathobiology, Switzerland

HPB 2018, 20 (S2), S182eS294 S236 Free Orals (FP01-FP31) > > age 45 yrs., CIT 8 hours and recipient previous FP16-06 abdominal surgery. The largest study to demonstrate a negative impact of THE OUTCOMES OF ADULT LIVER prolonged HT on outcomes on DCD LT and although TRANSPLANT RECIPIENTS IN KOREA HT 53 mins is not a contraindication for utilisation it USING KOREAN NETWORK FOR should be taken into a multifactorial assessment with ORGAN SHARING DATA established prognostic donor factors such as age K. C. Yoon, K. -W. Lee, S. K. Hong, J. -M. Lee, (>45yrs) and CIT (>8hrs) for an appropriately selected J. -H. Cho, N. -J. Yi and K. -S. Suh recipient. Department of Surgery, Seoul National University College of Medicine, Republic of Korea In Korea since the first liver transplantation in 1988, the FP16-05 number of liver transplants has increased due to the DUNBAR SYNDROME IN LIVER advancement in immunology and surgical procedures. The TRANSPLANTATION: RESULTS OF A Korean Network for Organ Sharing(KONOS) was estab- SINGLE CENTER MATCHED-PAIR lished in 2000 to match the brain death donor to the ANALYSIS AND A EUROPEAN SURVEY recipient We analyzed the survival rate and risk factors using national based data from January 2000 to February STUDY 2017. 1 1 1 D. A. Morales Santana , Z. Czigany , J. Böcker , A total of 14826 liver transplantation were performed in 1 1 1 1 W. Schöning , J. Bednarsch , I. Amygdalos , F. Meister , 57 hospitals from January 2000 to February 2017. Among 2 1 1 P. Isfort , U. P. Neumann and G. Lurje them the number of adult LT recipient was 13971(94.2%) 1 RWTH Aachen, Department of Surgery and Trans- while the number of pediatric transplants was 855(5.8%). 2 plantation, and RWTH Aachen, Institute for Radiology The underlying cause in adults was as follows: Background: Median-arcuate-ligament or Dunbar-syn- HBV5697(40%), alcoholic LC1503(10.8%), drome (MALs) potentially causes arterial complications in HCV577(4.1%), and the total of malignancy cases was orthotopic liver transplantation (OLT). This study aimed to 2140(15.3%). The 1, 5, 10year overall survival for adult investigate MALs incidence and its impact on outcome liver transplant was 87.1%, 78.5%, 73.5% respectively. after OLT with additional information on the management Two third of total LT patients received LDLT(75.1%) and of MALs in OLT, based on a survey-study within 52 Eu- the overall survival was superior than DDLT. The 1, 5,10 ropean centers. year overall survival was 89.8%, 80.8%, 75.7% in LDLT Methods: Between 2010-2017 MALs was diagnosed in 32 while 78.7%, 71%, 66.4% in DDLT respectively. The pa- patients (32/302; 10 %). A matched-pair-analysis (32:64; tient with malignant disease has better survival until 2years matching based on BAR-score, recipient age, transplant after LT but 5, 10year survival has better outcomes in pa- indication) was carried out comparing postoperative com- tient without malignant. The 1, 5,10year survival was plications as assessed by the Clavien-Dindo-Score (CD) 89.2%, 75.5%, 68.7% in malignant patient and 86.6%, and the comprehensive-complication-index (CCI). 81.6%, 79.1%, 74.5% respectively. The overall survival Furthermore, an online-survey with open-ended multiple- was improved in the latest period(2011-2015) than early choice questions on the management of MALs in OLT was period(2000-2005). The 5year survival was 82% and 78% sent to 52 European centers. respectively. Results: Characteristics (demographics, indications, MELD, SOFT, PSOFT, DRI, cold- and warm-ischemic time, distribution of extended criteria grafts) showed no FP16-07 fi signi cant differences. Early allograft dysfunction SEROLOGICAL TUMOR VIABILITY occurred in 40% in the MALs and 38% in the non-MALs groups (p=0.767). Incidence of major complications BASED ON ALPHA-FETOPROTEIN AND (CD3-CD5) and CCI-scores were not significantly C-REACTIVE PROTEIN MAY INDICATE different between the matched groups (p=0.398; 0.704, FUTILE LIVER TRANSPLANTATION IN respectively). ICU- and hospital stay showed no signifi- PATIENTS WITH HEPATOCELLULAR cant between group differences (p=0.240; 0.219, respec- CARCINOMA tively). Five-year graft (77% vs. 76%; p=0.839) and A. Kornberg1, M. Schernhammer1, K. Müller2, H. Friess1, patient (81% vs. 80%; p=0.820) survival did not differ 3 4 significantly between groups. 42/52 (81%) centers gave a J. Kornberg and K. Thrum 1Technical University Munich, 2FSU Jena, 3Ludwig- valid response to our survey. Strategies regarding man- 4 agement of MALs were heterogeneous. Only 29% of the Maximilian-University, and Institute of Pathology, Helios centers reported division of MAL during OLT as routine Klinikum Berlin, Germany procedure. Background: The Milan criteria (MC) are too restrictive Conclusions: Although, the European approach on in selecting suitable liver transplant (LT) patients with handling MALs in OLT is heterogeneous, dividing MAL in hepatocellular carcinoma (HCC). Apart from alpha-feto- Dunbar-syndrome, seems to provide a feasible approach to protein (AFP), parameters of inflammation are discussed to avoid vascular complications. improve the selection process. The aim of this

HPB 2018, 20 (S2), S182eS294 Free Orals (FP01-FP31) S237 retrospective study was to analyze the prognostic impact of incidence of anastomotic strictures were 11.2%, 16.1% combining AFP and C-reactive protein (CRP) in LT for and 17.2% after one year, three and five years after LT. HCC. Multivariate analysis revealed that biliary leak was in- Patients and methods: 119 patients following LT for HCC dependent risk factor (p< 0.001). The incidence of were analyzed. Tumors were classified as Milan In or Milan nonanastomotic biliary strictures was 9.5% five years Out. The impact of clinical, serological and histopatho- after LT. logical features on posttransplant outcome was analyzed by Conclusion: Considering that the biliary leak appears to be uni- and multivariate analysis. the most clinically significant risk factor for anastomotic Results: Only AFP >100ng/ml (Odds ratio [OR] = 13.31), stricture, actions to reduce this complication should also CRP >0.8mg/dl (OR = 13.97) and microvascular invasion lead to a reduction in the frequency of anastomotic (MVI; OR = 15.77) were identified as independent pre- strictures. dictors of HCC relapse. Posttransplant HCC recurrence rates were 2.3%, 14% and 84% in patients with low (AFP 100 ng/ml + CRP 0.8 mg/dl), intermediate (AFP >100 FP16-09 > > ng/ml or CRP 0.8 mg/dl) and high (AFP 100 ng/ml + VIABILITY OUTCOMES FOR DONOR CRP >0.8 mg/dl) serological tumor viability (STV). In the Milan Out cohort, the actuarial 5-year recurrence-free LIVERS DISCARDED DUE TO survival rates were 91.7% in low and 83.6% in interme- STEATOSIS FOLLOWING EXTRA- diate STV (log rank = 0.496), and thus not different from CORPOREAL MACHINE PERFUSION OF Milan In patients (88.4%; 82.5%). In contrast, it was 0% in THE LIVER high STV beyond MC patients. High STV was identified Y. Longatto Boteon1,2, R. Laing1,2, A. Schlegel1, as the most powerful clinical predictor of MVI (OR = 7.5; J. Attard1,2, L. Wallace2, R. Bhogal1, D. Mirza1, p =0.002). H. Mergental1, D. Neil1 and S. Afford2 Conclusion: STV based on standard variables AFP and 1Liver Unit, Queen Elizabeth Hospital Birmingham, and CRP may indicate futile LT and, thereby, safely increase 2National Institute for Health Research, Birmingham Liver the number of eligible transplant candidates. Biomedical Research Centre, University of Birmingham, United Kingdom Introduction: Steatosis is the main reason for discarding FP16-08 donor livers worldwide. Normothermic machine perfusion RISK FACTORS FOR BILIARY (NMP) permits viability testing and hypothermic machine COMPLICATIONS AFTER LIVER perfusion (HMP) may improve the function of marginal TRANSPLANTATION organs. We aimed to analyse the outcomes of human livers declined based on surgical macroscopic assessment (SMA) M. Krasnodebski, W. Patkowski, M. Krawczyk, M. Grat, of steatosis submitted to NMP alone or a merged protocol E. Bik, K. M.Î Wronka, J. Stypułkowski and K. ZieniewiczÎ of HMP+NMP. Department of General, Transplant and Liver Surgery, Methods: 33 donor livers discarded for SMA of steatosis Medical University of Warsaw, Poland followed by a period of cold storage had NMP alone for 6 Introduction: Biliary complications (BC) occur in 10.0% hours or a combined protocol of HMP for 2 hours followed to 30.0% of patients after liver transplantation (LT). The by 4 hours of NMP. Viability was assessed at the end of aim of the study was to identify risk factors for BC NMP using the lactate clearance criteria. Assessment of following LT. macrosteatosis rates was undertaking on paraffin sections Methods: The study group comprised 653 patients treated (< 5%: absent; 5-30%: mild; 30-60%: moderate; >60%: with LT in our Department between 2008 and 2013. The severe). fl in uence of the analyzed variables was determined based Results: Histological assessment showed that 12 livers on the logistic and Cox proportional hazards regression (36%) were not steatotic, 10 (30%) mildly, 7 (21%) methods. Cut-off points for analyzed risk factors were moderately and 4 (12%) severely steatotic. Of the 21 his- determined based on the analysis of receiver operating tologically macrosteatotic livers, 15% (n=5) met our characteristics (ROC) curves. Incidence of BC was esti- viability criteria against 50% (n=6) of the non-steatotic, mated using Kaplan-Meier method. p=0.049. Within steatotic livers 52% (n=11) underwent Results: Biliary leak occurred in 28 (4.3%) patients in NMP alone and failed to reach viability (5 mildly, 3 the postoperative period. The multivariate analysis moderately and 2 severely); 48% (n=10) had HMP+NMP revealed that the number of transfused packed red blood and 50% (n=5) (2 mildly and 3 moderately) met viability cells (p=0.003), preoperative gamma-glutamyl-trans- (p=0.006). All severely steatotic livers failed to reach peptidase activity (p=0.013) and bilirubin concentration viability criteria. (p=0.003) were independent risk factors. The analysis of Conclusion: Surgical assessment of steatosis does not the ROC curves showed that the optimal cut-off points correlate with histology. Steatotic livers often fail to for the prediction of biliary leak was intraoperative achieve viability on NMP alone. HMP prior to NMP may  transfusion 8 units of packed red blood cells, pre- be a potential approach to increase the rescue of mildly to  operative gamma-glutamyl-transpeptidase activity 112 moderately steatotic livers. IU/ml and bilirubin concentration 3.5 mg/dl. The

HPB 2018, 20 (S2), S182eS294 S238 Free Orals (FP01-FP31)

FP17 - Free Papers 17 (mini oral) - HPB 2 routinely performed in case of planned extended right hemihepatectomy. The aim of this study was to examine the FP17-01 functional versus volumetric response of the FRL and S4 RANDOMISED CONTROLLED TRIAL after right PVE. ON BENEFITS OF PREOPERATIVE Method: Patients that underwent both CT-volumetry and CARBOHYDRATE LOADING IN liver function test before and after PVE for planned hepa- PATIENTS UNDERGOING ELECTIVE tectomy were included. Assessment of liver function was MAJOR HEPATOBILIARY SURGERY performed with mebrofenin hepatobiliary scintigraphy (HBS). Functional and volumetric response of FRL and S4 1 1 2 N. K. Mohd Roslan , P. C. Tah , K. Ramayah , were calculated. Multiple regression analysis was con- 2 2 1 3 P. S. Koh , J. K. Koong , V. Rai and B. K. Yoong ducted to explore predictors of functional and volumetric 1 2 3 University Malaya, University Malaya, and Surgery, response. University Malaya, Malaysia Results: 90 patients underwent right PVE. Of the 66 Objective: Surgery induces a catabolic response which is resected patients, 32 underwent additional S4 resection. further augmented with prolonged preoperative fasting . After 22 days (21-25) [median (interquartile-range)], the Compared to fasting from midnight, preoperative carbo- functional increase was significantly higher than the volu- hydrate(CHO)-loading attenuates this response and en- metric increase for both segments 2-3 (69.5% (42.6-108.5) hances recovery. We aim to assess the effect of CHO- vs. 41.8% (29.6-56.3), P< 0.01) and segments 2-4 (54.7% loading on length of hospital stay (LOS), preoperative (30.0-92.6) vs. 36.8% (24.0-60.7), P< 0.01). The func- wellbeing, resting energy expenditure(REE) nutritional tional and volumetric contribution of S4 were respectively status, morbidity and mortality following elective major 41.5% (31.7-48.7) and 41.6% (37.4-48.5) of the non- hepatobiliary operations. embolized lobe. Multiple regression model revealed that Methods: 35 patients were recruited and randomised into type of liver disease and baseline liver function were sig- control group (fasted) or intervention group (preoperative nificant predictors for functional, but not for volumetric, CHO). 24 patients (12 from each group) were analysed. response. Preoperative wellbeing on thirst, hunger, dry mouth, anx- Conclusions: After right PVE, the functional and volu- iety and weakness assessed with VAS (0-10 score). REE, metric contribution of S4 to the FRL is substantial. The muscle grip strength(MGS), midupper arm circum- functional response is higher than the volumetric and cor- ference(MUAC) measured at baseline (preoperative), relates with liver pathology and baseline function. This postoperative day 1, 4 and 7. advocates for a functional rather than a volumetric Results: Patients in control group had mean LOS 13.8 days assessment. (SD=5.06) compared with 8.1 days (SD=2.47) for inter- vention group(P=0.004). The mean rank of thirst, hunger and dry mouth scores were significantly higher in the fasted FP17-04 group compared to intervention (15.96 vs 9.04 (P=0.014), ANALYSIS OF 90-DAY READMISSION 16.46 vs 9.04 (P=0.005) and 17.5 vs 7.5(P=0.000) respectively). REE was significantly higher in the fasted RATES AND HEALTHCARE COST IN compared to intervention group at postoperative day 1 by PANCREATIC SURGERY: WHAT IF mean of 332kcal (P=0.018). Comparing with baseline this BUNDLED PAYMENT SYSTEMS WERE rise was significant in the fasting group(P=0.015) but not APPLIED TO PANCREATIC SURGERY? for intervention group. No difference for trends in MGS M. LeCompte, J. Wright, A. Maiga, G. Edwards, and MUAC, morbidity and mortality. V. Tiwari and K. Idrees Conclusion: Preoperative oral carbohydrate significantly General Surgery, Vanderbilt University Medical Center, reduced LOS, improved preoperative patient comfort and United States reduced metabolic response to surgery when compared with conventional fasting from midnight in patients un- Introduction: National efforts to curb healthcare spending, dergoing major hepatobiliary operations. reduce waste and improve quality have led to the devel- opment of bundled payments systems by the Centers for Medicare and Medicaid for certain surgical procedures, e.g. joint replacements. Integral components of bundled pay- FP17-02 ment metrics include length of stay (LOS), post-operative FUNCTIONAL AND VOLUMETRIC major complication rates (PMCs) and unplanned 90-day RESPONSE OF LIVER SEGMENT 4 readmission. The goal of this study was to quantify 90-day AFTER RIGHT PORTAL VEIN readmission and overall in-patient hospital costs in patients EMBOLIZATION undergoing pancreatic surgery. fi 1 1 2 3 Methods: A retrospective review of clinical and nancial F. Rassam , P. Olthof , K. van Lienden , R. Bennink and data was conducted on 413 patients undergoing pancreatic T. van Gulik1 1 2 3 resection at a single academic institution. Data were Surgery, Radiology, and Radiology and Nuclear Med- analyzed using non-parametric testing to assess association icine, Academic Medical Center Amsterdam, The between LOS, readmission rates and total in-patient hos- Netherlands pital cost. Introduction: Portal vein embolization (PVE) is Results: 104 patients (25%) required 90-day readmission performed in patients with insufficient volume or function after surgery and PMCs occurred in 15%. 87 required a of the future remnant liver (FRL) before planned hepatec- single readmission (SRA) while 17 required multiple tomy. Additional embolization of segment 4 (S4) is not readmissions (MRA). Median LOS was 7 days in both SRA

HPB 2018, 20 (S2), S182eS294 Free Orals (FP01-FP31) S239 patients and those who did not require readmission (NRA). FP17-06 Median total index admission cost for patients was $24,788 (IQR=$14,189). Total hospital costs increased to $43,268 AN EARLY ASSESSMENT FOR THE USE (IQR=$22,770) and $55,322.00 (IQR=$47,069) for the OF HOLOLENS DURING SRA and MRA groups, respectively. Patients requiring LAPAROSCOPIC LIVER RESECTIONS readmission were found to have higher index hospitaliza- E. Pelanis1,2, R. P. Kumar1, Å. A. Fretland1,2,3, tion costs compared to NRA patients, $29,566 vs. $23,947 O. J. Elle1,4 and B. Edwin1,2,3 (p=0.01). 1The Intervention Centre, Oslo University Hospital, Conclusions: Our results demonstrate that increased index 2University of Oslo, Institute of Clinical Medicine, hospitalization costs after pancreatic surgery are associated 3Department of HPB Surgery, Oslo University Hospital, with higher 90-day readmissions with doubling of total Rikshospitalet, and 4Department of Informatics, University global (90-day) in-patient hospital costs. With changing of Oslo, Norway reimbursement models, health-care providers performing Introduction: The goal of this work was to improve pancreatic surgery need to pay attention to these trends and visualization and interaction during surgery by presenting quality metrics. the patient specific anatomy in mixed reality. Also assess the use of this work under sterile conditions. Methods: We make use of existing volumetric data, such FP17-05 as CT/MR images, to create patient specific 3D models. VISUAL GAZE PATTERNS REVEAL This is performed by initially segmenting the volumetric SURGEONS’ ABILITY TO IDENTIFY image data and creating 3D surface models, which can then RISK OF BILE DUCT INJURY DURING be visualized and interacted with by using our internally LAPAROSCOPIC CHOLECYSTECTOMY developed mixed reality platform for the Microsoft Holo- Lens. The model includes the parenchyma, tumor, and 1 1 1 C. Sharma , M. Sodergren , H. Singh , hepatic and portal vessels. 2 1 F. Orihuela-Espina and A. Darzi The HoloLens allow users to place holograms in real 1 2 Imperial College London, United Kingdom, and Na- environment and presents it through a transparent display. tional Institute for Astrophysics, Optics and Electronics Our program allows users to, (INAOE), Mexico a) Import patient specific 3D models and CT/MR images, Introduction: Bile duct injury is the main serious b) Interact with models through moving, scaling and complication of laparoscopic cholecystectomy. This study rotating, solely using gestures, uses eye-tracking technology to identify distinct visual c) Create and adjust a resection plane, gaze patterns employed by surgeons able to promptly d) Share the models between surgeons. detect bile duct injury risk during laparoscopic Results: Surgeons planned surgery preoperatively using cholecystectomy. HoloLens. During the laparoscopic liver resection, sur- Methods: Gaze tracking was performed using a remote geons placed and interacted with models, which contains eye-tracking system. Subjects were shown 3-minute the resection plane and CT/MR images. Multiple surgeons videos of a laparoscopic cholecystectomy that led to a shared and interacted with the model using gestures, under serious bile duct injury (Video A) and of an uneventful sterile conditions. procedure (Video B), while logging if they perceived This technology was qualitatively evaluated by response errors in surgical approach that could result in bile duct to a questionnaire, where surgeons answered questions on a injury. Outcome parameters include fixation sequences on six-point Likert scale. Results will be presented at the anatomical structures and eye tracking metrics. Partici- conference. pants were stratified based on performance and compared Conclusion: We created and presented the patient specific using conventional statistical and machine learning liver models in our application, which is shown to be highly approaches. recommendable and relatively comfortable for use during Results: 29 surgeons were recruited. There was no rela- surgery. tionship between experience and speed of risk identifica- tion. The group with early-detection of the risk of injury displayed reduced dwell time [ms] on the common bile duct (6632.50 [IQR 3600.3] vs 9249.30 [IQR 3924.8]; p = 0.041) in the first half of video A. They also displayed increased cystic duct dwell time [ms] (15665.30 [IQR 7999.5] vs 12715.65 [8348.7]; p= 0.026) and increased Calot’s triangle glances count (3.00 [IQR 2] vs 1.50 [IQR 1]; p = 0.018) viewing the dissection of Calot’s triangle in video B. Hidden Markov Model based classification of fixation sequences demonstrated clear separability between early and late identifier groups. Conclusions: There are discernible differences in visual gaze patterns of surgeons during laparoscopic cholecys- tectomy associated with early recognition of impending bile duct injury. Figure [Liver model with menu on the HoloLens]

HPB 2018, 20 (S2), S182eS294 S240 Free Orals (FP01-FP31)

FP17-07 Background: Pylorus preserving Pancreaticoduodenec- tomy (PPPD) is a standard treatment for patients with THE COMPREHENSIVE pancreatic head and periampullary tumors. With the COMPLICATION INDEX (CCI) VS THE development of laparoscopic surgery, interest in laparo- CLAVIEN DINDO (CD) DRADING FOR scopic PPPD has continued to grow. However, limited MEASURING SEVERITY OF POST- studies about laparoscopic PPPD have been reported. The OPERATIVE COMPLICATIONS. A purpose of this study is to evaluate the perioperative sur- gical outcomes after laparoscopic versus open pylorus PROSPECTIVE STUDY IN 1000 preserving pancreaticoduodenectomy. PATIENTS Methods: Between September 2012 and June 2017, 217 S. Ray, N. Mehta, V. Mangla, S. Mehrotra, S. Lalwani, patients underwent open or laparoscopic PPPD at Yonsei A. Yadav and S. Nundy University Severance hospital by single surgeon. Patients Sir Ganga Ram Hospital, India were divided into two groups, who underwent open PPPD Introduction: The Clavien Dindo (CD) grading is a method and laparoscopic PPPD. Both groups were compared in of assessing the severity of surgical complications, wherein, terms of clinical and oncologic variables. only the worst post-operative complication is taken into ac- Results: The clinical factors including age, gender, BMI count. However a new measure, the Comprehensive did not differ significantly in the two groups.The median Complication Index(CCI) has been proposed in which all the operative time in the open and laparoscopic PPPD was complications are included. We compared the CCI and CD in 451.3 min and 472.8 min, respectively (p = 0.081). Lapa- patients undergoing GI and HPB surgery. roscopic PPPD patients had a significantly lower blood loss Methods: Between June 2015 and December 2016 we compared with open PPPD patients (p < 0.001). Overall recorded the complications in terms of CD grading and CCI in complication rates did not differ significantly between open 1000 consecutive patients who underwent surgical procedures PPPD and laparoscopic PPPD (p = 0.519). In the open and in our unit. The outcome variables studied were post-operative laparoscopic groups, clinically relevant postoperative lengthofstay(LOS),ICUstayandtimetoreturntonormal pancreatic fistula (CR-POPF) rates were 18.8% and 13.5% activity. Correlation of the variables was done using the (p = 0.311).Both groups showed similar rates of negative Spearman’s test and the strength of prediction was expressed margins (p = 0.196). as beta coefficient using linear regression analysis. The pre- Conclusions: In this study, laparoscopic PPPD have dictive accuracy of CD and CCI were expressed as area under similar perioperative surgical outcomes as compared to the curve (AUC) using ROC curve analysis and compared. open PPPD. Therefore, laparoscopic PPPD is safe and Results: There were 600 males and 400 female patients feasible procedure. Multicenter study and randomized (M:F 3:2) with a mean age(range) of 50 (3-90) y. The overall controlled trial are needed to better assess the clinical and 30-day/in-hospital mortality was 7.9%. Both CD and CCI oncologic outcomes. showed a good correlation with LOS, ICU stay and time to normal activity with comparable predictive accuracy (Table 1). Results were similar on subgroup analysis of the popu- lation based on nature and route of the operation.

FP17-07 Table 1 Variables Correlation coefficient (r) b coefficient Area under the curve (AUC) CD CCI CD CCI P value CD CCI P value LOS 0.582 0.577 1.967 0.136 <0.001 0.906 0.890 0.001 ICU stay 0.623 0.618 1.967 0.139 <0.001 0.874 0.851 0.001 Time to normal activity 0.460 0.485 2.003 0.169 <0.001 0.771 0.762 0.001

Correlation and predictive accuracy (AUC) of CCI and CD with outcome variables.

Conclusion: Both CD and CCI are equally accurate in FP17-09 measuring the severity of complications and ability to predict the length of ICU and hospital stay in patients un- PERIOPERATIVE RESTRICTIVE dergoing elective and emergency GI and HPB surgeries. INTRAVENOUS FLUID THERAPY IN ERAS IN PANCREATICODUODENECTOMY FP17-08 T. Evans1,2, S. Koek1,3 and M. Ballal1,3 LAPAROSCOPIC VERSUS OPEN 1Upper Gastrointestinal Surgery, Fiona Stanley Hospital, PYLORUS PRESERVING 2School of Medicine, University of Notre Dame Australia, 3 PANCREATICODUODENECTOMY: and School of Surgery, University of Western Australia SINGLE-SURGEON EXPERIENCE Introduction: The optimal perioperative intravenous hy- dration (IVH) regimen for patients undergoing pancreati- S. H. Han, I. Kang, J. U. Chong, H. K. Hwang, coduodenectomy (PD) in enhanced recovery after surgery C. M. Kang, D. S. Yoon and W. J. Lee Yonsei (ERAS) remains unclear. Literature suggests that salt and Yonsei, Republic of Korea water overload can lead to oedema, thereby increasing

HPB 2018, 20 (S2), S182eS294 Free Orals (FP01-FP31) S241 complication rates and length of stay (LOS); and a dysplasia (IGD), high-grade dysplasia (HGD), or invasive restrictive regimen may lead to better outcomes. carcinoma (IPMC). Patients were categorized as benign Methods: Between 2009 and 2017 169 PD patients, (LGD and IGD) or malignant (HGD and IPMC). partitioned into pre-ERAS (29), transition (14), and ERAS Results: Pathological diagnoses were benign in 91 cases (126) phases, were reviewed. Postoperative IVH was pre- and malignant 65. In multivariate analyses, MN size (p< scribed to administer a total requirement of 20ml/kg/day 0.0001) and BPD cyst size (p=0.0004) were independent and a sodium requirement of 0.5mmol/kg. Total IVH predictors of malignancy, and in ROC analysis AUC was volume with relative salt and chlorine loads per kg on POD 0.74 and 0.72. With 5 mm taken as the cutoff value for MN 0, 1 and 2 were collected. Associated morbidity was size, diagnosis of malignant IPMN had sensitivity of 72%, assessed in univariable and multivariable analysis along- specificity 65% and accuracy 68%. side other ERAS elements. Carcinoma without MN was present in 15patients (15/ Results: ERAS groups were comparable in their de- 65=23%). Pathological findings were HGD in 12 (12/ mographics with mean age (Æ SD) 64 Æ 11.3 years. Mil- 65=18%) and IPMC 3 (3/65=5%). limoles IV sodium and chloride administered were reduced Conclusions: MN size measured with EUS showed high in all days from the pre-ERAS to ERAS periods (p< .001). predictive ability. However, about 20% of malignant pa- Chloride load per kg was a significant (p< .036) predictor tients have no nodules, that is a problem for future. in reducing LOS, and predictive of all grades post-operative pancreatic fistula (POPF; p< .011). A reduction in sodium and chloride load per kg were univariate predictors of FP18-02 < reduced chyle leak incidence (p .006). Restrictive IVH PREVALENCE OF INCIDENTAL saw no significant reduction in ileus or overall morbidity. Conclusion: Aggressive IVH as a predictor of increased INTRADUCTAL PAPILLARY LOS, chyle leak and POPF affirms a literature trend to- MUCINOUS NEOPLASM: BOLOGNA wards restrictive IVH. Sodium and chloride load per kilo- EXPERIENCE fi gram deserves speci c consideration in larger studies above C. Ricci1, M. Migliori2, A. Imbrogno2, E. Mazzotta2, volume-based approach alone in evaluating ERAS C. Felicani2, C. A. Pacilio2, L. Calculli2, F. Minni2 and complication risk. R. Casadei2 1DIMEC, University of Bologna, and 2S. Orsola Malpighi Hospital FP18 - Free Papers 18 (mini oral) - Pancreas: Background: Several studies evaluated intraductal papil- Miscellaneous lary mucinous neoplasms (IPMNs) prevalence using FP18-01 selected patients. The “true” prevalence of incidental IPMN SURGICAL INDICATION FOR BD-IPMN still remain unknown. e MURAL NODULE IN ALGORITHM OF Aim: To evaluate the prevalence of IPMN in a non selected 2012 & 2017 INTERNATIONAL population and its correlation with other diseases. Methods: Data from 6.357 abdominal ultrasonography CONSENSUS GUIDELINES (US) performed from 2012 to 2015 were collected. The Y. Shimizu1, H. Yamaue2, H. Maguchi3, S. Hirono2, inclusion criteria were: age > 18 years and absence of T. Sano4, Y. Senda1, S. Natsume1 and A. Yanagisawa5 pancreatic disease. All patients with suspected IPMN were 1Gastroenterological Surgery, Aichi Cancer Center Hos- further investigated with magnetic resonance cholangio- pital, 2Second Department of Surgery, Wakayama Medical pancreatography (MRCP). The prevalence of IPMN was University, 3Center for Gastroenterology, Teine-Keijinkai calculated. The association with others disease were studied Hospital, 4Gastroenterological Surgery, Aichi Medical with multivariate analysis. University, and 5Pathology, Japanese Red Cross Kyoto Results: A total of 240 pancreatic cystic lesions were Daiichi Hospital, Japan detected. MRCP confirmed that 224 (93.3%) were IPMN. Introduction: To identify high risk of malignancy, mural The prevalence of IPMN was 3.5% (224/6357): IPMN type nodule (MN) size 5mm was added in algorithm of 2017 II and type I/III had a prevalence of 3.3 % and 0.2 %, Guidelines for the management of BD-IPMN. respectively. At univariate analysis, IPMN were more Methods: The present study was a retrospective investi- frequent in female patients (P< 0.001), age between 40-50 gation of predictors of malignancy in BD-IPMN, paying or >70 years (P< 0.001), patients whit chronic disorder special attention to MN. 466 patients with IPMN under- (P=0.007) as well as cirrhosis (P=0.011), non cirrhotic liver went pancreatic resection at 3 hospitals between 1996 and disease (P< 0.001) or inflammatory bowel disease 2014. 156 patients with BD-IPMN were enrolled in this (P=0.003). Multivariate analysis showed that the only study. Preoperative findings and pathological diagnosis factors related to IPMN were female gender (OR 2.4; 95% were analyzed. Endoscopic ultrasonography (EUS) mea- CI 1.8-3.3; P< 0.001) and age between 40-50 years (OR surements were essential for MN size, and computed to- 5.0; 95% CI 2.3-10.7; P< 0.001) and over 70 years (OR mography (CT) for MPD diameter and BPD cyst size. BD- 2.1; 95% CI 1.5-3.0; P< 0.001). IPMN was defined cystic dilatation of BPD and MPD Conclusions: The prevalence of IPMNs in a non-selected diameter was considered < 5mm(2012 Guidelines). Ac- population was 3.5%. The factors related to this prevalence cording to WHO 2010 classification, diagnoses are classi- were gender and age. No association with chronic or fied as low-grade dysplasia (LGD), intermediate-grade neoplastic disease was found.

HPB 2018, 20 (S2), S182eS294 S242 Free Orals (FP01-FP31)

FP18-03 assess the potential role of adjuvant chemotherapy in terms of improvement of survival. OPTIMAL SURVEILLANCE AFTER Methods: Retrospective evaluation of all patients with SURGICAL RESECTION FOR resected IPMNs with high grade dysplasia (HGD) or INTRADUCTAL PAPILLARY invasive carcinoma (iCa) at the Department of General and MUCINOUS NEOPLASM OF PANCREAS Pancreatic Surgery - The Pancreas Institute, University of Verona, with univariate and multivariate analysis for de- S. Hirono, M. Kawai, K. -I. Okada, M. Miyazawa, terminants of disease specific survival (DSS) were Y. Kitahata, R. Kobayashi, S. Hayami, N. Suzaki, performed.Kaplan-Maier curves were used to compare M. Ueno and H. Yamaue patients. Second Department of Surgery, Wakayama Medical Uni- Results: From 467 resected IPMNs, we identified 195 versity, Japan patients: 110 (56%) with iCa and 85 (44%) with HGD.As Introduction: This study aimed to evaluate recurrence risk expected, none of the patients with HGD received adjuvant factors after surgical resection for intraductal papillary therapy, while 22.7% of patients with IPMN-iCa did.The mucinous neoplasm (IPMN), to suppose an appropriate most used chemotherapy regimen was Gemcitabine postoperative surveillance. (72%).Overall, the median DSS for IPMN-iCa was 208 Methods: This study included 257 consecutive IPMN pa- months.The median DSS of IPMN treated with adjuvant tients undergoing surgery from 1999 to 2014. Pathological therapy was not reached, while it was 94 months for sur- diagnosis showed low- or intermediate-grade dysplasia in gery alone (p< 0.01).Predictors of DSS were adjuvant 85 patients, high-grade dysplasia in 87, and invasive therapy (HR=0.42; CI=0.26-0.67; p< 0.01), tubular inva- intraductal papillary mucinous carcinoma (IPMC) in 85. sive component (HR=5.79; CI=2.69-12.45; p< 0.01) and The median postoperative follow-up period was 53.5 pancreatobiliary epithelial type (HR=2.85; CI=1.3-7.87; months. The risk factors for postoperative recurrence were p=0.03). At multivariate analysis, adjuvant therapy resulted analyzed. an independent predictor of DSS (HR=0.45; CI=0.20-0.98; Results: Fifty-six IPMN patients (21.8%) had post- p< 0.04).Stratifying for pathological features,adjuvant operative recurrence, including those with remnant therapy was independently associated with DSS only in pancreatic recurrence (n=14) and extra-pancreatic recur- case of tubular invasive component. rence (n=42). Remnant pancreatic recurrence had no in- Conclusions: Adjuvant therapy for IPMNs with iCa is fl uence on overall survival (OS), whereas, patients with associated with improved DSS compared with surgery fi < extra-pancreatic recurrence had signi cantly worse OS (P alone, especially in case of tubular invasive component. 0.001). Five patients (35.7%) experienced remnant Future controlled trials are needed to improve the level of pancreatic recurrence more than 5 years after surgery. All evidence. extra-pancreatic recurrences occurred within 5 years. The OS tended to be better in patients undergoing second resection for remnant pancreatic recurrence than in those without, although the difference did not reach statistical FP18-05 significance (P=0.081). We found that positive pancreatic SAFETY AND OUTCOMES OF transection margin was only independent risk factor for PANCREATECTOMY WITH ARTERIAL remnant pancreatic recurrence (P< 0.001; OR, 8.92), RESECTION FOR PANCREATIC whereas the risk factors for extra-pancreatic recurrence ADENOCARCINOMA were invasive IPMC (P< 0.001; OR, 29.41), mixed-type (P=0.008; OR, 6.41), elevated serum CA19-9 (P=0.019; P. Bachellier, P. Addeo, F. Faitot and P. Dufour OR, 3.57), and transfusion (P=0.025; OR, 3.33). HPB Surgery and Liver Transplantation, University of Conclusion: We suggest that continuous postoperative Strasbourg, France surveillance for more than 5 years is needed for IPMN Background: The high postoperative mortality rate and the patients to evaluate remnant pancreatic recurrence, and uncertain oncologic benefits have limited the use of strict 5-year surveillance is necessary for IPMN patients at pancreatectomy with simultaneous arterial resection (AR) risk for extra-pancreatic recurrence. for locally advanced pancreatic adenocarcinoma( LAPC). This study assesses the safety and the outcomes of a large cohort of pancreatectomies with simultaneous AR for FP18-04 Methods: A retrospective review of a prospectively maintained database of patients undergoing pancreatec- ADJUVANT CHEMOTHERAPY AFTER tomies with AR Between January 1990 and November SURGERY FOR INTRADUCTAL 2017 was performed. PAPILLARY MUCINOUS NEOPLASMS Results: There were 118 consecutive resections including OF THE PANCREAS: PROMISING 51 pancreaticoduodenectomies, 18 total pancreatectomies, OBSERVATIONAL DATA CALL FOR and 49 distal pancreatectomies. Resected arterial segments included the coeliac trunk (50), the hepatic artery (29), the CONTROLLED STUDIES superior mesenteric artery (35), or other segments (4). The C. Dal Borgo, S. Andrianello and G. Marchegiani overall mortality and morbidity were 5.0% and 41.5%, Department of Surgery, University of Verona, The respectively. Eighty-four (75.4%) patients received Pancreas Institute, Italy neoadjuvant chemotherapy, 105 (89%) simultaneous Introduction: Little is known about the role of adjuvant venous resection, and 101 (85.5%) arterial reconstruction. treatment after surgery for malignant Intraductal Papillary The rate of R0 resection, pathologic invasion of venous and Mucinous Neoplasms (IPMNs).The aim of this study is to arterial walls was 52.4%, 74.2%, and 58%, respectively.

