The use of noninvasive tests in the monitoring of NAFLD/NASH is a rapidly evolving field. In my view, the evolution to this point has been interesting in that the development of diagnostic tests seemed to await the arrival of US Food and Drug Administration–approved therapies, and, conversely, those who were working to develop therapeutics appeared to be waiting for an easier way to diagnose NASH. The diagnosis requires a biopsy, which is not practical as a public health measure considering that approximately 24% of US adults have NAFLD and approximately 1.5% to 6.5% of US adults have NASH, and there would be an insufficient number of people who are trained to perform and read all of those biopsies. Additionally, liver biopsies are invasive, uncomfortable, and not without risks, and the identification of steatohepatitis on biopsy is subject to interobserver variability. While we have learned a lot from liver biopsies and histologic assessment, this approach is not suitable for large-scale use in day-to-day clinical practice, and this is one of the reasons why only a very small fraction of patients with NAFLD have been evaluated for NASH.

Most patients with fatty liver are relatively asymptomatic until liver decompensation occurs, and, unless clinicians in primary care are already thinking about NAFLD/NASH, they will not be looking for it. This underscores the need to develop simple yet robust ways of identifying those who have NAFLD/NASH in day-to-day clinics. As clinicians, we really want to know the following: (1) whether the patient has NAFLD, and (2) whether the patient is at risk for liver-related events and mortality from a progression of their liver disease. It is critical to have simple tests, such as the ones we use to screen for diabetes and heart disease, that can identify individuals with NAFLD who are at high risk for adverse liver-related outcomes in the future. We therefore need a combination of diagnostic and prognostic information.

The recent wave of noninvasive tools that are under development are being tested not only against liver histology but also against clinical outcomes. The Fibrosis-4 (FIB-4) index was developed to identify advanced fibrosis in those with HIV and hepatitis C. This test is simple, yet very robust, and it has been translated across multiple liver diseases. There are data suggesting a link between FIB-4 2.67 or higher and all-cause mortality, as well as the progression to NASH, cirrhosis, end-stage liver disease, hepatocellular carcinoma, and liver transplantation. Additionally, data have emerged from the Non-Invasive Biomarkers of Metabolic Liver Disease (NIMBLE) project, which aims to standardize and validate noninvasive tests for the diagnosis and staging of NASH. In NIMBLE, NIS4 had superior diagnostic accuracy relative to alanine transaminase for the diagnosis of NASH. Further, the Enhanced Liver Fibrosis (ELF) test was very good for the detection of fibrosis. Researchers reported that ELF was superior to FIB-4, and ELF showed increasing diagnostic accuracy for increasing stages of fibrosis.

I would suspect that, in the long-term, the need for a biopsy will be minimized as these types of evaluation paradigms emerge. If a patient has risk factors for NAFLD (eg, obesity, type 2 diabetes, and hypertension), clinicians will get either a FibroScan (Echosens) or another test to confirm the presence of fat in the liver. This information, combined with information from serum-based noninvasive tests, will enable clinicians to risk stratify patients into low-, intermediate-, and high-risk categories. Robust data showing that these evaluation paradigms predict mortality and liver outcomes are important. The principal reason for performing biopsies is to determine if a patient has NASH, because, if they do, they are more likely to have fibrosis and progress to cirrhosis.

Ultimately, these noninvasive evaluation strategies will allow clinicians to decide which patients require only observation, which should be followed more actively with more aggressive lifestyle interventions, and which will be the best candidates for drug therapy. Lastly, biomarkers will eventually be needed for disease monitoring.