Preoperative HBsAg seroclearance affects long-term outcomes for hepatitis B virus-related hepatocellular carcinoma after liver resection: a multicenter analysis

Abstract

Background & aims

We aimed to investigate the impact of HBsAg seroclearance on the long-term prognosis of HBV-related HCC following curative liver resection.

Methods

Patients treated with curative-intent hepatectomy for HBV-related HCC in 4 Chinese hospitals were enrolled. Patients were stratified into HBsAg-positive and HBsAg-seroclearance groups. Overall survival (OS) and time-to-recurrence (TTR) were compared between HBsAg-positive and HBsAg-seroclearance patients using propensity score matching (PSM). Independent risk factors of OS, cancer-specific survival (CCS) and TTR were identified by univariable and multivariable Cox regression and competing risk regression analyses.

Results

Among 1008 HBV-related HCC patients, 81 (8.0%) patients achieved HBsAg seroclearance before surgery. Patients in HBsAg-seroclearance groups had better liver function reserve. Postoperative morbidity was comparable between the two groups before and after PSM. In the entire cohort, the Kaplan–Meier curves showed that HBsAg-seroclearance patients had better OS (p = 0.022) and TTR (p < 0.001). However, there was no significant difference between the two groups of patients regarding OS (p = 0.134) in the PSM cohort. Multivariable Cox regression and competing risk regression analyses demonstrated that HBsAg seroclearance was independently associated with lower TTR (p < 0.05), but not with better OS and CSS (p > 0.05).

Conclusions

Preoperative HBsAg seroclearance is associated with a lower risk of TTR for HBV-related HCC among patients undergoing curative hepatectomy.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12885-025-14869-4.

Introduction

Hepatocellular carcinoma (HCC) is the most common type of primary liver cancer, and is responsible for the fifth-highest incidence of tumors and the second-highest mortality rate among all cancer types globally [1]. Partial liver resection is the primary curative treatment modality for appropriately selected patients with HCC [2, 3]. Hepatitis B virus (HBV) infection remains a leading risk factor for HCC, particularly in regions with high endemicity such as China [4, 5]. A primary cause of HBV-related HCC is the integration of the HBV genome into the genomes of host hepatocytes, while chronic hepatitis B (CHB) also triggers immune dysregulation through various mechanisms promoting HCC development [6]. Despite curative-intent treatment, long-term outcomes for HCC patients remain suboptimal due to high rates of tumor recurrence and mortality. [7, 8]. Previous studies have shown that clearance of hepatitis B surface antigen (HBsAg) is always from the serum, spontaneously or after treatment [9–11]. Achieving HBsAg loss, considered as the functional cure, is a realistic goal for CHB patients [12, 13]. The rates of spontaneous HBsAg seroclearance range from 0.50% to 2.26% per year, depending on the studied cohort and the inclusion criteria [14]. Currently available treatments yield low rates of HBsAg loss; approximately 3–5% of patients treated with PEG IFN and 0–3% of those treated with NAs achieving HBsAg seroclearance each year [15]. HBsAg seroclearance is associated with favorable long-term outcomes, such as reduced risk of disease progression, HCC development, and mortality compared to patients who remain HBsAg positive [16–18]. According to the report by Qiu and Huang, patients with low preoperative serum levels of HBsAg had better overall survival (OS) and recurrence-free survival (RFS) compared to those with high levels [19, 20]. Another report from Yoo et al. involving 2520 HBV-related HCC patients further demonstrated that seroclearance of HBsAg is associated with a reduced risk of late recurrence (≥ 2 years) of HBV-related HCC after liver resection [21]. However, the possible impact of HBsAg seroclearance on long-term oncological prognosis including OS, time to recurrence (TTR) and cancer-specific survival (CCS) after HCC resection for patients with chronic HBV infection is little known until recently.

Therefore, the current study aimed to investigate the clinicopathological features, perioperative outcomes and long-term oncologic prognosis of HBsAg seroclearance versus HBsAg positive with HBV-related HCC by using a large multicenter database about HCC after liver resection from China where HBV is endemic.

