Resolved hepatitis B is associated with higher reactivation risk than chronic infection after chemoembolization for hepatocellular carcinoma

Abstract

BACKGROUND

Transarterial chemoembolization (TACE) is a cornerstone treatment for hepatocellular carcinoma (HCC), yet the hepatitis B virus (HBV) reactivation risk, particularly among patients with prior infection, remains incompletely characterized, with international guidelines providing discordant risk classifications. We hypothesized that patients with resolved HBV infection undergoing TACE face a higher reactivation risk than currently estimated by international guidelines.

AIM

To determine the incidence and risk factors of HBV reactivation in patients undergoing TACE for HBV-related HCC.

METHODS

We retrospectively analyzed patients who underwent TACE for HBV-related HCC at a tertiary hospital during 2007-2025. Patients receiving baseline antiviral therapy were excluded. Patients with definitive serostatus (n = 774) were stratified into chronic-infection [hepatitis B surface antigen (HBsAg)-positive, n = 727] and resolved-infection (HBsAg-negative/anti-hepatitis B core antibody-positive, n = 47) groups. Cumulative incidences and risk factors were assessed using Kaplan-Meier analysis and Cox regression, with robustness evaluated through multiple imputation, competing-risks, and antiviral-naive sensitivity analyses.

RESULTS

During a median follow-up of 16.7 months, HBV reactivation occurred in 60 patients (6.1% overall). Patients with resolved HBV status had a higher reactivation rate (21.3% vs 6.9%; P < 0.001). Resolved HBV status (adjusted hazard ratio: 3.98, 95% confidence interval: 1.99-7.97; P < 0.001) and elevated alpha-fetoprotein level (1.33 per log₁₀ ng/mL, 1.04-1.69; P = 0.024) were independent predictors. Median time to reactivation was 12 months.

CONCLUSION

HBV reactivation occurred in approximately 6% of patients undergoing TACE, with resolved infection status conferring > 3-fold higher risk and reactivation incidence exceeding 20%. These findings support prophylactic antiviral therapy rather than monitoring alone for HBsAg-negative/anti-hepatitis B core antibody-positive patients undergoing TACE.

Key Words: Hepatitis B virus reactivation; Transarterial chemoembolization; Hepatocellular carcinoma; Resolved hepatitis B infection; Anti-hepatitis B core antibody positive

Core Tip: Current international guidelines classify resolved hepatitis B virus (HBV) infection as moderate-risk for reactivation during transarterial chemoembolization, yet our study of 774 patients with confirmed HBV serostatus reveals a 21.3% reactivation rate in hepatitis B surface antigen-negative/anti-hepatitis B core antibody-positive patients, substantially exceeding the moderate-risk threshold. Resolved HBV status was the strongest independent predictor of reactivation (adjusted hazard ratio: 3.98), paradoxically conferring higher risk than chronic infection. These findings challenge existing risk classifications and support prophylactic antiviral therapy rather than monitoring alone for this vulnerable population.

INTRODUCTION

Hepatitis B virus (HBV) reactivation is a well-documented and potentially life-threatening complication in patients receiving immunosuppressive or cytotoxic therapies[1-3]. This process results from the loss of immune control over HBV, leading to increased viral replication and subsequent hepatic injury that may range from asymptomatic virological changes to fulminant hepatic failure[4-6]. Given that HBV-related liver disease is the predominant etiology of hepatocellular carcinoma (HCC) in endemic regions including Asia[7], understanding and preventing HBV reactivation in this population is of paramount clinical importance. In patients with HCC, most of whom have an underlying chronic HBV infection, HBV reactivation is particularly concerning because the resulting hepatitis or hepatic decompensation can lead to an interruption in cancer treatment and ultimately worsen survival outcomes[8-12].

Transarterial chemoembolization (TACE) is a cornerstone treatment for unresectable HCC, particularly for intermediate-stage disease[13,14]. Although TACE is delivered locally rather than systemically, it can induce HBV reactivation through mechanisms including immunosuppression owing to hepatic ischemia, the release of chemotherapeutic agents into the systemic circulation via arteriovenous shunts, and disruption of the hepatic microenvironment[8,15]. International guidelines have stratified the reactivation risk to guide decision-making regarding prophylactic antiviral therapy. The 2025 American Gastroenterological Association guideline classifies TACE as carrying a high risk (> 10%) in hepatitis B surface antigen (HBsAg)-positive individuals and a moderate risk (1%-10%) in HBsAg-negative/anti-hepatitis B core antibody (anti-HBc)-positive patients (i.e., those who previously had an infection)[16]. Similarly, the 2025 European Association for the Study of the Liver guideline recommends risk-based antiviral prophylaxis according to the patient’s serological status[17].

