Hepatocellular carcinoma incidence is reduced in cirrhotic chronic hepatitis B patients with HBsAg seroclearance comparing to those with viral suppression

include DM as a factor.We aimed to compare the performance of HCC risk scores among DM and non-DM patients on nucleos (t)ide analogue (NA) treatment.Method: Adult CHB patients on at least 6 months of entecavir or tenofovir treatment from January 2005 to March 2020 were identified using a territory-wide database in Hong Kong.DM was defined by any use of non-insulin antidiabetic agents, continuous use of insulin for ≥28 days, haemoglobin A 1c ≥6.5%, fasting glucose ≥7 mmol/L, and/or diagnosis codes.Two HCC risk scores with DM as a factor for treated CHB patients (i.e., cirrhosis, age, male sex, and DM [CAMD] score and Real-world Effectiveness from the Asia Pacific Rim Liver Consortium for hepatitis B virus [REAL-B] score) were studied; 2 other HCC risk scores without DM as a factor (i.e., PAGE-B and modified PAGE-B [mPAGE-B] scores) were also examined.The discriminatory power of the scores was assessed by area under the time-dependent receiver operating characteristic curves (AUROCs) with death as a competing event.Comparisons were done based on 1, 000 bootstrap samples.Results: Of 48,706 patients included, the mean age was 54.3 ± 13.6 years; 62.1% were male and 12.7% had cirrhosis.The prevalence of DM rose steadily from 15.5% to 24.3% in those who started NA treatment in 2005-08 and 2017-20 respectively.At a median (25th percentile-75th percentile) follow-up of 4.4 (2.2-5.0)years, 2, 157 (4.4%) patients developed HCC.All the 4 HCC scores were less predictive in DM patients than non-DM patients (all p < 0.001; Figure).DM was an independent risk factor for HCC on top of the risk groups of PAGE-B score (adjusted hazard ratio [aHR] 3.85, 95% CI 2.06-7.18,p < 0.001), and had an interaction with PAGE-B risk groups (aHR [95% CI]: 0.40 [0.21-0.77]for intermediate-risk group and 0.27 [0.14-0.51]for highrisk group as compared to non-DM patients in low-risk group).DM was however found to be not associated with HCC after adjusting for mPAGE-B score (aHR 1.04, 95% CI 0.95-1.14,p = 0.423).Figure: The area under the time-dependent receiver operating characteristic curves (AUROCs) of A. PAGE-B score; B. mPAGE-B score; C. CAMD score; and D. REAL-B score for predicting the development of hepatocellular carcinoma in nucleos (t)ide analogue-treated chronic hepatitis B patients with and without diabetes mellitus (DM).Conclusion: HCC risk scores are less accurate in NA-treated diabetic CHB patients than their non-diabetic counterparts, with a drop of AUROCs from around 0.8 to 0.7 regardless of whether DM is a component in the risk scores or not.