Introduction
Hepatitis B virus (HBV) infection, a major global health burden, can lead to hepatic cirrhosis (annual incidence of 2.1%–6.0%),1 hepatocellular carcinoma (HCC; annual incidence of 3%–6% and a five-year cumulative incidence exceeding 20%),1,2 acute-on-chronic liver failure (ACLF), and portal hypertension complications. Although a favorable trend in hepatitis B mortality was observed between 1990 and 2019, a recent analysis indicated that this pace of decline remains insufficient, as persistent gaps in diagnosis and treatment continue to fuel the burden of advanced liver disease.3,4 Consequently, China faces formidable challenges in achieving the WHO’s goal of a 65% reduction in HBV-related mortality by 2030.3
Nucleos(t)ide analog (NA) therapy has become the cornerstone of chronic hepatitis B (CHB) treatment, achieving potent viral suppression, preventing hepatic histological progression, and reducing the incidence of adverse events.5–8 The effectiveness of NA treatment in achieving initial virological suppression and improving clinical outcomes is well established.9,10 However, the differences in long-term dynamic virological response patterns among patients receiving NA antiviral therapy in relation to clinical outcomes remain inadequately defined.
Therefore, this study aimed to investigate the associations between dynamic virological response patterns and clinical outcomes in a large, 10-year real-world cohort of patients with HBV-related cirrhosis.
Methods
Study population
This retrospective–prospective cohort included patients with HBV-related cirrhosis treated at Beijing You’an Hospital from January 2009 to December 2019. The inclusion criteria were as follows: age 18–75 years; confirmation of liver cirrhosis by histopathology or a composite of clinical criteria, including manifestations and signs of portal hypertension (such as esophagogastric varices, variceal bleeding, ascites, or hepatic encephalopathy), laboratory test abnormalities, abdominal imaging findings (ultrasonography or enhanced computed tomography/magnetic resonance imaging), or liver stiffness exceeding 20 kPa, in accordance with the Guidelines for the Prevention and Treatment of CHB (2022 Edition)11; chronic HBV infection, defined by current serum hepatitis B surface antigen (HBsAg) positivity; serum HBV DNA positivity at baseline and consecutive antiviral therapy with first-line NA (entecavir, tenofovir disoproxil fumarate (TDF), or tenofovir alafenamide); and at least two follow-up visits per year. The exclusion criteria were comorbid chronic liver diseases other than HBV and the development of HCC within six months of enrollment.
Clinical data
Patient data were collected from the electronic medical record system, including demographic characteristics, clinical and laboratory data, and imaging findings. Laboratory markers of liver function (alanine aminotransferase, aspartate aminotransferase, total bilirubin, and albumin) and renal function (blood urea nitrogen and creatinine) were measured using an automatic biochemical analyzer (AU5400, Olympus, Tokyo, Japan). Coagulation parameters (prothrombin time, thrombin time, and activated partial thromboplastin time) were measured using an automatic coagulation analyzer (ACLTOP 700, Beckman Coulter, USA) and presented as the international normalized ratio (INR). A full blood count was conducted using the XE-5000 instrument (Sysmex, Kobe, Japan). Serum HBV DNA levels were measured using either the Roche (Basel, Switzerland) Cobas/TaqMan qPCR assay (limit of detection [LOD] = 10 IU/mL) or the Roche magnetic bead–based HBV DNA viral load assay (LOD = 20 IU/mL). Serum HBsAg levels were quantified using a Roche E601 chemiluminescence analyzer (LOD: 0.05 IU/mL). The management of esophageal and gastric variceal bleeding, hepatic encephalopathy, ascites, hepatorenal syndrome, and spontaneous bacterial peritonitis in patients with cirrhosis and portal hypertension followed guideline recommendations.12–14
Definitions of virological response patterns
Complete virological response (CVR) was defined as undetectable serum HBV DNA levels or levels below the LOD, whereas partial virological response (PVR) was considered if HBV DNA remained detectable or above the LOD after the first two years of NA treatment.
Maintained virological response (MVR) was defined as serum HBV DNA levels remaining below the LOD or undetectable at any time during follow-up.