Despite this framework, current guideline recommendations regarding TACE-related reactivation in HBsAg-negative/anti-HBc-positive patients are based on limited evidence. A recent systematic review of HBV reactivation during HCC therapies highlighted the paucity of data for HBsAg-negative/anti-HBc-positive patients, with evidence derived largely from small, single-center, retrospective studies[18-20]. Furthermore, the risk of reactivation in HBsAg-negative/anti-HBc-positive patients undergoing TACE remains incompletely characterized; existing guidelines categorize this population as having a moderate risk largely because of biological plausibility rather than robust empirical data[15,16]. This gap underscores the need for larger cohort studies in which serological subgroups are directly compared in order to refine risk stratification and inform prophylactic strategies. Herein, we investigated the incidence of HBV reactivation among patients undergoing TACE for HBV-related HCC. We also identified independent risk factors and specifically evaluated whether a resolved HBV infection confers a higher risk than a chronic HBV infection.

MATERIALS AND METHODS

Study population

We retrospectively identified patients who underwent TACE for HCC at Severance Hospital (Seoul, Korea) from September 2007 to March 2025. Patients were included if they had: (1) A confirmed diagnosis of HCC based on histopathological or non-invasive criteria according to international guidelines[21]; (2) Evidence of a current or past HBV infection, defined as HBsAg positivity, isolated anti-HBc positivity, or a clinical diagnosis with detectable HBV DNA; and (3) At least two HBV DNA measurements obtained during the follow-up period. Patients receiving antiviral therapy at baseline were excluded to assess the natural risk of HBV reactivation in the absence of such protection. Patients for whom baseline serological markers were missing were retained in the overall cohort because all had undergone HBV DNA testing, allowing assessment of the HBV reactivation incidence despite incomplete serostatus information. To ensure accurate risk stratification based on HBV serological status, 202 patients (20.7%) with insufficient serological data for definitive classification were excluded from the comparative analysis. Subsequently, the final analytic cohort (n = 774) was classified into two HBV-infection groups: Chronic (HBsAg-positive; n = 727) and resolved (HBsAg-negative/anti-HBc-positive; n = 47). Antiviral therapy, initiated during follow-up at the discretion of the treating physician, was permitted. This study was approved by the Institutional Review Board of Severance Hospital, Yonsei University Health System, No. 4-2020-1081, and the requirement for obtaining informed consent was waived owing to the retrospective design of the study and the use of de-identified data. The study protocol adhered to the ethical principles of the Declaration of Helsinki.

Definitions and outcomes

The primary outcome was HBV reactivation, defined in accordance with the 2021 Asian Pacific Association for the Study of the Liver guidelines[22] and the 2022 Korean Association for the Study of the Liver guidelines[21]. For patients with detectable baseline HBV DNA (predominantly those with a chronic HBV infection), reactivation was defined as an increase of ≥ 2 log₁₀ IU/mL over baseline. For patients with undetectable baseline HBV DNA (all patients with a resolved HBV infection and a subset with chronic HBV), reactivation was defined as the reappearance of HBV DNA to a level of > 100 IU/mL. In addition, reverse seroconversion of HBsAg (negative to positive) was considered reactivation in the HBsAg-negative/anti-HBc-positive (resolved) subgroup. These stratified criteria reflect biological differences in baseline virological status between serological populations and are concordant with major international guidelines.

Baseline characteristics were assessed at the time of the first TACE and included demographic data (age and sex), comorbidities (hypertension, diabetes mellitus, and dyslipidemia), liver function parameters (albumin and total bilirubin concentrations, the international normalized ratio, and the platelet count), cirrhosis status and Child-Pugh classification, HBV serology (HBsAg, hepatitis B e antigen, and anti-HBc concentrations), HBV DNA levels, and tumor characteristics [maximal tumor size and alpha-fetoprotein (AFP) and protein induced by vitamin K absence or antagonist-II concentrations]. The number of TACE sessions during the follow-up period was also recorded. HBV DNA levels were measured using real-time polymerase chain reaction assays with a lower limit of detection of 20 IU/mL. Patients were monitored via HBV DNA measurements at baseline and every 1-3 months during and after TACE treatment, per our institutional protocol. Details regarding the frequency and duration of HBV DNA measurements during follow-up are provided in Supplementary Table 1.

Statistical analysis

Baseline characteristics were presented as medians with quartiles 1-3 for continuous variables and as n (%) for categorical variables. Comparisons between patients with and those without HBV reactivation were performed using the Mann-Whitney U test for continuous variables and the χ² or Fisher’s exact test for categorical variables, as appropriate. The cumulative incidence of HBV reactivation was estimated using the Kaplan-Meier method, and comparisons between groups were performed using the log-rank test. Patients were censored at the last follow-up visit, death, or end of the study period, whichever occurred first. The time to event was defined as the period from the date of the first TACE to the date of HBV reactivation or censoring. The distribution of time to reactivation was analyzed to evaluate the temporal pattern of viral reactivation. Subgroup analyses were performed according to the presence/absence of baseline HBV DNA (detectable vs undetectable), number of TACE sessions (at least three vs less than three), and HBV status (chronic vs resolved infection). To identify independent predictors of HBV reactivation, univariate Cox proportional-hazards regression analysis was first performed for all candidate variables. Variables with P < 0.05 in the univariate analysis were subsequently entered into a multivariable Cox regression model. Hazard ratios (HRs) with 95% confidence intervals (CIs) were calculated, and results were visualized using a forest plot. Changes in HBV DNA levels over time were analyzed by comparing the mean values at baseline and at 0-3, 3-6, and 6-12 months after the first TACE between patients with and those without reactivation. Comparisons at each time point were performed using the Mann-Whitney U test.

