Surveillance failure in ultrasound for hepatocellular carcinoma: a systematic review and meta-analysis

Abstract

Recently, various advances have been achieved in the diagnosis and treatment of hepatocellular carcinoma (HCC). We read with great interest the paper by De Toni et al demonstrating the importance of early diagnosis of HCC by surveillance with respect to survival benefits, and that by Zeng et al emphasising risk stratifications in HCC surveillance. As these papers suggest, effective surveillance to detect early stage HCC is crucial in the management of patients atrisk. Although ultrasound (US) is recommended as a standard surveillance modality, 5 the sensitivity of US for detecting early stage HCC is suboptimal, being only 47%. Updated guidelines suggest alternative surveillance tools including MRI in patients likely to have inadequate US examinations. 5 Recently, MRI surveillance, including abbreviated MRIprotocols, has been drawing attention because of its high diagnostic performance compared with US. 8 However, considering its cost and availability, MRI surveillance cannot be applied to all atrisk patients, but should be done in patients who are prone to US surveillance failures and benefit from MRI surveillance. Thus, we conducted a metaanalysis to evaluate the incidence of US surveillance failures in detection of early stage HCC, and to determine risk factors for surveillance failure. We searched MEDLINE, EMBASE and Cochrane databases for original studies published before 6 July 2020, investigating the US surveillance failure for detecting of early stage HCC. Eligibility criteria were (1) population, patients at risk of HCC without prior history of HCC; (2) index test, US; (3) reference standard, pathological diagnosis or imaging followup; (4) outcomes, the incidence and risk factors of surveillance failure; (5) study design, observational studies and clinical trials. Surveillance failure was defined as a tumour detected beyond an early stage according to Barcelona Clinic Liver Cancer stage or the Milan criteria, or a tumour missed on US but later identified on CT or MRI. We calculated the pooled incidence of surveillance failure and the pooled OR for risk factors using a randomeffects model. Study quality was evaluated using the NewcastleOttawa Scale (online supplemental table 1). In our metaanalysis, we included 18 studies reporting the incidence of surveillance failure and five reporting risk factors for surveillance failure. Among 21 467 patients who underwent surveillance, 2670 developed HCC, with 833 (31.2%) Letter