HPB 2018, 20 (S2), S182eS294 Free Orals (FP01-FP31) S243

Overall survival(OS) was 59%, 14%, and 9.6% at 1, 3, 5 Conclusions: This is the largest series analyzing PDAC- years, respectively with a median overall survival after NBT who underwent surgical therapy. Improved survival is resection of 13,60 months (CI 95%:11.8-16.9 months). The associated with favorable tumor biology and operative longest OS was observed in patients with R0N0 resection outcomes. (n=15) (30.78 months). In multivariate analysis, R0 resec- tion (HR: 0.62; 95% CI: 0.40-0.95; p=0.01) and venous invasion (HR: 1.65; 95% CI: 1.05-2.61; p=0.02) were FP18-07 fi identi ed as independent prognostic factors for overall EFFICACY AND FEASIBILITY OF survival. Conclusions: In a specialized setting pancreatectomy COMBINING FOLFIRINOX AND with AR for LAPC can be performed safely with limited STEREOTACTIC RADIOTHERAPY mortality and morbidity. Negative resection margin and FOR PATIENTS WITH LAPC (LAPC-1 absent venous invasion might predict favourable TRIAL) outcomes. M. Suker, J. Nuyttens, B. Groot Koerkamp, F. Eskens and C. van Eijck Erasmus MC, University Medical Center Rotterdam, The FP18-06 Netherlands A MULTI-INSTITUTIONAL ANALYSIS Introduction: Patients with locally advanced pancreatic OF FACTORS AFFECTING SURVIVAL cancer (LAPC) rarely undergo resection with curative FOR LEFT-SIDED PANCREATIC intent and controlling disease progression should be the ADENOCARCINOMA goal of the treatment. We conducted a multicenter phase II fi 1 2 3 4 trial (ClinicalTrials.gov Identi er: NCT02292745) to F. Alemi , Z. Jutric , J. Grendar , A. M. Roch , investigate the efficacy and feasibility of combining A. L. Cheng5, P. D. Hansen3, E. P. Ceppa4, H. J. Asbun6, 7 8 FOLFIRINOX and Stereotactic Radiotherapy (SBRT) for S. Warner and A. A. Alseidi patients with LAPC (LAPC-1 trial). 1St Vincent Medical Center, 2University of California, 3 4 Methods: All eligible patients with biopsy-proven LAPC Irvine, Providence Portland Cancer Center, Indiana were included between January 2015 and June 2017. These University, 5University of Missouri, 6Mayo Clinic, 7City of 8 patients underwent a staging laparoscopy to exclude occult Hope Medical Center, and Virginia Mason Medical metastasis and were treated with FOLFIRINOX (8 cycles) Center, United States followed by SBRT (5 fractions/8 Gy) if there was no tumor Background: Pancreatic ductal adenocarcinoma of the progression on imaging. Primary outcome was overall neck, body and tail (PDAC-NBT) often presents at an survival (OS). Secondary outcomes were progression free advanced stage given lack of discrete clinical symptoms. survival (PFS), treatment-related toxicity, and resection This study compiles PDAC-NBT resections from multiple rates. institutions and aims to characterize clinicopathologic and Results: In total, 53 patients were included in the study. surgical factors associated with survival after The preliminary survival data showed a median OS of 18 pancreatectomy. months (95% CI 16-19) and median PFS of 12 months Methods: Patients who underwent distal or subtotal (95% CI 11-13). Thirty (60%) events of a grade 3 or 4 pancreatectomy for PDAC-NBT at five high-volume cen- adverse event occurred during FOLFIRINOX. Thirty-nine ters from 2001 to 2016 were retrospectively collected. (74%) patients had no tumor progression after the chemo- Clinicopathologic factors were correlated with the primary therapy andreceived the full dose of SBRT. One (3%) pa- outcome of overall survival in univariate and multivariate tient had a grade 5 adverse event three months after SBRT, analyses. while no grade 3 or 4 adverse event occurred after SBRT. Results: 330 patients were surgically treated between Seven (14%) patients underwent a resection, all being a 2001 and 2016. The majority of patients presented with radical resection. early stage disease (Stage I-15.9%, Stage II- 81.4%, Conclusion: FOLFIRINOX combined with SBRT in pa- Stage III-1.7%). Twenty-one patients (10.3%) received tients with inoperable LAPC is feasible and effecitve. neoadjuvant therapy and 221 (72.5%) completed adju- Forteen percent of the patients became operable and 6% vant therapy. Median survival was 26.8 months, with 1- had a complete response. ,3-,and 5- year overall survival of 77.1%, 41.2%, and 28.4%. Median follow-up time was 21.7 months. Path- fi < ologic ndings of lymphovascular invasion (p 0.0001), FP18-08 perineural invasion (p=0.0089), and lymph node invasion (p< 0.0001) were significantly associated with worse PANCREATIC NEUROENDOCRINE survival on univariate analysis. Tumor size >1.5 cm TUMORS: A MULTICENTER (p=0.0063), lymph node ratio >0.1 (p=0.0003), and non- INTERNATIONAL EVALUATION OF R0 resection (p=0.0002) were also significantly associ- GRADING, STAGING AND OUTCOMES ated with worse survival. Tumor stage, major surgical P. Worth1, J. Leal1, Z. Soonawalla2, P. Athanasopoulos3, complications (Clavien-Dindo 3), and local tumor C. Toumpanakis4, P. Newell5, R. Parks6, B. Lawrence7, recurrence were independently associated with worse E. Vicente Lopez8, B. Visser9 and International Pancreatic survival on multivariate analysis (p< 0.05). There were Neuroendocrine Tumor Collaborative no significant associations between survival and opera- 1Department of Surgery, Stanford University, United tion type, preoperative presenting symptoms, or neoad- States, 2Department of Surgery, University of Oxford, juvant or adjuvant therapy. 3Department of Surgery, University of Cambridge and

HPB 2018, 20 (S2), S182eS294 S244 Free Orals (FP01-FP31) ’ 4 Addenbrooke s Hospital, Department of Surgery, Uni- FP18-09 versity College London, Royal Free Hospitals, United Kingdom, 5Department of Surgery, Providence Portland PANCREATODUODENECTOMY WITH Cancer Center, United States, 6Department of Surgery, VENOUS RESECTION FOR University of Edinburgh, United Kingdom, 7Department of PANCREATIC CANCER DOES NOT Surgery, University of Auckland, New Zealand, 8Depart- IMPROVE THE R0-RATE ment of Surgery, Hospital Universitario Madrid, Spain, D. Kleive1,2, K. J. Labori1, P. -D. Line2,3, and 9Department of Surgery, Stanford University, United I. P. Gladhaug1,2 and C. S. Verbeke2,4 States 1Department of HPB Surgery, Oslo University Hospital, Background: Predicting the outcomes of pancreatic Rikshospitalet, 2University of Oslo, Institute of Clinical neuroendocrine tumors (PNETs) has been complicated by Medicine, 3Department of Transplantation Medicine, and rarity, staging systems, and variable biology. Tumor ac- 4Department of Pathology, Oslo University Hospital, tivity (Ki67) and mitoses have emerged as a grading based Rikshospitalet, Norway on international consensus, however adequate validation Introduction: Pancreatoduodenectomy with venous has been limited to small, single institution series. We resection is considered standard of care in case of tumor sought to validate the WHO grading of PNETs in a large, involvement of the portal and/or superior mesenteric vein. multi-center, international cohort. It is performed with the aim of achieving an R0-resection. Materials and methods: PNETs undergoing surgery Methods: This prospective observational study included from 1996-2016 were collected from eight major all patients undergoing pancreatoduodenectomy with or referral centers on three continents were collected. A without venous resection for pancreatic cancer between total of 696 patients were identified. We excluded pa- January 2015 and October 2017 at Oslo University Hos- tients with diagnosed genetic syndromes predisposing to pital. Patients who received neoadjuvant chemotherapy PNET, and those with incomplete staging. Preoperative were excluded. Detailed microscopic mapping of the characteristics, operative, pathologic, staging, and pa- tumour at the superior mesenteric vein (SMV)-groove was tient survival data and outcomes were collected. Data undertaken and results were compared between specimens were analyzed using R and compared on the basis of with and without venous resection. overall survival. Results: Ninety-five patients were included, of whom 25 Results: Median follow-up was 36.4 months (range 1-192); had pancreatoduodenectomy with venous resection. R0- median age was 59yrs and 50.6% female; 40.5% of pro- resection, based on >1 mm clearance, was achieved in 15 cedures involved the head; 57.7% tail; 36.7% had N1 or M1 of 70 patients (21.4 %) without venous resection and in one disease. Metastatic disease, LN positivity, lymphovascular of 25 patients (4 %) with venous resection (p=0.061). The invasion, WHO G3 tumors, R1 resection, and necrosis on SMV margin adjacent to the tethered vein was the most path specimen were associated-with worse odds of sur- frequently involved margin in both groups (24 of 25, 96 % vival. On multivariate analysis, LVI and necrosis on final with venous resection, 35 of 70, 50 % without venous pathology were associated with worse survival. Overall resection; p< 0.001). Microscopic mapping revealed that survival was worse in WHO G3 tumors, however no dif- the broad invasive tumour front at the SMV-groove as well ference between G1 & G2 at current WHO values were as the absence of peripancreatic adipose tissue at the SMV- identified. margin were responsible for these findings. The transection margins of the venous resection itself were involved in only three of the 25 patients (12 %). Conclusion: R0-resection following pancreatoduodenec- tomy with venous resection can rarely be achieved due to involvement of the SMV-margin immediately adjacent to the resected vein.

FP18-10 THE CONUNDRUM OF < 2CM PANCREATIC NEUROENDOCRINE TUMORS: A PREOPERATIVE RISK SCORE TO PREDICT LYMPH NODE METASTASES AND GUIDE SURGICAL MANAGEMENT A. Lopez-Aguiar1, M. Zaidi1, F. Rocha2, G. Poultsides3, 4 5 6 7 Figure [km_Plot]. M. Dillhoff , R. Fields , K. Idrees , C. Cho , D. Abbott8 and S. Maithel1 1Division of Surgical Oncology, Department of Surgery, 2 Conclusions: In a large international cohort of resected Emory University Hospital, Department of Surgery, Vir- 3 PNETs with robust follow-up, survival differences between ginia Mason Medical Center, Department of Surgery, 4 G1 & G2 tumors was not seen. G3 tumors had poor sur- Stanford University Medical Center, United States, Di- vival, similar to poorly differentiated neuroendocrine vision of Surgical Oncology, Ohio State University 5 carcinomas. Comprehensive Cancer Center, Department of Surgery,

HPB 2018, 20 (S2), S182eS294 Free Orals (FP01-FP31) S245

6 Washington University School of Medicine, Division of FP18-11 Surgical Oncology, Department of Surgery, Vanderbilt University Medical Center, 7Division of Hepatopancrea- EPIGENETIC INTERFERENCE IN tobiliary and Advanced Gastrointestinal Surgery, THE INTERFERON AND Department of Surgery, University of Michigan, and GEMCITABINE RESPONSE IN 8Department of Surgery, University of Wisconsin School of PANCREATIC CANCER Medicine and Public Health, United States A. Blaauboer, P. van Koetsveld, C. van Eijck and < Introduction: Management of 2cm pancreatic neuro- L. Hofland endocrine tumors(PanNETs) is controversial. Although Erasmus MC, The Netherlands often indolent, the oncologic heterogeneity of these tumors, Introduction: Patients with resectable pancreatic cancer particularly related to lymph node(LN) metastases, poses are indicated for surgery and adjuvant gemcitabine (GEM). challenges when deciding between resection vs Studies have indicated a role of interferon-bèta (IFN-b)in surveillance. counteracting dysregulated immune invasion and to sensi- Method: All pts who underwent curative-intent resection tize tumor cells for GEM treatment. Epigenetic modifica- of primary non-functional < 2cm PanNETs at 8 institutions tions seem to modify the downstream pathway of both IFN- of the US Neuroendocrine Tumor Study Group from 2000- b and GEM, by silencing gene expression, resulting resis- 2016 were analyzed. Tumors with poor-differentiation and tance towards these therapies. Importantly, epigenetic Ki-67>20% were excluded. Primary aim was to create a modifications are reversible and can be targeted by epi- Lymph Node Risk Score(LNRS) that accurately predicted drugs. This study aimed to investigate the potential sensi- LN metastases for < 2cm PanNETS utilizing preoperative tizing effect of different epi-drugs on GEM and IFN-b data. therapy. Results: Of 695 pts with resected PanNETs, 309 were < Method: KPC-3 cells were treated with increasing con- 2cm. 25% of tumors were proximal(head/uncinate), 23% centrations GEM or IFN-b to evaluate the anti-tumor effect had a Ki-67>3%, and only 8% were moderately-differen- of monotherapy. To assess the sensitizing effect of the tiated. 9% of all < 2cm tumors were LN(+), which was histone deacetylase inhibitor valproic acid (VPA), cells associated with worse 5-yr recurrence-free survival were treated with GEM or IFN-b combined with IC25 or compared to LN(-) disease (80%vs96%;p=0.007). Preop- IC50 growth inhibitory concentrations of VPA for seven eratively known factors associated with LN metastases days. were proximal location (OR4.0;p=0.002) and Ki-67>3% Results: The anti-proliferative effect of GEM and IFN-b (OR2.7;p=0.05). Moderate-differentiation was not associ- monotherapy is dose- and time-dependent. A shift in IC50 ated with LN(+) disease. Location and Ki-67 were assigned concentration of GEM was found, and declined when a value weighted by their odds ratio:(distal:1,proximal:4; combined with VPA. An additional inhibition of approxi- Ki-67< 3%:1,Ki-67>3%:3), which formed a LNRS mately 50% was found when IC50 GEM was combined ranging from 1-7. Scores were categorized into low(1-2), with IC50 VPA. Furthermore, 100 IU/ml IFN-b combined intermediate(3-4), and high(5-7) risk groups. Incidence of with IC50 VPA resulted in an additional inhibition of cell LN metastases increased based on risk group: Low:3.2%; growth as well. Intermediate:13.8%; High:20.5%(Table). Only 3.4% of Ki- 67< 3% tumors in the distal pancreas were LN(+) compared to 21.4% of Ki-673% tumors in the head/ uncinate. Conclusions: This simple and novel LN risk score utilizes readily available preoperative factors to accurately stratify risk of LN metastases for < 2cm PanNETs and may help guide management strategy.

Table 1 Lymph Node Positive Rates for <2cm PanNETs

Percent p-value OR (95%CI) p-value Figure [Enhanced gemcitabine reponse by VPA] Lymph Node <0.001 Risk Score Low (Score 1-2) 3.2% Ref – Conclusion: Combining GEM or IFN-b with VPA, we can (n=195) achieve a significantly stronger anti-tumor effect of GEM Intermediate 13.8% 4.9 (1.5-15.7) 0.007 and IFN-b. Other epi-drugs, such as 5AZAcytidine, will be (Score 3-4) tested. Based on the in vitro experiments, the optimal (n=61) combination will be studied in an in vivo study with High (Score 5-7) 20.5% 7.9 (2.5-24.9) <0.001 immune competent mice. (n=53)

HPB 2018, 20 (S2), S182eS294 S246 Free Orals (FP01-FP31)

FP18-12 Methods: We queried a prospectively maintained robotic database to identify patients who underwent robotic DOES NEOADJUVANT THERAPY pancreatic resection by a single surgeon between 2012 and AFFECT THE POSTOPERATIVE 2016. Patient demographics and operative outcomes were COURSE OF PATIENTS SUFFERING compared using Mann-Whitney U, Kruskal Wallis and PANCREATIC DUCTAL Pearson’s Chi-square test as appropriate. fi ADENOCARCINOMA? e A Results: We identi ed 118 patients; 65 whipples (RW), 42 distal pancreatectomies, 4 total pancreatectomies, 6 SYSTEMATIC REVIEW WITH META- pancreatic enucleations, and 1 robotic cyst gastrostomy with ANALYSIS a median age of 70 (24-94), median BMI of 27.5 (16.8-40.2), S. Schorn, E. I. Demir, L. Volovnik-Jose, H. Friess and and ASA of 3. The median EBL was 125(25-800) and LOH G. O. Ceyhan 6(1-34). Mean operative time for RW decreased after 15 Department of Surgery, Klinikum rechts der Isar, Tech- cases (578 vs 457 minutes p< 0.004). Conversions to open nical University of Munich, Germany occurred in 5(4.2%) patients. 116(98.3%) patients under- Background data: Neoadjuvant therapy/NTx is increas- went R0 resections and the median LN harvest was 16(0-37). ingly recognized as a therapeutic option in patients with The 30 and 90 day mortality was 1(0.8%). Major compli- pancreatic cancer/PC especially in borderline resectable PC cations (Clavien-Dindo Grade 3-5) were seen in 16(13.6%) and locally advanced PC. However, the impact of NTx on cases (20.3%) but decreased steadily as volume increased the postoperative course of PC patients have not been (case 30). Pancreatic leaks occurred in 13(11%): A: 8(6.8%), systematically assessed, yet B 1(0.8%), and C: 4(3.4%) Methods: For this purpose, a systematic review with meta- Conclusion: Robotic assisted approaches to pancreatic analysis was conducted according the Preferred-Reporting- resections is feasible. However, it takes approximately 15 Items-for-Systematic-review-and-Meta-Analysis/PRISMA- cases before a decrease in operative time and 30 cases guidelines by screening medical data-bases for predefined before major complications are decreased. These trends in search terms. complications associated with surgeon experience are Results: Thereby, 4,210 studies could be identified analy- critical to consider in robotic pancreatic surgery. sing the effect of NTx on the postoperative outcome in PDAC. Data of 22 studies with comparative data of NTx versus upfront surgery could be pooled in meta-analysis. FP18-14 No effect of NTx could be observed on overall morbidity VENOUS RECONSTRUCTION WITH (RR 0.96 95%-CI: 0.83-1.11; p=0.57), clinically relevant THE PARIETAL PERITONEUM. THE pancreatic fistula (RR 0.72 95%-CI: 0.48-1.09; p=0.12), biliary leakages (RR 0.61 95%-CI: 0.20-1.84; p=0.38), LONG RESULTS IN 106 PATIENTS postoperative bleeding (RR 1.04 95%-CI: 0.67-1.60; S. Dokmak, B. Aussilhou, F. Cauchy, O. Soubrane and p=0.87), cholangitis (RR 0.90 95%-CI: 0.34-2.42; p=0.84), A. Sauvanet delayed gastric emptying (RR 1.01 95%-CI: 0.81-1.25; HPB and Liver Transplantation, Beaujon Hospital, p=0.96) or reoperations (RR 1.06 95%-CI: 0.76-1.48; Assistance Publique Hôpitaux de Paris, University Paris p=0.73). Moreover, NTx was associated an increased risk VII, France of mortality (RR 1.54 95%-CI: 1.07-2.22; p=0.02) which Introduction: We recently described venous reconstruc- vanished after stratifying meta-analysis into subgroups tion during HPB surgery with the parietal peritoneum (PP). according the distribution of vascular resections. Our aim is to evaluate the long term results. Conclusions: Accordingly, NTx seems not to be associated Method: Between 2010-2017, 106 patients underwent with an increased risk for postoperative morbidity and no pancreatic (n=76) or liver (n=30) resections for malignancy obvious increased risk for postoperative mortality in PDAC (n=104) with reconstruction of the mesentericoportal vein patients. Prospective studies are urgently needed to address (84), the vena cava (15), or hepatic veins (7) with the PP. The this concern. PP (mean length=25.2 mm; 10-100) was harvested from the falciform ligament (n=48), hypochondrium (n=23), dia- FP18-13 phragm (n=19), or prerenal (n=16) area. Reconstruction was OUTCOMES ASSOCIATED WITH lateral in 101 patients, including 14 (14%) with resection of ROBOTIC APPROACHES TO > 75% of the vascular lumen, and tubular in 5 patients. In 14 PANCREATIC RESECTION patients the need for a graft was urgent. Postoperative anti- coagulation was standard and venous patency and stenosis 1 1 2 3 K. Meredith , P. Briceno , J. Huston and R. Shridhar was assessed by routine CT scans. The mean radiological 1 Gastrointestinal Oncology, Florida State University / follow-up was 19 (2-65) months. 2 Sarasota Memorial Institute for Cancer Care, Gastroin- Results: The mean age was 62 (31-84), 38 (36%) were testinal Oncology, Sarasota Memorial Institute for Cancer females and 28 (26%) were transfused. One non related 3 Care, and Radiation Oncology, University of Central mortality, overall morbidity (n=52; 49%) and the mean Florida, United States hopsital stay was 19 (6-75) days. There was no PP-related Introduction: Minimally invasive techniques have or haemorraghic complications and only one reintervention improved post-operative outcomes, however, the majority for symptomatic venous thrombosis. R1 resection in 34 of pancreatic surgery, known for its complexity, is still (32%) patients including 6 (6%) R1 venous (< 1mm). The performed via open approaches. The development of ro- patency rate was 97/106 (92%) and complete thrombosis botics has improved dexterity which may allow for appli- (n=9) was mainly observed after tubular reconstruction cation in more complex surgeries. (n=3) and distal pancreatectomies (n=4).

HPB 2018, 20 (S2), S182eS294 Free Orals (FP01-FP31) S247

Conclusions: The PP showed a high patency rate without pancreaticoduodenectomy (PD). We aim to evaluate any significant related complications and should be the first the problem and develop a tool to identify patients who choice when a lateral venous reconstruction is needed. will fail to comply with the ERp. Methods: From 1/1/2014 to 01/01/2016 205 consecutive patients underwent a PD in our center and were managed FP18-15 accordingly to an ERp.Postoperative complications were fi DISTAL SPLENORENAL AND graded using Clavien-Dindo classi cation (grade3=major).Perioperative data were prospectively MESOCAVAL SHUNTING AT THE TIME collected. Items failed were define as: no active deambula- OF PANCREATECTOMY: AN tion postoperative day (POD) 1; < 4 hours out of bed POD2; IMPORTANT ADJUNCT TO THE removal of NGT and bladder catheter after POD1 and 3 OPERATIVE MANAGEMENT OF respectively; reintroduction/suspension of oral feeding after SELECTED PATIENTS POD4;intravenous infusions after POD4. Univariate/multi- 1 1 1 1 variate analysis and regression models were conducted.Data M. I. Chavez , S. Tsai , M. Aldakkak , C. N. Clarke , are presented as percentage, mean (CI). P. Tolat2, D. B. Evans1 and K. K. Christians1 1 2 Results: Observing the mean number of failed items in Surgical Oncology, and Radiology, Medical College of patients without or with postoperative complications Wisconsin, United States (1.13(0.88-1.37) vs 2.53(2.25-2.81);p< 0,001), major Background: We previously provided proof that pancre- complications (1.66(1.446-1.88) vs 3.15(2.64- atic resection in the setting of complex vascular involve- 3.67);p=0,001) and prolonged length of stay, we define the ment can be safely performed. When the tumor involves the failure of the ERp as the failure of two of more items. portal-superior mesenteric-splenic vein (PV-SMV-SV) 56.6% failed the ERp. We create a predictive model confluence with abutment of the SMA, a mesocaval shunt considering age,BMI,operative time and the consistency of (MCS) may be created with/without distal splenorenal the pancreatic stump (OR 1.031,p =0.012;OR shunt (DSRS). This achieves wide exposure of the vascular 1.111.p=0.031;OR 1.004,p=0.022;OR 2.892,p=0.002 structures involved and allows for a safe pancreatectomy. respectively). We attributed a score from regression co- We report our experience with these procedures. efficients to each variable.Overall patient’s score predict the Methods: All patients with pancreatic resection and either failure of the ERp (AURC 0,7476(0,678-0,816)). The score DSRS and/or MCS were identified from our institutional showed a good agreement between predicted and observed database. Demographics, stage, neoadjuvant treatment, proportions of failures (Hosmer-Lemeshow test;p=0.7127). operative details and outcomes were reviewed. Conclusion: Despite the advantages of ERps,still half of Results: Twenty-eight patients were resected with the aid of patients after PD are not able to comply with an ERp.ERp shunting procedures (25 adenocarcinomas, 2 neuroendo- failure seems to be predictable.Can high risk patients crine cancers, 1 solid pseudopapillary tumor), including 23 benefit from a specific ERp? standard or extended pancreaticoduodenectomies and 5 total pancreatectomies. Shunts consisted of 11 MCS, 15 DSRS and 2 MCS/DSRS. Eight MCS were temporary. Median FP18-17 operative time was 8.8 hours (6.3-13), median estimated blood loss was 925 mL (100-3500) and median length of A VALIDATION OF COMPREHENSIVE hospital stay was 11.5 days (7-35). Two patients experi- COMPLICATION INDEX FOR enced Clavien-Dindo grade III or above complications. No POSTOPERATIVE COMPLICATION 90-day mortalities or reoperations encountered. Median SEVERITY COMPARED TO THE follow up was 12.2 months (0.5-97). Overall survival was 29 CLAVIEN-DINDO CLASSIFICATION months for patients with adenocarcinoma. Of the permanent S. H. Kim1,2, H. K. Hwang1, K. S. Kim1, W. J. Lee1 and shunts performed, all but one DSRS was patent at follow up. 1 Conclusion: Shunting facilitates safe resection in pancre- C. M. Kang 1Hepatobiliary and Pancreatic Surgery, Yonsei University atic tumor patients who may otherwise be deemed inoper- 2 able. In the era of more effective systemic therapy, it’s College of Medicine, and Surgery, Armed Forced Capital critically important to know which patients are candidates Hospital, Republic of Korea for extended resections and how to safely perform them. Background: Complication system is one of the important index for evaluation of immediate postoperative outcomes. FP18-16 Up till now, Clavien-Dindo Classification (CDC) have been PREDICT THE FAILURE OF AN widely used for complication system. Recently, Compre- hensive Complication Index (CCI) was introduced. Unlike ENHANCED RECOVERY PROTOCOL CDC, CCI incorporates all complication severities. The aim AFTER of study is to validate CCI compared to CDC. PANCREATICODUODENECTOMY Methods: Records from 222 patients who underwent G. Capretti1, M. Cereda2, F. Gavazzi2, C. Ridolfi2, radical pancreatectomy from April 2015 to October 2016 G. Nappo2, B. Branciforte2 and A. Zerbi3 were prospectively reviewed. CCI was calculated and the 1Humanitas Research Hospital, 2Humanitas Research relation with postoperative day (POD) and cost were Hospital, and 3Humanitas University, Humanitas Clinical assessed, comparing to the relation of CDC. A Hotelling’s and Research Center, Italy two-sample dependent test for correlations was used to fi Introduction: Few is known about patients’ compli- calculate the signi cant difference between the correlation fi ance to enhanced recovery protocols (ERp) for coef cient of the parameters with CDC and CCI.

HPB 2018, 20 (S2), S182eS294 S248 Free Orals (FP01-FP31)

Result: Complications occurred in 211 patients (95.0%). and 7 months for patients who received NT + AT, NT, AT, The correlation between CCI and CDC was r = 0.938, p < and SA, respectively (p< 0.0001). 0.001. Comparing CDC, CCI was shown significantly Conclusion: In contrast to recommended guidelines, 57% of stronger relation with POD and Cost. (POD: CCI vs. CDC, patients with LAPC underwent upfront surgical therapy. Uti- r = 0.725 vs. r = 0.630, p < 0.001; Cost: CCI vs. CDC, r = lization of neoadjuvant therapy was associated with patho- 0.774 vs. r = 0.723, p < 0.001). logic downstaging and an improved survival. Surgery should Conclusion: CCI could be a sophisticated scoring system be utilized cautiously in patients with LAPC and only in the that could be used to identify risks in surgical patients. context of a favorable response to neoadjuvant therapy.

FP18-18 Table 1Pathologic Stage Based on the American Joint Committee on Cancer, 7th edition Total NT NT + AT AT SA p-value n [ 1114 n [ 326 n [ 150 n [ 393 n [ 245 T Stage, n (%) <0.001 T0 4 (<1) 4 (1) 0 (0) 0 (0) 0 (0) T1 78 (7) 55 (17) 19 (13) 4 (1) 0 (0) T2 60 (5) 34 (11) 12 (8) 10 (3) 4 (2) T3 393 (35) 180 (55) 92 (61) 75 (19) 46 (19) T4 578 (52) 52 (16) 27 (18) 304 (77) 195 (80) N Stage, n (%) <0.001 N0 473 (43) 217 (67) 65 (43) 119 (30) 72 (29) N1 640 (57) 108 (33) 85 (57) 274 (70) 173 (71) Anatomic Stage, n (%) <0.001 0 4 (<1) 4 (1) 0 (0) 0 (0) 0 (0) I 109 (10) 80 (25) 20 (13) 7 (2) 2 (1) II 417 (37) 185 (57) 101 (67) 84 (21) 47 (19) III 583 (52) 56 (17) 29 (19) 302 (77) 196 (80)

FP18-18 OPERATIVE MANAGEMENT OF LOCALLY ADVANCED PANCREATIC FP19 - Free Papers 19 (mini oral) - Biliary: Surgical & Technical Outcomes CANCER: ANALYSIS OF THE FP19-01 NATIONAL CANCER DATABASE PERIOPERATIVE AND LONG-TERM C. Barnes, C. Clark, K. Christians, B. George, P. Ritch, W. Hall, B. Erickson, M. Aburajab, D. Evans and S. Tsai OUTCOMES OF LAPAROSCOPIC Medical College of Wisconsin, United States VERSUS OPEN LYMPHADENECTOMY Introduction: Practice guidelines recommend nonopera- FOR BILIARY TUMORS: A tive management for patients with locally advanced PROPENSITY-SCORE BASED CASE- pancreatic cancer (LAPC). However, surgery has been MATCHED ANALYSIS increasingly incorporated in the multimodality treatment F. Ratti, G. Fiorentini, F. Cipriani, M. Paganelli, schema for LAPC. M. Catena and L. Aldrighetti Methods: Patients with LAPC (clinical T4) who underwent Hepatobiliary Surgery Division, IRCCS San Raffaele fi surgery were identi ed from the National Cancer Database. Hospital, Italy Patients were classified based on the receipt of neoadjuvant Background: The aim of this study was to compare pa- therapy (NT), neoadjuvant therapy and adjuvant therapy tients undergoing MILS and open liver resection with (NT + AT), adjuvant therapy (AT) or surgery alone (SA). associated lymphadenectomy for biliary tumors (intra- Clinicopathologic characteristics and survival outcomes hepatic cholangiocarcinoma and gallbladder cancer) in a were compared. case-matched analysis using propensity scores. Results: Surgery was performed on 1,113 patients with Methods: 95 patients underwent liver resection with LAPC; 325 (29%) received NT, 150 (13%) received NT + associated locoregional lymphadenectomy by laparo- AT, 393 (35%) received AT, and 245 (22%) received SA. scopic approach constituting the study group (MILS Margin negative resections were achieved in 378 (80%) of group). The MILS group was matched in a ratio of 1:2 the 475 patients treated with neoadjuvant therapy as with patients who had undergone open resection for pri- compared to 275 (43%) of the 638 patients treated with mary biliary cancers (Open group, constituting the con- surgery first (p< 0.001). Patients who received NT or NT + trol group). Short and long term outcomes were evaluated AT had lower pathologic stage as compared to patients who and compared, with specific focus on specificdetailsof received SA or AT (p< 0.001; Table 1). The median overall lymphadenectomy. survival (OS) for all patients was 18 months; 28, 26, 14,

HPB 2018, 20 (S2), S182eS294 Free Orals (FP01-FP31) S249

Results: Preoperative characteristics of patients and disease FP19-03 were comparable between MILS and Open Group. Lapa- roscopic series resulted in a statistically significant lower ROBOTIC blood loss (200 vs 350, p=0.03), minor intraoperative blood PANCREATICODUODENECTOMY: transfusions (3.2% vs 7.9%, p=0.04) and postoperative SURGICAL OUTCOMES OF blood transfusions (10.5% vs 15.8%), other than shorter CONSECUTIVE 30 CASES lenght of stay (4 vs 6 days, p=0.04). Number of retrieved J. H. Lee, S. H. Shin, K. B. Song, D. W. Hwang, nodes was 8 vs 7 (p=ns); particularly, percentage of patients S. C. Kim and K. M. Park who achieved the recommended AJCC cutoff of six lymph Asan Medical Center, Republic of Korea nodes harvested were 93.7% vs 85.8% (p=0.05). Morbidity was lower in MILS group (16.8% vs 22.1%, p=0.02). Robot surgery is a novel platform that maintains ad- Median disease free survival was 33 vs 36 months and vantages and overcomes disadvantages of conventional fi disease recurrence occurred in 45.3% vs 55.3% of patients methods, even pancreaticoduodenectomy is a dif cult and in MILS and Open Groups respectively. complex operation. This study aimed to evaluate safety and Conclusion: Laparoscopic approach for lymphadenectomy feasibility of robotic pancreaticoduodenectomy (RPD). is a valid option in patients with biliary cancers since it Retrospective review of a prospective database of RPD for allows to maintain the advantages of minimally invasive periampullary lesions between December 2015 and January approach, without compromising the accuracy and the 2018. Thirty patients with mean age of 55.3 years under- outcomes of nodal dissection. went attempted RPD. No one required conversion to open procedure. Totally mean operative time was 461.8 min (range 350-754) with 514 min of initial 10 cases and 410 min of last 10 cases. Perioperative transfusion was required FP19-02 in 2 patients. For pancreatojejunostomy, all were performed QUALITY OF LIFE AFTER SINGLE duct-to-mucosa anastomosis, with internal stent (28 cases, PORT LAPAROSCOPIC 93.3%). Diagnosis was ampullary carcinoma (n= 12), CHOLECYSTECTOMY: A pancreatic adenocarcinoma (n = 7), and bile duct adeno- RANDOMIZED CLINICAL TRIAL carcinoma (n=5), intraductal papillary mucinous neoplasm (n = 3), neuroendocrine tumor (n = 2), and metastatic renal E. Ito, A. Takai and Y. Takada cell carcinoma (n = 1). Median tumor size was 2.4 cm and Hepato Biliary Pancreatic and Breast Surgery, Ehime the median number of lymph nodes harvested was 16 University, Japan (range, 7-32). The margin-negative resection rate was Introduction: New single port laparoscopic cholecystec- 93.3% and metastatic lymph nodes were found 11 patients tomy (SILC) offers a cosmetic benefit from reduced ports. (36.6%). Pancreatic fistula occurred in 7 patients (23.3%) Meanwhile, although there has been controversy, its large of grade A and 3 (10%) of grade B. Delayed gastric incision in the umbilicus compared with 4 port laparoscopic emptying occurred in 1 patient and wound problem in 2. No cholecystectomy (4PLC) may lead to worse postoperative one occurred mortality. Median length of postoperative quality of life (QOL) and pain. stay was 11 days (range, 7-34 days). RPD can be performed Methods: This study was a multicenter, parallel-group, with safety and oncologic outcomes. This procedure holds open-label, randomized clinical trial. One-hundred and promise for providing advantages seen with minimally twenty patients who planned to undergo elective chole- invasive approaches in other procedures. cystectomy were randomized 1:1 into SILC or 4PLC group. The primary outcome was QOL: defined as the time to recover daily activities. Patients were assessed continu- FP19-04 ously for 2 weeks during the perioperative period. The secondary outcome includes postoperative pain, and other VALIDATION OF DYNAMIC AIR clinical outcomes. INSUFFLATION TEST (DAIT) IN Results: Fifty-eight patients in SILC and 53 patients in TESTING INTEGRITY OF RYHJ 4PLC (n = 111, 47 male, mean age 57) were analyzed. The R. Saxena, V. Reddy, A. Mishra and SAXENA CHBDT time to recover daily activities was 1.4 day longer in SILC Surgical Gastroenterology, Sanjay Gandhi Post Graduate group (10.2 days in SILC vs 8.8 days in 4PLC, P = 0.12). Institute of Medical Sciences, India SILC group also showed slower releases from post- Introduction: Hepaticojejunostomy(HJ) is technically operative pain, though statistically insignificant. Mean challenging anastomosis. HJ leak is associated with sig- operative time of SILC was slightly longer than that of nificant morbidity and mortality. DAIT is a simple and easy 4PLC (99.3 vs 91.8 minutes). Hospital stay, time to resume method which can be used intraoperatively to test the working, and inflammatory response, complication rate integrity of HJ (ACS 2017 video). We are reporting results were similar between the groups. Two incisional hernia of DAIT done in 63 patients undergoing Roux-en-Y occurred in SILC group. Hepaticojejunostomy (RYHJ). Conclusion: The results suggest that SILC retards recovery Methodology: A prospective study at a tertiary centre in of QOL, especially for patients under the age of 60, females India. All patients undergoing RYHJ for Benign Biliary and full time workers. Stricture (BBS) and Choledochal cyst (CDC) excision were

HPB 2018, 20 (S2), S182eS294 S250 Free Orals (FP01-FP31) included in the study. Patients requiring concomitant liver resection were excluded from the study. Intraoperatively DAIT was done as described (video). Any air leak was repaired and DAIT repeated till no further air leak was present. Postoperative bile leak and other complications were noted and graded with Clavien Dindo system. Results: 181 patients undergoing RYHJ were divided into 2 groups - group 1 DAIT done (n=63) and group 2 (n=118) - DAIT was not done. In group 1, 5 patients tested positive on DAIT - only two had post-operative bile leak (3.2%) and only one (1.6%) had Grade 3a complication. None of the patients with negative DAIT experienced post-operative bile leak (Specificity 100%). In group 2, 11 (9.3%) expe- rienced postoperative bile leak and 9/118 (7.6%) had grade 3a or above complications. Conclusion: DAIT is reproducible, inexpensive, requiring no special instruments, very little operating time, and re- duces the incidence of bile leak to a third and morbidity to a fifth of those where DAIT was not done, ensuring safer RYHJ with 100% specificity.

