External validation of an algorithm combining multi‐analyte blood tests (FibroTest‐LCR1‐LCR2) to identify subjects at risk of hepatocellular carcinoma in patients with chronic liver disease

Summary Background and Aims The early non‐invasive prediction of hepatocellular carcinoma in patients with chronic liver disease, without or with cirrhosis, is needed. The algorithm FT‐LCR1‐LCR2 which sequentially combines the FibroTest and two multi‐analyte‐tests (LCR1‐LCR2) and increased the performance of alfa‐foeto‐protein (AFP) alone, was constructed and internally validated in the prospective FibroFrance cohort. The first aim was to externally validate the sensitivity and the 5‐year prognostic value of FT‐LCR1‐LCR2. Methods We selected patients included in HCC surveillance programs from the ongoing prospective cohort at Bondy‐Hospital, France. All patients had paired serum samples drawn before incidental cases of HCC detected during follow‐up (cases). Cases were blindly matched to controls who had not developed HCC during a similar follow‐up based on gender, age and fibrosis stages. The performance of the FT‐LCR1‐LCR algorithm was assessed and compared with the standard AASLD surveillance (F4+/+‐AFP). Results 159 patients were followed prospectively (median 5.1 yr), including 51 who had developed HCC (31 with cirrhosis and 20 without) and 108 controls (56 with cirrhosis and 52 without). Sensitivity of the FT‐LCR1‐LCR2 algorithm was higher than standard surveillance 51/51; 100% (95% CI 93.0‐1.00) vs 31/51; 60.8 (46.1‐74.2; P  < 0.001), and 5 yr‐survival without HCC was 90% (81‐99) vs 77% (67‐86; P  = 0.003) according to the FT‐LCR1‐LCR2. When the algorithm was repeated in the same 159 patients 1.5 years later, cases who developed cancer had a significant yearly increase in LCR2 = +0.014(SE = 0.02) compared to 0.00 (0.01) in controls P  = 0.003). Conclusion The FT‐LCR1‐LCR2 algorithm was validated in an independent external cohort for the prediction of HCC.