Virologic breakthrough (VBT) was defined as either serum HBV DNA becoming re-detectable or exceeding the LOD after CVR for at least three consecutive measurements during follow-up.15
“Functional cure” was defined as meeting the criteria for MVR and having serum HBsAg levels below the LOD (0.05 IU/mL, denoting HBsAg loss) for at least two consecutive measurements separated by a minimum interval of three months during ongoing NA therapy.
Follow-up and endpoints
Patients were followed at intervals of three to six months in the inpatient or outpatient department. The follow-up endpoints were the development of HCC or ACLF, liver-related in-hospital death, loss to follow-up, and the end of the study (December 31, 2024).
Diagnosis of clinical outcomes
The diagnosis of HCC was based on either liver histology or characteristic findings on dynamic contrast-enhanced computed tomography/magnetic resonance imaging in combination with serum levels of alpha-fetoprotein (AFP), AFP-L3, or protein induced by vitamin K absence or antagonist-II, in accordance with the China Liver Cancer Guidelines for the Diagnosis and Treatment of Hepatocellular Carcinoma (2024 Edition).16 If the clinical diagnosis of HCC was inconclusive, histological examination was imperative.
The diagnosis of ACLF was based on the Guideline for Diagnosis and Treatment of Liver Failure (2024 Edition)17 using the following criteria: INR ≥ 1.5 or prothrombin activity <40% and serum total bilirubin >12 mg/dL.
Hepatic recompensation was assessed according to the criteria of expanded Baveno VII,18 defined as having no complications of portal hypertension for ≥12 months (irrespective of the use of low-dose diuretics and/or lactulose/rifaximin), alongside stable liver function, defined as a Model for End-Stage Liver Disease score <10 and/or Child–Pugh class A (serum albumin >35 g/L, INR <1.50, and total bilirubin <34 µmol/L).19
Statistical analysis
Data were statistically analyzed using SPSS 26.0 (IBM, Armonk, NY, USA) and R 4.2.2 (The R Foundation for Statistical Computing, Vienna, Austria). Missing data were handled through multiple imputation via chained equations, with five imputations used to generate complete datasets. Normally distributed quantitative data were reported as the mean ± standard deviation. For two-group comparisons, the independent-sample Student’s t-test or Welch’s t-test was applied for data with homogeneous and heterogeneous variance, respectively. Similarly, for three-group comparisons, one-way ANOVA or Welch’s ANOVA was applied. Non-normally distributed data were expressed as the median (IQR) and analyzed using the Mann–Whitney U test for two-group comparisons and the Kruskal–Wallis test for comparisons across more than two groups. Categorical variables were presented as percentages (%) and assessed using the χ2 test or Fisher’s exact test.
The cumulative incidence of outcomes was plotted using Kaplan–Meier curves and compared by the log-rank test. The Benjamini–Hochberg method was used to adjust P-values for multiple pairwise comparisons. Associations between virological response patterns and outcomes were evaluated using Cox proportional hazards regression. A two-sided P < 0.05 was considered statistically significant.
Results
Baseline clinical characteristics
This study included 1,869 patients with HBV-related cirrhosis (Fig. 1). The mean patient age was 50.04 ± 10.87 years. The cohort included 1,302 men (69.7%), and 1,521 patients (81.4%) had decompensated cirrhosis. The baseline characteristics of the entire cohort are presented in Table 1 and Supplementary Tables 1–2.