FP19-05 PROGNOSTIC VALUE OF THE NUMBER OF METASTATIC LYMPH NODE IN PATIENT WITH BILIARY TRACT CARCINOMA Y. Miyata, R. Kogure, A. Nakazawa, T. Mitsui, R. Ninomiya, M. Komagome, A. Maki and Y. Beck Hepatobiliry Pancreatic Surgery, Saitma Medical Center, Figure [OS and DFS] Saitama Medical University, Japan Introduction: It was known that the prognosis of some carcinoma was correlated with the metastatic lymph node number (MLNN). About biliary tract carcinoma, relation- FP19-06 ship between MLNN and tumor prognosis was not clearly EVALUATION OF POSTADJUVANT known. The aim of this study was to investigate the prog- THERAPY FOR AMPULLA OF VATER nostic value of the MLNN in biliary tract carcinoma. CARCINOMA: MULTICENTER: A Method: This study was retrospective case-control study of patients who underwent surgical treatment for biliary tract RETROSPECTIVE STUDY carcinoma between January 2006 and July 2017. The J. S. Park1, H. S. Kim1, D. S. Yoon1, J. Y. Jang2, relationship between MLNN and overall survival (OS), W. I. Kwon2, S. W. Kim2, Y. S. Yoon3, H. S. Han3, disease free survival (DFS) and early recurrence within one S. J. Park4 and S. S. Han4 year after surgery were investigated. 1Department of Surgery, Pancreatobiliary Cancer Clinic, Results: During the study periods, 255 patients underwent Gangnam Severance Hospital, Yonsei University College the surgical resection for biliary tract carcinoma. Increasing of Medicine, 2Department of Surgery, Seoul National the MLNN revealed poor prognosis, especially MLNNS3 University College of Medicine, 3Department of Surgery, group (median OS; 10.8 months, median DFS; 6.5 months) Seoul National University Bundang Hospital, Seoul Na- was significantly poor than that of MLNN=1-2 group tional University, and 4Department of Surgery, Center for (median OS; 55.4 months, median DFS; 15.3 months) (P< Liver Cancer, National Cancer Center, Republic of Korea 0.001). Multivariate analysis revealed that lymph node Ampulla of vater carcinoma (AOV) is an infrequent metastasis and perineural invasion were independent tumor that account for about 0.2% of gastrointestinal ma- prognostic factor. About the multivariate analysis of lymph lignancies. The incidence of AOV carcinoma is steadily node metastasis positive (N1) patients (n=122), MLNNS3 increasing. But there are no guidelines from both the was poor prognosis factor (Hazard ratio 2.62, 1.40-4.86 NCCN and the ESMO for the treatment of advanced AOV 95%CI, P=0.003 in OS, Hazard ratio 2.51, 1.37-4.50 95% carcinoma. Only a few retrospective studies have focused CI, P=0.003 in DFS). Early recurrence was occurred on the adjuvant treatment in AOV carcinoma. So, we significantly frequently in N1 patients (n=61, P< 0.001). conducted multicenter retrospective review for evaluation Multivariate analysis of N1 patients revealed MLNNS3 of postadjuvant therapy for AOV carcinoma. Patients who was poor prognosis factor (Odds ratio 2.66, 1.08-6.77 95% underwent curative resection for AOV carcinoma at the CI, P=0.034). four hospital (Yonsei Gangnam severance hospital, Seoul Conclusion: MLNNS3 had significantly poor prognosis national university hospital, Seoul national university in biliary tract carcinoma. It was suggested that MLNN bundang hospital, National cancer center, n =715; 2002- could predict the risk of early recurrence. 2015) were reviewed. Patients with metastatic disease at

HPB 2018, 20 (S2), S182eS294 Free Orals (FP01-FP31) S251 surgery, or insufficient pathologic data were excluded. 5-year survival rate was significantly higher in patients Adjuvant treatment were chemotherapy or chemo- treated with adjuvant therapy (67%) than in patients treated radiotherapy. Overall survival and recurrence free survival with surgery alone (41%, p=0.0015). Adjuvant therapy was compared using Kaplan-Meier estimates. In univariate showed significant prognostic factor for survival on analysis, locally advanced T stage (T3-4) and node posi- multivariate analysis (Relative risk 0.346, p< 0.0001). tive, perineural invasion, lymphovascular invasion, Conclusions: Hepatectomy with adjuvant therapy im- moderately cell differentiation were associated with adju- proves survival in patients with ICC. vant treatment. (p = < 0.001) No adjuvant group (Group 1) were 437 patient and adjuvant treatment group were 272 patients (Chemotherapy,Group 2; 94 patients, Chemo- FP19-08 radiotherapy, Group 3; 184 patients). In the high risk pa- EVOLUTION OF BILE DUCT REPAIR: tients, such as patients with node-positive, Group 3 was significantly associated with increased recurrence free COMPARISON OF DIAGNOSIS, survival (p = < 0.05). Adjuvant treatment did not improve REFERRAL, MANAGEMENT AND survival in low risk patients. However adjuvant chemo- OUTCOMES IN REPAIR OF BILE DUCT radiation treatment maybe improved recurrence free sur- INJURY AFTER LAPAROSCOPIC vival in high risk patients (node-positive, advanced stage). CHOLECYSTECTOMY FROM 1991e2004 Prospective evaluation of appropriate adjuvant treatment e should be considered. AND 2005 2017 J. Lindemann1,2, E. Jonas1, U. Kotze1, C. Kloppers1, M. Bernon1, S. Burmeister1 and J. E. J. Krige1 1 FP19-07 Department of Surgery, University of Cape Town Health Sciences Faculty, South Africa, and 2Department of Sur- HEPATECTOMY WITH ADJUVANT gery, Washington University School of Medicine, United THERAPY IMPROVED SURVIVAL IN States PATIENTS WITH INTRAHEPATIC Bile duct injury continues to be a serious complication CHOLANGIOCARCINOMA after laparoscopic cholecystectomy (LC). Few studies have S. Ariizumi, Y. Kotera, S. Yamashita, A. Omori, T. Kato, analysed the evolution in the management of bile duct S. Nemoto, R. Higuchi, S. Katagiri, H. Egawa and injury (BDI) over time. In this study we compared pre- M. Yamamoto sentation, management and outcome in two time periods. Dept. of Gastroenterological Surgery, Tokyo Women’s A retrospective review of a prospectively maintained Medical Univ., Japan database for patients with BDI after LC was performed. All patients undergoing index hepaticojejunostomy repair from Background: Intrahepatic cholangiocarcinoma (ICC) is a 1991-2017 were included and divided into two groups fatal disease because of frequent recurrence despite curative (1991-2004 and 2005-2017) for comparison. Time of surgery. We investigated the usefulness of hepatectomy diagnosis, referral and repair, pre-operative interventions with adjuvant therapy for patients with advanced ICC. and imaging, post-operative re-interventions, and compli- Methods: We retrospectively studied 228 patients with cations (Accordion Classification) were analyzed. ICC who underwent surgery between 2000 and 2014. Of Three times more repairs were performed in the later these, 104 patients underwent adjuvant therapy after cura- period (93 vs 32). In the later group, there was a significant tive surgery (65 patients underwent immunotherapy and 39 increase in referral delay (p = 0.031), as well as an patients underwent chemotherapy with S-1). Surgical out- increased use of cross-sectional pre-operative imaging (p = comes were compared between 104 patients treated with 0.006, p < 0.001) and percutaneous intervention (p < adjuvant therapy and 90 patients treated with surgery alone. 0.001), and a significant increase in referral to repair time (p Results: The median CA19-9 did not differ between < 0.001). There was an increase in the number of com- groups, nor did the number of cases of macroscopic tumor plications in the later period (p = 0.038). type, the number of cases with lymph node metastasis. The In this analysis the number of BDIs requiring repair, as 5-year survival rate was significantly higher in patients well as pre-operative and operative management has treated with adjuvant therapy (56%) than in patients treated changed over time. The observed differences in the later with surgery alone (29%, p< 0.0001). In patients with N1, group are due to a combination of changes in referral pat- the 5-year survival rate was significantly higher in patients terns, and increased pre-operative investigations and in- treated with adjuvant therapy (39%) than in patients treated terventions, which influenced outcome. with surgery alone (5%, p=0.0002). In patients with N0, the

FP19-08 Table [BDI Repair 1991-2004 and 2005-2017] 1991-2004 2005-2017 p-value Pre-repair 1991-2004 2005-2017 p-value n[32 n[93 Intervention n[32 (%) n[93 (%) Time (days) - Diagnosis Median (IQR) 5 (0.75-8) 3 (0-8) 0.443 ERCP 20 (62.5) 46 (49.5) 0.206 Time - Referral 1.5 (0-10.25) 6 (1-15) 0.031 PTC 17 (53.1) 78 (81.3) <0.001 Time - Repair 2.5 (1-5.5) 8 (4-43) <0.001 Percutaneous Drain 7 (21.9) 17 (18.3) 0.659 Immediate repair (< 3 days) n (%) 7 (21.9) 11 (11.8) 0.328 Laparotomy 8 (25) 32 (34.4) 0.329 Early Repair (> 3,  14 days) 11 (34.4) 22 (23.7) 0.239 Laparoscopy 2 (6.3) 4 (4.3) 0.660 Intermediate Repair (> 14,  90 days) 11 (34.4) 38 (40.9) 0.521 MRCP 7 (21.9) 46 (49.5) 0.006 Late Repair (> 90 days) 3 (9.4) 22 (23.7) 0.083 CT 3 (9.4) 43 (46.2) <0.001

(continued on next page) HPB 2018, 20 (S2), S182eS294 S252 Free Orals (FP01-FP31)

(continued) 1991-2004 2005-2017 p-value Pre-repair 1991-2004 2005-2017 p-value n[32 n[93 Intervention n[32 (%) n[93 (%) Post-repair percutaneous intervention 3 (9.4) 4 (4.3) 0.163 Ultrasound 2 (6.3) 22 (23.7) 0.002 Post-repair operative intervention 2 (6.3) 2 (2.2) 0.259 Total Complications 18 (56.3) 50 (53.8) 0.038

Background: Hepatopancreaticoduodenectomy (HPD) is FP19-09 becoming more feasible to be performed safely due to HOW TO REPAIR AND MANAGE advancement of imaging diagnosis and surgical technique. INTRAOPERATIVE BILE DUCT However, there are few reports of HPD requiring combined resection and reconstruction of portal vein (PV) and hepatic INJURIES IN A HIGH-VOLUME artery (HA). REFERRAL CENTER IN 2018? Methods and results: We have performed 66 HPD from M. de Santibanes1, R. Sanchez Claria2, E. de Santibañes2, October 1996 to October 2017.2-stage pancreaticojejunos- O. Mazza2, M. Palavecino2, G. Arbues2, F. Alvarez2 and tomy was applied in all patients. In PV resection group, two J. Pekolj2 were diagnosed with gallbladder cancer and 11 bile duct 1Department of Surgery, Division of HPB Surgery, Liver cancer, in HA resection group one gallbladder cancer and two Transplant Unit, Hospital Italiano de Buenos Aires, and bile duct cancer, PV and HA resection patient was diagnosed 2Hospital Italiano de Buenos Aires, Argentina with gallbladder cancer. The hepatectomy procedures Introduction: Bile duct injury (BDI) remains the most consisted of extended right hepatectomy in 10 patients and serious complication of laparoscopic cholecystectomy others in 3 patients in PV resection group, left trisectionectomy (LC). The purpose of the current study is to review the in one patient and extended left hepatectomy in two patients in experience in the intraoperative repair of BDI sustained HA resection group, and inferior part of segment 4 and 5 during LC at a high-volume referral center. resection was performed in PV and HA resection patient. Methods: A retrospective analysis of a single-institution The median operative time and blood loss were 730 database was performed. Patients who underwent LC with minutes and 1100ml in PV resection group, 840 minutes routine intraoperative cholangiography between October and 3200ml in HA resection group, and 1020 minutes and 1991 and January 2018 were included. 900ml in HA and PV resection patient. fi Results: Among 15,473 LC performed during the study After operation, pancreatic stula of Grade B or more fi period, 23 (0.14%) patients had a BDI sustained during the was observed in all patients, but hepatic insuf ciency procedure. BDI was diagnosed intraoperatively in 21 patients beyond Grade B was not observed. The 90-day mortality (91.3%). According to the Strasberg classification of BDI, was 0. In PV resection group, two patients with bile duct there were 4 type C lesions, 17 type D lesions, and 2 type E2 cancer is surviving without recurrence for more than 5 lesions. There were no associated vascular injuries. Twelve years, and in HA resection group one patient with bile duct cases (52%) were converted to open surgery. The repairs cancer is surviving without recurrence for 4 years. included 13 primary biliary closures, 4 Roux-en-Y hepatico- jejunostomies, 2 end to end anastomosis, and 1 laparoscopic transpapillary drainage and 1 ERCP. Postoperative compli- FP19-11 cations occurred in 7 patients (30.4%). During the follow-up LONG-TERM RESULTS AFTER ALPPS period, early biliary strictures developed in 2 patients (11.7%) AND PVE FOR PATIENTS WITH and were treated by percutaneous dilation and a Roux-en-Y PERIHILAR CHOLANGIOCARCINOMA - hepaticojejunostomy with satisfactory long-term results. Conclusion: The current series represents one of the largest SINGLE CENTER EXPERIENCE single-center experiences in terms of intraoperative repair of J. Rolinger, I. Capobianco, A. Königsrainer and S. Nadalin BDI sustained during LC. The results suggest that a high level General, Visceral and Transplant Surgery, University of intraoperative diagnosis is possible, where intraoperative Hospital of Tübingen, Germany cholangiography is a useful tool. The intraoperative repair of Background: Surgical treatment for perihilar chol- BDI sustained during LC by experienced hepatobiliary sur- angiocarcinoma (ph-CCA) usually requires extended geons either by open or laparoscopic approach appears of resection after inducing hypertrophy of the future liver paramount importance to assure optimal results. remnant (FLR). Associating liver partition and portal vein FP19-10 ligation for staged hepatectomy (ALPPS) is associated with high morbidity and mortality, therefore portal vein embo- INDICATION AND SURGICAL lization (PVE) remains the gold standard in this context. TECHNIQUE OF Aim of this study was to compare outcomes in patients that HEPATOPANCREATICODUODE underwent ALPPS or PVE prior to right trisectionectomy NECTOMY WITH COMBINED for ph-CCA. RESECTION OF PORTAL VEIN AND Methods: Between January 2010 and December 2017 a total of 29 patients with ph-CCA were planned for right HEPATIC ARTERY FOR BILIARY trisectionectomy after augmentation procedure. Accord- CANCER ingly, 8 patients were treated with ALPPS and 23 with PVE Y. Kazami, Y. Sakamoto, T. Kokudo, N. Akamatsu, (2 ALPPS were salvage procedures after failed PVE). J. Arita, J. Kaneko and K. Hasegawa Retrospective evaluation focused on short- and long-term Hepato-Biliary-Pancreatic Surgery Division Department outcomes including morbidity, mortality, overall survival of Surgery, University of Tokyo, Japan and recurrence.

HPB 2018, 20 (S2), S182eS294 Free Orals (FP01-FP31) S253

Results: ALPPS could be completed in 87,5% (7/8) Hospital, 4Uijeongbu St. Mary’s Hospital, and 5St. Paul’s whereas resection after PVE in 60,9% (14/23; p=0,222). Hospital, Republic of Korea 3 Severe morbidity (Dindo-Clavien IIIa) after ALPPS 1 and Background: In cases of acute cholecystitis (AC), post- PVE was 37,5% (3/8) and 21,7% (5/23; p=0,393) respec- operative antibiotic prophylaxis is generally used for the tively. Moreover, severe morbidity after completed pro- purpose of preventing subsequent infections. However, cedure was comparable: 57,1% (4/7 ALPPS and 8/14 there is still no standardized guideline regarding antibiotic PVE). ALPPS was associated with 28,6% (2/7) 30-day administration after cholecystectomy. mortality, whereas PVE with 14,3% (2/14; p=0,574). Methods: A total of 200 patients at five participating Overall median survival and recurrence-free survival was hospitals who were admitted for cholecystectomy to treat 43,1 and 30,0 months for ALPPS and 12,1 and 11,3 months grade I or II AC were enrolled and randomly allocated to a for PVE. group given a placebo (group A) or a group given post- Conclusion: PVE impresses with sobering results operative antibiotics (group B). Surgical outcomes and concerning growth of Future river remnant (FLR) and a incidence of postoperative infectious morbidities were high preoperative drop-out rate. Despite comparable reviewed. morbidity ALPPS seems to offer better oncological results concerning radicality and long-term results.

Table 1: Demographics and surgical results Item ALPPS PVE p (n[8) (n[23) Median Age (years) 72 68 0,331 Charlson 37,5% (3/8) 39,1% (9/23) 1,0 Comorbidity Index  1 Median remnant liver 0,36 0,42 0,161 volume to body weight ratio (RLV-BWR) before augmentation (%) Median remnant liver 0,62 0,60 0,651 volume to body weight ratio (RLV-BWR) before completion of right trisectionectomy (%) Median time to surgery 14 45 0,005 after augmentation (days) Median kinetic growth 16,3 3,4 < 0,001 ratio (ml/day) Postoperative Staging 42,9% (3/7) 57,1% (8/14) 0,659  IIIa according to UICC 8th edition R0 57,1% (4/7) 42,9% (6/14) 0,659

N1 28,6% (2/7) 35,7% (5/14) 1,0 Figure [Flow chart of patient enrollment]

FP19-12 Results: A total of 188 patients (95 patients in group A IS THERE A REAL ROLE OF and93patientsingroupB)werefinally analyzed. The POSTOPERATIVE ANTIBIOTIC incidence rate of infectious complications (7 cases, ADMINISTRATION FOR MILD- 7.4%, in group A and 8 cases, 8.6%, in group B, MODERATE ACUTE CHOLECYSTITIS? p=0.794) and overall non- infectious complications (7 cases, 7.4%, in group A and 6 cases, 6.5%, in group B, A PROSPECTIVE RANDOMIZED p=1.000) showed no significant difference between the CONTROLLED TRIAL two groups. E. Y. Kim1, Y. C. Yoon2, H. J. Choi3,K.H.Kim4, Conclusions: Absence of postoperative antibiotic admin- J. H. Park5 and T. H. Hong3 istration did not lead to an increase in postoperative in- 1Department of General Surgery, Seoul St. Mary’s Hos- fections in cases of mild to moderate AC. Avoidance of pital, 2Incheon St. Mary’s Hospital, 3Seoul St. Mary’s unnecessary antibiotic use will reduce the adverse effects of

HPB 2018, 20 (S2), S182eS294 S254 Free Orals (FP01-FP31)

FP19-12 Table [Analysis of postoperative infectious morbidities] Characteristics Total (n[188) group A (No antibiotics, n[95) group B (Antibiotics, n[93) p-value Infectious complications (%) 15 (8) 7 (7.4) 8 (8.6) 0.794 Sustained fever after POD 2 (%) 7 (3.7) 3 (3.2) 4 (4.3) 0.719 Surgical site infection (%) 8 (4.3) 4 (4.2) 4 (4.3) 1.000 Superficial incisional SSI 6 (3.2) 3 (3.2) 3 (3.2) 1.000 Deep incisional SSI 1 (0.5) 0 1 (1.1) 0.495 Organ/space SSI 1 (0.5) 1 (1.1) 0 1.000

antibiotics and also allow for a tailored treatment strategy FP19-14 according to the severity of cholecystitis. SURGICAL RISK CALCULATOR DEVELOPMENT FOR POSTOPERATIVE FP19-13 OUTCOMES AFTER LAPAROSCOPIC LOW YIELD OF PRE-OPERATIVE CHOLECYSTECTOMY: A MRCP AND ERCP IN THE MULTICENTER PROSPECTIVE MANAGEMENT OF SUSPECTED COHORT STUDY CHOLEDOCHOLITHIASIS: A H. K. Lee1, H. Lee1, I. W. Han2, J. Cho3 and J. E. Choi4 1 2 CANADIAN EXPERIENCE Ewha Womans University College of Medicine, Sung- kyunkwan University School of MedicineK, 3Seoul Na- Y. Wang, D. Mergui, S. Sanders, S. Pallotta, E. Szirt, tional University Bundang Hospital, Seoul National J. -S. Pelletier and T. Vanounou University, and 4National Evidence-based Healthcare Jewish General Hospital, McGill University, Canada Collaborating Agency, Republic of Korea Introduction: The role of pre-operative magnetic reso- Background: Laparoscopic cholecystectomy (LC) is the nance cholangiopancreatography (MRCP) and endoscopic standard surgical treatment of benign gallbladder disease. retrograde cholangiopancreatography (ERCP) in the man- However, the incidence of postoperative complication agement of patients with low-intermediate suspicion of varies, and the risk factor was not well thoroughly inves- choledocholithiasis remains controversial. tigated. The aim of this study is to develop and evaluate the Methods: Single-center retrospective review of all pa- surgical risk calculator for postoperative outcomes after tients with suspected choledocholithiasis (bilirubin be- laparoscopic cholecystectomy. tween 30-70 umol/L and/or common bile duct dilatation Methods: A total of 3,002 patients were screened, and without evidence of choledocholithiasis on ultrasound 2,514 patients who underwent laparoscopic cholecystec- or computed tomography) undergoing non-elective tomy for benign gallbladder disease from 18 academic in- laparoscopic cholecystectomy between 2013-2016. Pa- stitutes in Korea were included in this prospective, fi tients were classi ed as upfront laparoscopic cholecys- multicenter cohort study. Preoperative or intraoperative tectomy (LC), laparoscopic cholecystectomy with intra- variables were evaluated as risk factors for various post- fi operative cholangiogram (LC-IOC), MRCP- rst or operative outcomes including overall complications, and fi ERCP- rst. tendency to increase use of medical facilities including Results: 105 patients were included: LC (n = 61); LC-IOC prolonged duration of hospital stay. After risk factor anal- fi fi (n = 10); MRCP- rst (n = 18); ERCP rst (n = 16). In the ysis, risk calculator after cholecystectomy was developed fi MRCP- rst group, 5 (27.8%) patients had a positive using multiple logistic regression analysis. MRCP, of which 1 subsequently had a positive ERCP. Of Results: The overall complication rate was 4.6%, and the fi the 16 ERCP- rst patients, 1 (6.3%) had a positive ERCP. surgery-related complication rate was 2.5%. A total of 310 Three (4.9%) LC patients required a post-operative ERCP patients (12.3%) had used additional medical facilities for retained stones. One (10%) LC-IOC patient had a including prolonged hospital stay, or unexpected re- fi positive cholangiogram. The LC group had a signi cantly admission. After multivariate analysis, risk calculator for shorter length of stay until OR (LOS-OR) compared to the overall complication (AUC 0.755) using gender, old age, fi fi MRCP- rst (2.59 vs. 4.83 days, p = 0.001) and ERCP- rst DM, preoperative ERCP for CBD stone, therapeutic anti- groups (2.59 vs. 4.25 days, p = 0.042). LOS-OR for LC- biotics usage was developed. Risk calculator for tendency fi fi IOC was also signi cantly shorter compared to MRCP- rst to increase use of medical facilities (AUC 0.777) was < fi (0.90 vs. 4.83 days, p 0.001) and ERCP- rst (0.90 vs. developed using old age, smoking history, preoperative 4.25 days, p = 0.002). There were no differences in 90-day comorbidity, emergent operation, and preoperative diag- complications. nosis of acute cholecystitis. Conclusions: Pre-operative MRCP and ERCP carry a low Conclusion: Using standard preoperative variables from diagnostic yield, and are associated with longer hospitali- this multi-institutional prospective database, we construc- zation, in patients with low-intermediate suspicion of ted a risk calculator for predicting adverse perioperative choledocholithiasis. Their judicious use is warranted to outcomes after laparoscopic cholecystectomy. Such infor- fi ensure delivery of resource-ef cient care. mation may be useful for risk stratification before laparo- scopic cholecystectomy.

HPB 2018, 20 (S2), S182eS294 Free Orals (FP01-FP31) S255

FP19-15 CENTRAL and Google scholar databases from inception to 29 march 2017. EFFICACY OF MELATONIN ON SLEEP Results: 89 studies were included in this systematic review. QUALITY AFTER LAPAROSCOPIC Overall, BDI preventive techniques can be categorized as CHOLECYSTECTOMY dedicated surgical approaches (Critical View of Safety fi D. Dahiya, V. Vij, L. Kaman and A. Behera (CVS), fundus rst, partial laparoscopic cholecystectomy), Surgery, PGIMER Chandigarh, India supporting imaging techniques (intraoperative cholangi- ography, intraoperative ultrasonography, fluorescence im- Introduction: Post-operative sleep and circadian rhythm aging) and others. Dedicated surgical approaches disturbances were associated with prolonged post-opera- demonstrate promising results, yet limited research is pro- tive convalescence, respiratory and cardiovascular vided. Intraoperative cholangiography and ultrasonography morbidity. Sleep disturbances have been shown to be due demonstrate beneficial effects in BDI prevention, however to decreased levels of circulating melatonin after surgery. the available evidence is low. Fluorescence imaging is in its If this sleep pattern and circadian rhythm is recycled, infancy, yet this technique is demonstrated to be feasible outcome after surgery could be improved. Therefore, aim and larger trials are in preparation. of this study was to observe the effect of melatonin on Conclusion: Given the low sample sizes and suboptimal quality of sleep in patients undergoing laparoscopic cho- study designs of the studies available, it is not possible to lecystectomy (LC). recommend a preferred method to prevent BDI. Surgeons Methods: Hundred patients of LC participated in this should primarily focus on proper dissection techniques, of randomized, placebo-controlled, double-blind, clinical which CVS is most suitable. Additionally, recognition of trial. Patients were randomized equally into Group A who hazardous circumstances and knowledge of bail-out tech- received 6 mg melatonin tablets 45 minutes before sleep niques is critical to complete surgery with minimal risk of for three days after surgery and Group B who received injury to the patient. placebo. Results: Melatonin usage results in decrease in sleep la- tency (SL) as compared to placebo (13.6 Æ 14.95 vs 20.10 Æ 16.18 minutes, p 0.04). There was also increased total FP19-18 sleep duration (TSD) on postoperative day (POD) 1 (p CONSERVATIVE MANAGEMENT OF 0.004) and POD 2 (p 0.001) in group A. There was decrease LIVER LESIONS IN SERIOUS HEPATIC in day time naps and night awakenings after surgery in TRAUMA: BILIARY COMPLICATIONS group A though it was not significant statistically. Sub- jective assessment of sleep on visual analogue scale (VAS) FOLLOWING HEPATIC ARTERIAL showed reduced sleep score (p 0.001) and decreased pain (p EMBOLIZATION 0.02) in group A on POD 1. Statistically significant dif- M. Massani, C. Nistri, L. Bonariol, E. Caratozzolo and ference was not observed in fatigue or general well being N. Bassi among groups. IV Dpt of Surgery, Regional Hospital, Italy Conclusion: Results of this study could demonstrate that Background: The aims of this study is to analyse the melatonin as a single agent could improve quality of sleep conservative treatment of liver trauma, focusing on biliary by decreasing sleep latency and increasing total sleep complications, and to compare the results with those in the duration. literature. Methods: This study analyses patients with serious hepatic trauma treated from January 2000 to September 2017 at the FP19-17 IV Department of Surgery, Regional Centre for HPB Sur- SAFE LAPAROSCOPIC gery. Liver injuries were evaluated according to the CHOLECYSTECTOMY: A SYSTEMATIC American Association for the Surgery of Trauma. Results: 104 patients with level III-IV-V liver injuries were REVIEW OF BILE DUCT INJURY included in the study (81 males and 23 females). Median PREVENTION age was 37 years (range 15-21), 76 (73,08%) patients had F. van de Graaf1, I. Zaïmi1, L. P. S. Stassen2 and haemodynamical stability, and overall mortality was J. F. Lange1 24,04%. Patients that underwent NOM were 68 (65,38%), 1Department of Surgery, Erasmus MC - University Medi- 64 of them (94,12%) were haemodynamically stable at cal Center Rotterdam, and 2Department of Surgery, admissions and 23 (33,82%) underwent hepatic arterial Maastricht University Medical Center (MUMC), The embolization. NOM success rate was of 77,94%, and only 4 Netherlands patients were treated with liver packing following NOM Background: Since the introduction of laparoscopic cho- failure. Mortality rate among conservative-treated patients lecystectomy (LC), a substantial increase in bile duct injury was 8,82%, with no liver-related mortality. Among patients (BDI) incidence was noted. Multiple methods to prevent with failure of NOM, there were 7 biliary complications this complication have been developed and investigated. (46,67% of failed NOM) and 57,14% of them occurred The most suitable method however is subject to debate. In after hepatic arterial embolization. Biliary complications fi this systematic review, the different modalities to aid in the after arterial embolization included 3 biliary stulas and 1 fi safe performance of LC and prevent BDI are delineated. acute cholecystitis; all biliary stulas needed a combination Methods: A systematic search for articles describing of endoscopic (nasobiliary drain) and surgical treatment methods for the prevention of BDI in LC was conducted (percutaneous drainage, biliostasis or cholecystectomy). using EMBASE, Medline, Web of science, Cochrane Discussion: NOM is an effective treatment modality in most cases of serious hepatic trauma. We must be aware of

HPB 2018, 20 (S2), S182eS294 S256 Free Orals (FP01-FP31) biliary complications occurring after hepatic angioemboli- unclear. We studied practice patterns in LN sampling rates zation and try to detect them in early stage to treat them during HCC resection and examined the appropriateness of correctly. LN sampling rates by analyzing LN positivity rates. Methods: Utilizing the National Cancer Data Base (NCDB), we identified HCC patients treated with surgical FP20 - Free Papers 20 (mini oral) - Liver: resection with complete tumor staging and survival infor- Primary Tumours mation between 2004-2013. Rates of LN sampling and FP20-01 node positivity rates were evaluated. Multivariate logistic regression was performed to find preoperative factors CLINICAL OUTCOMES OF associated with positive pathologic LN disease. LAPAROSCOPIC RADIO-FREQUENCY Results: Of the 8,598 HCC patients who met our inclusion OR MICROWAVE ABLATION FOR criteria, only 18.1% (n=1,559) had any LN sampling. LN HEPATOCELLULAR CARCINOMA sampling rates varied by clinical T-stage (T1:14.1%; T2:17.0%; T3:25.0%; T4:26.2%; p< 0.001). Among pa- M. Barabino, R. Santambrogio, M. Giovenzana, tients who had LN sampling, only 6.2% (n=97) had posi- M. Polizzi, F. Torri and E. Opocher tive LN metastasis, representing just 1.1% of the entire Chirurgia Epatobiliopancreatica e Digestiva, San Paolo national cohort. LN positivity rates were higher in Hospital, Italy advanced T-stages [T1:4.1%, T2:4.4%, T3:8.2%, T4- Introduction: Thermal ablation (TA) is widely used as a 18.8%; p=0.001]. In each clinical T-stage, the median fi rst-line treatment in patients (pts) with hepatocellular overall survival with LN metastasis was significantly worse carcinoma (HCC). If percutaneous approach is unfeasible, [T1:18.4mo vs. 80.2mo; T2:13.9mo vs. 37.0mo; laparoscopic thermal ablation (LTA) is a valid alternative. T3:13.9mo vs. 23.2mo; T4:3.7mo vs. 17.8mo]. On multi- fi Aim of the study is to con rm the effectiveness of the LTA variate analysis, preoperative clinical T3-4 stages were in challenging HCC nodules. associated with positive LN metastasis (OR:2.82; p< Methods: 503 LTA were offered to patients with single 0.001). nodule or up to three nodules smaller than 3 cm not suitable Conclusions: Based on national practice patterns, it ap- for surgical treatments or not suitable for percutaneous TA. pears that surgeons are appropriately limiting LN sampling fi Technical success was de ned as total necrosis in the TA to patients with high-risk HCC. Because overall LN posi- area at 1 month, technique effectiveness if this area was tivity is low, routine lymphadenectomy is not required for persistent at the follow-up. Local tumour progression (LTP) most HCC, but should be strongly considered in clinical if reappearance of enhancing tissue occurred around TA T3-4 tumors. area. We used: radiofrequency ablation (RFA) in 300 pts, microwave ablation (MWA) in 122, and RFA or MWS associated with intrahepatic vascular occlusion (IHVO) in 81. FP20-03 Results: Postoperative mortality was 0.4%, while major LAPAROSCOPIC RESECTION VS OPEN complications occurred in 2% of patients. Technical suc- RESECTION VS RADIOFREQUENCY cess was achieved in 467 patients (93%), with 100% after ABLATION FOR SINGLE IHVO. Median follow-up was 30.4 months and 67% of HEPATOCELLULAR CARCINOMA (< patients developed intrahepatic recurrence. LTP occurred in 5CM) IN ELDERLY CIRRHOTIC 15% of patients and this rate was not influenced by prob- lematic location of HCC. Technique effectiveness was PATIENT: A MULTICENTRIC STUDY primary influenced by HCC adjacent to vessels (67% vs R. Memeo1, F. Ratti2, M. Gelli3, Z. Cherkaoui4, T. Piardi5, 82%; p=0.01) and diameter >3 cm (69% vs 85%; p=0.001). F. di Benedetto6, P. Pessaux4, D. Cherqui7, L. Aldrighetti8, Conclusions: This study represents the largest single- L. Lupo1 and ELDHCC centre series on LTA for HCC on cirrhosis and shows that 1Policlinico di Bari, 2San Raffaele Hospital, Italy, 3Institut LTA is safe and achieves good results in patients with Gustave Roussy, 4Nouvel Hôpital Civil, 5Reims, France, challenging HCC nodules not suitable for percutaneous 6University of Modena and Reggio Emilia, Italy, 7Paul approach. Brousse Hospital, France, and 8San Raffaele Hospital, Italy Introduction: Hepatocellular carcinoma management is FP20-02 still a matter of debate in elderly cirrhotic patients (>70 SELECTIVE LYMPH NODE SAMPLING years old). IN RESECTABLE HEPATOCELLULAR Method: In order to evaluate the best therapeutic option for elderly patients, radiofrequency ablation (RFA), open liver CARCINOMA: NATIONAL PRACTICE resection (OLR) and laparoscopic liver resection (LLR) PATTERNS AND NODE POSITIVITY were compared for patient with a single nodule of RATES maximum 50 mm of diameter. A total of 601 patients were A. Lee, Y. -J. Chiang, Y. -S. Chun, C. Conrad, A. Kaseb, analysed from a multicentric retrospective database. K. Raghav, T. Aloia, J. -N. Vauthey and C. -W. Tzeng Globally, 202 patients underwent OLR, 169 LLR and 230 MD Anderson Cancer Center, United States RFA. Results: Postoperative complication were more frequent in Introduction: Although lymph node (LN) metastasis rep- OLR group compared to LLR and RFA (48% vs 31% vs resents Stage IVA disease in hepatocellular carcinoma 31%, p=0.0002) with more Dindo Clavien grade III-IV (HCC), the role of LN sampling during HCC resection is complication in OLR group, compared to LLR and RFA

HPB 2018, 20 (S2), S182eS294 Free Orals (FP01-FP31) S257

(11% vs 7% vs 8%, p=0.001. The postoperative course was FP20-05 longer in OLR compared to LLR and RFA (10.6 days vs 6.9 vs 3.3, p=0.0001). Five postoperative death occurred in DEVELOPMENT AND VALIDATION OF OLR group (2.5%, p=0.001). Treatment related blood NOMOGRAMS TO PROVIDE transfusion were more frequent in OLR group, compared INDIVIDUALIZED PREDICTIONS OF with LLR group and RFA group (16% vs 11% vs 0%, SURVIVAL BENEFITS FROM SURGERY p=0.0003). LLR group had a better 5-year overall survival, IN PATIENTS WITH INTERMEDIATE/ compared to OLR group and RFA group (78% vs 71% vs 47%, p=0.0001) and 5- year disease free survival (59% vs ADVANCED HEPATOCELLULAR 54% vs 45%, p=0.03) CARCINOMA Conclusion: LLR is feasible and safe in elderly cirrhotic W. T. Yan1,2, J. H. Wang1,2, H. Zhang1, J. H. Zhong3, patients with single (< 5cm) hepatocellular carcinoma, Y. H. Zhou4,W.M.Gu5, H. Wang6, T. H. Chen7, with better postoperative course, overall and disease free F. Shen1 and T. Yang1 survival. 1Department of Hepatobiliary Surgery, Eastern Hepato- biliary Surgery Hospital, 2Department of Clinical Medi- cine, Second Military Medical University, 3Department of Hepatobiliary Surgery, Affiliated Tumor Hospital of FP20-04 4 RADIOEMBOLIZATION WITH Y90 AS Guangxi Medical University, China, Department of Hepatobiliary Surgery, Pu’er People’s Hospital, 5First BRIDGE TO LIVER Department of General Surgery, Fourth Hospital of TRANSPLANTATION OR RESECTION Harbin, 6Department of General Surgery, Liuyang Peo- FOR INTERMEDIATE STAGE ple’s Hospital, and 7Department of General Surgery, HEPATOCELLULAR CARCINOMA Ziyang First People’s Hospital, China I. Labgaa1, P. Tabrizian2, J. Titano3, E. Kim3, S. Florman2, Introduction: According to the BCLC treatment guide- M. Schwartz2 and E. Melloul1 lines, surgery does not be recommended for intermediate/ 1Visceral Surgery, Lausanne University Hospital CHUV, advanced hepatocellular carcinoma (HCC). In real world, Switzerland, 2Liver Surgery, Recanati/Miller Trans- however, liver resections are often performed in patients plantation Institute, and 3Division of Interventional Radi- with intermediate/advanced but resectable HCC, especially ology, Department of Radiology, Icahn School of Medicine in the East. at Mount Sinai, United States Methods: We retrospectively evaluated multicentric data of 1,325 patients newly diagnosed with intermediate/ Introduction: To assess the safety and feasibility of radi- advanced HCC who underwent curative resection. We oembolization with Yttrium-90 (Y90) for the downstaging/ randomly divided the subjects into development (n = 875) downsizing of hepatocellular carcinoma (HCC), initially and validation (n = 450) samples. Multivariate Cox pro- unresectable or beyond Milan criteria. portional hazards models were developed and separately Method: Patients undergoing exclusively Y90 followed by validated on the basis of patients’ clinicopathological var- either orthotopic liver transplantation (OLT) or liver iables assessed for associations with 1-year recurrence and resection (LR) for HCC between 2012 and 2016 were 3-year mortality. The discriminatory accuracy of these included. Primary outcomes were postoperative morbidity models was compared with conventional tools by analyzing and mortality. Secondary outcomes were overall survival receiver operating characteristic (ROC) curves. (OS) and response to Y90. Response to Y90 was evaluated Results: The statistical nomograms built based on perfor- by radiology (mRECIST) and histology. mance status, Child-Pugh grade, portal hypertension, pre- Results: A total of 349 HCC patients were treated with operative alpha-fetoprotein level, tumor rupture, largest Y90 during the study period. Nine percent (n=32) un- tumor diameter, tumor number, macrovascular and micro- derwent either OLT (n=22) or LR (n=10). Major com- vascular invasion, and satellites had good calibration and plications and mortality were reported respectively in 5 discriminatory abilities, with c-indices of 0.70 (1-year (16%) and 1 (3%) patients. Median OS was 28 months recurrence) and 0.68 (3-year survival), respectively. These while survival rates at 1-, 3- and 5-years reached 97%, models showed satisfactory goodness-of-fit and discrimi- 86% and 86%, respectively. Median absolute increase of nation abilities in the validation cohort (c-index, 0.68 for 1- FLR/TELV was 8.9 (-2.5-13.6) at 3 months. Median size year recurrence and 0.69 for 3-year survival). The areas of nodules decreased from 4.2cm to 1.3cm (p< 0.001) under the ROC curve using these nomograms exceeded after Y90, while 50% of patients had no more viable those of traditional staging systems, indicating superior tumor. Based on mRECIST, 18 (56%) patients showed discriminatory capability (c-indices, 0.60-0.63 and 0.56- complete response. Histologically, complete necrosis was 0.62, respectively). observed on 11 (34%) specimen. Conclusions: Our proposed online nomograms, which Conclusions: Radioembolization with Y90 allows con- present graphically postoperative prognostic models for trolling the tumor in the treated and non-treated (naive) recurrence and survival in patients with intermediate/ liver lobes with sufficient downstaging/downsizing in 9% advanced but resectable HCC, offer valuable guidance to of patients initially not amenable to curative treatment. surgeons and hepatologists for individually predicting Y90 radioembolization allows appropriate liver hyper- survival benefits from surgery and planning recurrence trophy to safely perform LR with acceptable long-term surveillance and adjuvant therapy. survival.

