P52 Comparing the prognostic value for liver disease screening of three liver fibrosis markers in the Welsh population from 2000 to 2017: a general population-based cohort study

<h3>Introduction</h3> We investigated the prognostic utility of three non-invasive liver fibrosis markers calculated from routine blood tests in the Welsh primary care population for identification of those at risk of cirrhosis or hepatocellular carcinoma (HCC). <h3>Methods</h3> Using the Secure Anonymised Information Linkage (SAIL) Databank at Swansea University (2000–2017) we identified a cohort with opportunistic liver blood tests that allowed calculation of 3 established liver fibrosis markers: the aspartate transaminase to alanine transaminase (AST/ALT) ratio, the AST to platelet ratio index (APRI) and the fibrosis-4 index (FIB-4). Via Cox regression we calculated 10-year risks of cirrhosis/HCC and compared the performance of the markers over a range of risk thresholds via decision curve analysis (DCA). We also reassessed their performance in higher risk groups including those with diabetes. <h3>Results</h3> There were 203,005 people with available blood tests enabling calculation of FIB-4, APRI and AST/ALT. At the commonly utilised cut-points to detect advanced fibrosis/cirrhosis of 3.25, 1.5 and 1.0 for FIB-4, APRI and AST/ALT respectively, the 10-year risks of cirrhosis/HCC were 7%, 19% and &lt;1%. DCA demonstrated the APRI has the greatest net benefit for detecting 10-year risk of cirrhosis/HCC in a general population when compared to the AST/ALT ratio or FIB-4. For example, at cirrhosis risks of 1–3%, the APRI identifies more true positives via fewer referrals compared to the FIB-4. If implemented at the 3% risk threshold, the APRI would generate 6,952 referrals for every 100,000 patients screened and capture 488 (53%) potential cirrhosis cases, with 6,464 (93%) of those referrals being false positives. The diabetic sub-group included 17,784 patients. If we implemented the same 3% risk threshold but restricted to the 8.8% who would be expected to have diabetes (8760 per 100,000 of the population), the APRI would generate 1932 referrals and capture 116 (73%) potential diabetic cirrhosis cases, with 1816 (94%) of those referrals being false positives. Applying a similar approach to the obese and hazardous alcohol use sub-groups (47,818 and 29,139 patients), at the 3% threshold the APRI would generate 2,478 and 3,742 referrals and capture 57% and 75% potential cirrhosis/HCC cases in these populations respectively. <h3>Discussion</h3> Liver fibrosis markers in the general population are unlikely to be useful to identify those at risk of cirrhosis or HCC, due to too many false positive and false negative results. In higher risk populations, the markers have more acceptable sensitivity, identifying more true positives and generating fewer unnecessary referrals.