Table 1Baseline characteristics of all patients and patients stratified by dynamic virological response patterns
| Characteristic | Overall (N = 1,869) | MVR (n = 1,222) | PVR (n = 151) | VBT (n = 496) | P |
|---|
| Demographics |
| Age (years), mean ± SD | 50.04 ± 10.87 | 49.73 ± 10.78 | 51.53 ± 11.30 | 50.09 ± 11.03 | 0.152 |
| Male, n (%) | 1,302 (69.7) | 828 (67.8) | 107 (70.9) | 367 (74.0) | 0.037 |
| Cirrhosis status and complications |
| Decompensated cirrhosis, n (%) | 1,521 (81.4) | 1,004 (82.2) | 125 (82.8) | 392 (79.0) | 0.272 |
| Ascites, n (%) | 1,047 (56.0) | 685 (56.1) | 92 (60.9) | 270 (54.4) | 0.372 |
| EGVB, n (%) | 392 (21.0) | 254 (20.8) | 23 (15.2) | 115 (23.2) | 0.103 |
| HE, n (%) | 748 (40.0) | 456 (37.3) | 64 (42.4) | 228 (46.0) | 0.004 |
| Comorbidities |
| T2DM, n (%) | 409 (21.9) | 266 (21.8) | 29 (19.2) | 114 (23.0) | 0.622 |
| Hypertension, n (%) | 374 (20.0) | 232 (19.0) | 31 (20.5) | 111 (22.4) | 0.272 |
| Laboratory parameters, median (IQR) |
| ALT (U/L) | 41.60 (25.80–68.20) | 40.45 (25.22–65.65) | 44.20 (25.30–68.50) | 45.50 (30.28–75.67) | 0.001 |
| AST (U/L) | 49.30 (34.50–74.50) | 48.20 (34.00–72.68) | 49.50 (33.75–71.45) | 52.80 (36.48–84.15) | 0.012 |
| ALB (g/L) | 37.70 (33.30–42.40) | 37.80 (33.40–42.50) | 36.30 (31.75–40.95) | 37.90 (33.20–42.25) | 0.008 |
| Tbil (µmol/L) | 23.90 (16.20–36.10) | 23.90 (16.12–36.80) | 26.00 (16.25–36.65) | 22.70 (16.35–34.55) | 0.545 |
| BUN (µmol/L) | 4.77 (3.89–5.89) | 4.73 (3.84–5.85) | 4.91 (3.95–6.28) | 4.87 (4.03–5.89) | 0.282 |
| Scr (µmol/L) | 63.80 (54.40–74.20) | 63.50 (53.80–74.20) | 62.60 (55.90–72.75) | 64.80 (55.68–74.65) | 0.112 |
| INR | 1.16 (1.04–1.30) | 1.15 (1.05–1.30) | 1.17 (1.03–1.35) | 1.15 (1.02–1.29) | 0.347 |
| PLT (×109/L) | 89.00 (62.00–132.00) | 87.50 (61.00–131.00) | 95.00 (63.50–132.50) | 93.00 (63.75–134.00) | 0.348 |
| Virology and disease severity |
| HBeAg positivity, n (%) | 980 (52.4) | 592 (48.4) | 96 (63.6) | 292 (58.9) | <0.001 |
| HBV DNA (log IU/mL), median (IQR) | 4.82 (3.28–6.21) | 4.64 (3.19–6.02) | 5.56 (3.78–6.90) | 5.15 (3.50–6.51) | <0.001 |
| MELD score, median (IQR) | 9.51 (7.67–12.21) | 9.47 (7.70–12.20) | 9.97 (7.81–12.37) | 9.46 (7.50–12.18) | 0.606 |
Five- and ten-year clinical outcomes in the overall cohort
The median duration of follow-up was seven years (IQR, 5–11), and the five- and ten-year cumulative incidence of HCC (23.0% and 31.2%, respectively), ACLF (5.1% and 6.9%, respectively), and liver-related death (4.3% and 5.9%, respectively) was recorded, as presented in Table 2.