HPB 2018, 20 (S2), S182eS294 S258 Free Orals (FP01-FP31)

FP20-06 However, the role of MIH for patients with hepatocellular carcinoma (HCC) needs further investigation. SURGICAL RESECTION FOR BCLC Methods: Clinicopathological data of patients who un- INTERMEDIATE STAGE derwent liver resection for HCC between 2005 and 2016 HEPATOCELLULAR CARCINOMA: A were assessed. Postoperative outcomes und long-term COMPARISON WITH EARLY OR VERY survivals of patients following MIH were compared with EARLY STAGE HEPATOCELLULAR those of patients undergoing open hepatectomy (OH) after 1:1 propensity score matching. CARCINOMA Results: During the study period, 407 patients underwent J. Arita, N. Akamatsu, J. Kaneko, Y. Sakamoto and liver resection for HCC with curative intent. Fifty-six K. Hasegawa patients underwent MIH and were compared with a Hepato-Biliary and Pancreatic Surgery Division, Univer- matched cohort of 56 patients who underwent OH. The sity of Tokyo, Japan rate of patients with fibrosis/cirrhosis(82% vs. 86%, Introduction: Despite BCLC guideline, many patients p=0.959), multiple lesions(32% vs. 32%, p=1.00), tumor with intermediate stage HCC undergo surgical resection in size >30mm(61% vs. 55%, p=0.566), and major resec- Japan. tion(16% vs. 16%, p=1.00) was comparable between the Methods: Retrospective single-institutional analysis of was two groups(MIH vs. OH). MIH was associated with lower performed in patients who underwent initial radical surgery 90-day complication rate(32 % vs. 54%, p=0.022), lower for HCC between October 1994 and December 2015. They major complication rate(14% vs. 30%, p=0.041), lower were classified into 4 groups. A: Early or Very early stage liver failure rate(0% vs. 7%, p=0.042), lower 90-day (n = 606). B1: Solitary and > 5 cm (n = 198). B2: 2-3 mortality rate(0 vs. 7%, p=0.042), and shorter length of tumors and < 3 cm (n = 158). B3: 4 or more tumors (n = hospital stay(9 vs.12 days, p=0.009) compared to OH. 68). Long term results were analyzed. After a median follow-up time of 51 months, MIH and OH Results: In group A, 296 Pts (49%) had solitary tumor. In showed comparable 5-year overall(54 % vs. 41%, group B1, the median tumor size was 7.2 cm. In group B2, p=0.151), and 5-year disease-free survival rates(50% vs. 119/39 Pts had 2/3 tumors. In group B3, 31/11/7/19 Pts had 38%, p=0.956). 4/5/6/7- tumors, respectively. 5-year overall survival was Conclusion: MIH for HCC is associated with lower 71.7％ in group A, 63.7％ in group B1, 52.4％ in group postoperative morbidity and mortality and shorter B2, and 36.8％ in group B3, where statistically significant length of hospital stay, resulting in oncologic outcomes differences were seen between B1 and B2, B2 and B3, similar to those achieved with the established OH. Our although no significant difference was seen between A and findings suggest that MIH should be considered as the B1 (HR=1.10, 95%CI 0.86-1.42). Multivariate analysis preferred method for the treatment of curatively among group B3 patients using Cox proportional hazard resectable HCC. model revealed serum AFP value > 50 ng/mL was the only independent prognostic factor (HR = 1.93, 95%CI 1.04- 3.56). The 5-year overall survival of group B3 was 47.3% FP20-08 < > when serum AFP 50, and 27.6% when serum AFP 50 STEREOTACTIC BODY RADIATION (P = 0.033). Conclusion: Among intermediate stage HCC, Group B1 THERAPY FOR SMALL RECURRENT and B2 showed fair prognosis. However, group B3 showed HEPATOCELLULAR CARCINOMA rather poor prognosis unless serum AFP was < 50 ng/mL. Y. Shui1,W.Yu2, X. Bai2, L. Shen2,T.Ma2, T. Liang2 and Q. Wei3 1radiothrapy department, The Second Affiliated Hospital, FP20-07 2The Second Affiliated Hospital, and 3Department of Ra- fi MINIMAL-INVASIVE VERSUS OPEN diation Oncology, The Second Af liated Hospital, Zhejiang University School of Medicine, China HEPATECTOMY FOR Aim: To evaluate the efficacy of stereotactic body radiation HEPATOCELLULAR CARCINOMA: therapy (SBRT) in small hepatocellular carcinoma (HCC) COMPARISON OF POSTOPERATIVE patients. OUTCOMES AND LONG-TERM Methods: From March 2016 to September 2017, we SURVIVALS USING PROPENSITY treated 29 small HCC patients with SBRT. Eligibility SCORE MATCHING ANALYSIS criteria included longest tumor diameter5.0 cm; Child- Turcotte-Pugh (CTP) Class A or B; tumor recurrence after A. Andreou, B. Struecker, N. Raschzok, F. Krenzien, other local treatment. The dose of 32.5Gy-50Gy in 5 P. Haber, S. Wabitsch, G. Atanasov, R. Öllinger, fractions and 48 Gy in 4 fractions were prescribed. M. Schmelzle and J. Pratschke Results: Of the treated patients, 29 patients with 31 lesions, Department of Surgery, Campus Charité Mitte and 26 were classified as CTP A and 3 as CTP B. Median tumor Campus Virchow-Klinikum, Charité Universitätsmedizin size was 3.5cm(1.0-5.0cm). Median PTV size was Berlin, Germany 27.5cc(7.0cc-116.8cc). The local control rate at 1 years was Background: Minimal-invasive hepatectomy (MIH) has 96.5%. OS at 1 years was 100%. PFS at 1years was 89.6%. been increasingly performed with most promising results. Hepatic toxicity grade 3 was not observed. After three

HPB 2018, 20 (S2), S182eS294 Free Orals (FP01-FP31) S259 months of SBRT, assessments of HCC response showed complete response(CR) in 9 of 24 (37.5%) patients, partial response(PR) in 13 (54.1%), stable disease（SD）in 2 (8.3%), with an objective response rate of 91.6 %. Conclusions: SBRT is a promising alternative treatment for patients with small HCC who are unsuitable for surgical resection or local ablative therapy.

FP20-09 IMPACT OF PREOPERATIVE WEIGHT STATUS ON LONG-TERM PROGNOSIS IN PATIENTS UNDERGOING LIVER RESECTION OF HEPATOCELLULAR CARCINOMA: A MULTICENTER LARGE COHORT STUDY J. J. Yu1,21, T. H. Chen31, L. Liang11, H. Zhang11, Y. H. Zhou41, H. Wang51,W.M.Gu61, W. Y. Lau1,71, F. Shen11 and T. Yang11 1Department of Hepatobiliary Surgery, Eastern Hepato- biliary Surgery Hospital, 2Department of Clinical Medi- cine, Second Military Medical University, 3Department of General Surgery, Ziyang First People’s Hospital, Figure 4Department of Hepatobiliary Surgery, Pu’er People’s Hospital, 5Department of General Surgery, Liuyang Peo- ple’s Hospital, 6First Department of General Surgery, 7 Fourth Hospital of Harbin, and Faculty of Medicine, FP20-10 Chinese University of Hong Kong, China LIVER SURFACE NODULARITY Introduction: Overweight has been identified as a signif- icant risk associated with long-term prognosis after liver QUANTIFICATION FROM resection of hepatocellular carcinoma (HCC). PREOPERATIVE CT IMAGES FOR THE Although underweight is not unusual in HCC patients, its ASSESSMENT OF THE PERIOPERATIVE association with long-term outcomes after resection of RISK AFTER LIVER RESECTION FOR HCC remains unclear. Therefore, we investigated the HCC relationship between different preoperative weight status 1 2 2 3 and long-term outcomes in patients undergoing HCC C. Hobeika , R. Sartoris , M. Ronot , V. Paradis , N. Poté3, S. Dokmak1, O. Farges1, P. -E. Rautou4, resection. F. Cauchy1 and O. Soubrane1 Methods: Patients treated with curative liver resection of 1HPB Surgery and Liver Transplantation Unit, 2Depart- HCC at six Chinese institutions between 2001 and 2013 ment of Radiology, 3Department of Pathology, and were divided into 3 groups: underweight group (BMI < 4Department of Hepatology, Hôpital Beaujon, France 18.5), normal-weight group (BMI 18.5 w 25.0) and over- weight group (BMI > 25.0). Baseline characteristics and Introduction: While Liver surface nodularity (LSN) clinical variables were collected and compared. Univariate quantification on routine CT images allows to accu- fi fi and multivariate Cox-regression analyses were performed rately detect severe brosis and clinically signi cant to identify risk factors of overall survival (OS) and recur- portal hypertension (CSPH), its value in the anticipa- rence-free survival (RFS). tion of post-operative morbidity after liver resection Results: Of all 1,524 patients, there were 891 normal- for hepatocellular carcinoma (HCC) remains to be weight, 107 underweight, and 526 overweight. Under- assessed. weight and overweight patients had significantly poorer OS Methods: All patients undergoing hepatectomy for and RFS than normal-weight patients (all P < 0.01). HCC between 2012-2017 at a single center were Multivariate Cox-regression analysis identified that un- retrospectively analyzed. LSN was measured on the left derweight and overweight were significantly associated liver lobe on preoperative CT scans. The feasibility of with decreased OS (HR 1.217, 95% CI 1.19-1.56, P = 0.02, LSN measurement was assessed and the association and HR 1.59, 95% CI 1.36-1.81, P < 0.01), and RFS (HR between LSN and major postoperative morbidity was 1.30, 95% CI: 1.19-1.53, P < 0.01, and HR 1.34, 95% CI analyzed. 1.17-1.54, P < 0.01). Results: Among 199 resected HCC patients, 173 (86.9%) Conclusion: Apart from overweight, underweight was also had an underlying liver disease, including cirrhosis in 68 significantly associated with decreased OS and RFS in (35.1%) patients. CSPH was observed in 8 (13.1%) of the patients undergoing liver resection for HCC. 61 patients undergoing preoperative hepatic vein pressure

All authors contributed equally to this work.

HPB 2018, 20 (S2), S182eS294 S260 Free Orals (FP01-FP31) gradient. LSN measurement was feasible in 144 patients FP20-12 (72.4%) with a mean time of 88 (Æ29) seconds. Mean LSN was 2.46 (Æ0.37). LSN was significantly increased RECOMMENDATIONS FOR THE in patient with cirrhosis (2.68Æ0.40 vs. 2.33Æ0.29, MANAGEMENT OF WOMEN WITH p=0.001) and CSPH (2.82Æ0.29 vs. 2.40Æ0.29, p=0.001). SUSPECTED HEPATOCELLULAR Overall and major postoperative complications occurred ADENOMA AND CHILDBEARING in 118 (59.3%) and 40 (20.1%) patients, respectively. POTENTIAL Mean CCI was 17.18 (Æ21.04). The AUROC of LSN for M. Gaspersz1, A. J. Klompenhouwer1, M. Broker1, predicting major complications was 0.614 (p=0.044) with 2 2 1 3 a cut-off value of 2.62. On multivariate analysis, LSN  F. Willemssen , M. Tomeer , T. Terkivatan ,R.deMan, J. Ijzermans1 and J. van Vugt1 2.62 (HR=3.08, 95% CI: 1.21-7.81, p=0.018) and per- 1 2 3 operative transfusion (HR=5.89, 95%CI: 2.17-15.99, Surgery, Radiology, and Gastroenterology and Hepa- p=0.001) were independent risk factors for major tology, Erasmus MC, The Netherlands complications. Introduction: Hepatocellular adenoma(HCA) is a benign Conclusion: LSN measurement is feasible in the vast liver tumor occurring mostly in females. Pregnancy re- majority of patients and represents a valuable tool in the quires special consideration, as hormone induced growth anticipation of major postoperative complications and bleeding can occur, and is often discouraged or surgical following hepatectomy for HCC. resection is performed beforehand. Our aim was to assess the accuracy of diagnostics in these women to monitor HCA change during pregnancy. FP20-11 Method: Patients included were enrolled in an ongoing multicenter prospective cohort study investigating the COMPARISON OF LIVER RESECTION incidence of growth during pregnancy in HCA< VERSUS TACE FOR SHORT- AND LONG- 5cm(PALM study). Pregnant patients with suspected HCA TERM OUTCOMES IN PATIENTS WITH on imaging underwent extensive follow-up to assess INTERMEDIATE STAGE growth. Definitive diagnosis was established with state-of- HEPATOCELLULAR CARCINOMA: A the-art contrast-enhanced MRI(CE-MRI), preferably before fi SYSTEMATIC REVIEW AND META- pregnancy. Patients who did not have a de nitive diagnosis underwent CE-MRI after giving birth. ANALYSIS Results: Between 10-2011 and 10-2017, 57 patients were I. Labgaa, D. Clerc, D. Martin, N. Halkic, included. Median age was 34(IQR 32-37), median HCA- N. Demartines and E. Melloul size 25mm(IQR 20.8-41.5). Twenty-nine patients did not Visceral Surgery, Lausanne University Hospital CHUV, undergo CE-MRI before pregnancy. Of these patients, Switzerland HCA was suspected based on conventional MRI in 66%, Introduction: According to Barcelona Clinic Liver Cancer ultrasound in 24%, CT in 7% and contrast-enhanced ul- (BCLC) staging system, transarterial chemoembolization trasound in 4%. CE-MRI confirmed HCA diagnosis in 48 (TACE) is recommended in patients with intermediate patients(84%), whereas 9 patients(16%) had FNH. Of these stage B, while liver resection (LR) is restricted to patients patients, 8 were treated in non-referral hospitals and 1 in a with early stage A. This meta-analysis aimed to analyze tertiary referral center(p=0.018). None of the patients who long-term outcomes of LR compared to TACE in patients eventually were diagnosed with FNH underwent CE-MRI with intermediate stage HCC. before pregnancy. Method: A systematic review was conducted according to Conclusions: This study indicates that a large proportion of PRISMA guidelines. Only trials comparing LR with TACE childbearing women with suspicious HCA are misdiag- in intermediate stage patients were selected. Primary nosed and have FNH. Misdiagnosis may have major impact outcome was overall survival (OS) and secondary outcome on patients. Therefore, state-of-the-art imaging of a benign was treatment-related mortality. Heterogeneity of the liver lesion in younger fertile women should be performed included studies was assessed and sensitivity analyses were at the moment of first presentation. performed. Depending on heterogeneity, fixed-effects or random-effects models were used to analyze hazard ratios (HR). FP20-13 Results: A total of 8 eligible trials were analyzed, including IS R1 VASCULAR HEPATECTOMY FOR 2723 intermediate stage HCC patients who underwent HEPATOCELLULAR CARCINOMA either LR (n=1365) or TACE (n=1358). Comparison be- tween LR and TACE determined a pooled HR for 3-years ONCOLOGICALLY ADEQUATE? survival of 1.87 (95% CI 1.35-2.61, p< 0.001) and a pooled ANALYSIS OF 327 CONSECUTIVE HR for 5-years OS of 1.76 (95% CI 1.34-2.33, p< 0.001). PATIENTS Overall HR for treatment-related mortality was 1.26 (95% M. Donadon, A. Terrone, S. Darwish, F. Procopio, CI 0.66-2.40, p=0.48). M. Cimino, A. Palmisano, L. Viganò, D. Del Fabbro and Conclusions: In patients with intermediate BCLC-B stage G. Torzilli HCC, LR offers increased long-term survival compared to Department of Hepatobiliary and General Surgery, TACE, with comparable treatment-related mortality. Humanitas University, Humanitas Clinical and Research fi fi These results highlight the need to re ne the identi cation Center, Italy of patients with BCLC-B HCC who would benefit from Introduction: R0-hepatectomy is the standard-of-care in LR. patients with hepatocellular carcinoma (HCC). However,

HPB 2018, 20 (S2), S182eS294 Free Orals (FP01-FP31) S261 the performance of parenchymal-sparing-hepatectomy re- on the tumor microinvironment, a carcinogen induced quires tumor detachment from intrahepatic vessels, the so- tumor model with DEN in ST2 -/- mice was established. called R1-vascular (R1vasc) hepatectomy. The aim of the Results: Hepatocyte proliferation of non neoplastic liver study was to determine the oncological adequacy of R1- tissue in IL-33-/- and ST2-/- animals is significantly vasc in HCC patients. decreased compared to wildtype animals. HCC cells after Methods: A prospective cohort of patients who underwent knockout of IL-33 show a decreased proliferation, hepatectomy for HCC between 2005 and 2015 was disturbed colony fomation and migration capacity reviewed. R0 was resection with at least 1-mm of negative compared to the parental cells. ST2-/- mice develop less margin; R1vasc was resection with tumor detachment from hepatocellular carcinoma in a carcinogenic model with intrahepatic vessel; R1-parenchymal (R1par) was resection DEN. FACS analysis revealed a suppressive effect of IL33 with tumor exposure at parenchymal margin. The endpoints on the tumor microenvironment. were local recurrence, and the survival rates of R0, R1vasc Conclusion: Tumor secreted IL-33 has a dual pro tumori- and R1par. genic effect on HCC growth, a) by a direct effect on the Results: 327 patients with 532 HCC and 448 resection hepatocytes and b) by an ST2 dependent immunosuppres- areas were analyzed. 205 (63%) resulted R0, 56 (17%) sive effect on the tumor microenvironment. R1par, 50 (15%) R1vasc, and 16 (5%) both R1par and R1vasc. After a median follow-up of 33.5 months (range 6.1-107.6) the 5-year overall survival rates were 54%, 30%, FP20-15 65% and 36% respectively for R0, R1par, R1vasc and PREOPERATIVE STRATEGIES TO R1par+R1vasc (p=0.031-Fig.1). The 5-year disease-free survival rates were 27%, 15%, 28% and 0.9% respectively IMPROVE RESECTABILITY FOR for R0, R1par, R1vasc and R1par+R1vasc (p=0.003). Local HEPATOCELLULAR CARCINOMA: A recurrence rates were 3%, 14%, 4%, and 19% respectively SYSTEMATIC REVIEW AND META- for R0, R1par, R1vasc and R1par+R1vasc (p=0.001) per ANALYSIS patient, and 4%, 5.4%, 12.3%, and 17.4% respectively for F. Tustumi1, L. Ernani2, F. F. Coelho3, W. M. Bernardo4, R0, R1vasc, R1par and R1par+R1vasc (p=0.001) per S. Silveira Junior2, J. A. P. Kruger2, G. M. Fonseca2, resection-area. R1par and R1vasc+R1par were independent V. B. Jeismann2, I. Cecconello5 and P. Herman5 detrimental factors. 1University of Sao Paulo Medical School, 2Hospita das Conclusions: This study shows how R1vasc hepatectomy Clinicas of University of Sao Paulo School of Medicine, for HCC was not associated with increased local recur- 3Gastroenterology, Hospital das Clinicas of University of rence or decreased survival. HCC detachment from intra- Sao Paulo School of Medicine, 4University of Sao Paulo, hepatic vessels should be considered oncologically and 5Gastroenterology, University of Sao Paulo, Brazil adequate. Background: Preoperative strategies to enlarge future liver remnant are useful methods to improve resectability of patients with hepatocellular carcinoma (HCC) eligible for FP20-14 major liver resection. The aim of this study was to perform THE IL-33/ST2 AXES CONTRIBUTES TO a systematic review and meta-analysis of the main strate- HEPATOCYTE PROLIFERATION AND gies used for this purpose such as portal vein embolization PROMOTES DEVELOPMENT OF (PVE) or ligation (PVL), sequential transarterial emboli- HEPATOCELLULAR CARCINOMA zation and PVE (TACE+PVE), and associated liver parti- 1 2 2 2 3 tion and portal vein ligation for staged hepatectomy A. Kohler , I. Büchi , N. Melin , F. Bayer , K. McCoy , (ALPPS). D. Candinas1 and D. Stroka2 1 Methods: A systematic review was performed in Medline, Department of Visceral Surgery and Medicine, Inselspi- EMBASE, Cochrane, Scielo/LILACS and grey literature. tal, Bern University Hospital, 2Department for BioMedical 3 Perioperative morbidity and mortality, post-hepatectomy Research, Visceral Surgery, and Department for liver failure (PHLF), overall survival (OS) and disease-free BioMedical Research, Gastroenterology, University of survival (DFS) were evaluated. Bern, Switzerland Results: 40 studies were included in meta-analysis (1,250 Introduction: IL-33 is an IL-1 family cytokine that is patients). Hepatectomy was performed in 75.2% of patients typically secreted upon cellular stress or damage and has after PVE/PVL, 85.3% after ALPPS and 89.5% after been shown to have a tumorigenic effect outside the liver. TACE+PVE. Encompassing all strategies, PHLF and It is known that hepatocellular carcinoma (HCC) cells perioperative mortality were 7.3% and 6.0%, respectively. produce more IL-33 than healthy liver tissue. We therefore ALPSS was associated with higher PHLF rate (0.25; 95% hypothesized that activation of the IL-33/ST2 axis by CI 0.131,0.425) when compared with PVE/PVL (0.062; tumor induced IL-33 secretion facilitates development of 95%CI 0.029,0.130) and TACE+PVE (0.075; 95%CI HCC. 0.045,0.123). Similarly, ALPPS showed higher periopera- Method: In order to examine the effect of the IL-33/ST2 tive mortality (0.186; 95%CI 0.11, 0.297) vs. PVE/PVL axis on hepatocyte proliferation in healthy liver tissue, a (0.036; 95%CI 0.021,0.061) and TACE+PVE (0.05; 95% murine partial hepatectomy model was used to compare CI 0.029,0.086). Pooled OS and DFS at mean follow-up proliferation in IL-33-/- and ST2-/- animals to wildtype point was 64.1% and 48,3%, respectively. PVE/PVL animals. Using the Crispr/Cas9 method, an IL-33 knockout presented inferior OS and DFS in comparison with the line was established in the RIL175 mouse HCC cell line other strategies. and proliferation and migration of the cells were tested in Conclusion: Preoperative strategies to increase liver vitro. In order to assess the effect of tumor secreted IL-33 volume are effective to improve resectability in the setting

HPB 2018, 20 (S2), S182eS294 S262 Free Orals (FP01-FP31) of HCC. ALPPS is associated with higher risk of PHLF and the interaction between immune system components and mortality. Well-designed comparative studies are needed to tumor cells can contribute to the identification of new confirm our findings. mechanisms of development and survival of tumors under the immune system action.

FP20-15 Table [Risk difference (RD) between each strategy for the] Overall survival RD (95%CI) p-value PVE vs. TACE + PVE -10.9% (-18.85, -2.942) 0.01016 ALPPS vs. PVE 1.327% (-14.92, 17.57) >0.999 ALPPS vs. TACE + PVE -9.568% (-25.96, 6.825) 0.3152 Disease-free survival RD (95%CI) p-value PVE vs. TACE + PVE -26.84% (-36.44, -17.24) 0.000000214 ALPPS vs. PVE 21.88% (5.789, 37.96) 0.01373 ALPPS vs. TACE + PVE -4.964% (-22.12, 12.19) 0.6925

FP20-16 FP20-17 EVALUATION OF T CELL-MEDIATED LIVER RESECTION FOR ANTITUMOR IMMUNE RESPONSE IN TRANSPLANTABLE HCC ON LIVER TUMORS - AN APPROACH CHRONIC LIVER DISEASE: USING TUMOR AND PERIPHERAL PATTERNS OF RECURRENCE AND BLOOD SAMPLES IMPLICATIONS FOR TREATMENT R. Martins1,2,3, C. Martín-Sierra4, P. Laranjeira4, CHOICE 3,5 1,2,3 3,5 A. M. Abrantes , J. G. Tralhão , M. F. Botelho , M. Gelli1,2, M. Sebagh1, E. Romanelli1, E. Vibert1, 6,7 1 6,7 4 J. Leite , E. Furtado , F. Castro e Sousa and A. Paiva A. Sa Cunha1, D. Castaing1, R. Adam1 and D. Cherqui1 1 2 Liver Transplant Department, Surgery Department - 1Centre Hépato-Biliaire, Paul Brousse Hospital, and 3 HBP Unit, Coimbra University Hospital, Biophysics 2General Surgery, Gustave Roussy Institute, IBILI, Faculty of Medicine, University of Coim- Up-front liver resection (LR) followed by salvage bra, 4Department of Operational Management in Cytom- 5 transplantation (ST) in case of recurrence is considered the etry, Coimbra University Hospital, CIMAGO (Center of best option for transplantable HCC on chronic liver disease Investigation on Environment Genetics and Oncobiology), (CLD), but successful rate remains relatively low. The aim Faculty of Medicine, University of Coimbra, Portugal, of the study is to identify predictors of non-transplantable 6Surgery Department, Coimbra University Hospital, and 7 recurrence (NTR) in order to improve decision-making at Faculty of Medicine - University of Coimbra, Portugal initial diagnosis. Introduction: Hepatocellular carcinoma (HCC) and chol- All consecutive patients who underwent LR for trans- angiocarcinoma (CCA) are primary malignancies of the plantable HCC on CLD (age < 70 years, AFP-score2 and liver. The identification of highly diverse tumor-infiltrating fibrosis score F2-F4) between 2000 and 2015 were retro- leukocyte (TIL) subsets, and their distinct functions in the spectively analysed. Primary endpoint was to identify tumor niche, has been an important development in onco- preoperative predictive factors of NTR. immunology. One-hundred forty-eight patients underwent LR. The aim of this study is to identify potential correlations Median MELD score was 9. Major hepatectomy and between tumor type and stage and T cell-mediated anti- anatomical resection rates were 24.3% and 66.9%. After tumor immune responses in patients with these a median follow-up of 45.8 months, 5-year overall and malignancies. recurrence-free survivals were respectively 73.6% and Method: Peripheral blood (PB) and tumor samples (TS) 32.6%. The rate of ST was 35.4% among eligible pa- were colected from 14 patients with HCC (64Æ16.8 years) tients. The rate of NTR in the cohort was 38.2%. Pre- and 5 with CC (62Æ18.2 years) when performed a surgical operative predictors of NTR at multivariate analysis resection. Five controls used were (52Æ3.9 years). PB and were tumors>1 (OR 3.10;95%CI[1.049-9.185];p.041), TS were processed for the analysis of cytokine production AFP>100 ng/mL (OR 3.72;95%CI[1.129-12.282];p.031) (IL-17 and IFN-g) by T lymphocytes using flow cytometry. and cirrhosis (F4) (OR 2.98;95%CI[1.007-8.802];p.049). Results: HCC biopsies show a higher percentage of infil- In the presence of 0, 1 and >=2 factors, the 5-year RFS trated immune cells vs. CC (p< 0.05). However, the pro- were 52.6%, 32.4% and 0%(p< .0001). The rates of portion of TCD8 is higher in CC (p< 0.05). It is interesting NTR were 10.5%, 17.9% and 46.2%(p 0.002), respec- to note that in the same patient, tumor-infiltrating TCD4 has tively (Table 1). a higher percentage of cells producing IFNg and IL-17, Forty-six percent of the patients had no recurrence and compared to TCD4 from PB (p< 0.05). In PB, the per- possibly avoided unnecessary ST, however the successful centage of TCD4 IFNg+ IL-17+ and TCD8 IL-17+ is rate of a ST strategy remains low in case of transplantable decreased in HC and CC vs. control group (p< 0.05). recurrences. According to preoperative predictors, in case Conclusions: The analysis of the cytokines production of >=2 factors upfront liver transplantation should be profile in HCC and CC allows to understand the important preferred to LR. Simultaneously access to ST should be role of immune system in this type of tumors. Moreover, improved.

HPB 2018, 20 (S2), S182eS294 Free Orals (FP01-FP31) S263

FP20-18 Introduction: Laparoscopic liver resection (LLR) is considered standard of care for minor liver resections. For a BODY COMPOSITION IS ASSOCIATED safe implementation of major LLR, a stepwise approach WITH POOR OVERALL SURVIVAL AND and structured training are essential. The aim of this study ACCELERATED TIME TO was to evaluate the current implementation of major LLR in PROGRESSION IN HCC PATIENTS the Netherlands. TREATED WITH SORAFENIB Methods: We performed a nationwide retrospective anal- ysis of all adult patients who underwent LLR between T. Labeur1, J. van Vugt2, D. ten Cate2, B. Takkenberg3, 2 4 5 January 2011 and December 2016 in one of 27 centers. B. Groot Koerkamp ,R.deMan, O. van Delden , fi 2 6 1 Major LLR was de ned as all resections of more than 3 J. IJzermans , F. Eskens and H. -J. Klumpen segments. A 50-question survey was sent to all centers to 1Medical Oncology, Academic Medical Center Amster- 2 assess the use of LLR. dam, Surgery, Erasmus MC University Medical Center, Results: A total of 906 patients from 20 centers were 3Gastroenterology and Hepatology, Academic Medical 4 included. Most patients (n=854, 94%) underwent minor Center Amsterdam, Hepatology, Erasmus MC University LLR whereas 52 patients (6%) underwent major LLR. Medical Center, 5Radiology, Academic Medical Center 6 Median overall hospital stay was 5 (1-117) days, conver- Amsterdam, and Medical Oncology, Erasmus MC Uni- sion rate 13%, severe postoperative complication rate 9% versity Medical Center, The Netherlands and mortality 2%. Ten centers performed major LLR, Background: It remains challenging to predict outcomes in averaging 5 resections per center in the past 5 years (range patients with advanced HCC. Small and non-European studies 1-12). Median operating time for major LLR was 336 (91- have associated low muscle mass (sarcopenia) and visceral 720) minutes, blood loss 550 (100-3300) ml, postoperative adiposity with impaired survival and increased treatment stay 8 (4-33) days, conversion rate 23%, severe post- toxicity. However, large studies in European patients are lacking. operative complication rate 15% and mortality 6%. Methods: A retrospective analysis was performed in HCC Twenty-nine surgeons from 22 hospitals (81%) reported the patients treated with sorafenib at two tertiary referral cen- lack of structured training as most common cause for the ters between 2007-2016. Muscle mass and density and slow implementation of major LLR. adipose tissue areas were measured at baseline by Conclusion: Overall outcome of, mostly minor, LLR is computed tomography (CT) at the third lumbar vertebra good in the Netherlands. Major LLR is only rarely (L3) level. Sarcopenia and low muscle density were defined performed and outcomes may be improved. This study using published cut-off points. Body composition param- confirms the necessity of a safe and structured imple- eters were correlated with overall survival (OS), time-to- mentation program for major LLR. progression (TTP), response rate, and toxicity. Results: In total, 278 patients were included (79% male, median age 66) with a median OS of 9.5 (95%CI;8.1-11.0) FP21-02 months. At baseline, 52% had sarcopenia and 41% had low muscle density. Sarcopenia showed an independent association THE ROLE OF ANATOMIC VERSUS with TTP (HR 1.36;95%CI 1.03-1.77;p=0.04), whereas low PARENCHYMA SPARING RESECTION total adipose tissue index (TATI) was associated with reduced IN DETERMINING LONG TERM OS. Combined presence of sarcopenia and low total adipose PATTERNS OF RECURRENCE IN tissue index (TATI) was associated with reduced treatment CIRRHOTIC PATIENTS: A PROPENSITY duration (16 vs 11 weeks, p=0.029) and independently asso- SCORE ANALYSIS ciated with poorer OS (HR 1.68;95%CI 1.24-2.27;p=0.001). 1,2 3 1,2 3 None of the body composition parameters was independently S. Famularo , S. Di Sandro , A. Giani , A. Lauterio , 1,2 3 1,2 3 associated with response rate or treatment toxicity. F. Romano , V. Buscemi , F. Uggeri , R. De Carlis , 1,2 1,3 Conclusion: In European patients with HCC treated with L. Gianotti and L. De Carlis 1 sorafenib, low TATI and presence of sarcopenia are inde- School of Medicine and Surgery, University of Milano- 2 pendent prognostic factors of poorer OS and TTP respec- Bicocca, Department of Surgery, San Gerardo Hospital, 3 tively. Combined presence impairs survival to a greater and Department of Surgery and Trasplantation, Niguarda extent. CT-assessed body composition provided prognostic Ca’Granda Hospital, Italy information prior to sorafenib treatment. Introduction: Hepatocarcinoma recurrence is poorly un- derstood. Currently, anatomic resection (AR) is considered FP21 - Free Papers 21 (mini oral) - Liver: the standard treatment because of portal flow theory for Technical Surgery 1 recurrence. The aim of the study was to investigate the role FP21-01 of AR and parenchyma-sparing resection (PSR) on pattern OUTCOMES OF LAPAROSCOPIC of disease-free-survival, with regards to presence of microvascular invasion and early recurrence. MINOR AND MAJOR LIVER SURGERY Methods: 384 cirrhotic patients with a first diagnosis of IN THE NETHERLANDS (LAELIVE): hepatocarcinoma were stratified for AR (142; 37%) and NATIONWIDE RETROSPECTIVE PSR (242, 63%). To minimize selection bias, a 1:1 pro- COHORT pensity-score-matching was run. Results: After matching, 200 patients were enrolled, 100 M. J. van der Poel1, R. S. Fichtinger2, R. M. van Dam2, for each group. Fifty-nine patients (62.8%) had recur- M. G. Besselink1 and Dutch Liver Collaborative Group rence after AR, while fifty-eight (63.7%) after PSR 1Academic Medical Center Amsterdam, and 2Maastricht (p:0.891). Median disease-free survival (DFS) was 18.6 Universitair Medisch Centrum + months (95%CI: 14.1-23.2) and 20months (95% CI:

HPB 2018, 20 (S2), S182eS294 S264 Free Orals (FP01-FP31)

14.3-25.67) respectively (p:0.914). The rates of recur- Results: Twenty ’matched’ patients in each group Lap & rence on the resected margin were 15.3% for AR and open (total n=40) were identified. There was no difference 15.5% for PSR (p:0.968). The rate of recurrence within in the median age in the study population. The median 24months were 29% and 37% respectively (p:0.417). hospital stay was significantly shorter in the laparoscopic When microvascular invasion was considered, median group (lap=6 days, open=10 days, p=0.04). There was no DFS was 10.72 months (95%CI: 3.43-18.01) for AR, difference in the intra-operative blood loss (lap=100ml, and 9.44 months (95%CI: 5.01-13.86) for PSR open=300ml, p=0.42). Parenchymal transection time (lap. (p:0.607). No differences in median DFS were evident =100 min, open=35 min, p=0.0004) and pringle time (lap between groups when it is considered high histological =56 min, open=26.5min, p=0.0059) were significantly grading (p:0.520), more than 1 nodule (p:0.307), or shorter in the open surgery group. Lotus liver blade was Child B patients (p:0.679).Thus, the type of surgery was significantly more cost effective than CUSA (£5,300 vs not a risk factor for recurrence (p:0.428). £20,300 respectively, p< 0.001). Zero mortality was Conclusions: The excision of the anatomic segment does recorded in either group. not seem to reduce the recurrence rate in cirrhotic patients, Conclusion: Major laparoscopic liver surgery using a even in case of microvascular invasion. The type of surgery Lotus Liver blade compared to open liver resection is does not affect the risk of early-recurrence. associated with an increased operating & Pringle time but hospital stay is significantly decreased in the laparoscopic group. Major laparoscopic hepatectomy initially appears safe and feasible with a possible cost saving.

FP21-04 LAP LIVER RESECTION DECREASES MORTALITY AND COMPLICATIONS BY PROPENSITY SCORE-ADJUSTED REGRESSION J. Sulzer1, D. Monlezun2, P. McLaren3, M. Mvula2,C.Li2, A. Volk2, V. A. Khanh-Van Y2, G. Parker4, M. Darden5 and J. Buell6 1Louisiana State University Health New Orleans, 2Tulane University School of Medicine, 3Louisiana State University Health Science Center, 4Dartmouth College, 5George Washington University, and 6Department of Figure [pattern of recurrence] Surgery, Tulane University School of Medicine, United States Introduction: The use of laparoscopic liver resection has greatly increased over the last two decades with continuing FP21-03 expansion of indications. Prior observational studies COMPARATIVE ANALYSIS BETWEEN demonstrating improvements in morbidity and mortality MATCHED OPEN AND LAPAROSCOPIC have been largely limited by selection bias and small MAJOR LIVER RESECTIONS: THE sample size. Here we present a large single-center case control study using propensity score-adjusted multivariate MANCHESTER EXPERIENCE regression to account for potential confounders on mor- A. Mirza, M. Baltatzis, P. Stathakis, A. Tyurkylmaz, tality and complication rates following laparoscopic vs S. Jamdar, A. Siriwardena and A. Sheen open liver resection. Department of Hepatobiliary and Pancreatic Surgery, Methods: We used case-controlled retrospective data from Manchester University Hospitals NHS Foundation Trust, a single center we performed causal inference statistics of United Kingdom propensity score-adjusted multivariate regression Aims: Laparoscopic liver surgery is possibly associated augmented by forward and backward regression to examine with overall decreased morbidity and an earlier recovery. mortality and complications matched for demographics, However major laparoscopic liver surgery has been limited cirrhosis and type of resection. to a few centres of excellence. This study aimed to compare Results: 1147 subjects from 2000-2016 were analyzed. outcomes of a consecutive series of patients undergoing Ninety day mortality from surgery to date of death or end of major laparoscopic and open liver surgery performed in a follow-up was improved in lap (OR 0.24, P< 0.001) as well tertiary centre. as cumulative complication rate (OR 0.58, P=0.005) as Methods: Prospective data of patients undergoing liver compared to open controls. resection were analysed from 2014-2017 (48 months). All Conclusion: Laparoscopic liver resection significantly patients who underwent major liver resection (anatomical decreases the odds of post-op mortality and complications, right and left hepatectomies) were identified. Lotus ultra- according to a large single-center case-control study using sonic energy device with the specially designed liver blade causal inference statistics of propensity score-adjusted and CUSA were utilised in laparoscopic & open transec- multivariate regression. tions respectively.

