S2368 Detection of Hepatocellular Carcinoma via Blood-Based Testing in High-Risk Individuals

Introduction: Development and integration of reliable cancer detection tests into clinical care for patients with cirrhosis may aid in the detection of hepatocellular carcinoma (HCC) when it can be more effectively treated. Currently, physicians rely on ultrasound and alpha-fetoprotein testing. The introduction of a more convenient blood test could enhance adherence and improve patient outcomes. To be clinically useful, blood tests must demonstrate relatively high sensitivity to effectively screen for cancer in this higher-risk population while also demonstrating high specificity to minimize false positives and avoid unnecessary follow-on procedures. Here we present the findings when retrospectively using a plasma cell-free DNA methylation-based blood test (Shield) to differentiate patients with fibrosis or cirrhosis from patients who have been diagnosed with HCC. Methods: An epigenomic classifier was developed for the detection of advanced and early-stage HCC in plasma derived cell-free circulating tumor DNA (ctDNA) using clinical late-stage HCC samples as cases (N = 551) and samples determined to be cancer-free as controls (N = 8578). To evaluate performance of this classifier, patients who received a diagnosis of liver fibrosis, liver cirrhosis, or HCC (based on International Classification of Diseases, Tenth Revision, Clinical Modification codes) were identified in a HIPAA-compliant database that links health claims (combination of open and closed claims) with genomic and epigenomic information. Two separate cohorts were created: (1) patients who had any diagnosis of liver fibrosis or cirrhosis prior to testing and no evidence of HCC after testing; and (2) patients of who received a diagnosis of HCC after testing but no HCC diagnosis before. The epigenomic classifier was applied to each cohort. Results: 152 patients had liver fibrosis or cirrhosis with no evidence of HCC after testing (cohort 1), and 10 had a new diagnosis of HCC after testing (cohort 2). The AUC for the HCC epigenomic classifier using the 2 groups was 0.76. Fixing the specificity at 90%, the corresponding sensitivity was 60%. Conclusion: This feasibility study using real-world evidence and retrospective analysis of patient information demonstrated the blood-based epigenomic classifier has 60% sensitivity and 90% specificity for HCC detection. Integrating such a test in clinical practice may increase adherence to screening and lead to improved clinical outcomes.