Portal de Periódicos da CAPES

How to identify patients with advanced liver disease in the community?

2017; Lippincott Williams & Wilkins; Volume: 66; Issue: 1 Linguagem: Inglês

10.1002/hep.29197

1527-3350

Terry Cheuk‐Fung Yip, Vincent Wai‐Sun Wong,

Liver Disease and Transplantation

Potential conflict of interest: Vincent Wong has served as an advisory board member for AbbVie, Gilead Sciences, Janssen, and Perspectum Diagnostics; is a consultant for Allergan and Pfizer; and is a speaker for Bristol‐Myers Squibb, Echosens, Gilead Sciences, and Merck. Vincent Wong was supported by the General Research Funds of the Research Grant Council of the Hong Kong SAR Government (project reference 477813 and 14108916). See Article on Page 84 End‐stage liver disease and hepatocellular carcinoma (HCC) cause much suffering and are often incurable. Although professional societies such as the American Association for the Study of Liver Diseases have provided guidance on the management of various chronic liver diseases, the majority of such patients are seen at primary care settings and may remain undiagnosed. This is particularly true for nonalcoholic fatty liver disease (NAFLD) and alcoholic liver disease as patients are usually asymptomatic for the liver condition and yet burdened by comorbidities. If our system cannot identify at‐risk patients for further management, no amount of therapeutic advance and surveillance algorithms can make an impact at the society level. In this issue of Hepatology, Unalp‐Arida and Ruhl1 report the prognostic implication of four noninvasive tests of liver fibrosis (aspartate aminotransferase–to‐platelet ratio index [APRI], fibrosis‐4 [FIB‐4] index, NAFLD fibrosis score, and the Forns index) in the general United States population. They performed a post hoc analysis on 14,841 adults with negative hepatitis B surface antigen and anti–hepatitis C virus antibody in the Third National Health and Nutrition Examination Survey (NHANES III) conducted from 1988 to 1994. The NHANES III was a cross‐sectional population study with detailed questionnaire and laboratory assessments in a broad range of subjects. In this study, the outcomes were censored in December 2011 using the National Death Index. At a median follow‐up of 19 years, the cumulative mortality was 28% from all causes and 0.8% with liver disease. The cumulative mortality for cardiovascular disease, cancer, and diabetes was 11%, 9%, and 3%, respectively. By age‐adjusted Cox regression analysis, all four fibrosis scores were consistently associated with all‐cause and liver‐related mortality. Compared with subjects with low fibrosis probability, the age‐adjusted hazard ratios for subjects with liver‐related mortality were 16 for intermediate‐high APRI, 3.5 for intermediate FIB‐4, 42 for high FIB‐4, 2.7 for intermediate NAFLD fibrosis score, 9.8 for high NAFLD fibrosis score, 5.8 for intermediate Forns index, and 117 for high Forns index. Because chronic hepatitis B and C had been excluded, one may expect that most cases of liver‐related death were secondary to NAFLD or alcoholic liver disease. As usual, we can learn a lot from the NHANES cohorts. The cumulative mortality from liver causes was 0.8%, which was much lower than that from cardiovascular disease, cancer, and diabetes. This is consistent with other natural history studies in the general population or NAFLD cohorts.2 Although NAFLD is rapidly becoming an important cause of end‐stage liver disease and HCC, hepatologists are seeing the extreme end of the spectrum. Despite the undoubted importance of these liver complications, their relatively low incidence makes it unrealistic to expect primary care physicians to systematically screen for advanced liver disease using sophisticated tools such as transient elastography and magnetic resonance elastography. The use of simple fibrosis scores is thus a step in the right direction. Of course, the diagnostic yield in the study by Unalp‐Arida and Ruhl would have been higher if they did not exclude patients with chronic viral hepatitis or focused on patients with known liver diseases or high risk factors, as has been done by other investigators.3 In the study by Unalp‐Arida and Ruhl, all four fibrosis scores were able to predict overall and liver‐related mortality. Although the authors did not compare the accuracy of the scores, a few observations from their data may inform practice. First, the incidence of liver‐related death in the low‐risk group as defined by any of the four fibrosis scores was less than 1% in 25 years. In other words, all four scores have excellent negative predictive value in identifying patients who may not require specific liver monitoring or treatment. In addition, the Kaplan‐Meier analysis suggests that the NAFLD fibrosis score was the least discriminating. This is probably because the NAFLD fibrosis score was specifically designed for the assessment of NAFLD patients, whereas the current cohort included patients with different liver diseases. Another factor to consider is the proportion of subjects classified as having low risk according to fibrosis scores. Given that the scores had equally high negative predictive values for liver‐related mortality, those that classify more patients in the low‐risk group would reduce unnecessary further investigations and may be more cost‐effective. In the NHANES III cohort, APRI, FIB‐4, NAFLD fibrosis score, and the Forns index classified 96%, 79%, 70%, and 80% of subjects, respectively, in the low‐risk group.1 On the other hand, it has been demonstrated that these scores have modest correlation with each other.5 Their low positive predictive values, particularly when applied in the low‐risk community setting, also mean that additional investigations are required to confirm the presence of advanced fibrosis. Because it is still a lot to ask for liver assessment in 20%‐30% of the entire population (as classified by FIB‐4, NAFLD fibrosis score, or the Forns index), APRI may be the best choice for an initial assessment. In contrast, for high‐risk groups such as patients with type 2 diabetes, the use of more robust investigations may be preferred. Previous screening studies using transient elastography and magnetic resonance elastography identified NAFLD and advanced fibrosis in more than 60% and 7% of diabetic patients, respectively.4 Magnetic resonance elastography has higher accuracy and success rate than transient elastography.8 This is particularly relevant to the Western population; the diagnostic accuracy of transient elastography may decline in more obese patients. Although the study by Unalp‐Arida and Ruhl has the strengths of a large sample size, long duration of follow‐up, and the use of population data, there are still unanswered questions. First, the analysis was solely based on baseline parameters. In the clinic, patients are seen repeatedly, and clinical parameters may change over time (weight reduction in NAFLD patients, for example). Future studies should use time‐dependent analysis to determine the role and meaning of follow‐up data. Competing risk analysis should also be considered because the other causes of death are much more common than liver‐related mortality. Second, although mortality is a more reliable outcome in registry studies, the role of fibrosis scores in predicting HCC and other liver‐related complications is also important. The study by Unalp‐Arida and Ruhl provides important longitudinal data and is a crucial step toward the detection of advanced liver disease in the community. Although it may not be realistic to expect primary care providers to actively look for advanced liver disease, the widespread use of electronic health records can make this happen (Fig. 1). In some systems, the computer can generate prompting messages based on patient characteristics. It should not be difficult to automatically draw relevant parameters from electronic records and calculate fibrosis scores such as APRI and to remind doctors of the possibility of advanced liver disease. Workup for the underlying liver disease and confirmation of advanced fibrosis may be done either at the primary care and referral settings. This may identify patients who may benefit from treatment of specific liver diseases as well as HCC and varices surveillance.Figure 1: Pathway toward detection of advanced liver disease in the community and prevention of adverse outcomes. FibroScan machine photo © Echosens. Reprinted by permission.The last decade has seen pivotal development in noninvasive tests of liver fibrosis. We are now moving from histological correlation to prediction of clinical outcomes. Ultimately, we must use the knowledge to not only predict but also prevent liver‐related complications through effective case identification.