To evaluate potential selection bias due to the exclusion of 202 patients with indeterminate serostatus, baseline characteristics between the analytic and excluded cohorts were compared. Thereafter, we also performed a multiple imputation by chained equations sensitivity analysis with 20 imputed datasets, incorporating clinical, virological, and tumor-related covariates along with survival information (the Nelson-Aalen cumulative hazard estimator and event indicator); results were pooled using Rubin’s rules. To address potential informative censoring of death, the cumulative incidence of HBV reactivation was re-examined using the Fine-Gray competing-risks framework, using death without HBV reactivation as the competing event. Additional Cox analyses were performed by incorporating the tumor number (as a continuous variable) or multifocal HCC status (at least two nodules vs solitary) as covariates, by restricting the analysis to patients who remained antiviral-naive throughout follow-up, and by applying alternative reactivation thresholds (the American Gastroenterological Association-based criterion of an HBV DNA level of ≥ 1000 IU/mL and a uniform ≥ 2 log₁₀ increase).

The HBV DNA monitoring intensity (the number of post-TACE measurements) was compared between serological groups. Clinical and biochemical outcomes (alanine aminotransferase elevation and hepatic decompensation indicators) were compared between patients with and those without HBV reactivation; these comparisons were descriptive and exploratory, and no formal adjustment for multiplicity was applied.

The proportional-hazards assumption was tested using scaled Schoenfeld residuals, and model discrimination was assessed using Harrell’s concordance statistic. All statistical analyses were performed using R software version 4.5.0 (R Foundation for Statistical Computing, Vienna, Austria); the survival package was used for Kaplan-Meier estimation and Cox regression, cmprsk for competing-risks analyses, mice for multiple imputation, and ggplot2 and forestplot for data visualization. A two-sided P value < 0.05 was considered statistically significant.

RESULTS

Patient characteristics

In total, 976 patients who underwent TACE for HBV-related HCC were included in the initial analysis. Their baseline characteristics are summarized in Table 1. The median age was 60.3 years (Q1-3, 53.3-67.6), and 788 (80.7%) patients were male. The majority had underlying cirrhosis (844 patients, 86.5%) with well-preserved hepatic function (Child-Pugh class A in 891 patients, 91.3%). Regarding HBV serology, 541 (55.4%) patients had detectable baseline HBV DNA, with a median level of 1.65 log₁₀ IU/mL (Q1-3, 0.00-3.75). The median tumor size was 2.4 cm (Q1-3, 1.6-4.0), and patients underwent a median of two TACE sessions (Q1-3, one to three) during the study period.

Table 1 Baseline characteristics of study population, median (interquartile range)/n (%).

Characteristics	Total, n = 976	Reactivation (+), n = 60	Reactivation (-), n = 916	P value
Demographics
Age, years	60.3 (53.3-67.6)	60.4 (52.7-65.4)	60.3 (53.3-67.7)	0.463
Male sex	788 (80.7)	49 (81.7)	739 (80.7)	0.985
Comorbidities
Hypertension	245 (25.1)	13 (21.7)	232 (25.3)	0.631
Diabetes	192 (19.7)	11 (18.3)	181 (19.8)	0.919
Dyslipidemia	116 (11.9)	8 (13.3)	108 (11.8)	0.879
Cirrhosis	844 (86.5)	55 (91.7)	789 (86.1)	0.308
Child-Pugh score A	891 (91.3)	55 (91.7)	836 (91.3)	> 0.999
Albumin, g/dL	4.0 (3.5-4.3)	4.0 (3.6-4.3)	4.0 (3.5-4.3)	0.744
Total bilirubin, mg/dL	0.7 (0.5-1.0)	0.7 (0.5-0.9)	0.7 (0.5-1.0)	0.315
INR	1.02 (0.97-1.09)	1.04 (0.97-1.10)	1.02 (0.97-1.09)	0.592
Platelets, × 109/L	130 (89-171)	128 (82-164)	131 (91-171)	0.332
HBV serology and virology
Chronic HBV (HBsAg+)	727 (74.5)	50 (83.3)	677 (73.9)
Resolved HBV (HBsAg-/anti-HBc+)	47 (4.8)	10 (16.7)	37 (4.0)
Indeterminate	202 (20.7)	0	202 (22.0)
HBeAg status				0.214
Positive	224 (23.0)	13 (21.7)	211 (23.0)
Negative	598 (61.3)	42 (70.0)	556 (60.7)
Unknown	154 (15.8)	5 (8.3)	149 (16.3)
HBV DNA detectable	541 (55.4)	30 (50.0)	511 (55.8)	0.460
HBV DNA, log10 IU/mL	1.7 (0.0-3.8)	1.3 (1.0-3.2)	1.7 (0.0-3.8)	0.290
Tumor characteristics
Maximal tumor size, cm	2.4 (1.6-4.0)	2.5 (2.0-3.7)	2.4 (1.6-4.0)	0.446
AFP, ng/mL	15.8 (4.8-166.6)	50.2 (5.1-218.7)	15.0 (4.7-163.4)	0.277
PIVKA-II, mAU/mL	43 (24-193)	38 (18-168)	44 (24-197)	0.241
TACE-related
TACE sessions	2 (1-3)	2 (1-3)	2 (1-3)	0.044

P values were calculated using the Mann-Whitney U test for continuous variables and the χ² or Fisher’s exact test for categorical variables, as appropriate. Indeterminate hepatitis B virus serostatus: Patients with incomplete serological data insufficient for classification as chronic or resolved hepatitis B virus infection.