HPB 2018, 20 (S2), S182eS294 Free Orals (FP01-FP31) S265

FP21-05 Introduction: ALPPS is two stage hepatectomy with a high morbidity and mortality rate. The original technique COMPLEX LIVER RESECTION UNDER has been modified to minimize the surgical severity of both TOTAL VASCULAR EXCLUSION AND stages. The two dominant modifications besides Classic HYPOTHERMIC PERFUSION WITH ALPPS are Partial ALPPS and Tourniquet ALPPS. The aim VERSUS WITHOUT VENOUS BYPASS of this study is to assess the safety of the 3 variants and efficacy to induce hypertrophy. J. Navez1, F. Cauchy1, S. Dokmak1, E. Faivre2, E. Weiss2, Methods: An observational retrospective study of all pa- O. Scatton3, C. Paugam2 and O. Soubrane1 tients in the International Registry (NCT01924741) was 1Department of Hepato-Pancreatico-Biliary Surgery, performed. Only surgical variants with more than 40 cases 2Department of Anesthesiology and Critical Care, and were included. Only patients from centers with more than 3Department of Hepatobiliary and Liver Transplantation 10 cases were taken into account. Main endpoint was 90- Surgery, Pitié-Salpêtrière Hospital, France day mortality. Secondary endpoints were: complications, Background: While hypothermic perfusion of the liver has liver growth and time between stages. been shown to improve parenchymal tolerance to complex Results: 3 variants of ALPPS were identified: Classic resections in patients requiring prolonged hepatic vascular ALPPS (n=263), Partial ALPPS (n=75) and Tourniquet fi exclusion (HVE), the bene t of associated veno-venous ALPPS (n=45). Classic ALPPS (14,4%), had a high mor- bypass (VVB) in this setting remains poorly evaluated. tality than both Partial ALPPS (4%) and Tourniquet Methods: All patients undergoing liver resection requiring ALPPS (6,7%) (p=0,024). Severe complications ( IIIB) HVE for at least 60 minutes were retrospectively reviewed. after stage 1 were similar for all three techniques. Severe Perioperative outcomes were compared between patients complications after stage 2 were significantly higher for with (VVB+) or without VVB (VVB-). Classic ALPPS (30,4%) than for Tourniquet ALPPS Results: Twenty-seven patients had liver resection with (11,1%) and Partial ALPPS (17,3%) (p=0,004). Liver HVE and hypothermic perfusion of the liver between 2006 growth was lower for Classic ALPPS (62,5%; IQR:22%- and 2017, including 13 (48%) VVB+ and 14 (52%) VVB- 100,6%) than Tourniquet ALPPS (76%; IQR:57%-99%) patients. Demographic characteristics and indications for and Partial ALPPS (77,5%; IQR: 48,6%- liver resection were similar between the two groups. 110,9%)(p=0,043). Time between stages was longest for Median HVE exclusion durations were similar in (96 vs. 75 Tourniquet ALPPS (12 days; IQR 11-15), followed by min, p=0.72) VVB+ and VVB- patients. VVB+ patients Partial ALPPS (11 days; IQR 7-15,5), and Classic ALPPS had increased operative time (460 vs. 375 min, p=0.05) but (10 days; IQR 8-18) and (p=0,029). decreased amount of transfusion (p=0.05). . Five (19%) Conclusion: Partial and Tourniquet ALPPS have a lower patients died postoperatively from liver failure (n=4) or mortality and complication rate than classic ALPPS, while fi sepsis (n=1), without signi cant difference between VVB+ achieving superior liver hypertrophy. and VVB- patients (p=0.56). Postoperative major morbidity rate (Dindo-Clavien 3-4, 30% vs. 50%) was not different between the two groups. The rates of liver failure, haemorrhage, renal insufficiency, and sepsis were not FP21-07 different between the two groups but VVB- patients AN AUGMENTED REALITY experienced increased rates of respiratory complications NAVIGATION SYSTEM FOR than VVB+ patients (64% vs. 15%, respectively, p=0.01). LAPAROSCOPIC LIVER SURGERY e Conclusion: During liver resection under HVE and hypo- FIRST EXPERIENCES thermic perfusion of the liver, the use of VVB allows 1 1 2 2 reducing red blood cells transfusions and postoperative I. Paolucci , B. Eigl , G. A. Prevost , A. Lachenmayer , M. Peterhans3, S. Weber1, D. Candinas2 and G. Beldi2 respiratory complications. VVB should be recommended in 1 case of liver resection with prolonged HVE. ARTORG Center for Biomedical Engineering Research, 2Department of Visceral Surgery and Medicine, Inselspi- tal, Bern University Hospital, University of Bern, and 3Surgical Navigation, CAScination AG, Switzerland FP21-06 Introduction: Augmented reality (AR) is ubiquitous in PARTIAL AND TOURNIQUET ALPPS industries like entertainment, but it also has the potential to DECREASE MORTALITY OF ALPPS. A emerge into medicine. Especially during laparoscopic sur- COMPARATIVE STUDY BETWEEN gery, fusing virtual or preoperative information with the THREE TECHNIQUES FROM THE live laparoscopic video could improve intraoperative INTERNATIONAL ALPPS REGISTRY visualization. In this study, we evaluated technical feasi- bility of an AR navigation system (CAS-One AR, CASci- R. Brusadin1, R. Robles Campos1, A. López Conesa1, 1 2 2 nation AG, Switzerland) for laparoscopic liver surgery. V. López López , M. Linecker , H. Petrowsky , Methods: For navigation, a 3D laparoscope (Einstein E. Schadde3, E. De Santibañes4, P. Parriila Paricio1, 2 Vision 3, Aesculap, Germany), a grasper and an overholt P. A. Clavien and on behalf of International ALPPS were equipped with optical markers and calibrated. Registry Study Group Registration of the preoperative 3D model was performed 1 ’ ’ Hospital Clinico Universitario Virgen de la Arrixaca , using four surface landmarks. A fused view of the under- Spain, 2Univerity Hospital Zurich, 3Cantonal Hospital 4 lying anatomy (tumors, liver segments, hepatic/portal vein Winterthur, and Hospital Italiano de Buenos Aires, and hepatic artery), the instruments and the live video was Argentina shown on the 3D screen.

HPB 2018, 20 (S2), S182eS294 S266 Free Orals (FP01-FP31)

of 76.9% (95% CI: 46.2%-95.0%) and negative predictive value of 92.3% (95% CI: 64.0%-99.8%). Conclusions: Post-PVE HVPG is a simple and reproducible tool, which accurately predicts FRL hypertrophy after PVE and allows early detection of patients who may benefit from more aggressive procedures inducing FRL hypertrophy.

FP21-09 INTRAOPERATIVE NAVIGATION FOR LIVER RESECTION BY TABLET Figure [The navigation screens with the fused 3D view] TERMINAL BASED ON AUGMENTED REALITY Results: To date, the system was applied in one patient S. Satou and Y. Harihara with three lesions selected for non-anatomical resection. Department of Surgery, NTT Medical Center Tokyo, Japan Time required for laparoscope and instrument calibration Introduction: Most of recent surgical navigation systems was 1:39 minutes. In total, four registration attempts were employ expensive machines. A simple navigation system required with gradually decreasing registration errors from for liver resection by marker-based augmented reality 22.9 mm to 6.1 mm and an average duration of 4 minutes (AR) which runs on a tablet terminal has been newly each. developed. Conclusions: The AR system was mainly used during the Method: Three-dimensional (3-D) images of hepatic pa- initial phase of the surgery, where the 3D view of the un- renchyma, vessels and tumors were constructed by a derlying anatomy allowed fast localization of the tumors simulation software (SYNAPSE VINCENT, FUJIFILM, and was helpful for resection planning. We consider the Japan) utilizing preoperative CT data, and those were additional time required for setup and registration to be exported to a tablet terminal (ICONIA W4-820/FP, Acer, acceptable for the first use in clinics. Based on our previous Taiwan). In operation, the 3-D images were presented on a experience, we expect that the additional efforts and display of the tablet terminal by real scale and in appro- registration errors will decrease over time. priate position, by visual recognition of a small plastic AR FP21-08 marker which was sutured on the liver surface. The navi- gation system was employed in 18 cases of liver resections. HEPATIC VENOUS PRESSURE Position gap was measured in each operation by comparing GRADIENT FOLLOWING PORTAL VEIN the liver edge of gallbladder bed or umbilical fissure with EMBOLIZATION: AN ACCURATE those of 3-D reference images. PREDICTOR OF FUTURE REMNANT Results: The 3-D images were successfully displayed in LIVER HYPERTROPHY all cases. The system clearly indicated not only the lo- cations of intrahepatic tumor and vessels but also the K. Mohkam, A. Rode, B. Darnis, A. -F. Manichon, extent of the liver parenchyma to be resected. The L. Boussel, C. Ducerf, P. Merle, M. Lesurtel and median gap of image position ranged 0.3 to 3.2 cm (1.6 J. -Y. Mabrut cm in median). Croix Rousse Hospital, France Conclusions: This study showed a sufficient accuracy of Background: The impact of portal hemodynamic varia- the navigation system which is simple and easily-obtain- tions after portal vein embolization (PVE) on liver regen- able. It is a promising and evolvable tool for image-guided eration remains unknown. We studied the correlation liver surgery. between hepatic venous pressure parameters measured before and after PVE and future remnant liver (FRL) hy- pertrophy following PVE. Methods: Between April 2014 and August 2018, patients who were eligible for major hepatectomy and who had PVE underwent simultaneous measurement of portal venous pressure (PVP) and hepatic venous pressure gradient (HVPG) before and after PVE by direct puncture of portal vein and inferior vena cava. Discrimination of these pa- rameters to predict FRL hypertrophy was assessed. Results: Twenty-six patients were included. After PVE, PVP increased from 15 (9-24) to 19 (10-27) mmHg and HVPG increased from 5 (0-12) to 8 (0-14) mmHg. Median FRL volume was 513 (299-933) ml before PVE vs. 724 (499-1279) ml 3 weeks after PVE, representing a 35.4% (7.4-83.6) median hypertrophy. Post-PVE HVPG was the most accurate parameter to predict failure of FRL to reach a 30% hypertrophy (c-statistic: 0.882 [95% CI: 0.727-1.000], p< 0.001). A cut-off value of HVPG of 8 mmHg showed a sensitivity of 90.9% (95% CI: 58.7%-99.8%), specificity of Figure [Navigation by tablet terminal] 80.0% (95% CI: 51.9%-95.7%), positive predictive value

HPB 2018, 20 (S2), S182eS294 Free Orals (FP01-FP31) S267

FP21-10 Methods: All patients undergoing major hepatectomy on a “ ” non-cirrhotic liver were included in this prospective study THE DORSAL APPROACH TO since October 2017. Measurement of PVP and PVF were LAPAROSCOPIC MAJOR performed before and after liver resection a using Transit HEPATECTOMY ENABLES EXPANDED Time Flow Measurement (Medistim, Oslo, Norway). SAL INDICATIONS FOR LAPAROSCOPIC was performed in patients with a Liver-remnant-to-Body- < > LIVER RESECTION weight-ratio (LR/BW-ratio) 1%, or a PVP 5mmHg and/or suboptimal macroscopic liver parenchyma appear- ’ R. Bryant, D. Cavallucci and N. O Rourke ance, due to previous chemotherapy. ’ Royal Brisbane and Women s Hospital, Australia Results: Sixteen cases were analysed with a median LR/ Introduction: This new technique, a modification of the BW-ratio of 1.2% (range 0.7-1.8%). Median PVP increased caudal approach of Soubrane, involves a posterior to from 4 to 7 mmHg after resection, while PVF decreased anterior transection of the liver for laparoscopic major from 851 to 710 ml/min. The peak lactate was higher in hepatectomies. It was conceived to enable a standardised patients with an increase in PVP of >5mmHg (Figure 1). technique, and to broaden the indications for laparoscopic SAL effectively decreased the median PVP and PVF of 3 resection particularly for larger tumours and with anatom- mmHg and 100 ml/min, respectively. ical variations of the porta. Conclusion: Elevated pressures in the portal system have a Methods: As the initial step for a right hemihepatectomy the negative impact on the initial function of the liver remnant retrohepatic tunnel of Belghiti is developed and the caudate after major hepatectomy. Splenic artery ligation appears process divided as far superiorly as possible. Liver paren- easy to modulate the portal inflow. This study is currently chyma is dissected away from the posterior aspect of the right recruiting further patients and the entire data set will be hepatic inflow from medial to lateral, enabling safe stapling. available at the conference. Development of the retrohepatic tunnel and division of the posterior parenchyma is continued as the leading edge of the parenchymal transection, maintaining good surgical orienta- tion especially for tumours close to the midline or cava. Similar concepts apply for a left hemihepatectomy, with dissection beginning in the in the Arantius groove. Results: 8 cases (6 right hemihepatectomy, 1 extended right hepatectomy, 1 extended left hepatectomy) have utilised this technique. I required conversion to a hybrid approach, and 1 case required transfusion. R0 resection was achieved in all cases. Average length of stay was 5 days. Comparison to 50 consecutive previous laparoscopic major hepatectomies demonstrates larger tumours and fewer conversions with the dorsal approach. Schematic and case videos will be presented. Conclusion: The dorsal approach to laparoscopic major hepatectomy is a novel technical variation that enables a safe, standardised technique and an expanded set of in- dications for a laparoscopic approach. Figure 1

FP21-11 FP21-12 THE FIRST RESULTS OF A PROSPECTIVE EVALUATION OF THE PROSPECTIVE STUDY INVESTIGATING XY CLASSIFICATION TO EVALUE THE THE POTENTIAL BENEFIT OF PORTAL RESECTABILITY OF LOCALIZED INFLOW MODULATION BY SPLENIC PERIHILAR CHOLANGIOCARCINOMA ARTERY LIGATION IN MAJOR A. Mimmo, A. Robert, S. Tzedakis, H. Jeddou, M. Rayar, HEPATECTOMIES L. Sulpice and K. Boudjema Service de Chirurgie Hépatobiliaire et Digestive, CHU A. Schlegel, M. Kalisvaart, C. Coldham, K. Roberts, Rennes Pontchaillou, France R. Marudanayagam, R. Sutcliffe, B. Dasari, M. Abradelo, Introduction: Surgical resection is the best treatment for J. Isaac and P. Muiesan perihilar cholangiocarcinoma (PHC). The XY classifica- Queen Elizabeth Hospital Birmingham, United Kingdom tion, based on the invasion (X) or non-invasion (Y) of the Background: Posthepatectomy liver failure (PLF) is one of B2B3 convergence, distinguishes the Y PHC, always the most feared complications in major hepatectomies. resectable without arterial invasion, from the X CPH fl Elevated portal venous pressures (PVP) and ow (PVF) resectable with a complex surgery (possible arterial inva- after resection are associated with development of PLF. sion). The aim of this work is to prospectively evaluate the Splenic artery ligation (SAL) is a potential strategy to validity of this classification. fl modulate the portal venous in ow. The aim of our study is Methods: Prospective unicentric study concerning all the therefore to investigate the course of portal pressure and PHC operated between 01/2012 and 12/2016. Y were fl ows during major hepatectomy and the impact of SAL. resectable with extended right hepatectomy; X lesions with

HPB 2018, 20 (S2), S182eS294 S268 Free Orals (FP01-FP31) a difficult excision (extended left hepatectomy, arterial 48 patients from group 1 and group 2 each were compared. reconstructions). No differences were observed between the two matched Results: The analysis involved 74 patients, of median age groups. Operative and postoperative course were similar in 66, including 40Y and 34X. In group Y, 97% received a patients with HCC in favorable and unfavorable LLR lo- pre-operative work-out (left biliary drainage / right portal cations. Surgical margins were found to be similar before embolization). Excision was possible in 100% of cases and after PS. (80% R0) with portal resection in 16 cases. No left arterial Conclusions: These results show that LLR in patients with invasion. 90 days morbidity (Clavien-Dindo3) was 55%, HCC can be safely performed in all segments because of mortality was 10%. In group X, excision was possible in 27 the extensive experience of all surgeons from multiple cases (80% vs 100% (X / Y), p = 0.003). 71% of R0 centers in performing traditional open liver surgery as well (p>0,05) were found. Vascular invasion required portal as laparoscopic surgery. resection-reconstruction in 9 cases (p>0,05), arterial in 6 cases (p = 0.05). 90 days morbidity was 48%, mortality was 11% (p>0,05). 4 yrs overall and recurrence-free survival FP21-14 rates were respectively 55.7% (Y) vs. 49.7% (X) and 39.4% > SPLENIC ARTERY LIGATION AFTER (Y) vs.37.3% (X) (p 0,05). e Conclusion: The XY classification is reliable for predicting ALPPS USING AN ANIMAL MODEL the absence of left arterial invasion and a 100% resection EVALUATION OF ITS IMPACT ON rate of Y types. LIVER FUNCTION, VIABILITY AND REGENERATION R. Martins1,2,3, R. Nemésio2, K. Cardoso3, FP21-13 R. Caetano Oliveira4, A. C. Gonçalves5,6,7, LAPAROSCOPIC LIVER RESECTION OF A. B. Sarmento Ribeiro5,6,7, A. M. Abrantes3,5, HEPATOCELLULAR CARCINOMA M. F. Botelho3,5, J. G. Tralhão1,2,3 and F. Castro e Sousa2,8 1Liver Transplant Department, 2Surgery Department - LOCATED IN UNFAVAROBLE 3 SEGMENTS: A PROPENSITY SCORE- HBP Unit, Coimbra University Hospital, Biophysics Institute, IBILI, Faculty of Medicine, University of Coim- MATCHED ANALYSIS FROM THE bra, 4Pathology Department, Coimbra University Hospi- ITALIAN GROUP OF MINIMALLY tal, 5CIMAGO (Center of Investigation on Environment INVASIVE LIVER SURGERY REGISTRY Genetics and Oncobiology), Faculty of Medicine, 6 1 2 3 Oncobiology and Hematology Lab, Faculty of Medicine, G. B. Levi Sandri , L. Aldrighetti , U. Cillo , 7 R. Dalla Valle4, A. Guglielmi5, V. Mazzaferro6, University of Coimbra, CNC (Center of Neuroscience and 7 8 9 Cellular Biology), Faculty of Medicine, University of A. Ferrero , F. Di Benedetto , S. Gruttadauria and 8 G. M. Ettorre1 Coimbra, and Faculty of Medicine, University of Coim- 1San Camillo Hospital, 2Hepatobiliary Surgery, IRCCS bra, Portugal San Raffaele Hospital, IRCCS San Raffaele Hospital, Introduction: ALPPS (’’Associating Liver Partition and Milan, 3University of Padua, 4Parma University Hospital, Portal vein ligation for Staged hepatectomy’’) has been 5Department of Hepatobiliary Surgery, G.B. Rossi Hos- recently described as a revolutionary strategy in hepato- pital, University of Verona, 6HPB Surgery and Liver biliary surgery, becoming widely used. However, the Transplantation, University of Milano, 7Ospedale Maur- complex mechanisms for rapid hepatocellular regeneration iziano Umberto I, 8University of Modena and Reggio associated with ALPPS are not well known. On the other Emilia, and 9Mediterranean Institute for Transplantation hand, the technique bounds to a high rate of morbidity and and Specialization Therapies (IRCCS-ISMETT), Italy mortality due, mainly, to postoperative liver failure. Objective: Laparoscopic liver resection (LLR) for Hepa- Method: Thirty-one rats were submitted to portal ligation tocellular Carcinoma (HCC) is one of the most important and in situ splitting (ALPPS) with (n=15) and without indication of minimally invasive approach. The aim of our splenic artery ligation (n = 16). The control group (lapa- study is to analyze the experience of the Italian Group of rotomy and/or pedicle transection) included 14 animals. fi Minimally Invasive Liver Surgery with laparoscopic sur- After sacri ce (12, 24, 48 and 120h after surgery), blood gical treatment of HCC, with a focus on tumor location and and hepatic tissue samples were collected to evaluate he- how it affects morbidity and mortality. patic function, regeneration and viability. Methods: 38 centers participated in this study; 372 cases of Results: The animals submitted to splenic artery ligation at LLR for HCC were prospectively enrolled. Patients were 12h revealed better hepatic function and less reactive spe- divided into two groups according to the HCC nodule cies production. When evaluated at 48h, the group sub- location. Group 1 favorable location and group 2 unfa- mitted to splenic artery ligation had a higher percentage of vorable location. Perioperative outcomes were compared cell death by apoptosis and a lower reactive species pro- between the two groups before and after a propensityscore duction. 120h after surgery there is higher cell viability and match (PS) 1:1. lower reactive species production in the group submitted to Results: Before PS in group 2 surgical time was longer; splenic artery ligation. conversion rate was higher, postoperative transfusion and Conclusions: This experimental study suggests that splenic fl comprehensive complication index were also greater. artery ligation in ALPPS, by modulating the portal ow, PS was performed with a cohort of 298 patients (from 18 promotes an increase in hepatocellular viability and centers), with 66 and 232 patients with HCC in unfavorable regeneration, with no function impairment, probably and favorable locations, respectively. After PS matching, related to a decrease in oxidative stress.

HPB 2018, 20 (S2), S182eS294 Free Orals (FP01-FP31) S269 ’ FP21-16 determined on the basis of landmark vessels and surgeon s educated guesswork, which is technically demanding and COMPLEX LIVER RESECTION WITH IN not always precise. The fusion indocyanine green fluo- SITU HYPOTHERMIC PORTAL resence imaging (IGFI) offers a potential solution by PERFUSION attaining a strong and real-time contrast visible on the R. Rhaiem1, M. Chetboun1, T. Piardi1, F. Fleres1, intersegmental plane. M. Jaussaud2, R. Kianmanesh1 and D. Sommacale1 Method: Between December 2015 and January 2018, a 1Department of General, Digestive, and Endocrine Sur- serial 41 patients underwent fusion IGFI-guided laparo- gery, Université Reims Champagne Ardennes/Robert scopic anatomical hepatectomy for HCC. The de- Debré University Hospital, and 2Department of Anesthe- mographic parameters and operative data were statistical siology, Université Reims Champagne Ardennes/Robert analyzed. Debré University Hospital, France Results: The clinicopathologic features of the patients and their perioperative outcomes were comparable to those of Introduction: Total vascular exclusion (TVE) of the liver the open and traditional laparoscopic hepatectomies in our might be required for complex liver resection. Standard institute. Our initial series involved laparoscopic anatom- TVE is contraindicated in some situations harboring a high ical sectionectomy, segmentetomy, sub-segmentectomy, risk of prolonged normothermic ischemia related hemihepatectomy and multiple segmentectomies. The morbidity. In situ hypothermic portal perfusion (HPP) can Fusion IGFI herein offers a “GPS” in the “jungle” of liver be used to overcome prolonged TVE consequences. parenchyma. Through a fusion monitor mounted on the Methods: This study is a prospective observational case laparoscopic system, surgeons could view the ordinal and series from January 2012 to December 2016, including fusion IGFI images simultaneously and unremittingly patients with malignant tumors involving the vena cava without averting one’s gaze or the operating light. Some and/or the hepatocaval confluence. Two surgical techniques tricky parts about this technique need to be noted, such as were used: standard TVE with HPP and venovenous bypass the preoperative 3D image reconstruction and the practice and HPP with preservation of the caval flow using a tem- of laparoscopic intraoperative ultrasound. porary porto-caval shunt. Results: During the study period, 443 liver resections were performed in our department. Among them, fourteen (3.1%) were performed using HPP. Mean age was 55 years. Twelve (86%) patients had underlying liver disease. Mean tumor size was 87 mm (40-170mm). Mean operative time was 470 mn (360-580mn). Vascular reconstruction was performed in 4 (28.6%) patients. The 90-day mortality rate was 7.1% (1 patient). Four

(28.6%) patients experienced grade III of Clavien-Dindo Figure 1 classification postoperative morbidity. No cases of post- operative liver failure or renal insufficiency were observed. Conclusion: Fusion IGFI is technically feasible in laparo- Conclusion: HPP should be considered in patients scopic anatomical hepatectomy. Further studies regarding requiring complex liver resections for centrally located its oncological outcomes are needed in larger prospective tumors involving the hepatocaval confluence as it allows cohorts. preservation of the future liver remnant function. Mean- while, such resections should be indicated in selected pa- tients with low comorbidity and good performance status because of a high morbidity rate. FP21-18 EARLY DRAIN REMOVAL IMPROVES OUTCOMES AFTER WHIPPLE FP21-17 PROCEDURE FUSION INDOCYANINE GREEN K. A. Zorbas1, J. M. Daly1,2, N. Esnaola2, S. Reddy2 and FLUORESCENCE IMAGING: THE NEW A. Karachristos1,2 “GPS” IN LAPAROSCOPIC 1Department of Surgery, Temple University Hospital, and 2 ANATOMICAL HEPATECTOMY - Surgical Oncology, Fox Chase Cancer Center BASED ON AN INITIAL LARGE Introduction: The timing of drain removal after pancre- PRACTICE IN A SINGLE CENTER atectomy has been poorly studied. Only three single- center studies have been reported with relatively small 1 1 1 2 1 Y. Xu , H. Wang , M. Chen ,Y.Pu, X. Meng , sample sizes. The aim of this analysis was to determine 1 1 1 W. Duan , Y. Luo ,S.Lu and Chinese Academy of the significance of early drain removal after a Whipple Hepatobiliary Intraoperative Ultrasound procedure utilizing a large prospective multi-institutional 1 Hepatobiliary Surgery, Chinese PLA General Hospital, database. 2 and Chinese Academy of Medical Sciences Methods: The 2014-15 NSQIP Pancreatectomy Participant Background: Anatomical resection has become an estab- Use Files were queried for patients undergoing a Whipple lished procedure for hepatic malignancies. Various land- procedure. Patients undergoing other pancreatic procedures mark techniques have been adopted to achieve the precisely (total or distal pancreatectomy, and pancreatic enucleation “anatomical”. The parenchymal transection plane has to be or debridement), were excluded. Patients without

HPB 2018, 20 (S2), S182eS294 S270 Free Orals (FP01-FP31) information on the intraoperative drain placement, drain cancer two years ago. During follow-up a CT scan revealed amylase level on postoperative day one (POD1), post- a 9-mm deep lesion in segment 5 with elevation of CEA operative day of last drain removal and patients with POD1 level to 9 ng/mL. This lesion was confirmed with a PET/CT amylase level > 5000 U/L were also excluded. Patients scan but was not visible on liver ultrasound. Considering with early drain removal (3 days) were compared to those this as a non-palpable isoechoic lesion difficult for intra- with late drain removal (4 days). Multi-variable regres- operative detection, a preoperative tumor marking was sion models were used to evaluate the possible benefitof decided. Before surgery, a CT-guided titanium "coil" early drain removal after adjustment for multiple placement was performed 2 mm away from the liver me- confounding factors. tastases with local anesthesia. A laparotomy was performed Results: 1066 patient were eligible for analysis. Patients using a right subcostal incision and using intra-operative with early drain removal had significantly lower mean rates ultrasound the "coil" was easily visualized. An in-situ liver of serious postoperative complications (p< 0.001), overall splitting was performed through the Cantle plane aiming morbidity (p< 0.001), pancreatic fistula (p< 0.001), organ for a parenchymal preserving approach. Sequential intra- space infection (p=0.007), delayed gastric emptying (DGE) operative ultrasounds were extremely helpful to detect the (p=0.026) and shorter mean in-hospital stay (p< 0.001). marked lesion. Finally, the lesion was carefully resected After adjustment for many confounding factors with multi- with a negative margin. Saline solution through the cystic variable regression models, the early group continued to duct reveals no bile leaks and intra-operative cholangiog- have a significantly lower risk of all noted complications, raphy confirmed absence of intra and extra-hepatic bile duct except postoperative DGE. injuries. The patient was discharged on the second post- Conclusion: Early removal (3 days)of the drain after operative day. To conclude, preoperative CT-guided coil Whipple procedures is associated with lower rates of marking for unique small isoechoic liver metastases is a postoperative adverse outcomes when POD 1 drain helpful tool and should be incorporated by HPB surgeons amylase levels are < 5000 U/L. dealing with injured livers.

FP22-02 Table. [Multiple Logistic Regression] THE DIFFICULT ABDOMINAL Multiple Logistic Regression Analysis: Early* vs Late CLOSURE AFTER PAEDIATRIC Drain removal MULTIVISCERAL TRANSPLANTATION: Complications aOR 95% CI p-value USE OF ABDOMINAL RECTUS MUSCLE Mortality 0.49 0.125-1.873 0.294 FASCIA Serious Postoperative 2.03 1.5-2.76 <0.001 Complications N. Cassar, M. Cortes-Cerisuelo, C. Bambridge, A. Ali, Pancreatic Fistula 6.76 2.42-18.87 <0.001 N. Heaton and H. Vilca-Melendez Institute for Liver Studies, King’s College Hospital NHS Delayed Gastric 1.39 0.85-2.26 0.188 Foundation Trust, United Kingdom Emptying Background: Primary abdominal wall closure after Overall Morbidity 2.13 1.59-2.86 <0.001 multivisceral transplantation may not be possible because Organ Space 2.46 1.18-5.12 0.016 of loss of abdominal domain or graft size/abdominal Infection cavity mismatch. Traditional closure techniques for the *Reference group, aOR: adjusted odds ratios after adjustment open abdomen may not be valid in these circumstances for history of COPD, preoperative chemotherapy, preoperative because of severe scarring of the abdominal wall from radiotherapy, preoperative jaundice, preoperative albumin <3 multiple previous surgeries. We hereby present our initial and intraoperative pancreatic duct size and gland texture experience with non-vascularised abdominal rectus muscle fascia in two patients who underwent combined liver and FP22 - Free Papers 22 (mini oral) - Tricks of the small bowel transplantation who couldn’t be closed Trade primarily. FP22-01 Technique: The abdominal rectus muscle fascia is A USEFUL METHOD TO BETTER retrieved using the standard thoraco-abdominal incision, IDENTIFY A UNIQUE SMALL which is deepened to the fascia and the overlying tissue is ISOECHOIC LIVER METASTASES mobilised as far as the lateral margin of the rectus muscle. The rectus muscle is then incised subcostally, lateral to the DURING SURGERY: HOW TO MAKE IT rectus muscle on both sides and excised by dividing it EASIER suprapubically. J. Devoto, M. Balmer and L. McCormack During benching, the rectus muscle is then removed Hospital Aleman of Buenos Aires, Argentina leaving a dual layer of fascia (Fig. 1). Either at the end of Intraoperative ultrasound is a relevant tool for liver sur- the recipient transplant operation or when a suitable fascia gery. However, dealing with small isoechoic lesions can be becomes available, the prepared abdominal rectus muscle challenging in patients with injured livers after chemo- fascia is sutured to the remaining abdominal muscle of the therapy or steatosis. We present the case of a 57 years-old recipient. If possible, skin and subcutaneous fat should be male with history of laparoscopic sigmoidectomy for colon approximated over the rectus muscle fascia. Vacuum

HPB 2018, 20 (S2), S182eS294 Free Orals (FP01-FP31) S271

Assisted Dressing is applied to the area. As the wound FP22-04 heals, it contracts and granulation tissue appears (at 5 and 7 “ ” weeks for our cases). Skin grafting can also be considered THE DORSAL APPROACH TO but in our cases it wasn’t necessary as both healed over the LAPAROSCOPIC MAJOR fascia, This approach is simple, effective and avoids mul- HEPATECTOMY tiple operations. R. Bryant, D. Cavallucci and N. O’Rourke Royal Brisbane and Women’s Hospital, Australia This new technique is a modification of the caudal approach of Soubrane. After a long experience of over 600 laparoscopic hepatectomies, this posterior to anterior pro- cedure has enabled us to perform major hepatectomy more easily, via a standardised technique, for larger tumours, and for anatomical variations of the porta. As the initial step for a right hemihepatectomy the retro- Figure 1. The abdominal rectus muscle fascia after rectus muscle is hepatic tunnel of Belghiti is developed and the caudate removed and facial edges trimmed. process divided as far superiorly as possible. Liver paren- chyma is dissected away from the posterior aspect of the right hepatic inflow from medial to lateral. A small hepa- totomy anterior to the right inflow then allows stapling of the fl FP22-03 right hepatic in ow without blind force or risk of compro- mise to the confluence. Development of the retrohepatic DISTAL SPLENO-RENAL SHUNT FOR tunnel and division of the posterior parenchyma is continued RECONSTRUCTION OF THE GASTRIC as the leading edge of the parenchymal transection, with AND SPLENIC VENOUS DRAINAGE subsequent division of the anterior parenchyma. This DURING PANCREATECTOMY WITH maintains good surgical orientation, especially for tumours RESECTION OF THE PORTAL VENOUS close to the midline or cava, and for the middle hepatic vein. CONFLUENCE This also aids haemostasis by allowing the application of pressure in a postero-anterior direction. After completion of O. Strobel, T. Hackert and M. Buechler the parenchymal transection, the right hepatic vein is Heidelberg University Hospital, Germany divided, followed by the hepatocaval ligament, accessory Background and problem: Extended pancreatectomies hepatic veins, and peritoneal attachments. The same con- with portomesenteric vein resection are increasingly cepts can be applied to left and extended left hepatectomy. performed for locally advanced pancreatic cancer. This technique has enabled expanded indications for a Segmental resections of the portal vein (PV) and superior laparoscopic approach to major hepatectomies as well as a mesenteric vein (SMV) can be easily reconstructed by standardized operative technique. direct re-anastomosis, ensuring venous drainage of all Schematic and case videos will be presented. involved organs. In contrast, resections involving the portal venous confluence can result in gastric and splenic congestion requiring additional venous reconstruction or FP22-05 extension of the resection towards subtotal gastrectomy, RADICAL PANCREATECTOMY WITH total pancreatectomy and splenectomy. The occurrence of left-sided venous congestion after resection of the venous SUPERIOR MESENTERIC VEIN confluence depends on venous anatomy (localization of the RESECTION AND RECONSTRUCTION coronary and inferior mesenteric veins) and can be IN PATIENTS WITH TUMOR LOCATED addressed by reinsertion of the splenic vein (SV) or coro- IN UNCINATE PROCESS INVADING nary vein in the reconstructed portomesenteric axis. How- MESENTERIC ROOT ever, frequently the distance between these veins and the reconstructed portomesenteric axis is wide and reinsertion E. Vicente López, Y. Quijano, R. Caruso, H. Duran, may result in tension and compromised portal venous blood E. Diaz, V. Ferri, B. Ielpo, E. Barzola, C. Plaza and flow. R. Isernia Creation of a distal splenorenal shunt is a technically Sanchinarro University Hospital, Spain straight forward, effective solution for this problem. Pancreatic neoplasm may be deemed locally unresectable Technique and solution: During extended pancreatectomy because they appear to encase the superior mesenteric ves- with resection of the portal venous confluence the SMV, sels. A radical pancreatectomy with venous resection and PV, and SV are clamped and divided. Mesenteric venous reconstruction offer an important benefit in selected patients. drainage is reconstructed by direct end-to-end anastomosis At present, superior mesenteric vein /portal vein resection is of SMV and PV (or by interposed conduit). Stomach and performed in up to 25% of patients in our center. spleen are examined and in case of venous congestion the A variety of venous reconstruction have been described: SV is mobilized and venous drainage of stomach, spleen, Longitudinal or transverse venorrhaphy with a small ellipse and (in partial pancreatoduodenectomies) pancreatic vein excised or Segmental vein resection with or without remnant is reconstructed by creation of a distal spleno-renal splenic vein preservation. shunt with end-to-side anastomosis of the SV to the left A primary end to end suture or interposition graft conduit renal vein, as done in left-sided portal hypertension in (internal yugular vein or renal vein graft) can be performed benign disease. to re-establish the venous continuity. For short segment

HPB 2018, 20 (S2), S182eS294 S272 Free Orals (FP01-FP31) resection (< 3 cm), primary end to end reconstruction and popular among surgeons. Postoperative pancreatic fistula transverse venorrhaphies provide superior outcome to the (POPF) remains the most feared specific complication after other alternatives. For longer segmental resections, inter- LPD. Blumgart Anastomosis (BA) has been established as position grafting with an aprópiate size match achieve a one of the safest anastomosis for reconstruction in PD, with greater long-term patency. low rates of POPF and postoperative complications. The more important surgical challenge is always in Adaptation of the BA to the laparoscopic approach seems a tumors located in uncinate process with mesenteric root good way for increasing safety in LPD. Objectives:To invasión. In these cases, after pancreatic and vascular describe our technique of LPD with laparoscopic-adapted resection, a patch venoplasty with 2, 3 or 4 native veins can BA (Lap-BA) and present the results obtained. be performed with or without interposition grafts. Description: Through an additional 5-mm trocar inserted in We discuss the different surgical steps used to restore vein the epigastrium, in front of the planned pancreatic anasto- continuity in patients with this complex tumor location. mosis, transpancreatic stitches are externalized. Applying traction, the jejunum fits in behind the posterior face of the pancreas. The orifice in the jejunum is done in the most fitted FP22-06 point. The duct-to-mucosa is performed relaxing the exter- SELECTIVE HEPATIC ARTERY nalized sutures, so the posterior face of this anastomosis, the most difficult, is not attached while doing, and this facilitates CLAMPING IN LAPAROSCOPIC its creation. Once the duct-to-mucosa is completed, the HEPATECTOMY: IT PRETTY MUCH externalized threads are reintroduced and the jejunum is WORKS again crossed on its anterior face. When all the stitches are R. Cañada Surjan1, M. Autran Cesar Machado1, tied, the neck of the pancreas is sandwiched into the jejunum. T. de Castro Basseres1 and F. Ferrari Makdissi2 Results: From February 2013 to October 2017, we oper- 1Surgery, University of Sao Paulo Medical School, Brazil, ated 22 patients of Lap-BA. Three patients presented a and 2Gastrointestinal Surgery Department, University of clinically relevant POPF. Severe complications (grades Sao Paulo Medical School, Brazil IIIa) were observed in two patients. No reoperations and deaths were observed. Bleeding control during parenchymal transection in any Conclusions: The LapBA technique we propose can liver surgery is the keystone for technical success. Since the facilitate the pancreatic reconstruction after LPD and beginning of the 20th century, when James Hogarth Pringle decrease the POPF rate and severe complications. described a maneuver in witch the hepatoduodenal liga- ment in clamped in order to interrupt the blood flow through the hepatic artery and the portal vein and thus helping to control bleeding from the liver, the so called FP22-08 Pringle maneuver is virtually the only technique used with SYSTEMATIC BISEGMENTECTOMY 7e this purpose. Although being very effective, it has the 8 PRESERVING HEPATIC VENOUS disadvantage of inducing warm hepatic ischemia and the OUTFLOW OF RIGHT LOBE consequent need of being applied in an intermittent fashion in order not to motivate postoperative hepatic failure. D. S. Kim, H. S. Jo and S. R. Kim Since it has been previously demonstrated that under Division of HBP Surgery & Liver Transplantation, pneumoperitoneum the hepatic portal flow is reduced up to Department of Surgery, Korea University College of 60% with no compensatory hepatic artery flow (hepatic Medicine, Republic of Korea artery buffer effect), we proposed the selective hepatic When large tumor is involving both segment 7 and 8, it artery clamping during laparoscopic hepatectomies. This may be difficult to decide optimal extent of resection. Right clamping technique that can be performed in a continuous hemihepatectomy can be performed easily, but significant manner not resulting in hepatic ischemia and is also very amount of non-tumor bearing liver volume has to be un- effective in reducing blood loss during parenchymal tran- necessarily removed. On the other hand, removing both section. We have performed this technique in several pa- segments will sacrifice right hepatic vein (RHV), which tients submitted to laparoscopic hepatectomies (from non- may potentially cause serious congestion in segment 5 and anatomical wedge resection to right anatomic hepatec- 6. This situation can be especially problematic in patients tomies) and it proved to be an effective and safe technique. with marginal liver function. In case that patient has a significant sizable right inferior hepatic vein (RIHV), systematic bisegmentectomy of FP22-07 segment 7-8 without RHV reconstruction would not cause a significant congestion in remnant segment 5-6. LAPAROSCOPIC-ADAPTED However, in case that there is no good RIHV, recon- BLUMGART struction of RHV is mandatory. For this procedure, author PANCREATICOJEJUNOSTOMY IN prefer direct anastomosis between RHV stump on liver cut- LAPAROSCOPIC surface and that on IVC-side by approximating both ends PANCREATICODUODENECTOMY through folding of segment 5-6 towards caudal direction. This technique has several advantages over technique of I. Poves, O. Morató, F. Burdio and L. Grande RHV reconstruction using artificial graft or cryopreserved Surgery, Universitat Autònoma de Barcelona - Hospital cadaveric iliac vein, which are 1) only one anastomosis is del Mar, Spain required, which is easier and quicker, 2) avoidance of using Introduction: Laparoscopic pancreaticoduodenectomy foreign material, 3) decreased risk of kinking or distortion of (LPD) is a challenging procedure becoming increasingly anastomosis when remnant liver was placed back in position.