Patients and methods

Patients and study design

The study cohort comprised patients from the Fifth Affiliated Hospital of Wenzhou Medical University, Zhejiang Provincial People’s Hospital, Eastern Hepatobiliary Surgery Hospital, and Shaoxing Municipal Hospital, all of which are located in China. The diagnosis of HCC was definitively established through postoperative pathological analysis. Curative liver resection refers to the surgical removal of HCC, with a microscopically confirmed negative margin, categorized as R0 resection. The criteria for inclusion were as follows: 1) HBV-related HCC; 2) resection with the intention of cure; 3) patients older than 18 years; 4) no prior history of other cancers; 5) a score of less than 2 on the Eastern Cooperative Oncology Group (ECOG) status. A list of exclusion criteria was as follows: 1) recurrent HCC; 2) co-infection with HBV and HCV; 3) incomplete preoperative or postoperative follow-up medical records; 4) spontaneous tumor rupture before surgery; 5) with a history of neoadjuvant chemotherapy. For small HCCs, treatment options include partial hepatectomy and radiofrequency ablation (RFA). The preoperative assessment for surgical candidates should comprehensively evaluate the patient’s liver function, tumor location, and residual liver volume. If the criteria for surgical resection were not met, RFA was the preferred treatment option. During the conduct of this study, all participating hospitals adhered strictly to the provisions outlined in the Declaration of Helsinki and the Ethical Guidelines for Clinical Studies (NO.2024347), with informed consent obtained from patients who agreed to allow their data to be used for research purposes.

Clinicopathological characteristics and operative variables

Our analysis encompassed an assessment of survival and recurrence risk factors, incorporating the variables of patient demographics, tumor characteristics, and therapeutic approaches. There were several patient-related variables, including the age, gender, family history, American Society of Anesthesiologists (ASA) score, body mass index, alcohol consumption, diabetes mellitus, hypertension, presence of clinical symptoms (abdominal pain, distension, mass, and weight loss, etc.), presence of cirrhosis or portal hypertension, Child–Pugh grade, preoperative serum alanine aminotransferase (ALT), aspartate aminotransferase (AST), preoperative HBV DNA level, preoperative antiviral therapy, and preoperative regular surveillance. The presence of either esophageal varices or splenomegaly, accompanied by a platelet count of ≤ 100 × 10⁹/L was defined as a marker of portal hypertension [22]. The tumor-related variables investigated included preoperative serum α-fetoprotein (AFP) levels, maximum tumor diameter, tumor multiplicity (solitary or multiple), microvascular invasion (MVI) and Macrovascular invasion status (positive or negative), presence of satellite lesions (yes or no), degree of tumor differentiation, and tumor encapsulation status (incomplete or complete). The operative variables encompassed intraoperative blood loss (≤ 400 mL versus > 400 mL), requirement for intraoperative blood transfusion (absent or present), the magnitude of liver resection (minor or major), the type of hepatectomy performed (anatomical or non-anatomical), and the status of the resection margin. The diagnosis of cirrhosis and tumor-related variables involves the utilization of abdominal imaging modalities, including ultrasonography, computed tomography (CT), and magnetic resonance imaging (MRI). By definition, major hepatectomy involves removing three or more Couinaud’s segments, whereas minor hepatectomy involves removing fewer than three segments. Specifically, anatomical resections were classified using the Brisbane 2000 nomenclature for liver anatomy, whereas non-anatomical resections encompassed wedge resections and limited resections [23]. Preoperative regular surveillance was defined as performing HCC screening twice a year [24, 25]. HBsAg seroclearance was defined as undetectable serum HBsAg levels at preoperative testing, with persistent negativity confirmed postoperatively.

