4% vs. 6.4%). The regression of liver fibrosis due to the long-term use of antiviral agents may explain the improved long-term prognosis [38], [39]. However, because not all patients receiving antiviral treatment experience liver fibrosis regression [40], hepatic decompensation 17-AAG HSP and HCC can eventually occur in some patients despite antiviral treatment. Baseline HBV DNA level is the most important risk factor for HCC development without antiviral treatment [41]. However, we identified a significant predictive role for LSM with 19 kPa as an optimal cutoff value when appropriate suppression of HBV DNA using antiviral treatment was available. Similarly, Zakareya et al. identified a significant association between LSM and the development of cirrhosis-related complications in patients with CLD, and concluded that LSM values >32 kPa was associated with HCC development [42].

Because 19�C32 kPa for predicting LRE development is much higher than the generally accepted cutoff LSM value for cirrhosis (10.3�C11.0 kPa) [43], [44], it can be postulated that cirrhosis can be further sub-classified, which indicates that all patients with cirrhosis do not have identical prognoses according to severity. Because our study enrolled only patients with available histology before starting antiviral treatment, our results cannot be directly applied to patients who will receive antiviral treatment without baseline LBs. However, the purpose of this study was to evaluate the additional clinical implications of LSM value, when compared with histologic data as a reference standard, for CHB patients before starting antiviral treatment using NUCs.

We demonstrated that LSM with an optimal cutoff value might be useful in assessing the risk of LRE development in these patients, which is impossible using histologic data alone. Thus, LSM is not only a noninvasive tool for the one-time evaluation of the degree of liver fibrosis, but also a significant predictor of LRE development, which should be checked before antiviral treatment regardless of LB data, in CHB patients. Our results also suggest that LSM is more useful than LB, and that the incidence of LREs could only be identified using LSM, not histologic data. Since LSM values perform better than histology, further studies investigating the predictive ability of LSM in patients undergoing antiviral treatment using NUCs without baseline LB data are needed.

In our study, there was a significant overlap in LSM values between patients with F3 (5.7�C19.8 kPa) and F4 (4.4�C57.1 kPa), although the overall LSM values were significantly higher in F4 fibrosis stage. This can be explained Anacetrapib in part by the overestimating influence of necroinflammatory activity on LSM [45]. Indeed, the proportion of high activity (A3�C4) in F3 showed a trend to be higher than those of F4 (35.0% vs. 26.9%).