HPB 2018, 20 (S2), S182eS294 Free Orals (FP01-FP31) S273

This technique can preserve segment 5 and 6 without guided surgical navigation system of deep learning using compromise in oncologic principle while maximizing mixed reality (MR) and artificial intelligence (AI) for ho- remnant liver volume and function. lographic special surgical navigation in 11 HPB surgeries. We developed a semi-automatic slice-by-slice segmentation method of the organs from CT using cloud-based AI. By FP22-09 reconstructing the patient-specific 3D surface polygons of IMPROVED SUPERIOR MESENTERIC tumors, vessels, and each organ out of the CT data, we registered and overlaid them onto the real world (in the air ARTERY EXPOSURE AND REDUCED space above the surgical field) through a head-mounted BLOOD LOSS RISK USING TEMPORARY holographic smart glasses computer HoloLens in the oper- MESOCAVAL SHUNT FOR LOCALLY ating theater to enhance spatial awareness during actual ADVANCED PANCREATIC CANCER surgery. The HoloLens features an inertial measurement unit which includes an accelerometer and a gyroscope for envi- E. Simoneau, C. Goumard, J. E. Lee, J. -N. Vauthey, ronment understanding sensors, an depth camera, a photo- T. Aloia, Y. S. Chun, C. Conrad, M. P. Kim, M. Katz and graphic video camera, and an ambient light sensor. The C. -W. Tzeng accurate surgical anatomy of size, position, and depth of the MD Anderson Cancer Center, United States tumors, surrounding organs, and vessels during surgeries Introduction: Patients with locally advanced pancreatic could be measured using build-in dual infrared light sensors. cancer (LAPC) represent a challenging group to resect given The gesture controlled manipulation by surgeons’ hands their putative unreconstructible nature of vascular encase- with surgical groves was useful for intraoperative anatom- “ ” ment. In selected patients with anatomy beyond borderline, ical references under sterilized environment. It allowed the temporary mesocaval shunt can facilitate the pancreatectomy several surgeons to share and manipulate the spatial attri- by enhancing exposure to the superior mesenteric vessels. butes of the virtual and real anatomies. This could help Methods: This is a video of a pancreaticoduodenectomy complex surgical procedures including dissection, suturing, (PD) with temporary mesocaval shunt with left internal and hemostats. The total operation time and discussion time jugular (LIJ) vein conduit. were decreased. This has illustrative benefits in surgical Results: A 65 year-old woman presented with LAPC planning, simulation, education, and navigation. originating in the uncinate causing complete occlusion of the superior mesenteric vein (SMV) and encasement of the first jejunal artery. After neoadjuvant therapy, decision was FP23 - Free Papers 23 (mini oral) - Liver: made to perform a PD with temporary mesocaval shunt to divert SMV flow, reduce blood loss and prevent bowel Surgical Outcomes 2 ischemia during the anticipated longer time needed for the FP23-01 retroperitoneal-superior mesenteric artery (SMA) portion ASSOCIATION OF FAMILY HISTORY of the PD. During the procedure, the main mesenteric WITH CANCER RECURRENCE AND fl collateral (lieocolic vein) was divided ush to use as SURVIVAL IN PATIENTS WITH landing zone to sew the shunt to the inferior vena cava with the LIJ interposition. The remaining SMV tributaries HEPATITIS B-RELATED encased by the tumor were divided. The uncinate dissection HEPATOCELLULAR CARCINOMA: A was performed with an SMA-first approach with the SMV PROPENSITY SCORE MATCHING divided and out of the way. Once the resection was ANALYSIS completed, the shunt was stapled from the IVC graft and 1,2 1 1 1 1 J. J. Yu ,Z.L.Li, J. Han , H. Xing , H. Zhang , transposed to the upper SMV for final reconstruction. 1 1 1,3 1 1 H. Wu ,C.Li, W. Y. Lau , F. Shen and T. Yang Standard reconstructions were performed. Estimated blood 1Department of Hepatobiliary Surgery, Eastern Hepato- loss was 250cc. No perioperative complications occurred. biliary Surgery Hospital, Second Military Medical Uni- Conclusion: 2 In LAPC, PD with temporary mesocaval versity, China, Department of Clinical Medicine, Second shunt can facilitate margin-negative resection and trans- Military Medical University, China, and 3Faculty of fusion-free venous reconstruction in patients with other- Medicine, Chinese University of Hong Kong, China wise unreconstructible complete PV/SMV occlusion. Introduction: A family history of liver cancer is regarded as a risk factor for hepatocellular carcinoma (HCC) development. We investigated the association between FP22-10 family history and cancer recurrence and survival in pa- HOLOGRAPHIC MIXED REALITY AND tients with hepatitis B virus (HBV)-related HCC. ARTIFICIAL INTELLIGENCE FOR Methods: Patients who underwent curative resection of SPECIAL NAVIGATION IN IMAGE HBV-related HCC between 2003 and 2013 from a tertiary GUIDED HPB SURGERY hepatobiliary center in China were enrolled in this study. A family history was defined as a self-reported history of 1,2 M. Sugimoto primary liver cancer in the first-degree relatives. Propensity 1 International University of Health Welfare, and score matching (PSM) and multivariable Cox-regression 2 HoloEyes, Japan analyses were performed to compare the overall survival Hepato-pancreato-biliary (HPB) surgery requires a high (OS) and recurrence-free survival (RFS) between patients degree of spatial awareness and orientation because of with and without a family history of liver cancer. complex anatomies, a surgeon’s spatial awareness is diminished. We evaluated the efficacy of our novel image- All authors contributed equally to this work.

HPB 2018, 20 (S2), S182eS294 S274 Free Orals (FP01-FP31)

Results: Of 1,112 patients, 183 patients (16.5%) had a months, the number of patients free of local recurrence family history of liver cancer. A family history was not were similar in R1Vasc (90,8 %) and R0 (86,9%), but not in associated with OS and RFS (P = 0.994 and 0.428) in the R1Par (66,1%). Overall survival after 3 years was similar in entire cohort. Using PSM, 179 pairs of patients with and the R1Vasc (81,2%) and the R0 (92 %), but significantly without a family history but with comparable baseline less in R1Par (51%). Local recurrence was more frequent in characteristics and operative variables were created. A R1Par (27,4%) compared to R0 (10,5%) and R1Vasc family history was associated with decreased OS and RFS (P (14,3%). = 0.042 and 0.006) in the PSM cohort. On multivariable Conclusion: R1 resection of colorectal liver metastases Cox-regression analyses, a family history was significantly next to major intrahepatic vessels has a similar rate of associated with decreased OS (HR: 1.574, 95% CI: 1.171- tumor recurrence and early survival as after R0 resection. 2.116, P = 0.003) and RFS (HR: 1.534, 95% CI: 1.176- In contrast, intraparenchymal R1 shows considerably 2.002, P = 0.002) after adjusting for other prognostic factors. worse results. The presented data sugggest that extension Conclusion: A family history of liver cancer was associ- of the radicality of liver resections in a situation where ated with decreased OS and RFS rates after curative metastases are situated adjacent to major blood vessels resection in patients with HBV-related HCC. does not convey a survival benefit and should not be advocated.

FP23-03 PERIOPERATIVE SPLANCHNIC VASOREGULATION WITH TERLIPRESSIN IN PATIENTS UNDERGOING MAJOR LIVER RESECTION: A RANDOMIZED CONTROLLED TRIAL A. Kohler, S. Perrodin, A. De Gottardi, D. Candinas and G. Beldi Department of Visceral Surgery and Medicine, Inselspital, Bern University Hospital, University of Bern, Switzerland Introduction: Extensive liver resection leads to an increase Figure Comparisons of overall survival (A, in the entire cohort; C, in in portal pressure. This is potentially associated with the propensity score matching cohort) and recurrence-free survival impaired liver regeneration and postoperative complica- tions. Terlipressin, a splanchnic vasoconstrictor, decreases (B,intheentirecohort;D,inthepropensity score matching cohort) portal hypertension and improves renal function in patients curves between patients with and without a family history (FH) of liver with hepatorenal syndrome. Our hypothesis was that peri- cancer. operative administration of terlipressin could prevent postoperative complications and protect kidney function. Method: A randomized double-blind placebo-controlled FP23-02 trial was performed, including patients undergoing elective SIMILAR OUTCOME OF major liver resection. In the intervention group, terlipressin PERIVASCULAR R1- TO R0-, BUT NOT was administered prior to resection and repeated every eight hours for a period of five days. The primary outcome TO PARENCHYMAL R1-RESECTED was the incidence of a composite endpoint, consisting of COLORECTAL LIVER METASTASES the following postoperative complications: liver failure, M. Enger1, E. Wardelmann2, C. Kuhnen3, E. Allemeyer1, ascites, bile leakage, intraabdominal abscess and operative F. Fischer1 and M. W. Hoffmann1 mortality after 30 days. A secondary endpoint was post- 1Allgemein- und Viszeralchirurgie, Raphaelsklinik Muen- operative kidney function. ster, 2Pathology, Gerhard-Domagk-Institut, and 3Pathol- Results: A total of 150 patients were included in the study. ogy, Institut für Pathologie am Clemenshospital, Germany Baseline criteria were similar between the two groups. There Introduction: Colorectal liver metastases are sometimes was no difference in the occurrence of the composite endpoint situated next to major blood vessels. This study aimed to in the intervention group compared to the control group determine how R1 resection next to major blood vessels (22.6% versus 24.1%, p=>0.99, Fisher exact test). In patients compares to R0 or intraparenchymal R1 resection. with preexisting liver disease, kidney function was improved Methods: In a retrospective study, 268 colorectal liver in the postoperative phase if terlipressin was administered. In metastases were analyzed in 85 patients. R0-resected me- this population, the change in glomerular filtration rate (GFR) tastases were compared to R1 next to blood vessels was statistically different after terlipressin administration (R1Vasc) or in the parenchyma (R1Par). Overall survival compared to the control group (delta GFR +2.8 vs. -3.93 ml/ and local recurrence rates were determined. min/1.73m2 respectively, p=0.031, student’s t-test). Results: 26 % of the patients were in the R1Vasc, 33% in Conclusions: Administration of terlipressin does not pre- the R1Par and 41% in the R0 group. Within the R1Vasc vent liver specific complications after major liver resection. group (32 %) and the R1Par (39%) the incidence of mul- However, terlipressin has a protective effect on renal tiple metastases (>3) was higher than in R0 (9%). After 24 function in patients with preexisting liver disease.

HPB 2018, 20 (S2), S182eS294 Free Orals (FP01-FP31) S275

3 FP23-04 Assessment, University of Tokyo, Department of Health Policy and Management, School of Medicine, Keio WHO SHOULD NOT UNDERGO ALPPS University, 4Department of Gastrointestinal Surgery, FOR COLORECTAL LIVER University of Tokyo, 5Division of Gastrointestinal Sur- METASTASES? gery, Department of Surgery, Graduate School of Med- 6 P. Olthof1,2, J. Huiskens3, E. Schadde4,5,6, H. Lang7, icine, Kobe University, and Department of Surgery, ’ M. Malago8, H. Petrowsky9, E. de Santibanes10, Institute of Gastroenterology, Tokyo Women sMedical K. Oldhafer11 and T. van Gulik3 University, Jordan 1Department of Surgery, Academic Medical Center Introduction: In Japan, resection volume for hepatectomy Amsterdam, 2Department of Surgery, Reinier de Graaf is mostly determined according to indocyanine green Gasthuis, 3Academic Medical Center Amsterdam, The retention rate at 15 minutes (ICG15). However, it is un- Netherlands, 4Rush University Medical Center, United clarified how ICG15 is associated with postoperative States, 5Cantonal Hospital Winterthur, 6Center for Inte- complications. grative Human Physiology, University of Zurich, Method: Among 16780 patients registered for National Switzerland, 7Universitaetsmedizin Mainz, Germany, Clinical Database in 2015, 11686 patients with ICG15 8University College London, Royal Free Hospitals, United and information of hepatectomy were included in this Kingdom, 9University Hospital Zurich, Switzerland, study. There were 8156 male and 3530 female patients 10Hospital Italiano de Buenos Aires, Argentina, and with a mean age of 68Æ10.6 years (HCC; 5543, Meta; 11Asklepios Hospital Barmbek, Germany 4001, ICC; 839, other tumors; 1303). The patients were < ALPPS is a two-stage hepatectomy that induces more divided into four groups; group A (ICG15 10%; & < rapid liver-growth compared to conventional strategies. n=5661), group B (10% ICG15 20%; n=4381), group & < Morbidity and mortality are high in cohorts with multiple C(20% ICG15 30%; n=1173) and group D (30% & tumor-types. Outcomes in colorectal liver metastases ICG 15; n=463). Hepatectomy procedures were clas- fi (CRLM) are acceptable, but not in all cases. This report si ed as partial resection (n=3934), systematic subseg- aims to establish a risk score to predict adverse outcomes of mentectomy (n=2055), monosectionectomy (n=2043), ALPPS for patients with CRLM only. bisectionectomy (n=2993) and trisectionectomy (n=208). fi All patients with CRLM included in the ALPPS registry Surgical complications were classi ed using Clavien- fi were included. A risk score was generated for futile ALPPS Dindo classi cation (CD) and were evaluated according defined as death within 90 days after each stage, i.e. one for to ICG15. fi application before stage one, and one for application before Results: Complications over CD-III increased signi - stage two. Logistic regression analysis was performed to cantly as the operation time became longer and the < identify predictors of futility and a risk score was estab- intraoperative bleeding amount increased (p 0.001). lished based on the regression coefficients. As the ICG15 became bigger (group A to D), operative Of 1047 registered patients, 486 had CRLM with suffi- death in systematic subsegmentectomy, mono- and cient data for inclusion. 7% died 90-days after stage one or bisectionectomy, complications over CD-III in mono- two. In the stage-one risk score, age 67 years (OR3.7), and bisectionectomy, intractable ascites in all proced- future remnant-to-body-weight-ratio < 0.40 (OR2.9) and ures, SSI in partial resection and trisectionectomy, and center volume (OR2.4) were included. For the stage-two liver failure in mono- and bisectionectomy increased fi risk score age 67 years (OR3.7), future remnant to body signi cantly. While bile leakage was not associated with weight ratio < 0.40 (OR2.8), bilirubin 5 days after stage ICG15. one >50mmol/L (OR2.4), and stage-one morbidity grade Conclusions: Operation time and intraoperative bleeding fi IIIA (OR6.3) were included. Area under-the-curve amount are signi cantly associated with postoperative showed good predictive value for stage-one (0.70(0.62- severe complications. ICG15 is a good indicator for 0.79)) and two scores (0.70(0.62-0.79)). predicting operative death, complications over CD-III, The CRLM risk score for futile ALPPS demonstrates that intractable ascites, SSI and liver failure. ALPPS is especially associated with poor outcomes in older patients with small remnant livers in inexperienced centers, especially following morbidity after stage-one. The risk score may be used to restrict ALPPS to low-risk patients.

FP23-05 COMPLICATIONS AFTER HEPATIC RESECTION ACCORDING TO LIVER FUNCTION IN TERMS OF ICG15 USING NATIONAL CLINICAL DATABASE DATA K. Kubota1, T. Aoki1, H. Kumamaru2, T. Shiraki1, H. Miyata3, Y. Seto4, Y. Kakeji5 and M. Yamamoto6 Figure [Graph] 1Second Department of Surgery, Dokkyo Medical University, 2Department of Healthcare Quality

HPB 2018, 20 (S2), S182eS294 S276 Free Orals (FP01-FP31)

FP23-06 Introduction: Multidrug-resistant (MDR) bacteria are an emerging problem, however their influence on post- PATIENT-SPECIFIC PREDICTORS OF operative outcome and their correlation with preoperative FAILURE TO RESCUE AFTER MAJOR biliary drainage are still not clear. We sought to evaluate the HEPATECTOMY correlation between intraoperative MDR bacteria in bile fl E. Gleeson1, W. Bowne1 and H. Pitt2 and its in uence on postoperative short-term outcomes of 1Surgery, Drexel University College of Medicine, and patients who underwent bilio-enteric anastomosis after 2Surgery, Lewis Katz School of Medicine at Temple curative resection for hepato-pancreato-biliary University malignancies. Methods: One hundred and nineteen consecutive patients Introduction: Failure to rescue is a recently described between January 2013 and December 2016 were included outcome metric for quality of care. However, the predictors in the study. Intraoperative bile, postoperative abdominal of failure to rescue have not been adequately investigated, drains and blood cultures were prospectively submitted to particularly after major hepatectomy. The aim of this study microbiological examination. Results of microbiological is to identify predictors of failure to rescue for patients after analyses were classified as negative, positive with major hepatectomy. not-MDR bacteria and positive with MDR bacteria Methods: Failure to rescue was defined as those who died (MDR+). after developing a serious complication including post- Results: Among 119 patients included in the study, 68 hepatectomy liver failure (PHLF) and bile leak. All patients patients (72%) required a preoperative biliary drainage who developed a serious morbidity after major hepatectomy, (PBD). Microbiological cultures from intraoperative bile in defined by CPT codes 47122, 47125 and 47130, from 2014- PBD and Not-PBD patients were MDR+ in 29 (43%) and 1 2015 in the Hepatectomy Procedure-targeted ACS-NSQIP (5%) cases, positive not-MDR in 34 (51%) and 3 (16%) database were included. Those who underwent concurrent cases and negative in 4 (6%) and 15 (79%) cases, respec- biliary-enteric anastomosis or a major colectomy were tively (P< 0.001). Intraoperative bile cultures of MDR+ excluded. Univariable analysis was employed to identify correlates with more Clavien-Dindo >2 rates (23% vs 11% preoperative risk factors for 30-day mortality within the vs 0%) (P=0.046), more abdominal fluid collections rates group. Logistic regression was used to determine the factors (33% vs 16% vs 5%) (P=0.043) and longer median length- independently significant for failure to rescue. of-stay (17 (range 12-26) vs 15 (range 10-19) vs 14 (range Results: From 2014-2015, 2,041 major hepatecomies were 11-17) (P=0.030) when compared to positive not-MDR or performed. The 30-day failure to rescue rate was 7.1% for negative cultures. patients who developed serious morbidity after major Conclusion: MDR bacteria in intraoperative bile cultures hepatectomy (n=538). Seven patient-specific factors were are an emerging problem that can significantly worsen significant for failure to rescue on univariable analysis: postoperative course after bilio-enteric anastomosis for male sex, current smoker within one year, diabetes, hy- curative resection of hepato-pancreato-biliary malignancies. pertension, height, preoperative creatinine, and preopera- tive INR. After multivariable logistic regression three independent risk factors remained: preoperative INR, pre- operative creatinine, and current smoker within one year. FP23-06 Table [Significant risk factors after logistic regression] Risk Factor Beta Coefficients p value Odds Ratio (95% CI) Preoperative INR 2.33 0.023 10.29 (1.34-76.84) Preoperative creatinine 1.31 0.001 3.70 (1.75-7.81) Smoker within one year 1.10 0.007 2.99 (1.34-6.66)

Conclusions: Failure to rescue rates have previously been FP24-02 shown to be associated with hospital factors. Failure to rescue rates are also associated with potentially-modifiable LONG TERM FOLLOW-UP AFTER patient-specific factors, including preoperative INR, pre- SURGICAL TREATMENT OF operative creatinine, and current smoker within one year. INTRAHEPATIC BILIARY CYSTS (TYPES V - CAROLI’S DISEASE) IN FP24 - Free Papers 24 (mini oral) - Biliary: ARGENTINA Miscellaneous 1 G. Raffin1, J. Lendoire1, J. Grondona2, R. Russi3, FP24-01 R. Oddi4, O. Gil5, P. Sisco6, E. de Santibañes7, IMPACT OF MULTIDRUG-RESISTANT L. McCormak8 and O. Imventarza1 BACTERIA IN INTRAOPERATIVE BILE 1Liver and Transplant Unit, Hospital Dr Cosme Argerich, CULTURES ON SHORT-TERM 2UNACIR HPB, 3Serie Personal, 4CEMIC, 5Hospital Allende, 6Hospital Pirovano, 7Hospital Italiano de Buenos OUTCOMES AFTER HEPATO- 8 PANCREATO-BILIARY SURGERY Aires, and Hospital Aleman of Buenos Aires, Argentina Background: Caroli’s Disease (CD) is a rare congenital M. Piccino, A. Ruzzenente, T. Campagnaro, S. Conci, disorder characterized by cystic dilatation of the intra- A. Sartori, A. Dore, G. Di Filippo, A. Guglielmi and hepatic bile ducts. C. Iacono Objetive: Analyze the results of surgical treatment and Surgery, University of Verona, General and Hepatobiliary long term outcome of CD in a multicenter study. Surgery Unit, Italy

HPB 2018, 20 (S2), S182eS294 Free Orals (FP01-FP31) S277

Methods: Between 1991 and 2017, 55 patients with CD with multiple tumors were 75.1%, 40.4% and 25.6% were operated on at 13 HPB centers. Diagnosis was respectively. In multivariable analysis, positive resection performed by US, CT, ERCP and MRI. In all patients the margin (Hazard Ratio [HR]:1.57; p=0.001), lymph node diagnosis of CD was confirmed by histology. De- metastasis (HR:1.84; p< 0.001), invasion of adjacent mographic, clinical, surgical and pathological variables organs (HR:2.02; p< 0.001), and tumor size (per centimeter were analyzed. increase) (HR:1.06, p< 0.001) were each associated with Results: 32 patients were female, average age was 49,22 an increased hazard of death. In contrast, presence of years. 85,5% symptomatic. Chronic cholangitis was the multiple tumors was not a prognostic factor of survival main presented symptom (38,2%). The left hemiliver was (HR:0.94, 95%CI: 0.87-1.32; p=0.506). involved in 67,3%. Mean interval time from the onset of Conclusion: Five-year survival after complete resection of symptoms was 527,1 days. Surgical procedures performed patients with multiple ICC was 25.6%. Multi-focal ICC were: A) Hemiliver disease: left lateral sectionectomy 19, should not be considered an absolute surgical contra-indi- left hepatectomy (LH) 18 and right hepatectomy (RH) 6 B) cation and may provide survival benefit. Bilateral disease: RH 1, hepaticojejunostomy 5, LH 3 and liver transplant 3 . Vascular control was by ischemic preconditioning 5 (21,7%) and Pringle maneuver 18 (78,3%). Blood transfusion was required in 16,4%. Average hospitalization was 10,4 days. Morbidity was 32,7% (Dindo-Clavien type 1=4; 2=4; 3a=9 and 3b=1). Mortality was 0 %. A median follow-up of 66,6 months. Cholangiocarcinoma was present in two patients (3,6%). Conclusions: CD in Argentina is more common in females with left hemiliver involment. Surgical resection is the best curative option in unilateral disease with long term survival free of symptoms and complications. In cases of bilateral disease hepaticojejunostomy could be an optional treatment prior to liver transplantation.

FP24-03 SURVIVAL AFTER RESECTION OF MULTIPLE TUMOR FOCI OF INTRAHEPATIC CHOLANGIOCARCINOMA S. Buettner1, S. Alexandrescu2, H. P. Marques3, Figure. [Overall survival by lesion number] L. Aldrighetti4, S. K. Maithel5, M. J. Weiss6, G. A. Poultsides7, J. W. Marsh8, T. M. Pawlik9, B. Groot Koerkamp1 and International ICC Study FP24-04 Consortium PREOPERATIVE MANAGEMENT 1Surgery, Erasmus MC, University Medical Center 2 PROTOCOL FOR PERIHILAR Rotterdam, The Netherlands, Fundeni Clinical Institute, e Romania, 3Curry Cabral Hospital, Portugal, 4Ospedale CHOLANGIOCARCINOMA A SURVEY San Raffaele, Italy, 5Emory University Hospital, 6Johns OF 2778 CASES IN JAPAN Hopkins Hospital, 7Stanford University Medical Center,, R. J. Chaudhary1,2, R. Higuchi1, M. Nagino3, 8University of Pittsburgh Medical Center, and 9Ohio State K. Hasegawa4, I. Endo5, T. Wakai6, M. Ohtsuka7, Medical Center, United States M. Unno8, S. Hirano9 and M. Yamamoto1 1 Background: Bilobar lesions and multiple tumor foci of Tokyo Women’s Medical University, Institute of Gastro- 2 intrahepatic cholangiocarcinoma (ICC) are often consid- enterology, Japan, Institute of Liver Transplantation and ered a contra-indication for resection. We sought to define Regenerative Medicine, Medanta-The Medicity, Gurgaon, 3 long-term outcomes in this subset of patients. India, Nagoya University Graduate School of Medicine, 4 5 Methods: 1,023 patients who underwent resection for ICC University of Tokyo, Yokohama City University Grad- 6 between 1990-2015 were identified from 12 major HPB uate School of Medicine, Niigata University Graduate 7 centers. Satellite lesions, small lesions within 1 cm from the School of Medical and Dental Sciences, Chiba University, 8 9 index tumor or within the same segment of a larger lesion, Tohoku University Hospital, and Hokkaido University were not defined as multiple tumors. Hospital, Japan Results: On final pathology, 235 patients (23.0%) had Background: In Japan, strategies in preoperative man- multiple tumors. Patients with multiple tumors had a agement of perihilar cholangiocarcinoma (PHC) have median of 2 tumors (interquartile range [IQR]: 2-3). evolved over the last decade and the operative mortality has Recurrence occurred in 416 (52.8%) patients with a solitary significantly reduced to < 5%. tumor versus 135 (57.4%) patients with multiple tumors Methods: A Questionaire consisting of 20 questions was (p=0.209). Median survival of patients with a solitary sent to 10 institutions in Japan. tumor was 42.6 months (95%CI: 35.9-49.4), while the Results: The median years of experience and case volume median survival of patients with multiple tumors was 26.9 was 18 years (range17-43) and 226(range 100-889). n=8 months (p< 0.001) One-, 3-, and 5-year survival of patients institutions preferred ENBD(Endonasal-biliary-drainage)

HPB 2018, 20 (S2), S182eS294 S278 Free Orals (FP01-FP31) and n=2 EBS(Endoscopic-biliary-stenting) for biliary drainage(BD,p< 0.05). n=9 use bile replacement after external-biliary-drainage. n=4 use synbiotics preoper- atively.The median cutoff value for FRLV(Future-remnant- liver-volume) and S.Bilirubin, at which PVE(Portal-vein- embolization) is done,is < 40% and < 4mg/dl.The median interval between PVE and surgery is 3-4 weeks. To predict remnant liver function, ICGR15(ICG-retention-at-15- minutes,n=8), ICGK-F(ICG-clearance-of-the-future-liver- remnant,n=2),GEC(Galactose-elimination-capacity,n=1), 99mTc-GSA(galactosylated-serum-albumin-scintigraphy, n=2) and LTB(Level-of-total-bilirubin-in-drained-bile,n=2) are used.The nutritional status of the patient is improved with the use of Onodera prognostic score(n=1), immuno- nutrition(n=2) and peroral feeds(n=6). n=5 use Inchinkoto to improve liver function. n=9 use MDCT and direct cholangiography for surgical planning. Conclusions: The strong recommendations are- 1. ENBD-procedure of choice for BD, 2. Unilateral drainage(drainage of the FRL), 3. Bile Replacement should be given,immediately within 2-3 days of external biliary drainage, 4. Bile culture/sensitivity should be performed in all cases and its result should be used as guide for antibiotic usage in the perioperative period, 5. PVE should be performed for FRLV enhancement if Figure 1 - [Surgical Planning in PHC] the FRLV< 40% and when S.Bilirubin< 4mg/dl,and the interval between PVE and surgery should be 3-4weeks, 6. ICGR15(when S.Bilirubin< 2mg/dl)-most reliable test to predict postoperative liver function, 7. MDCT and direct cholangiography can comphre- hensively help in surgical planning in PHC.

FP24-04 Table [Results of Survey] Institution.. Number Period... Chief surgeon.. Type of Use of Indications Cutoff value Functional Use of 1.Nagoya of cases.. Jan Masato Drainage.. synbiotics ofPVE of S.Bilirubin assessment Inchinkoto University 889 1979 - Nagino et al ENBD preoperatively.. (Remnant (mg/dl) for ICGK-F Yes Mar 2017 Yes Volume).... PVE...... <5 <40% or ICGK-F <0.05 2.Tokyo Women’s 343 Jan 1973 - Ryota ENBD Yes <30%, <35% <10 ICGR15, LTB No Medical Dec 2016 Higuchi if High risk University et al 3.Tohoku 299 Feb 1991 - Michiaki ENBD No <40% <3 ICGR15 No University Dec 2016 Unno Hospital et al 4.Chiba 233 Jan 2000 - Masayuki ENBD - first No <40% <10 ICGR15, Yes University Mar 2017 Ohtsuka choice, EBS GEC,99m et al if more than Tc-GSA 2 required 5.Yokohama 231 Jan 1992 - Itaru Endo EBS Yes all major <3 ICGK No City University Dec 2016 et al resections 6.Hokkaido 220 Sep 1998 - Satoshi ENBD No <40% <3 ICGR15 No University Mar 2017 Hirano et al 7.Shizuoka 214 Jan 2000 - Katsuhiko ENBD No <40% <3 ICGR15 Yes Cancer Center Mar 2017 Uesaka et al 8.University 144 Jan 2000 - Kiyoshi EBS- first No <40% if ICGR <5 ICGR15, rate No of Tokyo Mar 2017 Hasegawa choice, 15<10%, of fall of et al ENBD if <50% if S.Bilirubin, evidence of ICGR15 <20% Plt, PT, LTB cholangitis 9.Niigata 105 Jan 2000 - Toshifumi ENBD Yes <40% <5 ICGR15 Yes university Mar 2017 Wakai et al 10.Kyoto 100 Jan 2000 - Shinji ENBD No <30%, <35% <3 ICGR15, Yes University Mar 2017 Uemoto if High risk 99mTc-GSA et al

HPB 2018, 20 (S2), S182eS294 Free Orals (FP01-FP31) S279

FP24-05 Introduction: Biliary tract carcinomas (BTCs) are het- erogeneous tumors in which analysis of genetic alterations SURGERY MAY NOT IMPROVE using next-generation-sequencing (NGS) plays an SURVIVAL FOR NODE POSITIVE increasing role in diagnosis and therapy. However, a sys- INTRAHEPATIC tematic overview of BTC mutation profiles is lacking. We CHOLANGIOCARCINOMA COMPARED performed a systematic review including a weighted over- fi TO CHEMOTHERAPY ALONE view of BTC mutation pro les, to enable the construction of BTC-specific NGS panels. S. Kizy, A. Altman, S. Marmor, J. Y. C. Hui, T. Tuttle, Methods: A systematic search was performed for articles J. W. Denbo and E. Jensen on whole-exome/whole-genome sequencing (WES/WGS) Department of Surgery, University of Minnesota, United and targeted sequencing (TS) of BTCs published between States 2000 and June 2017. Supplementary data were requested Nodal positivity is a strong predictor of poor survival when not available. For each study, the frequency of following resection for intrahepatic cholangiocarcinoma mutated and non-mutated genes was recorded. The (ICC). The objective of this study was to evaluate the weighted mutation-proportion was calculated per gene. impact of surgical resection on survival in patients with These were stratified by tumor localization and NGS lymph node positive (LN+) ICC. technique. We utilized an augmented version of the Surveillance, Results: Out of 1332 studies, 23 studies were included in Epidemiology, and End Results cancer database to identify the final analysis. The most frequently mutated genes for patients with LN+ ICC without distant metastases from intrahepatic cholangiocarcinoma were KRAS, IDH1, 2000-2014. Patients were stratified by treatment: chemo- ARID1A, TP53, MCL1 and TP53, ARID1A, MCL1, therapy alone or resection with/without chemotherapy. PBRM1, MLL3, respectively. For extrahepatic chol- Patients were excluded if they died within 4 months of angiocarcinoma these were MLL3, TGFBR2, TP53, KRAS, diagnosis. Survival was evaluated using Kaplan-Meier and CDKN2A and ALB, BCL3, CDKN2A, TGFBR2, TP53, Cox proportional hazard models. respectively. For gallbladder carcinoma these were TP53, We identified 169 patients who underwent treatment for C5DM1, EYS, ARID2, CSDM3 and TP53, CSDM1, LN+ ICC. Eighty-eight percent underwent surgical resection CDKN2A, MCL1, EYS, respectively. with/without chemotherapy and 12% underwent chemo- Conclusion: The gene mutations in BTC varied between therapy alone. There were no significant differences between different localizations in the biliary tree. These differing cohorts, although the chemotherapy-alone group trended mutational profiles may be the result of differences in the toward higher T stage and tumor size. The median survival for prevalence of molecular subtype for each location. NGS patients who underwent surgical resection was 17 months, results further differed depending on the used NGS which was not different from that of patients treated with technique (WES/WGS or TS), underscoring the need for chemotherapy alone, 16 months (p=.32) A cox-proportional custom-made panels for TS based on WGS/WES results hazard ratio model demonstrated that black race was associated for future studies and the development of diagnostic with worse survival (vs. white; p< .05; Table 1), while surgical tests. resection was not independently associated with survival. Surgical resection for patients with LN+ ICC did not improve survival compared to chemotherapy alone. Rigorous LN evaluation should be performed prior to surgical resection, in order to improve patient selection and ensure receipt of optimal therapy.

FP24-05 Table [Table 1] Characteristic Hazard Ratio 95% Hazard Ratio Confidence Interval p value Age: 18-49, 50-59, 60-69, 70+ Ref, 1.13, 1.88, 1.44, Ref, 0.60-2.11, 1.05-3.36, 0.79-2.63, Ref, 0.71, 0.03, 0.23, Gender: Female, Male Ref, 0.94 Ref, 0.65-1.36 Ref, 0.74 Race: Non-Hispanic White, Black, Other Ref, 2.11, 1.12 Ref, 1.01-4.38, 0.58-2.15 Ref, 0.046, 0.73 Grade: I, II, III, Ref, 1.05, 1.27 Ref, 0.53-2.10, 0.62-2.59 Ref, 0.88, 0.51 T Stage: T1, T2, T3 or T4, Missing Ref, 1.47, 1.76, 1.87 Ref, 0.82-2.65, 0.986-3.12, 0.97-3.62 Ref, 0.20, 0.06, 0.06 Treatment: Surgery +/- Chemo, Chemo Alone Ref, 1.23 Ref, 0.72-2.11 Ref, 0.46

FP25 - Free Papers 25 (mini oral) - Pancreas: FP24-06 Surgical Outcomes 3 (Fistula) MUTATION PROFILING OF BILIARY FP25-01 TRACT CARCINOMA: A SYSTEMATIC PREDICTION OF RISK FACTORS FOR REVIEW AND META-ANALYSIS POSTOPERATIVE PANCREATIC E. Roos1, E. C. Soer2, S. Klompmaker1, FISTULA USING A NOMOGRAM BASED T. M. van Gulik1 and M. J. van de Vijver2,3 ON THE UPDATED DEFINITION 1 2 3 Surgery, Pathology, and Pathology, VU University C. Guo1,2, Y. Shen1,2,3 and Q. Zhang1,2 Medical CenterUniversity Medical Center, Cancer Center 1The Second Affiliated Hospital Zhejiang University Amsterdam, The Netherlands School of Medicine, 2Key Laboratory of Pancreatic

HPB 2018, 20 (S2), S182eS294 S280 Free Orals (FP01-FP31)

Disease of Zhejiang Province, Hangzhou, China, and Outcomes were analyzed according to 2005-ISGPS, 2016- 3Department of Surgery, City of Hope National Medical ISGPS and Clavien-Dindo classifications. Center, United States Results: 522 Pancreatoduodenectomies (PD), 300 Distal Objective: To identify risk factors for postoperative Pancreatectomies (DP), 97 Central Pancreatectomies (CP) pancreatic fistula (POPF) in patients underwent pancreati- and 81 Pancreatic Enucleations (PE) were included. 90-day coduedenectomy (PD) based on the updated definition of mortality rate was 1.4% (2.4%, 0.7%, 0% and 0% following POPF, and to generate a nomogram to predict POPF. PD, DP, CP and PE respectively). Severe postoperative Methods: We retrospectively reviewed 298 patients who complication rates (Clavien-Dindo grades 3 and 4) were underwent PD from March 2012 to October 2017 and 16%, 12%, 23% and 19% in PD, DP, CP and PE respec- redefined their POPF statuses using the updated definition. tively. Compared to 2005-ISGPS, 2016-ISGPS classifica- Independent risk factors for POPF were identified from tion resulted in fewer C-POPF (3.9% vs. 10.6%) and more possible clinicophathological data using univariate and B-POPF (22.7% vs. 16.0%) in the entire population as well multivariate analyses. A predictive nomogram was estab- as in each surgical category (p< 0.001). 2016-ISGPS lished based on the independent risk factors, and was classification focused C-POPF on patients requiring more compared with the existing models. intensive care unit stay (66% vs. 36% in 2005-ISGPS) Results: Five independent risk factors including texture while B-POPF focused on patients requiring more inter- of pancreas (P < 0.001), use of pancreatic duct stent (P ventional radiology or endoscopic procedures (21.9% vs. = 0.007), size of the main pancreatic duct (P =0.007), 3.1% in 2005-ISGPS). tumor venous invasion (P = 0.016), and definitive pa- Conclusion: Postoperative mortality rate following thology (P = 0.014) were identified. The nomogram had pancreatic surgery is low, whereas morbidity rate remains a C-index of 0.805, and was internally validated. The high following all procedures expect for DP. The 2016- nomogram showed better performance than the other ISGPS classification resulted in both reduced C-POPF and two most cited models (C-indexes of 0.728 and 0.735, increased B-POPF rates. Worldwide adoption of this clas- respectively) in the current cohort. The nomogram also sification seems necessary in order to centralize pancreatic worked adequately using the original POPF definition surgery. (C-index of 0.757). In addition, patients can be assigned into low- (less than 10%), intermediate- (10% to 30%), and high-risk (equal or higher than 30%) FP25-03 groups based on the nomogram to facilitate personal- NOMOGRAM FOR PREDICTING ized management. POSTOPERATIVE PANCREATIC Conclusions: We generated a practicable nomogram for prediction of POPF in patients underwent PD using FISTULA (POPF) weighted independent risk factors for POPF based on the Y. You, I. W. Han, D. W. Choi, S. H. Choi, J. S. Heo, updated definition. Further prospective and external studies S. Han, D. J. Park and Y. Ryu are warranted to validate the predictive ability of the Department of Surgery, Samsung Medical Center, Sung- nomogram. kyunkwan University School of Medicine, Republic of Korea Introduction: Previous studies analyzed risk factors for FP25-02 postoperativce pancreatic fistula (POPF) and develop risk 2016 INTERNATIONAL STUDY GROUP prediction tool using scoring system. However, there was FOR PANCREATIC SURGERY (ISGPS) no study to build nomogram based on individual risk. This study evaluated individual risks of POPF and proposed a FISTULA DEFINITION: EFFECT ON nomogram for predicting POPF. POSTOPERATIVE PANCREATIC Methods: From 2007 to 2016, the medical records of 1771 FISTULA (POPF) DISTRIBUTION IN A patients at Samsung Medical Center undergoing pancreati- CONTEMPORARY COHORT OF 1000 coduodenctomy were reviewed retrospectively. Variables < PATIENTS with p 0.05 in the multivariate logistic regression analysis were included in the nomogram. Internal performance vali- H. Hermand, O. Benoit, N. Tabchouri, S. Dokmak, dation was executed using a 5-fold cross validation method. B. Aussilhou, F. Cauchy, S. Gaujoux, O. Soubrane and Results: The rate of POPF was 12.5%. On multivariate A. Sauvanet stepwise logistic regression analysis, body mass index Beaujon Hospital, Assistance Publique Hôpitaux de Paris, (BMI) University Paris VII, France (p < 0.001), preoperative level of albumin (p = 0.035), Introduction: Due to high postoperative morbidity pancreatic duct size (p = 0.002), sex (p = 0.004), ASA following pancreatectomies, consensual definitions of (American Society of Anesthesiologists) score (p = 0.039), complications are essential. The 2016 ISGPS update of and location of tumor (p < 0.001) were identified as in- POPF definition limited grade C to patients undergoing dependent predictors for POPF. A POPF nomogram was reoperation and/or developing organ-failure. This study based on these 6 variables. The area under the curve (AUC) aimed to analyze the effect of the modified classification on estimated from the receiver operating characteristic (ROC) a cohort of 1000 patients undergoing pancreatectomy graph was 0.709 in the train set and 0.652 the test set. procedures. Conclusions: The Samsung Medical Center POPF nomo- Methods: Data from all patients who underwent partial gram was developed to predict the POPF. This nomogram pancreatectomy procedures from 2012 to 2017 were may be useful in selecting patients who need more inten- collected from a prospectively maintained database. sified therapy and establishing effective treatment strategy.