INR: International normalized ratio; HBV: Hepatitis B virus; HBsAg: Hepatitis B surface antigen; anti-HBc: Antibody to hepatitis B core antigen; HBeAg: Hepatitis B e antigen; AFP: Alpha-fetoprotein; TACE: Transarterial chemoembolization; PIVKA-II: Protein induced by vitamin K absence or antagonist-II.

Based on HBV serological status, 727 patients (74.5%) were classified as having a chronic HBV infection, and 47 patients (4.8%) had previously had an HBV infection, whereas 202 patients had indeterminate serostatus due to incomplete serological markers. No significant differences in baseline demographics, liver function parameters, or tumor characteristics were observed between patients who developed reactivation and those who did not (Table 1). However, patients with reactivation underwent significantly more TACE sessions (P = 0.044). Of the 976 patients, 743 (76.1%) were initiated on antiviral therapy during follow-up, while 233 (23.9%) remained antiviral-naive throughout the observation period.

Incidence and timing of HBV reactivation

During a median follow-up of 16.7 months (Q1-3, 2.1-51.3), HBV reactivation occurred in 60 patients. Among the population with a confirmed serostatus, the incidence of reactivation differed substantially according to baseline HBV status: 6.9% (50/727) among patients with a chronic HBV infection vs 21.3% (10/47) among those who previously had an HBV infection (P = 0.002; Table 2). The median time from the first TACE to reactivation was 12.0 months (Q1-3, 6.7-22.2), with no significant difference between the chronic and resolved HBV groups (12.5 months vs 11.5 months, respectively; P = 0.820; Supplementary Figure 1). Approximately half of reactivations occurred within a year of TACE, and reactivation events continued to occur throughout the follow-up period. Patients with a resolved HBV infection underwent fewer HBV DNA measurements during follow-up than those with a chronic HBV infection (median: Three vs five; P < 0.001; Supplementary Table 2).

Table 2 Hepatitis B virus reactivation incidence and clinical characteristics according to hepatitis B virus status, median (interquartile range)/n (%).

Characteristics	Chronic HBV1, n = 727	Resolved HBV2, n = 47	P value
Reactivation incidence	50 (6.9)	10 (21.3)	0.002
Clinical characteristics
Time to reactivation, months	12.5 (6.9-21.7)	11.5 (3.7-22.4)	0.820
Peak HBV DNA, log10 IU/mL	5.3 (3.9-6.8)	1.5 (1.1-1.6)	< 0.001
Severity of reactivation
≥ 1 Log increase	47/50 (94.0)	7/10 (70.0)	0.052
Clinical outcome
All-cause mortality	29/50 (58.0)	5/10 (50.0)	0.733

P values were calculated using the Mann-Whitney U test for continuous variables and Fisher’s exact test for categorical variables.

¹Hepatitis B surface antigen positive.

²Hepatitis B surface antigen negative, antibody to hepatitis B core antigen positive.

HBV: Hepatitis B virus.

Cumulative incidence of HBV reactivation

Kaplan-Meier analysis revealed distinct patterns of reactivation across patient subgroups (Figure 1). In the analysis of the total cohort stratified according to the presence/absence of baseline HBV DNA, no difference was observed in the cumulative incidence of reactivation between groups (log-rank P = 0.689; Figure 1A). Similarly, the intensity of treatment, defined as the number of TACE sessions (three or more vs fewer than three), did not influence the cumulative reactivation risk (log-rank P = 0.413; Figure 1B). In contrast, in the comparison according to HBV serological status, patients who had previously had an HBV infection exhibited a higher cumulative incidence of reactivation than those with a chronic HBV infection (log-rank P = 0.001; Figure 1C).

Figure 1 Cumulative incidence of hepatitis B virus reactivation after transarterial chemoembolization. A: Cumulative incidence stratified according to presence/absence of baseline hepatitis B virus (HBV) DNA (detectable, n = 541; undetectable, n = 435); B: Cumulative incidence stratified according to number of transarterial chemoembolization sessions (at least three sessions, n = 302; fewer than three sessions, n = 674); C: Cumulative incidence stratified according to baseline HBV status: Chronic HBV infection (hepatitis B surface antigen-positive, n = 727) and resolved HBV infection (hepatitis B surface antigen-negative/antibody to hepatitis B core antigen-positive, n = 47). Patients for whom serological data were incomplete (n = 202) were excluded. Shaded areas represent 95% confidence intervals. HBV: Hepatitis B virus; TACE: Transarterial chemoembolization.