Follow-up and postoperative outcomes

At each hospital, a surveillance strategy was adopted for following up on the patients. After discharge from the hospital, the patients were followed up using serum AFP or Prothrombin induced by vitamin K Absence-II (PIVKA II) level, ultrasonography or CT scan or MRI of the abdomen once every 2 to 3 months for the first 6 months, once every 3 to 4 months for the next one and a half years, and then once every 3 to 6 months thereafter. The definition of HCC recurrence is the emergence of new nodules, either intrahepatic or extrahepatic, with or without elevations in serum AFP or PIVKA-II levels, and these nodules had the typical imaging features that are consistent with the characteristics of HCC on enhanced CT or MRI examinations. Patients experiencing tumor recurrence were actively managed with a range of therapeutic options, including re-resection, salvage transplantation, TACE or Hepatic Artery Infusion Chemotherapy (HAIC), percutaneous local ablation, radiotherapy, targeted therapy, and immunotherapy, either as monotherapy or in combination therapies. The perioperative outcomes included postoperative 30-day morbidity and hospital stays. Postoperative morbidity was categorized using the Clavien-Dindo classification system, where grades III and IV represent major morbidity, while grades I and II denote minor morbidity [26]. The key objectives of our study focused on survival rates and recurrence occurrences. The initial recurrence details encompassed the recurrence patterns observed and the primary treatment approach utilized. We recorded the dates pertaining to tumor recurrence, the final follow-up, and mortality.

Study endpoints

The key endpoints of our study centered on assessing long-term oncologic outcomes, specifically: overall survival (OS) and time to recurrence (TTR). The secondary endpoint was cancer-specific survival (CSS), which provided additional insights into the treatment’s efficacy. Cancer-specific mortality (CSM) was defined as deaths directly caused by recurrence, metastasis, or treatment of the recurrent tumor after surgery. The OS was assessed by calculating the interval from the date of liver resection to either the occurrence of CSM or non-CSM (indicating the end of the study event), or until the last follow-up date (considered a censored event). The TTR was determined, starting from the date of surgery, and ending on the occurrence of either the detection of initial recurrence of HCC (end event), non-CSM (censored event), or the date of the last follow-up (also a censored event). Based on the date of hepatectomy, the CSS was identified, either to the date of CSM (end event), the date of non-CSM (competing-risk event), or the date of last follow-up (censored event).

Statistical analysis

For continuous variables that exhibit normal distribution, we present them as mean ± standard deviation (SD). For those variables that do not follow a normal distribution, we employ the median and interquartile range (IQR) for representation. To compare the differences among these variables, we utilized the Student's t-test for normally distributed variables and the Mann–Whitney U test for non-normally distributed variables. The chi-squared test or Fisher’s exact test was employed, as appropriate, for categorical variables, which were expressed as frequencies (percentages). The OS and TTR rates were estimated using the Kaplan–Meier methodology and compared via the log-rank test. Propensity score matching (PSM) was used to reduce the confounding bias, with a ratio of 2 and a caliper value set at 0.2. The specific matching variables included age, ASA score, cirrhosis, portal hypertension, ALT, AST, AFP, and preoperative antiviral therapy. In order to reduce the influence of confounding factors, univariable and multivariable Cox proportional hazards regression analyses were performed in the PSM cohort, utilizing a forward stepwise approach for variable selection. Variables that demonstrated statistical significance (p < 0.05) in the univariable analysis were included in the multivariable Cox regression analysis. To assess the genuine oncologic influence on HBsAg seroclearance, we employed the Fine and Gray subdistribution hazard regression model in the entire cohort, a widely recognized tool for multivariable competing-risk regression analysis, which facilitated the identification of independent predictors for OS and TTR, while accounting for the competing risk of non-CSM [27]. Variables achieving statistical significance (p < 0.05) in the univariable analysis were included in the multivariable competing-risk regression models. p values < 0.05 were deemed statistically significant. A statistical analysis was carried out using SPSS version 25.0 (SPSS, Chicago, IL, USA) and R version 4.3.1 (R Foundation for Statistical Computing, Vienna, Austria).

