Editorial: Using HBcrAg and HBsAg Kinetics to Optimise Monitoring After Nucleo(s)tide Analogue Withdrawal

Nucleo(s)tide analogues (NAs) remain the cornerstone therapy for patients with chronic hepatitis B (CHB). While long-term NA therapy effectively reduces the risk of cirrhosis, hepatocellular carcinoma, and hepatic decompensation, the annual rate of hepatitis B surface antigen (HBsAg) loss during continuous therapy remains low (< 1%) [1]. Notably, several studies have demonstrated increased HBsAg loss rates after NA withdrawal, presumably due to immune reactivation triggered by viral rebounds [1-3]. However, these rebounds may provoke excessive immune responses, potentially leading to severe hepatitis flares, hepatic decompensation, and even death [4]. Consequently, careful risk stratification is essential when considering NA withdrawal. Although low HBsAg levels at end of treatment (EOT) are consistently associated with higher rates of HBsAg loss [1, 2], even these patients remain at risk of relapses and decompensation [4], highlighting the need for additional biomarkers. One proposed biomarker is Hepatitis B core related antigen (HBcrAg), which comprises a combination of hepatitis B core antigen (HBcAg), hepatitis B e antigen (HBeAg) and the 22 kDa precore protein (p22Cr) [5]. An advantage of HBcrAg over HBsAg is that it may more accurately reflects intrahepatic transcriptional activity, particularly in HBeAg negative patients, as HBsAg can also be produced from integrated HBV DNA [6]. Prior studies suggest that lower HBcrAg levels at EOT (e.g., < 3.0 log10 IU/mL) are associated with fewer relapses and less severe flares [1, 7-9]. Nevertheless, adverse outcomes can still occur in patients with the most favourable EOT characteristics and thus timely identification of potentially harmful relapses necessitating early retreatment is needed. In the current issue of AP&T, Tsai et al. addressed this question by evaluating off-treatment dynamics of ultrasensitive iTACT-HBcrAg and HBsAg assays after NA withdrawal and their utility in predicting flares (ALT > 5xULN) and HBsAg loss in a cohort of 206 patients [10]. They reported that lower EOT HBcrAg levels were associated with fewer flares, lower peak ALT levels and higher HBsAg loss rates. Additionally, off-treatment HBcrAg dynamics outperformed HBsAg dynamics in predicting flares (time-dependent AUROCs: 69.7%–71.3% vs. 57.8%–65.2%). Conversely, HBsAg dynamics were superior in predicting HBsAg loss (aSHR: 0.33 vs. 0.95, per log IU/mL). These findings highlight the potential of HBcrAg monitoring to identify patients at risk for flares who may benefit from early retreatment. Strengths of the study include the relatively large patient population and prospective design with strict off-treatment follow-up. A limitation, however, is that the cohort consisted exclusively of Asian patients, underscoring the need for validation in larger multiethnic populations. Moreover, HBcrAg was measured only every 6 months, whereas earlier or more frequent testing may be necessary, particularly given the risk of early flares, which are known to be more severe [4, 8]. Furthermore, other biomarkers, such as HBV RNA and anti-HBc, were not evaluated and their comparative or additive value remains unclear. In conclusion, this study provides important new insights into off-treatment HBcrAg and HBsAg dynamics after NA withdrawal, supporting their potential for patient monitoring. Future studies should validate these findings and assess the additive value of combining HBsAg and HBcrAg with other biomarkers to establish thresholds to guide decision making regarding timely reinitiation of antiviral therapy. Edo J. Dongelmans: conceptualization, writing – original draft. Milan J. Sonneveld: conceptualization, writing – review and editing. The authors have nothing to report. Edo Dongelmans received speakers' fees from Roche and Gilead Sciences; Milan Sonneveld receives speakers' fees and research support from Roche, Gilead Sciences, GSK and Fujirebio. This article is linked to Tsai et al. papers. To view this article, visit https://doi.org/10.1111/apt.70658. Data sharing not applicable to this article as no datasets were generated or analysed during the current study.