Risk factors for HBV reactivation

Univariable Cox regression analysis identified several factors associated with HBV reactivation (Table 3). A resolved HBV status (HR = 4.02, 95%CI: 2.04-7.93; P < 0.001), higher AFP level (HR = 1.43 per log₁₀ ng/mL increase, 95%CI: 1.12-1.82; P = 0.004), and larger tumor (HR = 1.14 per cm, 95%CI: 1.05-1.24; P = 0.002) were associated with an increased reactivation risk. Notably, the baseline HBV DNA level, presence/absence of HBV DNA, hepatitis B e antigen status, and number of TACE sessions were not significantly associated with reactivation in the univariate analysis. Upon multivariable Cox regression analysis, a resolved HBV status was the strongest independent predictor of reactivation (adjusted HR = 3.98, 95%CI: 1.99-7.97; P < 0.001), and the AFP level was also an independent predictor (adjusted HR = 1.33 per log₁₀ ng/mL, 95%CI: 1.04-1.69; P = 0.024). The tumor size was not an independent predictor (adjusted HR = 1.08 per cm, 95%CI: 0.98-1.19; P = 0.137; Figure 2). The multivariable model satisfied the proportional-hazards assumption (global P = 0.383) and showed acceptable discrimination for an explanatory model (Harrell’s C-index: 0.641).

Figure 2 Forest plot of multivariable Cox regression analysis for hepatitis B virus reactivation. Hazard ratios and 95% confidence intervals are shown for variables included in the final multivariable model. Variables with P < 0.05 in the univariable analysis were entered into the multivariable model. The vertical dashed line indicates hazard ratio = 1 (no effect). Orange denotes statistical significance (P < 0.05). HBV: Hepatitis B virus; AFP: Alpha-fetoprotein; CI: Confidence interval.

Table 3 Predictors of hepatitis B virus reactivation.

Variables	Univariate		Multivariable
	HR (95%CI)	P value	HR (95%CI)	P value
Age (years)	0.984 (0.958-1.010)	0.218
Male sex	1.159 (0.602-2.229)	0.659
Cirrhosis	1.432 (0.573-3.577)	0.443
Child-Pugh score, B/C	1.308 (0.523-3.269)	0.566
Baseline HBV DNA detectable	0.905 (0.545-1.501)	0.698
HBV DNA (log10 IU/mL)	0.924 (0.816-1.046)	0.210
TACE sessions, three or more	0.813 (0.483-1.369)	0.437
Albumin (g/dL)	0.810 (0.502-1.306)	0.387
Platelets (× 109/L)	0.998 (0.993-1.002)	0.361
AFP (log10 ng/mL)	1.429 (1.123-1.817)	0.004	1.325 (1.039-1.691)	0.024
PIVKA-II (log10 mAU/mL)	1.221 (0.815-1.830)	0.333
AST (U/L)	0.998 (0.988-1.008)	0.754
ALT (U/L)	1.000 (0.993-1.007)	0.986
HBsAg positive	1.757 (0.913-3.379)	0.091
Anti-HBc positive	1.891 (1.132-3.159)	0.015
HBeAg positive	1.103 (0.596-2.039)	0.755
Resolved HBV infection1	4.019 (2.038-7.925)	< 0.001	3.982 (1.990-7.968)	< 0.001
Total TACE sessions	0.925 (0.794-1.078)	0.320
Tumor size (cm)	1.141 (1.048-1.243)	0.002	1.077 (0.977-1.189)	0.137
Hypertension	0.682 (0.369-1.262)	0.223
Diabetes mellitus	0.849 (0.442-1.633)	0.624
Total bilirubin (mg/dL)	0.894 (0.475-1.684)	0.729
INR	3.437 (0.646-18.293)	0.148

Univariable and multivariable Cox proportional hazards regression analyses were performed. Variables with P < 0.05 in univariable analysis were included in the multivariable model.

¹Hepatitis B surface antigen negative, antibody to hepatitis B core antigen positive.

HR: Hazard ratio; CI: Confidence interval; HBV: Hepatitis B virus; TACE: Transarterial chemoembolization; AFP: Alpha-fetoprotein; PIVKA-II: Protein induced by vitamin K absence or antagonist-II; AST: Aspartate aminotransferase; ALT: Alanine aminotransferase; HBsAg: Hepatitis B surface antigen; anti-HBc: Antibody to hepatitis B core antigen; HBeAg: Hepatitis B e antigen; INR: International normalized ratio.

Viral dynamics during HBV reactivation

The pattern of viral reactivation differed substantially between the chronic- and resolved-HBV groups (Supplementary Table 3). Patients with a chronic HBV infection exhibited a greater viral surge during reactivation than those with a previous infection (Δlog₁₀ from baseline to peak: 3.24 log₁₀ IU/mL vs 1.00 log₁₀ IU/mL; P < 0.001; corresponding peak HBV DNA level: 5.33 log₁₀ IU/mL vs 1.49 log₁₀ IU/mL, P < 0.001). Longitudinal analysis of HBV DNA kinetics revealed divergent viral trajectories between patients with and those without reactivation (Figure 3). At baseline, HBV DNA levels were similar between groups. However, from 3-6 months onward, patients who developed reactivation showed progressively increasing viral loads, whereas those without reactivation maintained stable or declining HBV DNA levels. The difference became statistically significant at 3-6 months (P = 0.031) and more pronounced at 6-12 months (P < 0.001).

