52% of pts had multiple-sites ATS which was associated with FLI independent of age, tobacco use, hypertension and T2DM (OR = 1.004, 95%CI 1.001–1.007, p = 0.02). 64% of pts with NAFLD vs. 26% of those without had FRS ≥10%. FLI predicted FRS independent of the individual ATS sites or the number of CVRF (beta = 0.193, p < 0.001) in a model adjusted for age and sex. FLI alone or combined with CP, FP and CAC had a better AUROC for FRS ≥ 10% (0.763 95% CI 0.745–0.781, p < 0.001 and 0.771, 95% CI 0.752–0.789, p < 0.001) than a combination of CP, FP and CAC (0.675, 95% CI 0.655–0.696, p < 0.001). Conclusions: Fatty liver is a major and independent determinant of carotid and coronary ATS but not of femoral ATS. The multiple site involvement and quantitative tonic relationship suggest fatty liver could promote coronary and cerebrovascular events. The prediction of cardiovascular mortality risk is reinforced by fatty liver status over imaging-based measurements of atherosclerotic lesions. Aknowledgements: EPoS Consortium - funded by the EU H2020. SAT-306 Ethnicity and outcomes in non alcoholic fatty liver disease: prospective evaluation of 1500 cases R. Parker 1 , O. Cain 1 , C. Raymonde-Parker 1 , E. Aron 1 , S. Hubscher 2 , P. Newsome 1 . 1 NIHR Center for Liver Research, University of Birmingham; 2 Histopathology, University Hospitals Birmingham NHS Foundation Trust, Birmingham, United Kingdom E-mail: richardparker@nhs.net Background and Aims: Non-alcoholic fatty liver disease (NAFLD) is a common liver disease that can cause progressive liver disease and cirrhosis in some individuals. NAFLD is closely associated with the metabolic syndrome, where ethnicity plays an important role in susceptibility to complications of obesity. We examined the impact of ethnicity on liver-related outcomes in our cohort of NAFLD patients in Birmingham, UK. Methods: Data from patients with NAFLD seen at a single specialist clinic were collected prospectively onto a departmental database. Alternative causes of liver disease and secondary causes of fatty liver were ruled out by history and appropriate tests. Liver biopsy was performed if there was uncertainty regarding diagnosis or stage of disease, and described with the Brunt/Kleiner system. Ethnicity using UK census categories was self-determined for most participants (1,419, 91%), with a minority determined using avalidated software program (NamPechan) that assigned likely ethnicity based on surname. Groups were compared with student’s t-test for normally distributed data and Mann Whitney test for skewed data. Survival was analysed with univariable Kaplan-Meier curves and multi- variable Cox-regression analysis. Results: 1562 patientsseen between 2003 and 2015 were included; median follow-up was 52 months (range 12–153). 282 (18%) of patients underwent liver biopsy. 142 deaths were observed[N1] (9%) and 24 patients (1.5%) underwent liver transplantation. 991 (63%) were White, 336 (22%) Asian, 28 (2.0%) Black [N2] patients,175 (11%) unknown and 32 other ethnicities. In univariate analysis survival differed significantly by ethnicity with worse survival in White patients (Tarone-Ware p < 0.001). Taking into account differences in baseline factors (age, BMI, AST:ALT ratio, presence of diabetes), survival remained better in Asian patients than White patients (HR for death 0.225, 95% CI 0.077–0.658, p = 0.006). These differences in survival were not explained by differences in severity of disease which was similar in distribution between ethnicities on both non-invasive tests (FIB-4, one-way ANOVA p= 0.092) and biopsy (F3 or F4 fibrosis, Fisher’s exact test p = 0.066). Conclusions: The likelihood of death is greater in White patients with NAFLD than in Asian patients. The reasons for this are not clear but may be behavoural – some markers of alcohol use were higher in White patients although this is not reliable. The pathophysiological causes for this clinical observation need to be investigated. SAT-307 Development and validation of an automated system for assessment of liver steatosis and fibrosis in routine histological images in patients with Non-Alcoholic Fatty Liver Disease R. Forlano 1,2 , J. Maurice 1 , N. Angkathunyakul 3 , E. Goldin 3 , M. Yee 4 , G. Serviddio 2 , N. Giannakeas 5 , A.T. Tzallas 5 , M.G. Tsipouras 6 , R.D. Goldin 3 , M. Thursz 1 , P. Manousou 1 . 1 Department of Hepatology, Imperial College, London, United Kingdom; 2 Centro Universitario per Ricerca e la cura delle Epatopatie, Università degli Studi di Foggia, Foggia, Italy; 3 Department for Cellular Pathology; 4 Department of Endocrinology, Imperial College, London, United Kingdom; 5 Department of Computer Engeneering, Technological Educational Institute, Epirus; 6 Department of Informatics and Telecommunication Engineering, University of Western Macedonia, K, Greece E-mail: robertaforl@gmail.com Background and Aims: Although liver biopsy is the gold standard method for diagnosing and staging NAFLD, up to date, the steatosis grade and fibrosis stage are reported using semi-quantitative scores. Inter-observer variability in the scoring system may impact on the interpretation of responses in clinical trials and more objective systems are required. We aimed to develop an automated method for steatosis and fibrosis quantitation using routine histological images of NAFLD patients. We further assessed the correlation of steatosis quantitation with steatosis grade and fibrosis quantitation with stage (Kleiner NASH CRN scoring system). Methods: 118 consecutive patients with biopsy-confirmed NAFLD were retrospectively evaluated. Twenty liver biopsies with steatosis <5% were used as controls. Biopsies were stained with H&E and Sirius red, and then scored by two histopathologists. They were also digitalized at 2x magnification and areas of steatosis were annotated manually by experts, using the NDP.view2, to facilitate machine learning. The same method was followed for fibrosis quantitation. Each image was then analysed by the automated software in two stages: Machine learning clustering and Morphological Image Processing (Figure 1). Fat % and fibrosis % computed bythe software were compared with the manual annotation for all cases. Results: There was excellent concordance between manual and automatic measurements with inter-class correlation coefficient (ICC) 0.981, 95%CI (0.96–0.99), F value = 103.3, p = 0.0001. There was good correlation between fat% and steatosis grade but with significant overlap between groups: Anova 50.9, p < 0.001 (Figure 2a) grade 0: fat 0.6%(0–3), grade 1: 7.5%(1–22), grade 2: 12.5%(2.5–29), grade 3: 20.8%(11.5–32). POSTER PRESENTATIONS S589 Journal of Hepatology 2017 vol. 66 | S543–S750