Nonalcoholic fatty liver disease (NAFLD) is an increasingly common cause of cirrhosis and on pace to be the leading indication for liver transplantation in the United States.(1, 2) NAFLD presents as a spectrum of disease ranging from isolated steatosis, which portends little risk of significant morbidity, to nonalcoholic steatohepatitis (NASH), which is characterized by inflammation and cell death and has substantial risk of progression to cirrhosis and liver-related mortality.(3) Unfortunately, liver biopsy remains the only way to accurately discriminate between isolated steatosis and NASH; however the procedure is invasive and remains impractical to scale to the estimated affected population of 60 million adults in the United States. Attempts to use individual or small combinations of biomarkers to characterize risk in NAFLD have been largely unsuccessful leaving a tremendous need for non-invasive risk stratification. My central hypothesis is that distinct subtypes of NAFLD can be identified by combining multiple non-invasive biomarkers, genetic and clinical factors using advanced analytic techniques for high dimensional data. Through my collaboration with the NIH-funded, multicenter NASH Clinical Research Network (NASH CRN) I explored the association between 28 putative plasma biomarkers and NAFLD histology and found that small sets of biomarkers were limited in discriminating between clinically significant stages of histologic severity. However, by applying a novel statistical technique, latent class analysis (LCA), we generated preliminary data identifying distinct subgroups of patients with NAFLD that are strongly associated with histologic severity. The research goal of this application is to (1) combine clinical and dietary factors, genetic markers and an expanded set of plasma biomarkers to refine distinct phenotypes of NAFLD using LCA, (2) validate the association between LCA defined phenotypes and histologic severity in an independent cohort with biopsy proven NAFLD, (3) build on an existing longitudinal cohort and test the ability of these phenotypes to predict progression of fibrosis and inflammation. My long-term goal is to combine expertise in multimodal, non-invasive biomarkers of NAFLD with advanced analytic techniques to personalize the management and treatment of patients with NAFLD. In order to accomplish this goal, I have assembled an exceptional mentorship team including my primary mentor, Dr. Rohit Loomba, who is an internationally renowned expert in NAFLD and Director of the UCSD NAFLD Research Center. In addition, Dr. Ariel Feldstein, Chief of the Division of Pediatric Gastroenterology, and an expert in translating NAFLD pathophysiology into biomarker development will serve as a co-mentor. Professor Lily Xu, biostatistical director of the UCSD Clinical and Translational Research Institute, will serve as my biostatistical mentor. Together, we formed a four-fold career development plan to gain expertise in (1) cohort development, biobanking and advanced NAFLD phenotyping, (2) statistical analysis of genetic and high dimensional data, (3) NAFLD pathobiology and biomarker development, and (4) research dissemination and the development of national recognition in the non-invasive assessment of NAFLD.
Despite the increasing prevalence of nonalcoholic fatty liver disease (NAFLD), our ability to determine who has the progressive form of disease relies upon liver biopsy, which is an invasive test that is impractical to scale to the affected population. This project will build upon preliminary work demonstrating that applying advanced analytic techniques to a combination of multiple modalities of non-invasive markers including levels of plasma cytokines and clinical factors can help uncover distinct subtypes of NAFLD associated with disease severity. This study will non-invasively stratify the phenotypes of NAFLD into low versus high risk of disease progression, which is desperately needed in order to prevent excessive testing and potential morbidity for patients at low risk for progression and to focus interventions on the population at greatest risk.
