A switch from tenofovir to entecavir prior to hepatitis B treatment cessation is associated with a reduced risk of off-therapy relapse: An observational study

Predictors of hepatic flares after nucleos(t)ide analogue cessation - Results of a global cohort study (RETRACT-B study)

BACKGROUND & AIMS

Flares after nucleos(t)ide analogue (NA) cessation are common and potentially harmful. Predictors of flares are required for risk stratification and to guide off-treatment follow-up.

METHOD

This multicenter cohort study included virally suppressed patients with chronic hepatitis B (CHB) who were hepatitis B e antigen negative at NA cessation. Hepatic flares were defined based on ALT levels of ≥5x, 10x or 20x the upper limit of normal (ULN). Multivariable Cox regression analyses were performed with censoring at retreatment, HBsAg loss or loss to follow-up. A sub-analysis was performed including HBV DNA levels within the first 12 weeks as a time-dependent covariate.

RESULTS

Of the 1,552 included patients, 350 developed a flare (ALT ≥5x ULN), of whom 70.6% did within the first year. One-year cumulative incidences for ALT flares ≥5x, ≥10x, ≥20x ULN were 18.6%, 10.2% and 3.4%, respectively. Severity of flares decreased over time, but severe flares still occurred after 1 year. Thirteen patients decompensated after a flare, of whom three died. Flares did not seem to be associated with increased rates of HBsAg loss (adjusted hazard ratio [aHR] 1.42, p = 0.28). Multivariable analyses showed that older age (aHR 1.02, p = 0.001), male sex (aHR 1.57, p = 0.003), HBsAg levels at NA withdrawal (100-1,000 IU/ml; aHR 1.99, p <0.001; >1,000 IU/ml; aHR 2.65, p <0.001) and tenofovir disoproxil fumarate vs. entecavir therapy (aHR 2.99, p <0.001) were predictive of flares (≥5x ULN). Early HBV DNA levels >5log10 IU/ml were associated with the highest risk of flares (aHR 2.36, p <0.001).

CONCLUSION

Flares are common after NA withdrawal, especially within the first year and can result in hepatic decompensation and death. Older age, male sex, higher HBsAg levels at end of treatment and tenofovir therapy were associated with a higher risk of flares. Close monitoring and retreatment should be considered if HBV DNA levels exceed 5log10 IU/ml within the first 12 weeks.

IMPACT AND IMPLICATIONS

This is the first large global multi-centered cohort study which provides detailed data about flares after nucleos(t)ide analogue cessation in patients with chronic hepatitis B. Older age, male sex, higher HBsAg levels at end of treatment and tenofovir therapy were associated with a higher risk of flares. These results could guide follow-up after withdrawal, helping clinicians identify high-risk patients and decide when to restart anti-viral therapy, which we recommend if HBV DNA levels exceed 5log10 IU/ml within the first 12 weeks.

CLINICAL TRIAL NUMBER

not applicable.

Higher end-of-treatment HBsAg levels is associated with later onset but not severe relapse in HBeAg-negative chronic hepatitis B patients stopping antivirals

BACKGROUND

Quantitative hepatitis B surface antigen (qHBsAg) level at end-of-treatment (EOT) predict clinical relapse (CR) after nucleos(t)ide analogues (Nuc) in chronic hepatitis B(CHB) patients. It is unclear if higher EOT qHBsAg leads to earlier onset or more severe off-Nuc CR.

AIM

This large cohort study investigates the association between EOT qHBsAg and CR onset/severity.

METHODS

This study enrolled HBeAg-negative CHB patients who had achieved undetectable HBV DNA for over 1 year after receiving Nuc therapy before discontinuation. The EOT qHBsAg level was categorised into three groups: <100, 100-999, ≥1000 IU/mL. The study assessed the predictability of qHBsAg levels for CR, and analysed and compared the incidence, time to onset and severity of CR among these three groups.

RESULTS

Patients with higher EOT qHBsAg showed a higher incidence of CR (≥1000, 100-999, <100 IU/mL: 73%, 65%, and 38%, p < 0.01) but a later onset of CR (median time to CR: 35, 33 and 27 weeks, p < 0.01). The predictabilities of EOT qHBsAg for CR were greater in patients aged <50-year-old or with genotype C than in those aged ≥50-year-old or with genotype B. There's no correlation between EOT qHBsAg level and ALT folds at CR (Pearson correlation coefficient: r = -0.03, p = 0.35). EOT qHBsAg was neither a predictor for severe hepatitis flare nor a predictor for hepatic decompensation.

CONCLUSIONS

Predictability using EOT qHBsAg levels for CR differed in subgroups of age and genotypes. Higher EOT qHBsAg levels correlate with higher incidence but later onset of CR. No correlation between EOT qHBsAg and relapse severity was observed.