Results

Comparisons clinicopathologic characteristics

During the study period, a total of 1008 consecutive patients who underwent curative-intent liver resection for HBV-related HCC met the inclusion criteria (Fig. 1). Out of these patients, 81 (8.0%) patients achieved HBsAg seroclearance before surgery. Clinicopathologic characteristics among patients in the HBsAg-positive and HBsAg-seroclearance groups were compared (Table 1). In terms of clinicopathologic variables, the mean age of HBsAg-seroclearance group patients was 63.4 years old, whereas HBsAg-positive group patients were 53.9 years old. Furthermore, compared with HBsAg-positive group patients, HBsAg-seroclearance group patients had a higher proportion of individual age > 60 years (61.7% vs. 27.5%), ASA score > 2 (32.1% vs. 11.4%), hypertension (35.8% vs. 19.1%), often had lower preoperative ALT level (13.6% vs. 28.2%), preoperative AST level (9.9% vs. 22.9%), preoperative HBV DNA level (0% vs. 33.1%) and had lower proportion of cirrhosis and portal hypertension (all p < 0.05). The proportion of patients with preoperative antiviral therapy was higher in the HBsAg-positive group than in the HBsAg-seroclearance group (33.8% vs. 21.0%). Besides having higher levels of AFP in HBsAg-positive group patients (28.4% vs. 16.0%), there were no statistically significant differences between the two groups in terms of tumor-related variables and operative-related variables (all p > 0.05). After PSM, the comparisons of patients’characteristics and operative variables between the HBsAg-positive and HBsAg-seroclearance groups in the PSM cohort were shown in Table 1. Except for HBV-DNA, all variables of patient’s clinical characteristics were well-balanced between groups in the PSM cohort (all p > 0.05).

Fig. 1.

Selection of the study population

Table 1.

Baseline characteristics of the patients before and after propensity score matching