HPB 2018, 20 (S2), S182eS294 Free Orals (FP01-FP31) S281

nostic model incorporating these factors yielded an c-sta- tistic of 0.76. The prognostic nomogram demonstrated a success range from 98-14%. Conclusions: Male sex, higher age and respiratory failure are negative prognostic factors for successful catheter drainage in patients with pancreatic fistula after pancrea- toduodenectomy. Future prospective trials should deter- mine whether early detection and management of pancreatic fistula can prevent respiratory failure and ulti- Figure [SMC POPF nomogram] mately improve clinical outcome.

FP25-04 FP25-05 PREDICTING SUCCESSFUL CATHETER EFFECT OF SARCOPENIC OBESITY ON DRAINAGE IN PATIENTS WITH POSTOPERATIVE PANCREATIC SEVERE PANCREATIC FISTULA AFTER FISTULA AFTER PANCREATODUODENECTOMY: A PANCREATICODUODENECTOMY IN MULTICENTER COHORT STUDY PATIENTS WITH PANCREAS HEAD J. Smits and on behalf of the Dutch Pancreatic Cancer CANCER Group Y. Ryu, I. W. Han, D. W. Choi, S. H. Choi, J. S. Heo, UMC Utrecht, The Netherlands Y. H. You, S. Han and D. J. Park Introduction: Catheter drainage appears to be superior to Department of Surgery, Samsung Medical Center, Sung- relaparotomy in the management of postoperative pancre- kyunkwan University School of Medicine, Republic of atic fistula, but is not successful in all patients. The aim of Korea this study is create a prognostic model for successful Introduction: Preoperative nutritional status may reflect fi catheter drainage in patients with pancreatic stula after outcomes after pancreatoduodenectomy (PD) in patients pancreatoduodenectomy. with pancreas head cancer (PHC). Recently, several studies Methods: This is multicenter cohort study on patients un- have reported that preoperative sarcopenic obesity (SO), fi dergoing catheter drainage as rst intervention for pancre- which is a high visceral adipose tissue-to-skeletal muscle fi atic stula after pancreatoduodenectomy (January 2005 to ratio, could worsen postoperative complications in patients September 2013) in 9 Dutch Pancreatic Cancer Group with various periampullary diseases. The purpose of this centers. Possible prognostic factors for successful catheter study is to evaluate the effect of preoperative SO on POPF drainage (i.e. survival without relaparotomy) were evalu- following PD. ated using logistic regression and selected using the Akaike Method: Preoperative SO was assessed in 548 consecutive Information Criterion. The model was internally validated patients undergoing PD for PDAC at Samsung Medical fi and de nitive predictors were combined in a nomogram. Center between 2007 and 2016. The visceral adipose Results: Of 2196 patients undergoing pancreatoduode- tissue-to-skeletal muscle ratio was calculated from cross- fi nectomy, catheter drainage was the rst intervention for sectional visceral fat and muscle area on preoperative CT fi pancreatic stula in 227 patients. Primary catheter drainage imaging at the third lumbar vertebra level and normalized was successful in 175/227 patients (77%). Multivariable for height by an automatic calculation program. Overall logistic regression revealed the following negative prog- survival(OS) and the rate of POPF with ISGPF grade B or nostic factors for success: male sex (odds ratio [OR] 0.46, C among postoperative complications were extracted from fi 95% con dence interval [CI] 0.21-1.00, P=0.049), higher prospectively maintained databases. age (for every 5 years over 50; OR0.69 95%CI0.57-0.84; Results: Preoperative SO was present in 202 (36.9%) of the < P 0.001) and respiratory failure in 24h before catheter patients. After multivariate analysis, the presence of SO was < drainage (OR0.10, 95%CI0.03-0.33, P 0.001). A prog- the only independent risk factor for developing POPF.

FP25-05 Table [Risk factor analysis for CR-POPF] Variables Univariate Multivariate HR 95% CI P HR 95% CI P Bwt 1.030 0.999 - 1.063 0.059 0.997 0.943 - 1.053 0.997 BMI 1.106 1.003 - 1.221 0.044 1.066 0.889 - 1.279 0.489 Cardiac disease 2.334 1.189 - 4.582 0.012 2.017 0.949 - 4.286 0.068 CA19-9 1.000 1.000 - 1.000 0.090 1.000 1.000 - 1.000 0.149 PV invasion 0.289 0.087 - 0.957 0.031 0.326 0.088 - 1.200 0.092 P duct diameter < 4mm 2.235 1.079-4.628 0.027 2.238 0.992 - 5.050 0.052 Visceral obesity 2.836 1.432 - 5.616 0.002 0.962 0.249 - 3.714 0.956 Sarcopenic obesity 3.224 1.627 - 6.389 <0.001 2.561 1.179 - 5.564 0.018

HPB 2018, 20 (S2), S182eS294 S282 Free Orals (FP01-FP31)

Age over 63 years, poorly differentiated carcinoma, Results: Of the 3430 patients that were identified, 16.4% nodal metastasis, portal vein invasion and absence of were managed using GA drainage. On univariate analysis, a adjuvant treatment were identified as independent risk greater proportion of patients with CS drains experienced factors for OS, but preoperative SO was not significantly delayed gastric emptying (17.5% vs.13.7%, p=0.029), all associated with decreased OS. surgical site infections (SSI) (22.9% vs. 17.2%, p=0.003) and organ space SSI (16.5% vs. 12.4%, p=0.0017). Following CEM, 206 patients were matched with equal numbers in each drain group. There was no difference in POPF (p=0.61), SSI (p=0.49), or organ-space SSI (p=0.70) between the groups. Conclusion: There is no difference in the risk of POPF or SSI between patients managed with CS versus GA drains following PD.

FP26 - Free Papers 26 (mini oral) - Transplant: Living Donor Transplantation/Basic Science FP26-01 Figure [Risk factor analysis for overall survival] FEASIBILITY OF TOTAL LAPAROSCOPIC LIVING DONOR Conclusions: Preoperative CT-derived SO is the only predictive factor for CR-POPF after PD in patients with RIGHT HEPATECTOMY COMPARED PHC. Preoperative SO measures may stratify patients into WITH OPEN SURGERY: risk categories for developing POPF. For evaluation of the COMPREHENSIVE REVIEW OF 100 effect of SO on survival after PD, more observational LAPAROSCOPIC CASES studies will be needed. J. Rhu, C. H. Kwon, G. S. Choi, J. M. Kim and J. -W. Joh Samsung Medical Center, Sungkyunkwan University School of Medicine, Republic of Korea FP25-06 Background: We designed this study to analyze the DOES ACTIVE VERSUS PASSIVE feasibility of laparoscopic living donor hepatectomy DRAINAGE ALTER THE RISK OF compared to open surgery. POSTOPERATIVE PANCREATIC Methods: Donors who underwent living donor right or FISTULA (POPF) FOLLOWING extended right hemihepatectomy by laparoscopy or open PANCREATICODUODENECTOMY surgery from May 2013 to October 2017 were included. Comparisons between laparoscopy and open surgery were (PD)? performed. M. Lemke1, F. K. Balaa2, G. Martel2, J. Abou Khalil2 and Results: During the study period, 305 patients underwent K. A. Bertens2 living donor right or extended right hemihepatectomy. Of 1School of Medicine, Queen’s University, and 2Division of these, 100 underwent laparoscopy and 205 underwent open General Surgery, University of Ottawa, Canada surgery. The laparoscopy group (30.9Æ11.2 years) were Introduction: POPF remains a serious and highly significantly younger than the open group (34.5Æ12.3 morbid complication of pancreatic surgery. Prophylactic years, P=0.014). The laparoscopy group mostly had type 1 drainage following pancreatic surgery has been (95.0%) bile duct and 81% had single duct in graft livers, frequently employed to improve early detection of PF; compared with 59.5% type 1 bile duct and 59.5% single however, little evidence exists on whether closed-suction bile duct in the open group. The laparoscopy group had (CS) drains or gravity-assisted (GA) drains result in su- significantly longer operation time (378.2 Æ 93.5 minutes perior outcomes. vs. 329.1 Æ 68.0 minutes, P< 0.001) and warm ischemic Methods: Patients who underwent PD were selected time (median 271 minutes vs. 151 minutes, P< 0.001). using the 2016 ACS-NSQIP pancreas-specific PUF. However, estimated blood loss was smaller in the lapa- Analysis was restricted to patients with a surgical drain roscopy group (298.3 Æ 162.9 mL vs. 344.3 Æ 149.9 mL, placed. Patient characteristics, intraoperative details and P=0.015). There was no difference in complication rate postoperative complications were compared using Stu- (laparoscopy group 22.0% vs. open group 15.6%, P=0.170) dent’s T-tests, Wilcoxon rank sum test, and Chi square and the severity of complications classified by Clavien- test where appropriate. Patients with CS and GA drains Dindo system did not differ significantly between the were matched using coarsened exact matching (CEM) on groups (P=0.094). a priori selected variables (age, gender, BMI, ASA, race, Conclusions: When living donors are selected cautiously, wound classification, preoperative weight loss) and laparoscopic living donor hepatectomy can be performed variables with a p value < 0.1 on univariate analysis safely with similar outcome to open surgery. However, (diabetes, smoking, dyspnea, use of a wound protector, the procedure should be performed by a surgeon experi- pylorus preservation, reconstruction by pancreatojeju- enced in both liver transplantation and minimally invasive nostomy, surgical indication, gland texture, duct size, and surgery. vascular resection.)

HPB 2018, 20 (S2), S182eS294 Free Orals (FP01-FP31) S283

3 FP26-02 in grade /4 encephalopathy at presentation and 70% on ventilator preoperatively.Pre operative culture was positive PORTAL VEIN COMPLICATION AND in 41%.The median MELD was 36(32-40), admission to TECHNICAL REFINEMENT AFTER transplant time was 32.5 hours. The 1 year post-LT survival LIVE DONOR RIGHT HEPATECTOMY was 64 %. The survival was 21 %(29/133) in group who met ’ D. -H. Jung1, S. -G. Lee1, S. Hwang1, K. -H. Kim2, KCH but didn t undergo LT. On univariate analysis grade C. -S. Ahn1, D. -B. Moon1, T. -Y. Ha1, G. -W. Song1 and III-IV HE, SIRS, serum Na, need of intubation, culture G. -C. Park1 positivity, severe cerebral edema and on multivariate anal- 1Ulsan University and Asan Medical Center, Republic of ysis post-operative severe cerebral edema & SIRS were fi Korea, and 2Asan Medical Center, University of Ulsan signi cant factor predicting early mortality. There was no College of Medicine, Republic of Korea donor mortality. Major complication was seen in one donor in form of wound infection requiring secondary suturing. Background: This study intended to assess incidence, risk Conclusion: Emergency LDLT is a feasible in selected factors and treatment of PV complication after living donor patient with ALF with good donor safety and acceptable right hepatectomy (LDRH). recipient survival.Window for transplantation is narrow in Methods: This study analyzed 2979 cases of LDRH from ALF so timely intervention before severe cerebral edema July 1997 to December 2014 at Asan Medical Center and/or MODS developed can further improve outcome. regarding on PV complication. Results: Male and female were 2055 (69.0%) and 924 (31.0%), respectively. Mean donor age was 27.5 Æ 8.1 years old. Mean body mass index was 22.70 Æ 2.70. Type FP26-04 1, 2, 3, and other PV anomalies were 2727 (91.6%), 113 PREPRO TRIAL: RANDOMIZED (3.8%), 132 (4.4%), and 6 (0.2%), respectively. PV stenosis DOUBLE-BLIND PLACEBO (>50% narrowing of PV diameter) occurred 47 cases CONTROLLED TRIAL TO ANALYZE (1.5%). PV reconstruction (odds ratio 7.949; p=0.012), THE EFFECT OF SYNBIOTICS ON anomalous PV anatomy (OR 4.536; p< 0.001), acute angulation between main and Left PV (60-90 OR 2.214; INFECTIOUS COMPLICATIONS p=0.041, < 60 OR 7.690; p< 0.001), and no fixation of FOLLOWING LIVING DONOR LIVER falciform ligament (OR 2.213; p=0.010) were significant TRANSPLANTION. [CTRI NO. - CTRI/ risk factors for PV stenosis. Among 47 PV stenosis donors, 2017/09/009869] PV stent insertion was performed in 9 cases (0.3%) which S. Mallick, M. Kathirvel, M. Thillai, P. Sethi, occurred 1 in type 1 (0.1%), 2 in type 2 (1.8%), 6 in type 3 M. S. Durairaj, K. Nair, R. N. Menon, D. Balakrishnan, (4.5%), and 0 in other types (P< 0.001). All PV compli- U. Gopalakrishnan and S. Sudhindran cation donors had no long-term sequelae. Dept. of GI Surgery and Solid Organ Transplant, Amrita Conclusions: PV reconstruction and no fixation of falciform Institute of Medical Sciences, India ligament should be avoided to prevent PV complication during LDRH. Because donors with anomalous PV anatomy Background: Synbiotics, a combination of pro and pre or acute angulation between main and left PV have a higher biotics, is thought to prevent bacterial translocation, stim- tendency to occur PV complication after LDRH, those ulate epithelial growth, promote gut motility and enhance donors require meticulous surgical techniques during oper- innate immunity. Although, synbiotics have a role in ation and periodic image studies after operation. decreasing post-operative infections, their use in LDLT is still not recommended. Primary objective of this trial was to compare infectious complications occurring in the first 2 weeks following LDLT. FP26-03 Methods: This is a double-blind randomized placebo LIVE DONOR LIVER TRANSPLANT FOR controlled trial conducted between August 2016 to ACUTE LIVER FAILURE November 2017 at Amrita Institute of Medical sciences, A. Vagadiya, P. Sinha, S. Sasturkar, A. Choudhury, India. Total of 80 consecutive elective patients undergoing LDLT were randomized (1:1) to receive either capsules V. Pamecha and Acute Liver Failure Ò HPB Surgery & Liver Transplantation, Institute of Liver of Prowel [Lactobacillus Acidophilus-Bifidobacterium and Biliary Sciences, India (Probiotic) and Fructooligosacccharide (Prebiotic)] (Prepro group) or empty capsules (Placebo group), Introduction: Emergency liver transplant is life saving in th selected patients with ALF. In parts of the world where starting 2 days prior to surgery and continued till 14 post- deceased donation is sparse, emergency LDLT is the transplant day. dominant option. Results: Baseline characteristics like age, etiology, co- Method: Profile and outcomes of ALF undergoing LDLT morbidities, MELD score, blood loss, operative duration, between “March 2010 to October 2016” were analyzed. cold ischemia times and GRWR were comparable in both Selection criteria for LT was Kings College criteria (KCC) groups. Overall infectious complications (25% vs 47.5%; [ fi Results: Of the 301 patients, 188(62%) fulfilled KCC and p 0.036) and speci cally blood borne infections (10% vs [ fi referred for transplantation. 55 (29%) of these (33 male) 37.5%; p 0.004) were signi cantly lower in the Prepro underwent LDLT, the remaining (n-133,71 %) includes group compared to the Placebo. The mean ICU stay (8.1 vs those who didn’t undergo LT due to various reasons. The 9.2 days, p=0.41), mean hospital stay (18.7 vs 18.9 days, median age was 25.5 years(10.5-37).Indeterminate (29%) p=0.45) overall 30-day mortality (10%vs 15%, p=0.4) or fi was predominant etiology. The median jaundice to en- sepsis speci c 30-day mortality (2.5% vs 10%, p=0.16) was cephalopathy time was 15(9-28) days. Majority(70%) were similar in both groups (Prepro vs Placebo).

HPB 2018, 20 (S2), S182eS294 S284 Free Orals (FP01-FP31)

Table [Study end points] Background: The efficacy of transplanting chimeric Primary End-Points PREPRO P1.ACFBO organs lacks of any evidence, even in preclinical studies involving small animals. We challenged this idea by OVERALL Infectious No 30(75%) 21 (52.5%) transplanting mouse-rat chimeric livers into baby rats. complications [ Yes 10(25%) 19(47.5%) p 0.036 Method: Chimeric livers were created by transplanting Subgroup analysis Lewis rat hepatocytes into FRGÒ mice (C57Bl/6 Fah-/-/ Blood stream No 36(90%) 25 (62.5%) p-0.004 Rag2-/-/Il2rg-/-). These grafts were transplanted into newly infection Yes 4(10%) 15(37.5%) weaned Lewis rats (45Æ3g) without or with immunosup- Urinary infection No 37(92.5%) 37(92.5%) p» 1.000 pression (IS) (Tacrolimus 0.6mg/kg/day, for 56 or 112 days). Yes 3 (7.5%) 3 (7.5%) Rejection was assessed by graft biopsies and immune cells Intra-abdominal No 33(82.5%) 30(75%) p= 0.058 activation. Rat and mouse albumin was measured weekly. infection Yes 7(17.5%) 10(25%) Wild type C57Bl/6 mice were used as control donors. Results: All non-immunosuppressed recipients experi- Conclusion: Use of Synbiotics in the peri-operative period enced acute rejection and died from 8 to 11 days after fi signi cantly reduces overall as well as blood borne infec- transplantation. tious complications following LDLT. Under tacrolimus all chimeric liver recipients survived having normal development and weight gain. Banff score FP26-05 was 2-3, rejection being mostly driven by endothelium and LIVING DONOR LIVER cholangiocytes. Rat albumin production was within phys- TRANSPLANTATION FOR ALCOHOLIC iologic ranges. Chimeric livers grew on average 670% (from 1.3Æ0.2g to 8.7Æ1.4g), in line with syngeneic rat LIVER DISEASE; DOES ABSTINENCE controls. CORRELATE WITH RECIDIVISM? By contrast, xenogeneic controls showed impaired B. Pattnaik, P. Sinha, S. Sasturkar, N. Mohapatra and growth and died before or shortly after the immunosup- V. Pamecha pression was stopped, in the 56-day IS group (p=0.0014). HPB Surgery and Liver Transplantation, Institute of Liver In the 112-day IS group they all died while still on and Biliary Sciences, India Tacrolimus (p=0.0013). Banff score was never inferior to 6. Introduction: The traditional “6 months rule” for ALD is Four months after implantation into rats, chimeric livers becoming debatable with recent evidences showing com- were partially repopulated by portal endothelial cells and parable outcomes without a pre-fixed abstinence period. some rat hepatocytes had undergone complete trans- Method: Abstinence of minimum 3 months being the differentiation into cholangiocytes. standard, was made flexible in selected cases where pa- Conclusion: We show robust survival of orthotopically tients’ clinical condition with good social support merited transplanted chimeric livers. The transplanted organs not early Liver Transplant. Outcomes of all spectrums of ALDs only sustained normal animal growth, but also underwent undergoing LT between “June 2011 to September 2017" profound recipient-oriented remodelling. were analyzed. Recidivism was defined as any alcohol intake after LT and was compared between three groups (A up to 3 months, B 4 to 6 months and C > 6 months). FP27 - Free Papers 27 (mini oral) - Pancreas: Results: Out of total 159 ALD patients, 148 underwent Surgical Outcomes 4 Living donor LT ad 11 had diseased donor LT. The mean FP27-01 MELD Na was 24 and 65 patients (40.8%) had score > 25. EFFECT OF LENGTH OF PROXIMAL Amongst living donors, 97(65%) were female family mem- JEJUNAL RESECTION ON THE bers. The median abstinence period was 8 months (range 1 to 84) with 64(43.2%) patients had < 6 months and 23(15.5%) INCIDENCE OF DELAYED GASTRIC had < 3 months of sobriety. Out of 148 patients, 9 (6.08%) EMPTYING FOLLOWING PYLORUS had recidivism, out of which 5(55.5%) died within 6 months. RESECTING The recidivism rates were 2(8.7%) in Group A, 3(7.3%) in B PANCREATICODUODENECTOMY and 4(4.8%) in C (p value 0.72).The overall survival was (PRPD), A RANDOMISED CONTROLLED significantly lesser in the recidivism group compared to those who didn’t resume to alcohol (44.5% vs 89.8%, p = 0.004). TRIAL Conclusion: Abstinence didn’t predict recidivism, provided A. Chowdhury1, V. Gupta1, T. D. Yadav1, a promising social support along with strong commitment S. K. Sinha2 and R. Kochhar2 from patient. Specifically in LDLT settings, abstinence can 1General Surgery, and 2Gastroenterology be made more flexible to benefit patients with high MELD. Aim: To determine the role of length of proximal jejunal resection in the occurrence of Delayed Gastric Emptying FP26-06 following Pylorus resecting Pancreaticoduodenectomy (PrPD). GRAFT REMODELLING AFTER Methods: Of the 137 patients enrolled, 97 patients were TRANSPLANTATION OF CHIMERIC randomised to long segment (30cm) (n=46) and short MOUSE/RAT LIVERS INTO RATS segment(15cm)(n=51) proximal jejunal resection. Delayed G. Oldani1, A. Peloso2, S. Vijgen1, L. Orci1, V. Delaune1, Gastric Emptying (DGE) was defined as per ISPGS L. Rubbia-Brandt1, S. Lacotte3 and C. Toso1 guidelines. Operative time, removal of NG tube, resump- 1Geneva University Hospitals, 2Surgery, Geneva Univer- tion of oral feed, morbidity, mortality and overall hospital sity Hospitals, and 3Geneva University, Switzerland stay was compared.

HPB 2018, 20 (S2), S182eS294 Free Orals (FP01-FP31) S285

Results: Both the groups were comparable in terms of Introduction: In contrast with pancreatoduodenectomy, demographic and operative data. The overall incidence of the benefit of early drain removal (EDR) for distal delayed gastric emptying in this study was 58.8% with long pancreatectomy (DP) is debated. This study explores the segment 47.8% and short segment 68.6% (p=0.010).Inci- results of EDR following DP in light of the current evi- dence of Clinically relevant delayed gastric emptying was dence on clinically-relevant fistula (CR-POPF) prediction. lesser in the long segment group (21.7% vs 54.9%, Method: Outcomes of DPs performed at two institutions P=0.001). The incidence of Primary delayed gastric (2013-2017) were compared between de facto “early” emptying was significantly lower in long segment group (POD5) and “late” (>POD5) drain removal groups. (26.1% vs. 31.3%, P=0.029) and there was a trend towards Multivariable regression and propensity-score matching fewer incidence of secondary delayed gastric emptying were adjusted for previously identified CR-POPF risk (21.7% vs. 37.2%, P=0.096). Patients in the long segment factors after DP [age, BMI, albumin, pathology, vascular group had early removal of NG tube (5.13Æ4.39 vs. resection, splenectomy, and postoperative-day (POD) 1 8.28Æ5.59 days, P=0.046) and early resumption of solid drain fluid amylase (DFA)]. diet (7.6Æ4.51 vs. 10.4Æ5.61 days P=0.011). The post- Results: Drains were removed early in 261 DPs (57.2%). operative hospital stay was comparable in both groups. Post-operative outcomes, including CR-POPF, were worse Other post-operative complications were comparable in in the late-removal group (Table). At multivariable anal- both the groups. ysis, POD1 DFA>2000 (OR=1.96, p=0.037) and late drain Conclusion: Resection of 30 cm of proximal jejunum de- removal (OR=7.37, p< 0.001) were the only significant creases the incidence of delayed gastric emptying both predictors of CR-POPF. Propensity-score matching primary and secondary. This translates into early removal confirmed a lower CR-POPF rate in the early-removal of NG tube, and early resumption of oral feed. group (8% vs 41.4%, p< 0.001). Among patients who did not develop CR-POPF, complications (Accordion1: 36.8% vs 71.6%, p< 0.001; Accordion3: 3.8% vs 11%, FP27-02 p=0.012) and duration of stay (9 vs 7 days, p< 0.001) were  EXPLORING THE UTILITY OF EARLY improved in the early-removal group. A POD5 DFA 50 and >5000 had the greatest negative (86.8%) and positive DRAIN REMOVAL FOLLOWING (56.5%) predictive value for CR-POPF, respectively, while DISTAL PANCREATECTOMY: A a cut-off of 100 provided the best CR-POPF overall PROPENSITY SCORE MATCHED prediction. ANALYSIS Conclusions: This study represents the largest examination of drain management for DP, and substantiates EDR as a L. Maggino1,2, T. Seykora1, G. Malleo2, B. Ecker1, “best practice” following DP. POD5 DFA can aid in M. Lee1, R. Roses1, C. Bassi2 and C. Vollmer1 determining which patients will develop CR-POPF, and 1Surgery, University of Pennsylvania, United States, and benefit from longer drain duration. 2Surgery, University of Verona, Italy

FP27-02 TablePatients’ Characteristics and Outcomes of the Early and Late Drain Removal Cohorts - Univariable Analysis Variable Overall Removal £ Removal > p-value Outcomes (N - 456) POD5 (N [ 261) POD5 (N [ 195) Age (median, IQR) 62 (52-70) 64 (53-71) 59 (49-69) p [ 0.020 ASA I–II 62.7% (286) 66.1% (150) 73.9% (136) p = 0.086

III–IV 27.4% (125) 33.9% (77) 26.2% (48) BMI Normal (< 25) 45.6% (208) 45.1% (110) 51% (98) p = 0.463 Overweight (25-29.9) 37.1% (169) 40.6% (99) 36.5% (70) Obese (30) 12.9% (59) 14.3% (35) 12.5% (24) POD1 DFA 1813 (470-5095) 938 (319-2063) 5055 (2061-7500) p< 0.001 (median, IQR) POD1 DFA > 2000 46.8% (182) 26% (59) 75.9% (123) p < 0.001 POD Drain Removal (median, IQR) 5 (3-10) 4 (3-5) 13 (7-22) p < 0.001 Clinically-relevant fistula 20.8% (95) 6.9% (18) 39.5% (77) p < 0.001 Any complication (Accordion 1) 54.4% (248) 41.9% (95) 83.2% (153) p < 0.001 Severe complications (Accordion  3) 15.8% (72) 11.5% (26) 25% (46) p< 0.001 Duration of Stay (median, IQR) 8 (6-11) 7 (6-9) 10 (8-17) p < 0.001 Readmission 12.7% (52) 8.4% (19) 18% (33) p - 0.003 Reoperation 7.1% (29) 2.6% (6) 12.6% (23) p < 0.001 Death 1% (4) 0.4% (1) 1.6% (3) p = 0.222

POD: postoperative day, DFA: drain fluid amylase

HPB 2018, 20 (S2), S182eS294 S286 Free Orals (FP01-FP31)

4 FP27-03 Oncological Centre, Bulgaria, Department of Clinical Sci- ences, University of Barcelona, Spain, 5Department of KOREAN NATION-WIDE MULTI- Research, Netherlands Comprehensive Cancer Organisation INSTITUTIONAL VALIDATION STUDY (IKNL), The Netherlands, 6Pancreatic Surgery, Humanitas OF THE 8TH AMERICAN JOINT Research Hospital, Italy, 7Institute for Medical Information COMMISSION ON CANCER STAGING Processing, Biometry and Epidemiology, Munich Cancer 8 SYSTEM IN RESECTED PANCREATIC Registry, Germany, Epidemiology and Cancer Registry, Institute of Oncology Ljubljana, Slovenia, 9Taras Shev- CANCER chenko National University of Kyiv, Ukraine, and 10Ukrai- 1 2 3 4 5 J. U. Chong ,I.W.Han ,S.J.Park ,H.K.Lee ,Y.H.Kim , nian National Institute of Cancer, Ukraine 6 7 8 1 Y. S. Choi ,Y.-H.Roh,J.-Y.Jang andC.M.Kang The EUropean REgistration of Cancer CAre (EUREC- 1Department of Surgery, Yonsei University College of Medi- 2 CA) consortium aims to investigate differences in treatment cine, Department of Surgery, Sungkyunkwan University and to improve cancer care through Europe. The aim of this School of Medicine, 3Department of Surgery, National 4 study was to compare adjuvant chemotherapy treatment and Cancer Center, Department of Surgery, Ewha Womans outcomes after tumor resection for pancreatic adenocarci- University School of Medicine, 5Department of Surgery, 6 noma stage I&II in the EURECCA Pancreas consortium. Keimyung University Dongsan Medical Center, Depart- The eight collaborating national and regional partners ment of Surgery, Chung-Ang University College of Medi- shared their dataset. Patients diagnosed in 2012-2013 who cine, 7Department of Surgery, Dong-A University College of 8 underwent tumor resection were investigated with respect Medicine, and Department of Surgery, Seoul National to treatment and overall survival and compared using uni- University College of Medicine, Republic of Korea and multivariable logistic regression and Cox regression th Introduction: The objective of this study is to validate 8 models. Belgium was used as reference category due to the AJCC staging system using the KOTUS database. provided number of patients and high quality data. Bulgaria Methods: Between February, 1991 to December 2016, 1562 and Ukraine were excluded from adjuvant chemotherapy patients undergoing operation for pancreatic ductal adeno- treatment and survival analyses, since tumor resection carcinoma were identified in KOTUS database. Among could not be confirmed in their datasets. them, 159 patients with neoadjuvant therapy and 37 patients In total, 5212 patients were included. Stage II disease with R2 resections were excluded for analysis. Harrell’sc- was present in 77.4% of patients. Administration of adju- th index was used to validate 8 AJCC staging system. vant chemotherapy was different between the countries Results: The Harrell’s c-index was 0.591 [95% CI: 0.572- (range: 40.4%-70.0% (P< 0.001), even after adjustment for th 0.612] for AJCC 8 staging system and 0.571 [95% CI: potential confounders. Ninety-day mortality differed th th 0.554-0.591] for AJCC 7 staging system. 8 staging significantly and (range: 0.9%-9.8%, P=0.005). In Kaplan- system was not able to discriminate stage IIA with IIB Meier (Figure 1, stage II) and Cox regression analysis, (median 19 months [95% CI: 11.0-26.9] vs. median 20 patients from Milan (Italy) showed a better overall survival months [95% CI: 17.6-22.4], p=0.945) and III (median 19 (HR=0.663, 95%CI=0.517-0.851, P=0.001) compared to months [95% CI: 11.0-26.9] vs. median 16 months [95% the reference category, where other countries showed no th CI: 14.1-17.9], p=0.128) in disease-specific survival. In 8 differences compared to the reference category. AJCC staging system, patients were more evenly spread among stages, whereas in 7th AJCC staging most patients were IIA (422, 31%) or IIB (859, 63%). In a subgroup analysis, patients were divided according to tumor location (head vs. body/tail) and the survival analysis showed that for both stage I (median 31 months vs. 64 months, p< 0.001) and stage II (median 18 months vs, median 42 months, p< 0.001) of AJCC 8th staging system, left-sided pancreatic cancer showed significantly better survival. Figure 1 . Kaplan-Meier curves of overall survival Conclusions: New AJCC 8th staging system better predicts prognosis. But there are possibilities of further improvements. With results from this study, tumor location should be This study provides a clear insight in clinical practice in considered as another factor in advancing the staging system. different European countries. The differences observed in adjuvant chemotherapy treatment and short- and long-term FP27-04 outcome give us the chance to further investigate the best DIFFERENCES IN TREATMENT AND practices and improve outcome in PC. OUTCOME OF PANCREATIC FP27-05 ADENOCARCINOMA STAGE I & II IN THE HOSPITAAAL: A SCORE TO PREDICT EURECCA PANCREAS CONSORTIUM BLOOD TRANSFUSION IN PATIENTS J. V. Groen1, E. van Eycken2, Z. Valerianova3, 4 5 6 UNDERGOING J. M. Borras , L. van der Geest , G. Capretti , PANCREATICODUODENECTOMY A. Schlesinger-Raab7, M. Primic Zakelj 8, 1 2 1 A. Ryzhov9,10 and J. S. Mieog1 C. Garcia-Ochoa , E. McArthur , M. Tun-Abraham , 1 1 1Surgery, Leiden University Medical Center, The K. Leslie and A. Skaro 1 Netherlands, 2Belgian Cancer Registry, Belgium, 3Depart- General Surgery, London Health Sciences Centre, and 2 ment of Epidemiology and Cancer Control, National Statistics, London Health Sciences Centre

HPB 2018, 20 (S2), S182eS294 Free Orals (FP01-FP31) S287

Introduction: Blood transfusion is associated with worse coarsened-exact matching (CEM) on demographic vari- outcomes and increased resource utilization after pancrea- ables, stage and tumor specific variables as well as intra- ticoduodenectomy. The aim of this study is to develop a operative and technical variables. risk score to predict blood transfusion after Results: Of the 6206 patients who underwent PD, 1585 pancreaticoduodenectomy. (25.5%) had PD following NAC. Patients receiving NAC Methods: American College of Surgeons National Surgical had a higher T stage and were more likely to have under- Quality Improvement Program (ACS-NSQIP) registry from gone a vascular resection (Table1.) Pancreatic duct size did 2014-2016 was used to develop and validate a prediction not differ between the groups but more patients in the NAC model of blood transfusion within 30 days after pancrea- group had a hard gland texture. There were more PFs in the ticoduodenectomy. We used univariable analysis to deter- upfront surgery group compared to the NAC group (15% mine potential predictors and best subsets logistic vs. 8.7%, p< 0.001) In the 966 CEM matched patients, regression was used for variable selection, restricting to there remained more PF in the upfront surgery group eight variables. Model discrimination was assessed using c- compared to the NAC group (14.7% vs. 8.9%, p< 0.001.) statistic, and was adjusted using bootstrapping and esti- At multivariate logistic regression adjusting for t preoper- mation of Harrell’s optimism. We used a loess-based ative radiotherapy, the odds of a PF remained lower in the smoothing algorithm to graphically assess calibration. NAC group (OR=0.61, 95%CI 0.44-0.84, p=0.003). In Results: From 17,463 patients that received pancreatic patient undergoing DP16.2% and 16.8% in the upfront surgery, 11,183 patients underwent a pancreaticoduode- surgery and NAC groups respectively (p=0.78) developed nectomy, and 2,130 (19%) had a blood transfusion. Fifteen PF. pre-operative variables were associated with risk of trans- Conclusion: Patients undergoing PD after NAC experi- fusion and eight were selected for inclusion in the final enced less PF suggesting that prolongation of pancreatic model (Hematocrit < 38%, Open-pancreatectomy, pre- duct obstruction decreases the incidence of PF. operative Sepsis, Insulin-dependent diabetes, pre-surgical blood Transfusion, American society of anesthesiologists classification ‡ 3, Albumin < 4 g/dl, Age > 60 years; FP28 - Free Papers 28 (mini oral) - Liver: HOSPITAAAL). The eight-variable model had a c-sta- Surgical Outcomes 3 tistic of 0.68 (95% CI, 0.67-0.69) with an optimism of FP28-01 < 0.01. The model showed good calibration with concor- dance between the predicted and observed probabilities. SELECTIVE INHIBITION OF THE Conclusion: Blood transfusion is a prominent quality in- THROMBIN RECEPTOR PROTEASE- dicator in pancreatic surgery. The HOSPITAAAL score ACTIVATED RECEPTOR-1 shows moderate discrimination in predicting blood trans- ATTENUATES HEPATIC ISCHEMIA- fusion after pancreaticoduodenectomy. This could be used REPERFUSION INJURY IN MICE to augment risk-adjustment, standardize performance measurement, and to develop strategies pre-operatively to D. Noguchi, N. Kuriyama, T. Fujii, H. Kato, Y. Azumi, reduce the risk of blood transfusion. M. Kishiwada, S. Mizuno, M. Usui, H. Sakurai and S. Isaji Hepatobiliary Pancreatic and Transplant Surgery, Mie University, Japan FP27-06 Introduction: Development of liver diseases is associated with blood coagulation cascade, whose activation results in DO PATIENTS UNDERGOING a generation of thrombin. Thrombin not only cleaves NEOADJUVANT CHEMOTHERAPY FOR fibrinogen, but also triggers intracellular signaling in PANCREATIC ADENOCARCINOMA numerous hepatic nonparenchymal cells through an acti- HAVE LESS PANCREATIC FISTULA vation of the protease-activated receptor-1 (PAR-1). This AFTER PANCREATICO- activation has been shown to drive multiple liver pathol- fl DUODENECTOMY ? AN ANALYSIS OF ogies including hepatic in ammation and hepatocellular injury. However, there have been few studies on the THE PANCREATECTOMY-TARGETED thrombin activation of PAR-1 during hepatic ischemia- NSQIP USING COARSENED-EXACT reperfusion injury (IRI). This study focused on the MATCHING thrombin activation of PAR-1 and researched whether J. Abou Khalil, G. Martel, F. K. Balaa and K. A. Bertens vorapaxar, selective PAR-1 inhibitor, could attenuate he- The Ottawa Hospital, Canada patic IRI. Materials and methods: Using 60-min partial hepatic IRI Introduction: The use of neoadjuvant chemotherapy in the model of mice, PAR-1 expression and plasma thrombin- treatment of pancreatic adenocarcinoma (PA) carries many antithrombin complex (TAT) levels were evaluated. theoretical advantages. We hypothesize that the adminis- Additionally, IRI mice were divided into the two groups: tration of neoadjuvant chemotherapy (NAC) may decrease the one received intraperitoneal administration with vora- the incidence of pancreatic fistula (PF) due to pancreatic paxar (0.005mg/kg) twice before and after reperfusion and parenchymal changes caused by prolonged pancreatic duct the other without it. obstruction. Results: PAR-1 was expressed on sinusoidal endothelial Methods: Patients with PA undergoing elective pancrea- cells in liver. PAR-1 expression and plasma TAT levels, ticoduodenectomy (PD) and distal pancreatectomy (DP) in which were equivalent to thrombin levels, were signifi- the 2104-2016 ACS NSQIP pancreatectomy-targeted cantly increased after IRI compared to naïve liver (PAR-1/ Participant User Files were selected. We performed 1:1