Table 2Cumulative incidence of clinical outcomes in all patients and patients stratified by baseline cirrhosis status
| Population | Outcome | Cumulative incidence, % (95% confidence interval) [number of events]
|
|---|
| 1 Year | 3 Years | 5 Years | 10 Years |
|---|
| Overall (N = 1,869) | HCC | 8.3 (7.1–9.6) [n = 155] | 17.8 (15.9–19.5) [n = 333] | 23.0 (21.0–24.9) [n = 430] | 31.2 (28.9–33.5) [n = 583] |
| Liver-related death | 1.7 (1.1–2.3) [n = 31] | 3.5 (2.6–4.4) [n = 65] | 4.3 (3.3–5.3) [n = 80] | 5.9 (4.6–7.1) [n = 110] |
| ACLF | 2.9 (2.1–3.7) [n = 54] | 4.4 (3.4–5.3) [n = 82] | 5.1 (4.1–6.1) [n = 95] | 6.9 (5.5–8.3) [n = 129] |
| Compensated cirrhosis (n = 348) | HCC | 2.6 (0.9–4.2) [n = 9] | 8.6 (5.6–11.5) [n = 30] | 12.1 (8.6–15.4) [n = 42] | 20.0 (15.4–24.3) [n = 70] |
| Liver-related death | 0.3 (0.0–0.9) [n = 1] | 0.6 (0.0–1.4) [n = 2] | 0.6 (0.0–1.4) [n = 2] | 1.5 (0.0–3.0) [n = 5] |
| ACLF | 0.0 (0.0–0.0) [n = 0] | 0.3 (0.0–0.9) [n = 1] | 0.3 (0.0–0.9) [n = 1] | 1.0 (0.0–2.5) [n = 4] |
| Decompensated cirrhosis (n = 1,521) | HCC | 9.7 (8.1–11.1) [n = 148] | 19.9 (17.9–21.9) [n = 303] | 25.5 (23.3–27.7) [n = 388] | 34.0 (31.3–36.6) [n = 517] |
| Liver-related death | 2.0 (1.3–2.7) [n = 30] | 4.2 (3.1–5.2) [n = 64] | 5.2 (4.0–6.4) [n = 79] | 7.0 (5.5–8.6) [n = 107] |
| ACLF | 3.6 (2.6–4.5) [n = 55] | 5.3 (4.2–6.5) [n = 81] | 6.3 (5.1–7.5) [n = 96] | 8.7 (7.0–10.4) [n = 132] |
The annual incidence of HCC peaked (5.85 per 100 person-years) during the initial two years of NA therapy and declined to 3.08 per 100 person-years between years 3 and 6, remaining stable at 1.63–2.60 per 100 person-years thereafter (Fig. 2). The annual incidence of ACLF and liver-related death is presented in Supplementary Table 3.
Compared with patients with compensated cirrhosis, those with decompensated cirrhosis had a higher five-year (25.5% vs. 12.1%) and 10-year (34.0% vs. 20.0%) cumulative incidence of HCC (P < 0.001). Similarly, the incidence of ACLF and liver-related death significantly differed between these groups (Table 2).
Relationships of dynamic virological response patterns with five- and ten-year clinical outcomes
Overall, MVR, VBT, and PVR accounted for 65.4% (1,222/1,869), 26.5% (496/1,869), and 8.1% (151/1,869) of the cohort, respectively. When stratified by baseline compensation status, the MVR rate was lower in patients with decompensated cirrhosis than in those with compensated cirrhosis (64.0% [248/348] vs. 71.3% [974/1521], P = 0.029). Critically, during a median follow-up of 5.5 years after achieving initial CVR, 28.9% (496/1,718) of these patients subsequently developed VBT. The median time to VBT from initial CVR was 2.0 years (IQR, 1.0–4.0). The three- and five-year cumulative incidence of VBT after initial CVR reached 21.0% and 27.2%, respectively (Supplementary Table 4).
Compared with patients with VBT and PVR, those with MVR had a lower five- and ten-year cumulative incidence of HCC in both the compensated (five-year: 10.1% vs. 17.0%; ten-year: 14.2% vs. 33.6%) and decompensated cirrhosis subgroups (five-year: 19.5% vs. 36.7%; ten-year: 25.7% vs. 49.7%; all log-rank P < 0.001). Among patients with decompensated cirrhosis, those with VBT had a higher five- and ten-year incidence of HCC than those with PVR (five-year: 38.8% vs. 29.5%; ten-year: 53.2% vs. 32.2%; log-rank P = 0.022). Similarly, MVR was associated with a lower cumulative incidence of liver-related death in both the compensated and decompensated cirrhosis subgroups (Fig. 3, Table 3).