N	Entire cohort			PSM cohort
	HBsAg-positive (N = 927)	HBsAg-seroclearance (N = 81)	P	HBsAg-positive (N = 147)	HBsAg-seroclearance (N = 79)	P
Clinicalpathologic Characteristics
Age, years	53.9 ± 10.8	63.4 ± 10.2	< 0.001	61.8 ± 10.5	63.2 ± 10.3	0.336
Age > 60 years	255 (27.5)	50 (61.7)	< 0.001	80 (54.4)	48 (60.8)	0.438
Male	795 (85.8)	72 (88.9)	0.436	124 (84.4)	70 (88.6)	0.500
Family history	106 (11.4)	8 (9.9)	0.671	14 (9.5)	8 (10.1)	1.000
ASA score > 2	106 (11.4)	26 (32.1)	< 0.001	37 (25.2)	24 (30.4)	0.494
Body mass index > 24 kg/m2	405 (43.7)	40 (49.4)	0.322	65 (44.2)	38 (48.1)	0.675
Alcohol consumption	277 (29.9)	26 (32.1)	0.771	46 (31.3)	26 (32.9)	0.921
Diabetes mellitus	101 (10.9)	14 (17.3)	0.121	16 (10.9)	14 (17.7)	0.215
Hypertension	177 (19.1)	29 (35.8)	0.001	40 (27.2)	28 (35.4)	0.257
Clinical symptom	102 (11.0)	6 (7.4)	0.414	15 (10.2)	6 (7.6)	0.686
Cirrhosis	678 (73.1)	44 (54.3)	< 0.001	81 (55.1)	44 (55.7)	1.000
Portal hypertension	188 (20.8)	9 (11.1)	0.046	16 (10.9)	9 (11.4)	1.000
Child–Pugh grade B	59 (6.4)	3 (3.7)	0.473	6 (4.1)	3 (3.8)	1.000
Preoperative ALT level > 40 U/L	261 (28.2)	11 (13.6)	0.004	18 (12.2)	11 (13.9)	0.880
Preoperative AST level > 40 U/L	212 (22.9)	8 (9.9)	0.007	19 (12.9)	8 (10.1)	0.687
Preoperative AFP level > 400 μg/L	263 (28.4)	13 (16.0)	0.017	27 (18.4)	13 (16.5)	0.860
Preoperative HBV DNA level positive	560 (60.4)	0 (0.0)	< 0.001	91 (61.9)	0 (0.0)	< 0.001
Preoperative HBV DNA level > 104 IU/ml	293 (31.6)	0 (0.0)	< 0.001	54 (36.7)	0 (0.0)	< 0.001
Preoperative antiviral therapy	313 (33.8)	17 (21.0)	0.019	29 (19.7)	17 (21.5)	0.884
Preoperative regular surveillance	132 (14.2)	16 (19.8)	0.238	16 (10.9)	15 (19.0)	0.137
Largest tumor diameter > 5 cm	381 (41.1)	39 (48.1)	0.217	56 (38.1)	38 (48.1)	0.189
Multiple Tumors	121 (13.1)	7 (8.6)	0.253	20 (13.6)	6 (7.6)	0.258
Microvascular invasion	384 (41.4)	28 (34.6)	0.229	54 (36.7)	27 (34.2)	0.813
Macrovascular invasion	42 (4.5)	2 (2.5)	0.571	3 (2.0)	2 (2.5)	1.000
Satellites	95 (10.2)	6 (7.4)	0.414	13 (8.8)	6 (7.6)	0.943
Poor tumor differentiation	584 (63.0)	53 (65.4)	0.663	136 (92.5)	71 (89.9)	0.666
Incomplete tumor encapsulation	687 (74.1)	52 (64.2)	0.053	108 (73.5)	50 (63.3)	0.150
Postoperative TACE	413 (44.6)	38 (46.9)	0.682	70 (47.6)	37 (46.8)	1.000
Operative Variables
Operation time	127 (100–165)	130 (100–170)	0.534	120 (100–160)	130 (103–169)	0.529
Intraoperative blood loss > 400 ml	86 (9.3)	7 (8.6)	0.850	12 (8.2)	6 (7.6)	1.000
Intraoperative blood transfusion	58 (6.3)	6 (7.4)	0.684	6 (4.1)	6 (7.6)	0.417
Major liver resection	153 (16.5)	17 (21.0)	0.301	22 (15.0)	16 (20.3)	0.408
Non-anatomical liver resection	791 (85.3)	67 (82.7)	0.526	130 (88.4)	65 (82.3)	0.280
Resection margin < 1 cm	630 (67.7)	57 (70.4)	0.627	91 (61.9)	55 (69.6)	0.312
Perioperative Outcomes
Postoperative 30-day morbidity	293 (31.6)	24 (29.6)	0.713	42 (28.6)	23 (29.1)	1.000
Minor morbidity	243 (26.2)	20 (24.7)	0.765	36 (24.5)	19 (24.1)	1.000
Major morbidity	50 (5.4)	4 (4.9)	1.000	6 (4.1)	4 (5.1)	0.998
Postoperative hospital stays, days	7 (7–9)	8 (7–9)	0.522	7 (7–8)	7 (7–9)	0.334

Abbreviations: AFP Alpha-fetoprotein, ASA American Society of Anesthesiologists, AST aspartate transaminase, ALT alanine aminotransferase, HBsAg hepatitis B surface antigen, HBeAg hepatitis B e antigen

Comparisons of long-term outcomes

At a median follow-up of 52.8 months, death and recurrence were observed in 433 (43.0%) and 582 (57.8%) patients, respectively. Death occurred in 409 patients (44.1%) in the HBsAg-positive group and 24 patients (29.6%) in the HBsAg-seroclearance group (p = 0.012), while recurrence occurred in 551 (59.5%) HBsAg-positive patients and 31 (38.3%) HBsAg-seroclearance patients (p < 0.001). The CSM rates among the HBsAg-positive patients were higher than those among the HBsAg-seroclearance patients (37.8% vs. 23.5%, p = 0.010); but non-CSM rates were comparable between the two groups (6.4% vs. 6.2%, p = 0.946), as shown in Table 2. Meanwhile, Table 2 presents the results after PSM. The specific causes of death distribution of patients were shown in supplementary Table 1. Compared with the HBsAg-positive patients, the HBsAg-seroclearance patients had a better 5-year OS (58.8% vs. 72.2%, p = 0.022), and a lower 5-year TTR (58.4% vs. 38.6%, p < 0.001), respectively. After PSM, patients with HBsAg-seroclearance had a comparable 5-year OS (65.1% vs. 72.8%, p = 0.134), and a lower 5-year TTR (58.0% vs. 37.8%, p = 0.006), respectively. Using the Kaplan–Meier method, OS and TTR among the HBsAg-positive and HBsAg-seroclearance patients before and after PSM were compared by the log-rank test (Figs. 2 and 3).

