No abstract

ObjectiveThe investigation regarding the clinical significance of quantitative hepatitis B core antibody (anti-HBc) during chronic hepatitis B (CHB) treatment is limited. The aim of this study was to determine the performance of anti-HBc as a predictor for hepatitis B e antigen (HBeAg) seroconversion in HBeAg-positive CHB patients treated with peginterferon (Peg-IFN) or nucleos(t)ide analogues (NUCs), respectively.DesignThis was a retrospective cohort study consisting of 231 and 560 patients enrolled in two phase IV, multicentre, randomised, controlled trials treated with Peg-IFN or NUC-based therapy for up to 2 years, respectively. Quantitative anti-HBc evaluation was conducted for all the available samples in the two trials by using a newly developed double-sandwich anti-HBc immunoassay.ResultsAt the end of trials, 99 (42.9%) and 137 (24.5%) patients achieved HBeAg seroconversion in the Peg-IFN and NUC cohorts, respectively. We defined 4.4 log10 IU/mL, with a maximum sum of sensitivity and specificity, as the optimal cut-off value of baseline anti-HBc level to predict HBeAg seroconversion for both Peg-IFN and NUC. Patients with baseline anti-HBc ≥4.4 log10 IU/mL and baseline HBV DNA <9 log10 copies/mL had 65.8% (50/76) and 37.1% (52/140) rates of HBeAg seroconversion in the Peg-IFN and NUC cohorts, respectively. In pooled analysis, other than treatment strategy, the baseline anti-HBc level was the best independent predictor for HBeAg seroconversion (OR 2.178; 95% CI 1.577 to 3.009; p<0.001).ConclusionsBaseline anti-HBc titre is a useful predictor of Peg-IFN and NUC therapy efficacy in HBeAg-positive CHB patients, which could be used for optimising the antiviral therapy of CHB.

Commonsense reasoning aims to empower machines with the human ability to make presumptions about ordinary situations in our daily life. In this paper, we propose a textual inference framework for answering commonsense questions, which effectively utilizes external, structured commonsense knowledge graphs to perform explainable inferences. The framework first grounds a question-answer pair from the semantic space to the knowledge-based symbolic space as a schema graph, a related sub-graph of external knowledge graphs. It represents schema graphs with a novel knowledge-aware graph network module named KA GNE T, and finally scores answers with graph representations. Our model is based on graph convolutional networks and LSTMs, with a hierarchical path-based attention mechanism. The intermediate attention scores make it transparent and interpretable, which thus produce trustworthy inferences. Using ConceptNet as the only external resource for BERT-based models, we achieved state-of-the-art performance on the CommonsenseQA, a large-scale dataset for commonsense reasoning. We open-source our code 1 to the community for future research in knowledge-aware commonsense reasoning.