Table 3Cumulative incidence of clinical outcomes by baseline cirrhosis status with dynamic virological response patterns
| Population | Virological response patterns | Outcome | Cumulative incidence, % (95% confidence interval) [number of events]
|
|---|
| 1 Year | 3 Years | 5 Years | 10 Years |
|---|
| Compensated cirrhosis (n = 348) | MVR (n = 248) | HCC | 2.8 (0.7–4.8) [n = 7] | 8.9 (5.3–12.5) [n = 22] | 10.1 (6.2–14.0) [n = 25] | 14.2 (9.5–18.6) [n = 35] |
| | Liver-related death | 0.4 (0.0–1.2) [n = 1] | 0.4 (0.0–1.2) [n = 1] | 0.4 (0.0–1.2) [n = 1] | 0.4 (0.0–1.2) [n = 1] |
| | ACLF | 0.0 (0.0–0.0) [n = 0] | 0.4 (0.0–1.3) [n = 1] | 0.4 (0.0–1.3) [n = 1] | 0.4 (0.0–1.3) [n = 1] |
| PVR + VBT (n = 100) | HCC | 2.0 (0.0–4.7) [n = 2] | 8.0 (2.5–13.2) [n = 8] | 17.0 (9.3–24.0) [n = 17] | 33.6 (23.0–42.7) [n = 34] |
| | Liver-related death | 0.0 (0.0–0.0) [n = 0] | 1.0 (0.0–3.1) [n = 1] | 1.0 (0.0–3.1) [n = 1] | 4.3 (0.0–8.9) [n = 4] |
| | ACLF | 0.0 (0.0–0.0) [n = 0] | 0.0 (0.0–0.0) [n = 0] | 0.0 (0.0–0.0) [n = 0] | 2.8 (0.0–8.0) [n = 3] |
| Decompensated cirrhosis (n = 1,521) | MVR (n = 974) | HCC | 7.9 (6.2–9.6) [n = 77] | 15.8 (13.4–18.1) [n = 154] | 19.5 (16.8–22.1) [n = 190] | 25.7 (22.6–28.5) [n = 250] |
| | Liver-related death | 1.1 (0.4–1.7) [n = 11] | 2.7 (1.7–3.7) [n = 26] | 3.2 (2.0–4.3) [n = 31] | 3.9 (2.5–5.3) [n = 38] |
| | ACLF | 2.6 (1.6–3.5) [n = 25] | 3.7 (2.5–4.9) [n = 36] | 4.5 (3.2–5.8) [n = 44] | 5.1 (3.5–6.6) [n = 50] |
| PVR + VBT (n = 547) | HCC | 12.8 (10.0–15.6) [n = 70] | 27.5 (23.5–31.2) [n = 150] | 36.7 (32.4–40.8) [n = 201] | 49.7 (44.5–54.4) [n = 272] |
| | Liver-related death | 3.7 (2.0–5.3) [n = 20] | 7.0 (4.7–9.3) [n = 38] | 9.1 (6.4–11.8) [n = 50] | 13.6 (9.6–17.4) [n = 74] |
| | ACLF | 5.4 (3.4–7.2) [n = 30] | 8.2 (5.8–10.6) [n = 45] | 9.5 (6.7–12.3) [n = 52] | 16.5 (12.0–20.7) [n = 90] |
| PVR (n = 129) | HCC | 9.9 (4.4–15.0) [n = 13] | 23.2 (15.1–30.4) [n = 30] | 29.5 (20.6–37.4) [n = 38] | 32.2 (21.9–41.2) [n = 42] |
| | Liver-related death | 3.3 (0.1–6.4) [n = 4] | 6.1 (2.0–10.3) [n = 8] | 8.4 (3.0–13.5) [n = 11] | 10.2 (3.7–16.3) [n = 13] |
| | ACLF | 10.1 (4.8–15.2) [n = 13] | 12.7 (6.7–18.3) [n = 16] | 15.9 (9.0–22.3) [n = 21] | 21.5 (11.1–30.8) [n = 28] |
| VBT (n = 418) | HCC | 13.7 (10.3–16.9) [n = 57] | 28.8 (24.2–33.0) [n = 120] | 38.8 (33.8–43.4) [n = 162] | 53.2 (47.4–58.2) [n = 222] |
| | Liver-related death | 3.8 (1.9–5.6) [n = 16] | 7.3 (4.6–9.2) [n = 31] | 9.4 (6.2–12.3) [n = 39] | 14.2 (9.6–18.6) [n = 59] |
| | ACLF | 3.9 (1.9–5.7) [n = 16] | 6.8 (4.2–9.2) [n = 28] | 7.5 (4.8–10.1) [n = 31] | 16.4 (10.6–21.7) [n = 69] |
Among patients with decompensated cirrhosis, those with MVR had a lower five- and ten-year cumulative incidence of ACLF than those with PVR and VBT (five-year: 4.5% vs. 9.5%; ten-year: 5.1% vs. 16.5%; log-rank P < 0.001), and those with PVR had a higher cumulative incidence of ACLF than those with VBT (five-year: 15.9% vs. 7.5%; ten-year: 21.5% vs. 16.4%; log-rank P = 0.017). However, no significant difference in the cumulative incidence of ACLF was observed among these dynamic virological response patterns in patients with compensated cirrhosis (Fig. 3, Table 3). Furthermore, multivariable Cox regression analyses demonstrated that dynamic virological response patterns were independently associated with clinical outcomes. Detailed results are provided in Supplementary Tables 5–6.