HPB 2018, 20 (S2), S182eS294 S288 Free Orals (FP01-FP31) bactin: 2.45vs1.49, p=0.05; TAT: 4.946vs.0.336 ng/ml, (morbidity rates increase as the MILS volume decreases). p=0.002) (Figure). At 4 hours after reperfusion, vorapaxar Low-volume centers can safely perform MILS of antero- significantly improved serum transaminase levels and his- lateral segments with excellent outcomes. High-volume tological liver damage (AST: 2194vs.3113, p=0.046; centers approach more complex resections with adequate Suzuki’s score: 3.3vs.6.7, p=0.006). Furthermore, vora- safety. paxar markedly attenuated recruitment of neutrophil (Ly6G: 9vs.32/HPF, p=0.003) and decreased TNFa expression (1.27vs.3.46, p=0.032). FP28-03 Conclusion: Hepatic IRI increased a generation of IS HEPATECTOMY SAFE FOLLOWING thrombin, whose activation of PAR-1 drove hepatic inflammation and hepatocellular injury. Selective inhibition YTTRIUM-90 THERAPY? A MULTI- of PAR-1 by vorapaxar significantly attenuated hepatic IRI. INSTITUTIONAL INTERNATIONAL EXPERIENCE L. Melstrom1, G. H. Choi2, R. Salem3, G. Poultsides4, FP28-02 S. Shah5, S. Maithel6, F. Rocha7, I. M. Borel Rinkes8 and CENTER CASE VOLUME AND D. Abbott9 1Surgery, City of Hope Medical Center, United States, MINIMALLY INVASIVE LIVER 2 SURGERY: A STUDY BASED ON THE Yonsei University College of Medicine, Republic of Korea, 3Surgery, Northwestern University, 4Stanford ITALIAN GROUP OF MINIMALLY University, 5Surgery, University of Cincinnati, 6Surgery, INVASIVE LIVER SURGERY REGISTRY Emory University, 7Surgery, Virginia Mason Medical 8 L. Viganò1, M. Cimino1, L. Aldrighetti2, U. Cillo3, Center, United States, Surgery, University Medical 9 A. Ferrero4, G. M. Ettorre5, F. Giuliante6, R. Dalla Valle7, Center Utrecht, The Netherlands, and Surgery, University F. Calise8, G. Torzilli1 and Italian Group of Minimally of Wisconsin School of Medicine and Public Health, Invasive Liver Surgery (I Go MILS) United States 1Hepatobiliary and General Surgery, Humanitas Univer- Background: Previous reports from single institutions sity, Humanitas Clinical and Research Center, Milan, have shown variable safety profiles when liver-directed 2San Raffaele Hospital, 3Padua University Hospital, therapy with Yttrium-90 (Y90) is followed by hepatec- 4Mauriziano Hospital, 5San Camillo Hospital, 6Gemelli tomy. We hypothesized that in well-selected patients, Hospital, Catholic University of the Sacred Heart, Rome, hepatectomy after Y90 is feasible and safe. 7Parma University Hospital, and 8Pineta Grande Hospi- Methods: Nine institutions contributed clinical data for tal, Italy patients undergoing Y90 followed by hepatectomy (2008- Introduction: The outcome of liver surgery is associated 2017). Clinicopathologic and perioperative data were with the hospital case volume. To analyze the impact of the analyzed, with 90-day morbidity and mortality serving as center case volume on the minimally invasive liver surgery primary endpoints. (MILS). Results: Forty-seven patients met inclusion criteria. Methods: All patients enrolled in the prospective registry Median age was 59 (range 26-73) and 62% were male. of the Italian Group of MILS from November 2014 to Primary malignancies included hepatocellular cancer August 2016 were considered. Only centers with an (n=15;32%), colorectal cancer (n=9;19%), chol- accrual 6 months were included. The center case angiocarcinoma (n=8;17%), neuroendocrine (n=8;17%) volume was defined as the number of MILS performed and other (n=10). Eight patients (17%) had extrahepatic per month. disease. Three patients (6%) had prior liver resection. The Results: 919 MILS performed by 37 centers were distribution of Y90 treatment was: right lobe (n=30;64%), analyzed. Ten centers performed >2MILS/month bilobar (n=14;30%) and left lobe (n=3;6%). Median future (n=600 patients), 7 performed 1-2 MILS/month (n=142), liver remnant (FLR) following Y90 was 44% (range:30- and20performed< 1 MILS/month (n=177). Centers 78). Resections were primarily right (n=15;32%) and performing < 1MILS/monthresectedmorebenign extended right (n=13;28%) hepatectomies. The median tumors and tumors in anterolateral segments (28.2% vs. time to resection from Y90 was 82 days (range:13-798). On 20.9%, p=0.035; 87.0% vs. 72.6%, p< 0.001). Major pathology, 85% of tumors demonstrated some component hepatectomies increased with the case volume (3.4% if of necrosis; 23% had no viable tumor. The 90-day < 1 MILS/month, 7.7% if 1-2, 13.7% if >2, p< 0.001). complication rate was 43% and mortality was 2%. The The R1 resection rate was higher in centers performing number of Y90-treated lobes was associated with Clavien- >2 MILS/month (10.9% vs. 2.6%, p< 0.001). The Dindo grade >3 complications (OR1.78; 95% CI 1.13-2.8; conversion, overall morbidity, and severe morbidity rates p=0.012) whereas time from Y90 to surgery, number of were similar among groups. The overall and severe Y90 treatments or preoperative chemotherapy were not morbidity rates of major hepatectomies were higher in associated with complications. centers performing < 2 MILS/month (52.9% vs. 29.3%, Conclusions: Contrary to previous reports, these data p=0.060; 23.6% vs. 8.5%, p=0.092). demonstrate that hepatectomy following Y90 is safe in Conclusions: The center case volume does not impact the well-selected populations. This multi-disciplinary treat- outcome of minor hepatectomies; however, it impacts the ment paradigm should be more widely studied for patients outcome of minimally invasive major hepatectomies with inadequate FLR and few other treatment options.

HPB 2018, 20 (S2), S182eS294 Free Orals (FP01-FP31) S289

FP28-04 Conclusions: The ACHBT score represents an externally validated tool to predict severe PHBL. It is applicable for a EXTERNAL VALIDATION OF A RISK wide variety of patients and procedures, including laparo- SCORE TO PREDICT SEVERE scopic liver resections. POSTHEPATECTOMY BILIARY LEAKAGE FP28-05 K. Mohkam1, D. Fuks2, E. Vibert3, T. Nomi2, F. Cauchy4, HEPATIC RESECTION FOR Y. Kawaguchi2, E. Boleslawski5, J. -M. Regimbeau6, MALIGNANCY IN THE ELDERLY: B. Gayet2 and J. -Y. Mabrut1 NSQIP PREDICTORS OF MORBIDITY 1 2 Croix Rousse Hospital, Institut Mutualiste Montsouris, AND MORTALITY 3Paul Brousse Hospital, Université Paris Sud, 4Beaujon J. Pasko1, E. Dewey2, C. K. Enestvedt1, E. Maynard1 and Hospital, Assistance Publique Hôpitaux de Paris, Uni- S. L. Orloff1 versity Paris VII, 5CHU Lille, and 6CHU Amiens, France 1Abdominal Organ Transplant, Oregon Health and Sci- Background: Biliary leakage is a major contributor ence University, and 2General Surgery, Oregon Health to morbidity after hepatectomy. A score to predict and Science University, United States severe posthepatectomy biliary leakage (PHBL) was Background: Liver resection for malignancy in the elderly recently developed by the French Association of Hepato- is common given most malignancies occur in the 6th and pancreatobiliary Surgery and Transplantation (ACHBT). 7th decades. Previous studies show increasing age in sur- The aim of the study was to validate the score externally. gical patients is the most important predictor of morbidity Methods: Patients undergoing liver resection without and mortality. The aim was to evaluate patients > 65yrs hepaticojejunostomy between 1994 and 2016 at a single undergoing liver resection for malignancy and determine center were used as an external validation cohort. The risk factors for morbidity and mortality. score was tested upon that cohort for predicting severe Methods: The National Surgical Quality Improvement PHBL as defined by International Study Group on Liver Program (NSQIP) database (2014-2016) was queried for Surgery grade B or C. Association between the score, patients age>65 who underwent liver resection for malig- pre/intraoperative variables and severe PHBL was nancy. Odds ratios for factors associated with increased risk assessed. of post-hepatectomy complications and death were evalu- Results: Among 778 procedures performed (including 679 ated with Cox logistical regression. (87.3%) laparoscopic and 260 (34.3%) major hepatec- Results: 3,460 patients underwent liver resection with a 3% tomies), 31 (4.0%) complicated with severe PHBL. The mortality and 32% complication rate. Post-operative bile ACHBT score showed good discrimination (area under the leak occurred in 9.4% of patients, liver failure occurred in receiver operating characteristic curve [AUROC]: 0.747, 5.7%, and transfusions occurred in 20% of patients. 95%CI: 0.652-0.841), calibration and accuracy (diagnostic On univariate analysis, there was no association between odds ratio for a score 1: 6.217 [95% CI: 2.642-14.627], complications and age (p=.75), neoadjuvant therapy (p= for a score 2: 6.059 [95% CI: 2.858-12.846], for a score 0.13), diabetes (p=.44), and smoking (p=0.73). Factors 3: 9.589 [95% CI: 2.868-32.066]). On multivariable associated with complications: decreased BMI (p< 0.01), analysis, the ACHBT score was the only predictor of severe higher ASA (p< .01), pringle maneuver (p< .01), hilar PHBL. A model combining the ACHBT score and con- cholangiocarcinoma (p< .01), and open surgery (p< .01). version into open surgery was significantly more discrim- On multivariate analysis, factors associated with increased inant than the ACHBT score alone (AUROC=0.790 [95% risks can be seen in Figure 1. CI: 0.711-0.870], Delong’s test p=0.002).

FP28-05 Table [Table 1]Multivariable Adjusted Odds Ratios ond 95% Confidence intervals for Post Hepotectomy Compactions and Mortality for Elderly Patients Age  65 Term Multivariate model: at least one complication Multivariate model: modality Odds 95% CI Odds 95% CI Ratio Lower Upper p-value Ratio lower Upper p-value BMI (per [ 1) 0.98 0.97 1.00 0.0149 0.99 0.96 1.02 0.5036 Pre-operative platelet count 1.00 0.98 1.03 0.9568 0.93 0.87 0.99 0.0191 (per [ 25) Preoperative serum albumin 0.60 0.50 0.72 <.0001 0.64 0.43 0.96 0.0313 (per [ 1) Pre-operative serum creatinine 1.11 1.05 1.18 0.0002 1.07 1.00 1.16 0.0636 (per [ 0.25) Pre-operative total bilirubin 1.02 0.91 1.14 0.7783 1.08 0.96 1.23 0.2012 (per [ 1) Total operation time 1.08 1.07 1.10 <.0001 1.05 1.03 1.08 <.0001 (per [ 15 minutes) Biliary Stent (Yes vs No) 1.52 1.04 2.24 0.0327 1.69 0.92 3.11 0.0930 Open Approach vs Minimally 2.14 1.65 2.78 <.0001 1.59 0.77 3.28 0.2100 Invasive

(continued on next page)

HPB 2018, 20 (S2), S182eS294 S290 Free Orals (FP01-FP31)

(continued) Term Multivariate model: at least one complication Multivariate model: modality Odds 95% CI Odds 95% CI Ratio Lower Upper p-value Ratio lower Upper p-value Primary hepatobiliary cancer vs 1.26 1.05 1.50 0.0114 1.94 1.20 3.11 0.0064 Secondary (metastatic) tumor Inflow Occlusion (Pringle 1.00 0.83 1.21 0.9959 0.85 0.53 1.35 0.4906 Maneuver) During Resection ASA classification Severe vs None/Mild 1.04 0.81 1.32 0.7774 4.91 1.74 13.86 0.0026 Life-Threatening vs None/Mild 1.86 1.32 2.62 0.0004 6.29 2.00 19.77 0.0016 Life-Threatening vs Severe 1.80 1.36 2.37 <.0001 1.28 0.71 2.31 0.4110 Principal Operative Procedure TOTAL l£FT LOBECTOMY 1.04 0.78 1.38 0.8052 0.97 0.45 2.12 0.9470 vs PARTIAL LOBECTOMY TOTAL RIGHT LOBECTOMY 1.83 1.46 2.30 <.0001 2.29 1.35 3.88 0.0021 vs PARTIAL LOBECTOMY TOTAL RIGHT LOBECTOMY vs 1.77 1.28 2.44 0.0005 2.35 1.08 5.13 0.0319 TOTAL LEFT LOBECTOMY TRlSEGMENTECTOMY 2.00 1.49 2.70 <.0001 2.34 1.26 4.36 0.0071 vs PARTIAL LOBECTOMY TRlSEGMENTECTOMY vs 1.93 1.33 2.80 0.0005 2.41 1.04 5.55 0.0391 TOTAL LEFT LOBECTOMY TRlSEGMENTECTOMY vs 1.09 0.79 1.51 0.5872 1.02 0.56 1.87 0.9380 TOTAL RIGHT LOBECTOMY

Conclusions: Age was not associated with increased Results: SSI developed in 587 patients (9.6%), including complications. Decreased BMI was associated with com- superficial/deep incisional SSI in 357 patients (5.8 %), and plications. Increasing albumin was associated with organ/space SSI in 304 patients (5.0 %). Multivariate logistic improved mortality and morbidity. These findings suggest regression analysis showed that obesity, diabetes mellitus, that cachexia effects surgical outcomes in elderly and liver cirrhosis, re-hepatectomy, hepatoliathiasis, and intra- indirectly reflects the frailty of patients. operative blood transfusion were independent risk factors of overall SSI. However, incisional and organ/space SSI differed from each other with respect to risk factors. Among FP28-06 a variety of risk factors, hepatolithiasis, liver cirrhosis, and RISK FACTORS FOR SURGICAL SITE intraoperative blood transfusion were consistently associ- ated with both incisional and organ/space SSI. INFECTIONS AFTER LIVER Conclusions: SSI is a common complication after liver RESECTION: A MULTIVARIATE resection, and more caution should be taken in patients with ANALYSIS OF 6,132 PATIENTS hepatolithiasis or liver cirrhosis. Prevention strategies L. Y. Sun1,2, B. Quan1,2, H. Zhang1, Z. L. Li1, J. Han1, focusing on factors associated with SSI is necessary in C. Li1, M. D. Wang1, H. Xing1, F. Shen1 and T. Yang1 order to reduce SSI after liver resection. 1Department of Hepatobiliary Surgery, Eastern Hepato- biliary Surgery Hospital, Second Military Medical Uni- Table [Multivariate Analysis of Risk Factors of Surgical] versity, China, and 2Department of Clinical Medicine, fi Second Military Medical University, China Odds 95% con dence ratio interval Introduction: Surgical site infection (SSI) is a common postoperative complication and associated with an Low High P value increased morbidity, hospital stay, and overall cost. The Overall SSI aim of the present study was to identify risk factors for Obesity (Body mass 1.62 1.25 2.09 <0.01 SSIs after hepatic resection based on a large single-center index > 30) cohort. Diabetes mellitus 1.43 1.07 1.90 0.03 Methods: A retrospective study was conducted of 6,132 Cirrhosis 1.55 1.34 1.80 <0.01 patients who underwent liver resection without concomi- Re-hepatectomy 1.15 1.03 1.31 <0.05 tant biliary reconstruction or gastrointestinal procedures between 2014 and 2016 at the largest hepatic center in Hepatolithiasis 1.75 1.42 2.13 <0.01 China. The occurrences of SSI, classified as incisional SSI Intraoperative blood 1.57 1.32 1.98 <0.01 and organ/space SSI within 30 days after operation were transfusion investigated. Patient- and surgical-related risk variables Incisional SSI were collected using standardized data collection form. A Obesity (Body mass 1.55 1.30 1.97 <0.01 likelihood ratio forward regression model was used to index  30) assess the independent association of risk factors with SSI.

HPB 2018, 20 (S2), S182eS294 Free Orals (FP01-FP31) S291

Table (continued) equally high proportions of intestinal and pancreatobiliary Odds 95% confidence type. ratio interval Conclusion: As one of the prevalent nations of chol- angiocarcinoma, this is the first Korean multi-center Low High P value collaboration study of IPNB, which is thought to be a Diabetes mellitus 1.35 1.18 1.65 0.01 precursor lesion of cholangiocarcinoma. The ’modified Cirrhosis 1.61 1.35 1.94 <0.01 anatomical classification’ is a simple and intuitive criteria Hypoalbuminemia 1.19 1.03 1.58 0.03 which can help to determine the treatment strategy and Re-hepatectomy 1.17 1.03 1.32 <0.05 show some correlations with histologic subtype and path- ologic phenotypes. Hepatolithiasis 1.58 1.32 1.99 <0.01 Intraoperative blood 1.32 1.10 1.68 0.01 transfusion Organ/Space SSI FP29-02 Cirrhosis 1.70 1.41 2.12 <0.01 EXTENDED ANTIBIOTIC THERAPY Hepatolithiasis 1.66 1.46 1.97 <0.01 VERSUS PLACEBO AFTER LAPAROSCOPIC CHOLECYSTECTOMY Intra operative blood 1.71 1.45 2.11 <0.01 transfusion FOR MILD AND MODERATE ACUTE Length of abdominal 1.27 1.07 1.54 0.03 CALCULOUS CHOLECYSTITIS A drainage  5days RANDOMIZED DOUBLE-BLIND Postoperative bile 1.22 1.03 1.49 0.04 CLINICAL TRIAL leakage M. de Santibanes1, J. Glinka2, F. Alvarez2, C. Elizondo2, 2 2 2 2 Multivariate Analysis of Risk Factors of Surgical Site Infection D. Giunta , V. Ardiles , R. Sanchez Claria , O. Mazza , after Liver Resection E. de Santibañes2 and J. Pekolj2 1Department of Surgery, Division of HPB Surgery, Liver Transplant Unit, Hospital Italiano de Buenos Aires, and 2Hospital Italiano de Buenos Aires FP29 - Free Papers 29 (long oral) - Biliary: Background: Laparoscopic cholecystectomy (LC) is the gold standard treatment in mild and moderate forms of Miscellaneous 2 acute calculous cholecystitis (ACC). Currently there is FP29-01 consensus for the use of antibiotics in the preoperative CLINICOPATHOLOGIC phase of ACC. However, the need for antibiotic therapy CHARACTERISTICS OF INTRADUCTAL after surgery remains undefined with a low level of scien- PAPILLARY NEOPLASM OF BILE DUCT tific evidence. (IPNB) AFTER CENTRAL REVIEW: Objective: To determine the effect of extended antibiotic therapy on infectious complications in patients with mild KOREAN MULTI-CENTER and moderate ACC undergoing LC. COLLABORATION STUDY Methods: The CHART (Cholecystectomy Antibiotic J. R. Kim, W. Kwon, S. -W. Kim and J. -Y. Jang Randomised Trial) study is a single-center, prospective, Surgery, Seoul National University College of Medicine, double blind, and randomized trial. Patients with mild to Republic of Korea moderate ACC operated by laparoscopic cholecystectomy Introduction: Intraductal papillary neoplasm of the bile LC were randomly assigned to receive antibiotic (amoxi- duct (IPNB) is a recently defined concept and its clinico- cillin/clavulanic acid) or placebo treatment for 5 consecu- pathologic characteristics and diagnostic criteria have not tive days. Primary endpoint was postoperative infectious been well established yet. We performed this study to complications. Secondary endpoints were as follows: clarify the clinicopathologic features of IPNB and to find (1) length of hospital stay, out the optimal morphologic criteria. (2) readmissions, Methods: Between 1997 and 2016, 457 patients were (3) reintervention and included in this study with final diagnosis of IPNB from 23 (4) mortality. tertiary referral centers of Korea. Every case was recon- Results: In the per-protocol analysis, 6 of 104 patients firmed through central review by 3 specialized biliary- (5.8%) in the placebo arm and 6 of 91 patients (6.6%) in the pancreas pathologists. We applied ’Japan Biliary Associa- antibiotic arm developed postoperative infectious compli- tion (JBA) classification’ and ’modified anatomical classi- cations, absolute difference 0.82 (95% CI, -5.96 to 7.61, p= fication’ to compare the morphological features. 0.81). The median hospital stay was 3 days, without sig- fi Results: Patient’s mean age was 65.5 years. The 5-year ni cant difference between groups (p= 0.3). There was no overall survival rate of all the patients was 80.6% and 104 mortality. There were no differences regarding read- (24.6%) patients underwent recurrence during the follow- missions and reoperations between the two groups. up periods. Patients with the intestinal type (n=237; 52.2%) Conclusions: The use of antibiotics in the postoperative was the majority in the histologic subtypes of IPNB. period of laparoscopic cholecystectomy for mild and fi Among 457 patients, 451 (98.7%) had bile duct dilatation moderate acute calculous cholecystitis is not justi ed, and 430 (94.1%) had intraductal mass on preoperative since it was not associated with a decrease in the inci- images. We found out that the extrahepatic type had a dence of infectious and other types of morbidity in the higher rate of intestinal type and the intrahepatic type had present study.

HPB 2018, 20 (S2), S182eS294 S292 Free Orals (FP01-FP31) > FP29-03 Primary endpoint was re-intervention 90 days after the hepaticojejunostomy and secondary endpoints were severe WILL ROUX-EN-Y 90-day complications and liver-related mortality. HEPATICOJEJUNOSTOMY HELP Results: In total 913 patients with a median follow-up of PREVENT CALCULUS RECURRENCE IN two years from 48 centres were included in the analysis. In HEPATOLITHIASIS PATIENT WITH 465 patients (42%) the BDI was diagnosed intraoperatively SPHINCTER OF ODDI LAXITY: AN and 299 patients (27%) were subjected to an attempted immediate intraoperative repair, including 170 (15%) ONGOING RANDOMIZED hepaticojejunostomies. Hepaticojejunosstomies were CONTROLLED TRIAL performed early (n=339), intermediate (n=261) and late W. Su, J. Lou, S. Gao, X. Bai, R. Que and T. Liang (n=313). No difference in re-intervention after 90 days Department of Hepatobiliary and Pancreatic Surgery, could be demonstrated between groups. 90-day complica- Zhejiang University/Second Affiliated Hospital, China tions were higher in the early group (p=0.041) and liver- Introduction: Sphincter of Oddi laxity (SOL) has been related mortality lower in the late group (p=0.040). proven a high risk factor of hepatolithiasis recurrence and Conclusions: In this multicenter study, early repair was Roux-en-Y hepaticojejunostomy is the most widely used associated with a higher complication rate. Factors intervention. However, no prospective studies have contributing to this may include that patients in the early demonstrated its efficacy so far. group more often had complicated gallstone disease and Methods: In this single-center open-label randomized were subject to emergency cholecystectomy. Once recov- controlled trial, we randomly assigned hepatolithiasis patients ered from surgery, the need for further interventions was with SOL to receive either Roux-en-Y hepaticojejunostomy similar in groups. (intervention group) or merely calculus removal (control group). The primary endpoint was recurrence of calculus. Results: Of the 110 patients who underwent randomization, FP30 - Free Papers 30 (long oral) - Liver: 83 were included in the intention-to-treat analysis. At a median Technical Surgery 2 follow-up of 27 month, the primary-endpoint event occurred in FP30-02 11 of the 43 patients (25.6%) in the intervention group as ROBOTIC VS LAPAROSCOPIC compared with 10 of the 40 patients (25.0%) in the control group (relative risk[RR], 1.023; 95% confidence interval [CI], HEPATECTOMY: APPLICATION OF 0.488 to 2.145; P=0.951). Cholangitis attack occurred in 8 THE DIFFICULTY SCORING SYSTEM patients (18.6%) in the intervention group and in 7 patients C. Chong, H. -T. Lok, A. Fung, A. Fong, S. Cheung, (17.5%) in the control group (RR, 1.063; 95% CI, 0.424 to J. Wong, K. -F. Lee and P. Lai 2.664; P=0.896). Cholangiocarcinoma occurred in 1 patients Department of Surgery, The Chinese University of Hong (2.3%) in the intervention group and in 2 patients (5.0%) in the Kong, Hong Kong control group (RR, 0.465; 95% CI, 0.044 to 4.934; P=0.607). Background: The development of robotic system may Conclusion: Based on our current data, Roux-en-Y hepa- help to relieve the difficulties encountered during laparo- ticojejunostomy did not help prevent calculus recurrence in scopic hepatectomy. A difficulty scoring system (DSS) was hepatolithiasis patients with SOL in the short-medium term. developed to assess the difficulty of various laparoscopic A larger sample size and a longer follow-up may be needed liver resection procedures. The aim of this study is to fi to evaluate the ef cacy of Roux-en-Y hepaticojejunostomy compare the outcomes of robotic hepatectomy and lapa- in the treatment of hepatolithiasis. (ClinicalTrials.gov roscopic hepatectomy among different difficulty levels. number, NCT01459549). Methods: Clinical data from all consecutive patients who underwent robotic or laparoscopic hepatectomy at the Prince of Wales Hospital, Hong Kong, from 2003 to 2017 FP29-04 were prospectively collected and reviewed. The difficulty E-AHPBA MULTICENTRE STUDY: POST level of operations was graded using the DSS. Perioperative CHOLECYSTECTOMY BILE DUCT outcomes of robotic and laparoscopic hepatectomy were fi INJURY: EARLY VS LATE REPAIR compared at each dif culty level. Results: A total of 107 and 94 patients underwent robotic WITH HEPATICOJEJUNOSTOMY and laparoscopic hepatectomy during the study period J. Rystedt and on behalf of E-AHPBA respectively. Among them, 16 and 2 patients were operated Surgery, Lund University, Sweden for recurrent pyogenic cholangitis respectively and were Introduction: Treatment of bile duct injuries (BDI) during excluded because no mark for tumour location can be cholecystectomy depends on the severity of injury and the assigned. The mean DSS scores of robotic and laparoscopic timing of diagnosis. Standard of care for severe BDIs is hepatectomy were 4.5 and 3.6 respectively. (p=0.004) For hepaticojejunostomy. The aim was to assess the optimal cases with low (DSS 1-3) and intermediate (DSS 4-6) dif- timing for repair of BDI with hepaticojejunostomy. ficulty level, there was no significant difference in operative Methods: Members of the European-AfricanHepatoPan- blood loss, operation time and overall complications rate. creatoBiliary Association (E-AHPBA) were invited to Only 2 cases (2.2%) with high difficulty level were oper- participate and report all consecutive patients with hepati- ated with laparoscopic approach while 20% of patients cojejunostomy after BDI from January 2000 to June 2016. operated with robotic approach had DSS >6. Patients were stratified according to the timing of biliary Conclusions: Perioperative outcomes following robotic reconstruction with hepaticojejunostomy: early (day 0-7), and laparoscopic hepatectomy were similar in cases with intermediate (1 to 6 weeks) and late (6 weeks to 6 months). low and intermediate difficulty. However, robotic system

HPB 2018, 20 (S2), S182eS294 Free Orals (FP01-FP31) S293 allowed minimally invasive approach in cases with higher subset of patients may have complications mitigated by difficulty level. operative placement of drains. Therefore, the aim of this analysis is to determine which patients undergoing a major hepatectomy might benefit from drain placement. FP30-03 Methods: The 2014-15 NSQIP hepatectomy Participant PURE LAPAROSCOPIC RIGHT Use Files were queried for patients undergoing major hepatectomy (N=3,312). Patients who had operative drains HEPATECTOMY: A RISK SCORE FOR placed and those undergoing biliary reconstruction or CONVERSION FOR THE PARADIGM OF concomitant operations were excluded. Patients requiring DIFFICULT LAPAROSCOPIC LIVER postoperative intervention (percutaneous drain, biliary stent RESECTIONS or reoperation) were compared to those who did not require a postoperative procedure. Multivariable regression models F. Cipriani, F. Ratti, M. Catena, M. Paganelli and were used to develop a score to predict who might benefit L. Aldrighetti from placement of an operative drain. Hepatobiliary Surgery Division, San Raffaele Hospital, Results: Of 1,051 patients (35.1%) managed without a Italy drain, 991 were eligible for analysis. Factors associated with Introduction: Converted laparoscopic liver resection postoperative intervention were hypertension (p< 0.02), (LLR) are known to lose some advantages of the mini- ASA 3(p< 0.02), preoperative bilirubin  1.2 (p< 0.01), fi mally-invasiveness, and factors are identi ed to predict preoperative biliary stent (p< 0.04), cirrhotic or fatty liver fi patients at risk. Speci c evidence for laparoscopic right (p< 0.001) and bleeding requiring transfusion (p< 0.001). hepatectomy (LRH) is expected of uttermost usefulness in After multivariable regression, preoperative bilirubin, stent, clinical practice, given the peculiarities and technical dif- liver texture and bleeding/transfusions remained significant. ferences with other complex hepatectomies. In developing a Hepatectomy Drain Prognostic Score each ’ Methods: Data of LRH (2007- 17) were analysed. Risk factor was weighted 1. Higher scores were associated with a fi factors for conversion were identi ed by uni- and multivar- step-wise greater risk of need for an intervention (Table). iable analysis. A conversion risk score (CRS) was built Conclusions: Preoperative and intraoperative factors can assigning each factor 1point, and comparing the score with predict which patients undergoing a major hepatectomy the conversion status for each patient. Accuracy was assessed may benefit from operative drain placement. A Hepatec- by calculating the area-under-the-receiver-operator-charac- tomy Drain Prognostic Score may help guide surgeons in teristic-curve (AUROC). The outcomes of converted and management of their major hepatectomy patients. completed LRH were compared after propensity score matching with the risk factors identified acting as covariates. Results: Among 100 LRH, 18 procedures were converted Table [Hepatectomy Drain Prognostic Score*] (18%), and reasons were: 44.4%oncologic radicality, 33.3% bleeding, 11.1%adhesions, 11.1%impossibility to proceed, Cumulative Odds Ratio 95% Confidence p-value Score Interval 5.5%biliostasis. Risk factors for conversion (multivariable analysis) were: previous liver surgery (HR4.5,p0.013), pre- Score 0 Reference operative chemotherapy (HR6.7,p0.023), malignant diagnosis Score 1 1.921 (1.184-3.117) 0.008 (HR3.8,p0.018), closeness to hepatocaval confluence Score 2 5.421 (2.926-10.042) <0.001 (HR4.2,p0.038), tumour volume (HR3.4,p0.010). Conversion Score 3 21.825 (3.545-134.38) 0.001 ratescorrelatedwiththeCRS,raisingfrom0to100%when CRS increased from 0 to 5 (Spearman p0.020). The CRS *Bilirubin1.2=1, Biliary stent=1, Cirrhotic or Fatty liver=1, showed accuracy with a 0.76 AUROC. Converted patients Bleeding/Transfusion=1 exhibited longer operative time (376vs340min,p0.045); greater blood loss (619vs364ml,p0.002), transfusions FP31 - Free Papers 31 (long oral) - Pancreas: (27vs13%,p0.028) and morbidity rates (55vs19%,p0.003); Surgical Outcomes 5 longer hospital stay (8.5vs6.5days,p0.026). FP31-01 Conclusions: Specific risk factors for conversion are identified for LRH. Since the outcomes of converted pa- MINIMALLY INVASIVE VS. OPEN tients are less favourable than those completed lapa- DISTAL PANCREATECTOMY roscopically, this CRS may help in choosing the best (LEOPARD): MULTICENTER PATIENT- approach for such a challenging procedure. BLINDED RANDOMIZED CONTROLLED TRIAL FP30-04 1 2 3 4 WHEN TO USE A DRAIN AFTER MAJOR T. de Rooij , J. van Hilst , D. Boerma , R. van Dam , C. van Eijck5, M. Gerhards6, J. Klaase7, G. Kazemier8, HEPATECTOMY M. Luyer9, O. Busch1, M. Besselink1 and LEOPARD Trial K. A. Zorbas1, H. A. Pitt1, S. N. Jayarajan1, S. Reddy2 and Group, Dutch Pancreatic Cancer Group A. Karachristos1,2 1Academic Medical Center, Cancer Center Amsterdam, 1Department of Surgery, Temple University Hospital, and 2Surgery, Cancer Center Amsterdam, Academic Medical 2Surgical Oncology, Fox Chase Cancer Center, United Center, University of Amsterdam, 3St Antonius Hospital, States 4Maastricht University Medical Center (MUMC), 5Eras- 6 Introduction: Drain placement after a major hepatectomy mus MC - University Medical Center Rotterdam, OLVG, 7 8 is associated with higher rates of morbidity, bile leaks, Medisch Spectrum Twente, VU University Medical 9 secondary interventions and readmissions. However, a Center, and Catharina Hospital, The Netherlands

HPB 2018, 20 (S2), S182eS294 S294 Free Orals (FP01-FP31)

Introduction: Minimally invasive distal pancreatectomy Conclusion: A MPD > 5 mm represents an independent (MIDP) is gaining popularity, but it is unknown whether it predictor of PC in surgically resected IPMN. After is superior to open distal pancreatectomy (ODP) because including IPMNs under surveillance in the present analysis, randomized studies are lacking. the threshold to be an independent predictor of PC Methods: We randomized patients with left-sided pancre- increased to 3 10 mm. In the absence of other features of atic tumors to undergo either MIDP or ODP in 14 centers malignancy, an MPD of 5 to 9 mm alone should not be after completing a nationwide MIDP training program (Ann considered as an absolute indication for surgical resection Surg 2016). Patients were blinded for the type of procedure. in presumed IPMNs. The primary endpoint was postoperative time to functional recovery. Analysis was according to intention-to-treat. Results: Time to functional recovery was 4 (IQR 3-6) days FP31-03 in 51 patients after MIDP vs. 6 (IQR 5-8) days in 57 patients OUTCOMES AND RISK SCORE FOR after ODP (P< 0.001). Length of hospital stay was shorter after MIDP (6 vs. 8 days, P< 0.001). Eight percent of DISTAL PANCREATECTOMY WITH MIDPs were converted to ODP. Operative blood loss was CELIAC AXIS RESECTION (DP-CAR less after MIDP (150 vs. 400 mL, P< 0.001), but operative SCORE): AN INTERNATIONAL time was longer (217 vs. 179 minutes, P=0.005). There MULTICENTER ANALYSIS were no significant differences in Clavien-Dindo grade 3 S. Klompmaker1, N. Peters2,3, J. van Hilst1, C. Bassi4, complications (25% vs. 38%, P=0.21) and mortality (0% vs. U. Boggi5, W. Niesen6, H. Yamaue7, C. L. Wolfgang2, 2%, P>0.99) between MIDP and ODP, respectively. Grade M. E. Hogg8, M. G. Besselink1 and The E-AHPBA DP- B/C postoperative pancreatic fistula was seen in 37% after CAR Study Group MIDP vs. 21% after ODP (P=0.06), without difference in 1Surgery, Cancer Center Amsterdam, Academic Medical radiological drainage or surgical re-interventions. A Center, University of Amsterdam, The Netherlands, 2Sur- reduction in grade B/C delayed gastric emptying was seen gery, Johns Hopkins Hospital, United States, 3Surgery, after MIDP, as compared to ODP (4% vs. 20% (P=0.01). University Medical Center Utrecht, The Netherlands, Conclusions: MIDP was superior to ODP, indicating that 4Surgery, University of Verona - The Pancreas Institute, MIDP is the preferred treatment for patients with tumors Italy, 5Surgery, Cisanello University Hospital, Italy, confined to the left side of the pancreas. 6General, Visceral and Transplantation Surgery, Heidel- berg University, Germany, 7Second Department of Sur- gery, Wakayama Medical University, Japan, and 8Surgery, FP31-02 University of Pittsburgh Medical Center, United States REAPPRAISAL OF MAIN PANCREATIC Background: Distal pancreatectomy with celiac axis DUCT IN IPMN UNDER resection (DP-CAR) is performed in selected patients with SURVEILLANCE: DUCT DILATATION pancreatic cancer involving the celiac axis. A recent OF 5 TO 9 MM ALONE IS NOT multicenter European study reported a 16% 90-day mor- ASSOCIATED WITH HIGH CANCER tality, highlighting the importance of patient selection. We constructed a risk score to predict 90-day mortality and RISK assessed oncological outcomes. S. Andrianello1, G. Marchegiani1, G. Morbin1, Methods: Retrospective, multicenter study in patients un- E. Secchettin1,M.D’Onofrio2, R. De Robertis2, dergoing DP-CAR at 20 European centers (model design; G. Malleo1, C. Bassi1 and R. Salvia1 2000-2016) and three very-high volume international cen- 1General a Pancreas Surgery, The Pancreas Institute, and ters in the USA and Japan (model validation; 2004-2017). 2Radiology - The Pancreas Institute, University of Verona, We used the area-under-receiver-operator-curve (AUC) Italy and calibration plots for validation of the 90-day mortality Introduction: The correlation between the risk of pancre- risk model. Secondary outcomes assessed included onco- atic cancer (PC) and a dilated main pancreatic duct (MPD) logical outcomes such as resection margin status, (neo) in IPMNs is a matter of debate. The aim of this study was to adjuvant therapy use and survival. assess the role of MPD size in predicting PC in a large Results: Among 191 DP-CAR patients, 90-day mortality cohort of both resected IPMNs and surveilled IPMNs. was 5.5% (95CI: 2.2-11%) in 5 high-volume (1 DP-CAR/ Methods: All patients with a radiological or pathological year) and 18% (95CI: 9-30%) in 18 low-volume DP-CAR diagnosis of IPMN referred to the The Pancreas Institute, centers (P=0.020). A risk score with age, sex, BMI, ASA, University of Verona Hospital Trust, from 1985 to 2016 multivisceral resection, open versus minimally invasive were included. The primary endpoint was the occurrence of surgery, and low versus high-volume center performed well PC detected at surgery or during follow-up. in both, the design and validation cohort (AUC 0.79 versus Result: The final cohort consisted of 2134 patients, 439 0.74, P value = 0.642) and calibration was adequate after resected and 1695 under surveillance, with a median adjustment. In 174 patients with pancreatic ductal adeno- follow-up of 67 and 43 months respectively. MPD dilata- carcinoma, the R0 resection rate was 60%, neoadjuvant and tion was progressively associated with other features of adjuvant therapy was used in 69% and 67% of patients, and malignancy, but when presenting alone, the occurrence of median overall survival was 19 (95CI 15-25) months. PC was 17%. For resected IPMNs, both an MPD measuring Conclusions: In selected patients operated in high-volume 5-9 mm or 3 10 mm were independent predictors of HGD or centers, DP-CAR is associated with acceptable 90-day PC. Considering both resected and surveilled IPMNs, pa- mortality and overall survival. We proposed a 90-day tients with an MPD between 5-9 mm had more than 10 mortality risk score to improve patient selection and out- years of survival without developing PC. comes, with DP-CAR volume as dominant predictor.

HPB 2018, 20 (S2), S182eS294