Figure 3 Hepatitis B virus DNA kinetics after transarterial chemoembolization, according to reactivation status. Mean hepatitis B virus (HBV) DNA levels (log₁₀ IU/mL) with standard errors of the mean are shown at baseline and during follow-up intervals (0-3, 3-6, and 6-12 months after first transarterial chemoembolization) for patients with reactivation (orange) and those without reactivation (blue). Baseline values were obtained from the first available HBV DNA measurement. For each follow-up interval, the maximum HBV DNA value within that period was used when multiple measurements were available. The number of patients with data for each time point is shown below the X-axis (reactivation positive/reactivation negative). P values were calculated using the Mann-Whitney U test. HBV: Hepatitis B virus; TACE: Transarterial chemoembolization.

Clinical significance of HBV reactivation

Patients who developed HBV reactivation had higher rates of alanine aminotransferase elevation at the last available laboratory assessment than those without reactivation [greater than three times the upper limit of normal: 26.7% vs 11.8%, P = 0.002; severe flare (greater than five times the upper limit of normal): 16.7% vs 6.3%, P = 0.005]. However, hepatic decompensation indicators (any indicator: 56.7% vs 46.6%, P = 0.168) and all-cause mortality (56.7% vs 45.4%, P = 0.119) were not significantly different between the two groups. Within the reactivation cohort, the rate of biochemical injury was not significantly higher in patients with chronic HBV infection compared to those with resolved HBV infection (n = 10) (Supplementary Table 4).

Subgroup analyses

To assess the robustness of our findings, we performed subgroup analyses, evaluating the effect of a resolved HBV status on the reactivation risk across clinically relevant subgroups (Supplementary Table 5). The increased risk associated with a resolved HBV status was consistent across all subgroups examined. In patients with baseline undetectable HBV DNA, a resolved HBV status conferred a 4.6-fold increased risk (HR = 4.59, 95%CI: 2.15-9.82; P < 0.001). The association remained significant regardless of TACE session frequency (at least three sessions: HR = 4.43, P = 0.003; fewer than three sessions: HR = 3.86, P = 0.005) and was observed both in patients with cirrhosis (HR = 3.63, P < 0.001) and in those without cirrhosis (HR = 12.82, P = 0.023), although the latter subgroup was small.

Sensitivity analyses

Baseline characteristics of the analytic (n = 774) and excluded indeterminate (n = 202) cohorts were largely similar, with minor differences in the cirrhosis prevalence and liver function variables (Supplementary Table 6). The multiple imputation by chained equations sensitivity analysis, in which we incorporated all 202 indeterminate patients, confirmed that a resolved HBV infection remained an independent predictor of reactivation (adjusted HR = 2.97, 95%CI: 1.43-6.16; P = 0.004; Supplementary Table 7). Alternative reactivation thresholds, the American Gastroenterological Association criterion of an HBV DNA level of ≥ 1000 IU/mL and a uniform ≥ 2 log₁₀ increase, captured zero events in patients with a resolved HBV status (Supplementary Table 8), supporting the specificity of the primary definition for this serological population. In the Fine-Gray competing-risks analysis, the subdistribution HR for a resolved HBV infection was 4.91 (univariate) and 5.16 (multivariable; Supplementary Table 9), confirming that accounting for the competing risk of death strengthens rather than attenuates the association (Supplementary Figure 2).

Incorporating tumor number (as a continuous variable) or multifocal HCC status (at least two nodules vs solitary) into the multivariable model did not alter the effect estimates for a resolved HBV status or the AFP level (Supplementary Table 10). In the antiviral-naive subset (n = 233), a resolved HBV status remained an independent risk factor for reactivation (AFP-adjusted HR = 3.69, 95%CI: 1.30-10.43; P = 0.014; Supplementary Table 11).

DISCUSSION

In this retrospective cohort study of 976 patients who underwent TACE for HBV-related HCC, including 774 with a confirmed HBV serostatus, we observed an overall HBV reactivation rate of 6.1%, indicating that TACE yielded a moderate risk for such reactivation (1%-10%) according to current international guidelines[16,17]. The most notable finding was that patients who had previously had an HBV infection exhibited a substantially higher reactivation rate than those with a chronic HBV infection, and a resolved HBV status emerged as the strongest independent predictor of reactivation. An elevated AFP level was the only other independent risk factor. The median time to reactivation was 12 months, indicating a delayed onset pattern distinct from that observed with systemic immunosuppressive therapies.