Hepatic decompensation and recompensation
Among the 1,521 patients with decompensated cirrhosis at baseline, those with MVR had a higher clinical recompensation rate than those with VBT (34.1% vs. 22.5%; P < 0.001). Conversely, among the 348 patients with compensated cirrhosis at baseline, those with MVR had a lower rate of progression to decompensation than those with VBT or PVR (16.1% vs. 37.2% vs. 36.4%; P < 0.001; Table 4).
Table 4Association between dynamic virological response patterns and hepatic decompensation/recompensation
| Transition of compensation status | n/N (%) | P |
|---|
| Decompensated cirrhosis → recompensation |
| MVR | 332/974 (34.1%) | <0.001 |
| VBT | 94/418 (22.5%) | |
| Compensated cirrhosis → decompensation |
| MVR | 40/248 (16.1%) | <0.001 |
| PVR+VBT | 37/100 (37.0%) | |
| PVR | 8/22 (36.4%) | |
| VBT | 29/78 (37.2%) | |
“Functional cure” and clinical outcomes
The cumulative serum HBsAg clearance rate was 9.8% (52/935) during a median follow-up of seven years, and no significant difference was observed between patients with compensated and decompensated cirrhosis (8.2% vs. 10.5%; P = 0.467). Furthermore, the HBsAg clearance rates were 34.9% (19/72), 11.7% (15/278), and 5.7% (18/585) among patients with baseline quantitative HBsAg levels of <100, 100–1,000, and >1,000 IU/mL, respectively (Fig. 4A, Supplementary Table 7).
In the baseline HBsAg < 100 IU/mL subgroup, which exhibited the highest clearance rate, the median time to serum HBsAg clearance among the 19 responders was 4.0 years (IQR, 1.0–6.0). The clinical outcomes of this subgroup stratified by cirrhotic compensation status are presented in Supplementary Table 8.
Among patients with MVR, those achieving “functional cure” had the lowest five- and ten-year cumulative incidence of HCC (3.9% and 8.7%, respectively). Interestingly, a higher serum HBsAg level at the last measurement during follow-up was associated with an increased risk of HCC (Fig. 4B, Supplementary Table 9).
Discussion
This large, real-world, 10-year cohort study demonstrated that MVR to NA therapy is strongly associated with a lower incidence of HCC and liver-related death and higher recompensation rates in patients with HBV-related cirrhosis. These findings collectively emphasize the critical importance of sustained virological suppression and the pursuit of HBV functional cure.
Functional HBV cure is defined as undetectable HBsAg and unquantifiable serum HBV DNA levels for at least 24 weeks after a finite course of NA or pegylated interferon-α therapy.20 Functional cure is a critical goal in the management of chronic HBV infection,21 as it is significantly associated with a reduced risk of clinical outcomes.22–24 However, functional cure is rarely achieved with NA monotherapy.20 A meta-analysis by Jian et al. revealed that among patients with non-cirrhotic CHB, long-term TDF monotherapy for up to 240 weeks resulted in a serum HBsAg clearance rate near zero and a median HBsAg level reduction of only 21% by 168 weeks.25 Given this limitation, recent research has focused on achieving low HBsAg levels (<100 IU/mL) as an intermediate endpoint in non-cirrhotic cohorts.26 In our study, during a median follow-up of seven years, the cumulative HBsAg clearance rate was 9.8% for the entire cohort, and this rate was highest (34.9%) among patients with a baseline HBsAg level lower than 100 IU/mL. This observation is also consistent with the established association between low baseline HBsAg levels and a higher likelihood of HBsAg clearance in other CHB populations.27 Importantly, our finding converges with emerging evidence on finite NA therapy. Jeng et al. reported that patients with an end-of-treatment HBsAg level of <100 IU/mL achieved higher rates of HBsAg clearance.28 Another study also demonstrated that this threshold reliably identifies individuals most likely to maintain remission without relapse after stopping antiviral treatment.29 Therefore, a low HBsAg level robustly predicts a greater probability of achieving HBV functional cure.