Table 2.

Long-term outcomes in entire cohort and propensity score matching cohort after excluding postoperative early deaths

N%	Entire cohort			PSM cohort
	HBsAg-positive (N = 927)	HBsAg-seroclearance (N = 81)	P	HBsAg-positive (N = 147)	HBsAg-seroclearance (N = 79)	P
Period of follow-up, months	52.4 ± 23.0	56.9 ± 22.1	0.092	54.1 ± 21.0	56.9 ± 22.3	0.354
Recurrence during the follow-up	551 (59.5)	31 (38.3)	< 0.001	86 (58.5)	31 (39.2)	0.009
Intrahepatic recurrence	508 (54.8)	26 (32.1)	< 0.001	86 (55.8)	25 (32.9)	< 0.001
Extrahepatic recurrence	11 (1.2)	1 (1.2)	1.000	1 (0.7)	1 (1.3)	1.000
Intra- & extrahepatic recurrence	32 (3.5)	4 (5.0)	0.525	3 (2.0)	4 (5.0)	0.243
Mortality during the follow-up	409 (44.1)	24 (29.6)	0.012	59 (40.1)	23 (29.1)	0.134
Cancer-specific mortality	350 (37.8)	19 (23.5)	0.010	48 (32.7)	19 (24.1)	0.177
Non-cancer-specific mortality	59 (6.4)	5 (6.2)	0.946	11 (7.5)	4 (5.0)	0.486
Median overall survival (OS), 95% CI			0.022			0.134
1-year OS rate, %	92.1%	97.5%		94.6%	96.2%
3-year OS rate, %	73.1%	80.2%		81.0%	82.3%
5-year OS rate, %	58.8%	72.2%		65.1%	72.8%
Median time-to-recurrence (TTR), 95% CI			< 0.001			0.006
1-year TTR rate, %	32.1%	17.3%		24.5%	15.2%
3-year TTR rate, %	50.5%	27.3%		48.3%	28.0%
5-year TTR rate, %	58.4%	38.6%		58.0%	37.8%

Fig. 2.

Kaplan–Meier curves comparing the two groups of patients in the entire cohort

Fig. 3.

Kaplan–Meier curves comparing the two groups of patients in the propensity score matching cohort

Prognostic analyses of survival and recurrence in the PSM cohort and entire cohort

Cox regression was conducted in the PSM cohort, while the competing risk model was performed in the entire cohort. Tables 3 and 4 describe the prognostic predictors of survival (OS and CSS) following liver resection for HCC, which were identified by univariable and multivariable Cox regression and competing risk models regression analyses, respectively. Multivariable Cox model analyses demonstrated that HBsAg seroclearance was not an independent protective factor of OS (p = 0.085), and multivariable competing-risk model analysis identified HBsAg seroclearance to be not independently associated with CSS following hepatectomy for HCC (p = 0.100). In addition, both the multivariable Cox regression model (HR: 0.47, 95%CI: 0.31–0.73, p < 0.001) and the multivariable risk competing model (HR: 0.59, 95%CI: 0.40–0.86, p = 0.006) identified HBsAg seroclearance as an independent factor for better TTR.

Table 3.