Our findings extend previous studies on TACE-associated HBV reactivation in HBsAg-negative patients. Jang et al[18] reported an 11.0% reactivation rate among 109 HBsAg-negative patients undergoing TACE-based therapy, identifying treatment intensity and a history of chronic HBV infection as independent predictors. Peng et al[19] observed a 9.3% reactivation rate in 43 HBsAg-negative/anti-HBc-positive patients, with cirrhosis and cumulative TACE cycles as risk factors. Importantly, prior virological studies confirmed the persistence of intrahepatic covalently closed circular DNA and low-level HBV transcription even after HBsAg loss, supporting the biological plausibility of reactivation in patients who previously had an HBV infection[23,24]. The notably higher reactivation rate in our resolved-HBV cohort (21.3%) than those in previous studies likely reflects crucial differences in the surveillance duration and treatment context. Peng et al[19] limited their monitoring to only 6 months after treatment completion, a duration insufficient to capture the delayed reactivation pattern observed in our study (median: 12 months). Consequently, the risk in their cohort may have been underestimated owing to missed late-onset events.

Nonetheless, all three studies consistently demonstrate that patients with a resolved HBV status who undergo TACE face a clinically meaningful reactivation risk. The reactivation rate of 6.9% in our chronic HBV group is consistent with the 5%-7% rates reported in a systematic review of HBsAg-positive patients receiving locoregional therapies for HCC[20]. Therein, Papatheodoridi et al[20] demonstrated pooled TACE-associated reactivation rates of 23% without prophylaxis vs 1% with prophylaxis among HBsAg-positive patients, while notably highlighting the paucity of data for HBsAg-negative/anti-HBc-positive patients, a gap that our study helps address. Importantly, our findings suggest that the risk attributed to a prior HBV infection may be higher than previously presumed, underscoring the need to re-evaluate current risk classifications.

The higher reactivation risk observed in patients with a resolved HBV status compared to that in patients with a chronic HBV infection, despite neither group receiving prophylactic antiviral therapy, suggests an inherent immunological vulnerability in the resolved-HBV population. Patients with a chronic HBV infection maintain a dynamic equilibrium between viral replication and host immunity, with persistently activated HBV-specific T-cells providing ongoing immune surveillance. In contrast, patients with a resolved HBV infection rely on memory T-cells that remain relatively dormant in the absence of antigenic stimulation; this dormant immune state may be slower to respond when TACE-induced immunomodulation disrupts the equilibrium, thereby permitting viral escape from residual intrahepatic covalently closed circular DNA[8,25]. This interpretation is supported by recent evidence: Hoogeveen et al[26] showed that functionally cured patients harbor HBV-specific CD4-positive T-cells with a resting central memory phenotype, whereas chronic carriers maintain an activated effector memory profile indicative of ongoing functional engagement. Moreover, Heim et al[27] identified an “attenuated” CD8-positive T-cell subset enriched in patients with endogenously controlled HBV infections, reinforcing that a resolved infection is not immune-quiescent but depends on a fragile memory state that may fail under TACE-induced immunomodulation. As demonstrated by Werle-Lapostolle et al[28], HBV reactivation is fundamentally driven by the persistence of fully replication-competent covalently closed circular DNA within the hepatocyte nucleus, which serves as a reservoir capable of triggering viremia as soon as the host immune control is disrupted. This immune control, while effective under normal circumstances, may be more susceptible to disruption by TACE-induced immunomodulation. Additionally, the definition of reactivation differs between groups according to established guidelines: Chronic HBV requires a ≥ 2 log₁₀ increase from a detectable baseline, whereas viral reappearance (> 100 IU/mL) constitutes reactivation in those with a resolved HBV status. Although this differential threshold could theoretically inflate the reactivation rate in the resolved-HBV group, the clinical significance remains valid because any detectable viremia in HBsAg-negative patients represents a failure of immune control and warrants clinical attention. Moreover, 70% of patients with resolved HBV who experienced reactivation demonstrated a ≥ 1 log₁₀ viral surge, indicating genuine viral replication rather than merely crossing a low detection threshold.

The median time to reactivation of 12 months is notably longer than the 1-3 months typically reported with systemic chemotherapy or B cell-depleting agents[2,8] and also exceeds the median of 84.5 days reported in a prior TACE study[29]. This delayed onset likely reflects the distinct mechanism by which TACE affects immune surveillance. Unlike rituximab, which causes acute, profound B-cell depletion, TACE induces sustained immunological changes within the liver microenvironment through repeated hepatic ischemia, necroinflammation, and cytokine release[8,9]. Furthermore, TACE typically involves the localized delivery of anthracyclines (e.g., doxorubicin), which are classified as high-risk agents for HBV reactivation when used systemically[30], potentially contributing to the immunosuppressive milieu despite their locoregional application. Although the number of TACE sessions was not an independent predictor of reactivation in our multivariable model, the prolonged presence of a tumor combined with repeated procedures likely contributes to the gradual erosion of HBV-specific T-cell immunity, eventually permitting viral escape from the covalently closed circular DNA reservoirs.

