Hepatic decompensation (overt ascites, encephalopathy and variceal hemorrhage) and mortality in patients with nonalcoholic steatohepatitis (NASH) increases exponentially with the development of cirrhosis. Prevention of such liver-associated clinical events (LACE) is therefore a major goal of therapeutics. This requires identification of those at risk and targeting them with effective therapeutics. Current approaches to identify this population include the use of fibrosis markers such as FIB4, liver-stiffness measurement (LSM), assessment of hepatic venous pressure gradient and/or varices (endoscopy) and traditional measures of liver function. These are unfortunately limited by modest accuracy of some models, retrospective, single center nature of most data-sets published and lack of information on the impact of changes in these parameters over time with respect to dynamic changes in risk profile. This proposal innovates by novel application and integration of systemic, hepatic perfusion and function, and portal hypertension related parameters to holistically model the risk of LACE and generate a probability score that is sensitive to change. We will specifically focus on MRI-based measures of systemic metabolic dysfunction (metabo-phenotype), gadoxetate uptake and clearance (integrated measure of hepatic perfusion and clearance function) and spleen stiffness measurement (SSM) as a surrogate for portal hypertension. We further propose the novel hypothesis that holistic models including these measures and capturing the multi-factorial origin of hepatic decompensation events are superior to conventional tools used in routine practice to define the probability of LACE in patients with compensated cirrhosis due to NASH. The hypothesis will be tested in a longitudinal cohort study of compensated cirrhosis due to NASH. The specific measures of the metabo-phenotype will include visceral adipose tissue volume, fat-free muscle volume, muscle fat infiltration. Gadoxetate uptake and clearance will be measured in a 10-minute exam. The SSM probe was recently approved and will be deployed on Fibroscan 630 the flagship instrument for transient elastography. Metabo-phenotyping requires implementing an imaging protocol, based on a routine clinical imaging sequence, onto standard clinical scanners which adds less than 10 minutes of scan-time while gadoxetate is routinely used for liver-imaging. Sub-aim 1 will model baseline parameters, alone and in combination, to define outcome risk within 1-2 years. Sub-aim 2 will evaluate dynamic changes in values of the test measures over time and relate them to changes in the risk of outcomes. Both regression and machine learned approaches will be taken to generate probability scores of the outcomes. Sensitivity analyses will be performed to test the robustness of the models. The patient populations needed and all of the methods/expertise are available. The rationale of this proposal is that it will provide a method for risk-i...