In this study, we demonstrated that patients who achieved “functional cure” had the lowest cumulative incidence of HCC and highlighted the need for stratification of HCC risk by serum HBsAg level. As reviewed by Lok, a high circulating HBsAg load is a key driver of immune dysfunction in chronic HBV infection.20 This view is directly supported by the perspective that HBsAg clearance is fundamental to reversing HBV-induced immune dysfunction.30 Collectively, this supports the plausibility that reducing the antigenic burden facilitates immune restoration, thereby mitigating hepatocarcinogenesis.
It is well established that although NA therapy rapidly suppresses viral replication and leads to substantial histological improvement, reducing the risk of HCC is a more protracted process.6,8 Our findings delineate the dynamics of the annual HCC incidence in patients with HBV-related cirrhosis receiving long-term NA therapy. The results revealed that the highest incidence (5.85 per 100 person-years) was observed during the initial two years of antiviral treatment. This annual incidence exceeds the pooled annual incidence rate of 3.37 per 100 person-years reported for untreated patients with compensated cirrhosis in a large meta-analysis.31 This might be explainable by the high proportion (80%) of patients with decompensated cirrhosis in our cohort. Importantly, we observed that the annual incidence of HCC began to decline from the third year and decreased to lower than 2 per 100 person-years by the seventh year of NA treatment. Thus, the high risk of HCC remains pronounced during the initial two years of NA therapy, underscoring the critical importance of intensified surveillance in this early treatment window in patients with HBV-related cirrhosis.
In this real-world cohort study, MVR, VBT, and PVR accounted for 65.4%, 26.5%, and 8.1%, respectively, of the virological response patterns during a median follow-up of seven years. Indeed, the MVR rate observed in our cohort (65.4%) is higher than that (39.3%) reported in an entirely decompensated cohort treated with entecavir or lamivudine.10 This disparity may be partly explained by differences in disease severity between the two cohorts, as our cohort included both compensated and decompensated patients, and we confirmed a higher MVR rate in those with compensated cirrhosis (71.3% vs. 64.0%). Furthermore, this discrepancy may also be attributed to the inclusion of lamivudine in the earlier cohort, an agent associated with a high rate of drug resistance and weaker virological suppression compared with contemporary first-line NA regimens. In contrast, our MVR rate aligns closely with the 61.9% reported by Kim et al. in a cohort of CHB patients (50.6% with cirrhosis) who received entecavir monotherapy.32 Notably, their cohort had a high median baseline HBV DNA level (7.0 log10 IU/mL versus 4.82 log10 IU/mL in our cohort). Given that a high baseline HBV DNA load is a well-established predictor of suboptimal virological response to NA therapy,33,34 it is plausible that this elevated baseline HBV DNA load attenuated the potential MVR advantage in their cohort with less advanced liver disease. In summary, the MVR rates across studies underscore the multifactorial determinants of the response to long-term antiviral therapy, including the potency and historical context of the regimen, the duration of therapy, baseline HBV DNA load, and the severity of liver disease. Furthermore, the cirrhotic microenvironment is associated with a distinct virological profile, including a higher prevalence of specific HBV mutations,35,36 which may further compromise the efficacy of antiviral therapy.