The emergence of AFP as an independent predictor provides insight into the host-tumor-virus interaction. Elevated AFP levels correlate with a higher tumor burden and may serve as a surrogate marker for an immunosuppressive tumor microenvironment. High AFP levels have been associated with impaired dendritic cell function and T-cell exhaustion, creating an immune-tolerant environment that favors viral escape. This suggests that biologically aggressive tumors cause a host environment prone to loss of viral control. While the mechanistic link between AFP and the reactivation risk remains incompletely understood, these associations raise the possibility that tumor-related immune alterations may modestly influence susceptibility to HBV reactivation. Further studies are needed to clarify whether an elevated AFP level reflects a causal pathway or is simply a marker of overall host immune status. Recent translational evidence supports this framework: Tumor-derived AFP directly impairs natural killer-cell, monocyte, and dendritic cell function while promoting regulatory T-cell differentiation[31], and AFP-exposed dendritic cells exhibit a reduced T-cell stimulatory capacity with increased transforming growth factor-β secretion[32]. These findings suggest that high baseline AFP levels may mark a tumor-driven immunosuppressive milieu that weakens host control of latent HBV, consistent with our observation that the AFP level, but not baseline virological markers, predicted reactivation. However, the association with the AFP level dissipated in several sensitivity analyses (Supplementary Tables 7, 9, and 11), suggesting that the modest effect size is sensitive to modeling assumptions and cohort composition. An intricate relationship between HBV reactivation and HCC recurrence has been reported; Huang et al[33] demonstrated that posthepatectomy HBV reactivation independently predicted shorter disease-free survival (HR = 1.21, P = 0.037) and poorer overall survival (HR = 1.41, P < 0.001) in HBV-related HCC, with a bidirectional relationship between reactivation and tumor progression. Although we did not assess oncological outcomes in this study, their results suggest that the prevention of reactivation may have implications beyond hepatic stability, providing an additional rationale for proactive antiviral strategies in HBsAg-negative/anti-HBc-positive TACE recipients.

The 2025 American Gastroenterological Association guideline classifies TACE as conferring a moderate risk (1%-10%) in HBsAg-negative/anti-HBc-positive individuals, with a conditional recommendation (i.e., weak recommendation based on limited evidence) for prophylaxis, while monitoring remains an acceptable alternative[16]. The 2025 European Association for the Study of the Liver guideline acknowledges that TACE poses a substantial (> 10%) risk in this population based on retrospective data in Asian cohorts, yet maintains monitoring as the default strategy when HBV DNA is undetectable[17]. The reactivation incidence of 21.3% in patients in our study who had previously had an HBV infection suggests that the actual risk may exceed current estimates, supporting consideration of prophylactic antiviral therapy rather than monitoring alone. Notably, HBsAg-negative/anti-HBc-positive patients are often not subjected to routine HBV DNA monitoring in clinical practice, as they are perceived to be at low risk for reactivation; this surveillance gap may lead to delayed detection and missed opportunities for timely intervention. Based on our findings, anti-HBc testing should be routinely performed in all patients with HCC prior to TACE, and those with a resolved HBV status, particularly those with elevated AFP levels, may benefit from prophylactic nucleos(t)ide analogue therapy. For patients who are only monitored, virological surveillance should extend at least 12-18 months after TACE completion given the delayed reactivation pattern observed in this study.

Notably, several factors traditionally considered risk factors for HBV reactivation were not significantly associated with reactivation in our analysis, including the baseline HBV DNA level, presence/absence of HBV DNA, hepatitis B e antigen status, and number of TACE sessions. This lack of an association may reflect the complex interplay between viral and host factors in the HCC setting, in which tumor-related immune dysfunction may overshadow the contribution of baseline viral characteristics. Unlike systemic chemotherapy in which intensity correlates with immunosuppression, TACE exerts a more localized effect with variable systemic exposure depending on tumor vascularity. Taken together, these findings suggest that the immune alterations associated with HCC and TACE may reduce the predictive value of conventional virological markers.

This study has several limitations. First, its retrospective, single-center design precluded causal inference, and the resolved-HBV subgroup was small, though it remains one of the largest cohorts reported for this population and findings were consistent across multiple analytical approaches[18,19]. Second, as residual tumor-level confounding (e.g., microvascular invasion) and the unavailability of anti-HBs titers[34,35] precluded full adjustment, the prognostic role of AFP should be interpreted cautiously. Clinical-outcome comparisons were exploratory rather than confirmatory, as cause-specific attribution of biochemical injury was not feasible in this multifactorial TACE-treated population. Third, the modest C-index (0.641) reflects substantial residual variance not captured by conventional clinical variables; the incorporation of immune biomarkers such as HBV-specific T-cell function and quantitative anti-HBs titers may improve predictive discrimination. Finally, as our results were derived from one Korean genotype-C cohort, external validation in multi-ethnic populations is warranted, and prospective studies with protocolized outcome adjudication will be needed to quantify the independent clinical impact of resolved-HBV reactivation.

CONCLUSION

In conclusion, HBV reactivation occurred in approximately 6% of patients undergoing TACE for HBV-related HCC, and a resolved HBV status conferred a > 3-fold higher risk, with an incidence exceeding 20%. Our study indicates that current guidelines may underestimate the reactivation risk in this population and warrants consideration of prophylactic antiviral therapy rather than monitoring alone for HBsAg-negative/anti-HBc-positive patients undergoing TACE. Prospective validation in multi-ethnic cohorts and cost-effectiveness analyses are needed before this approach can be universally recommended.