Previous studies preliminarily revealed the prognostic value of MVR. For instance, Wong et al. demonstrated that MVR is associated with reduced risks of HCC and liver-related death.6 Jang et al. also established that patients with MVR had better five-year transplant-free survival than non-responders or patients with a suboptimal response.10,37 Our study confirmed that MVR was associated with a reduced risk of liver-related clinical outcomes, especially HCC, in HBV-related cirrhosis across all disease stages. Moreover, we found that MVR was also associated with a higher rate of hepatic recompensation and a lower risk of progression to decompensation. However, more than 30% of patients experienced PVR or VBT during long-term NA therapy in this cohort. By stratifying patients who did not achieve MVR into PVR and VBT subgroups, we further revealed a distinct HCC risk gradient: the lowest risk in patients with MVR and the highest in those with VBT, a finding that aligns with previous studies identifying VBT as an independent risk factor for HCC in patients with cirrhosis.38
We hypothesize that the markedly elevated HCC risk associated with VBT reflects mechanisms that differ from the chronic injury pattern observed in PVR. First, the cyclical nature of VBT, defined by initial viral suppression followed by rebound, might interrupt hepatic repair and promote a protumor microenvironment. This process can trigger renewed inflammation, disrupt regeneration, and drive dysregulated fibrogenesis and aberrant angiogenesis.39 Second, the rapid surge in viral replication during relapse can promote more complex viral DNA integration into hepatocyte genomes, facilitating clonal expansion of premalignant hepatocytes.40,41 Finally, in the compromised cirrhotic liver, VBT can induce intense inflammation and immune exhaustion. Reactivated CD8+ T cells and macrophages release inflammatory mediators and reactive oxygen species, generating oxidative DNA damage, and concurrent viral immune evasion mechanisms undermine antitumor surveillance.42,43 By contrast, PVR typically reflects a more stable, low-grade inflammatory milieu in which carcinogenesis progresses gradually. These findings highlight that VBT confers a particularly unfavorable prognosis relative to persistent low-level viremia. Therefore, maintaining MVR is the central therapeutic priority. For the substantial proportion of patients who achieve PVR or VBT, our results underscore the urgent need for a more proactive and stratified management strategy to improve their clinical outcomes.
We propose a structured management approach: (1) intensified virologic monitoring for patients who achieve initial CVR, involving more frequent HBV DNA testing (e.g., every three months) for early identification of treatment failure; (2) prompt salvage therapy upon confirmed VBT, entailing thorough evaluation and a switch to a non–cross-resistant NA regimen guided by genotypic resistance testing. The efficacy of this salvage therapy should be formally assessed by monitoring HBV DNA levels every three months until undetectable.15 Studies have demonstrated that patients managed with this salvage therapy strategy can achieve rapid virologic re-suppression, with undetectable HBV DNA in all patients by 48 weeks in one study and maintained efficacy in approximately 77% of patients through 144 weeks in another study44,45; and (3) intensified HCC surveillance in all patients who experience VBT or PVR, given their association with an elevated risk of HCC. This underscores the imperative for strict adherence to the surveillance protocol recommended by the China Liver Cancer Guidelines (2024 Edition),16 which emphasizes semiannual abdominal ultrasonography combined with serum AFP testing for high-risk individuals.
We acknowledge several limitations in this study. First, this was a single-center study conducted at a tertiary care hospital specializing in hepatology. Consequently, our cohort comprised a uniquely high proportion (81.4%) of patients with decompensated cirrhosis. Therefore, our findings should be interpreted with caution when generalized to the broader population of patients with HBV-related cirrhosis managed in general hospitals. Second, potential selection bias, unmeasured confounding, and incomplete data for some variables were inherent to the retrospective design. Third, although our study quantified the timing of CVR-to-VBT conversion, analysis of key risk factors for this conversion, such as medication adherence or viral mutations, was not feasible within the retrospective study design. Fourth, a detailed analysis of salvage therapy strategies after VBT or PVR was not conducted. Meanwhile, the potential effects of different NA regimens (such as entecavir and TDF/tenofovir alafenamide) on clinical outcomes, especially the occurrence of HCC, also remain unexplored. Fifth, mortality data were captured only during hospital admission, leading to an underestimation of the overall mortality rate. Finally, although standard guideline-based treatments were used to manage cirrhotic complications, variability in application across individual cases may have influenced the clinical outcomes.