Cox regression model and competing risk regression model analyses of risk factors for overall survival and cancer-specific survival

		Cox regression model				Fine and Gray’s Competing risk model
Variables	HR Comparison	UV HR (95% CI)	UV P	MV HR (95% CI)	MV P	UV sHR (95% CI)	UV P	MV sHR (95% CI)	MV P
Age	> 60 vs. ≤ 60 years	1.23 (0.79–1.92)	0.358			0.82 (0.65–1.03)	0.092
Sex	Male vs. Female	0.92 (0.51–1.67)	0.793			1.33 (0.97–1.81)	0.078
ASA score	> 2 vs. ≤ 2	1.25 (0.79–1.99)	0.347			0.65 (0.47–0.91)	0.012	NS	0.530
Body mass index	> 24 vs. ≤ 24 kg/m2	0.69 (0.44–1.08)	0.098			0.69 (0.56–0.85)	0.001	NS	0.210
Cirrhosis	Presence vs. Absence	1.59 (1.00–2.50)	0.048	NS	0.168	1.50 (1.17–1.92)	0.001	1.42 (1.09–1.84)	0.009
Portal hypertension	Presence vs. Absence	0.93 (0.46–1.86)	0.834			0.93 (0.72–1.20)	0.570
Child–Pugh grade	B vs. A	1.68 (0.68–4.15)	0.263			2.02 (1.48–2.75)	< 0.001	NS	0.620
Preoperative AFP level	> 400 vs. ≤ 400 μg/L	2.45 (1.50–4.01)	< 0.001	1.82 (1.04–3.15)	0.035	2.31 (1.87–2.86)	< 0.001	1.34 (1.06–1.71)	0.016
Preoperative HBV-DNA level	≥ 104 vs. < 104 IU/ml	1.51 (0.95–2.41)	0.081			1.65 (1.34–2.03)	< 0.001	NS	0.330
Largest tumor diameter	> 5 vs. ≤ 5 cm	1.91 (1.24–2.95)	0.003	1.62 (1.01–2.60)	0.044	2.70 (2.20–3.31)	< 0.001	1.86 (1.45–2.39)	< 0.001
Tumor number	Multiple vs. Solitary	1.45 (0.79–2.68)	0.233			1.81 (1.38–2.37)	< 0.001	1.59 (1.19–2.12)	0.002
Macrovascular invasion	Yes vs. no	5.89 (2.36–14.7)	< 0.001	1.94 (1.23–3.07)	0.004	7.09 (5.08–9.89)	< 0.001	2.78 (1.89–4.10)	< 0.001
Microvascular invasion	Yes vs. no	2.23 (1.44–3.44)	< 0.001	NS	0.862	2.51 (2.05–3.09)	< 0.001	1.89 (1.51–2.36)	< 0.001
Satellites	Yes vs. no	2.71 (1.46–5.02)	0.002	NS	0.058	2.97 (2.25–3.92)	< 0.001	1.49 (1.10–2.00)	0.010
Poor tumor differentiation	Yes vs. no	1.10 (0.51–2.39)	0.811			1.96 (1.60–2.47)	< 0.001	NS	0.140
Incomplete tumor encapsulation	Yes vs. no	1.31 (0.80–2.14)	0.279			1.84 (1.42–2.37)	< 0.001	NS	0.240
Intraoperative blood loss	> 400 vs. ≤ 400 ml	2.38 (1.26–4.50)	0.008	NS	0.121	2.34 (1.76–3.11)	< 0.001	NS	0.110
Intraoperative transfusion	Yes vs. no	2.00 (0.92–4.35)	0.080			2.08 (1.49–2.91)	< 0.001	NS	0.720
Extent of liver resection	Major vs. ≤ Minor	1.33 (0.78–2.27)	0.294			2.09 (1.64–2.65)	< 0.001	NS	0.960
Non-anatomical liver resection	Yes vs. no	0.85 (0.47–1.54)	0.600			0.75 (0.57–0.98)	0.033	NS	0.410
Resection margin	< 1 vs. ≥ 1 cm	1.68 (1.02–2.76)	0.041	NS	0.404	2.44 (1.87–3.19)	< 0.001	1.75 (1.32–2.32)	< 0.001
Preoperative antiviral therapy	Yes vs. no	0.62 (0.34–1.15)	0.131			0.58 (0.46–0.73)	< 0.001	0.71 (0.54–0.93)	0.013
HBsAg seroclearance	Yes vs. no	0.69 (0.43–1.12)	0.134	NS	0.085	0.57 (0.36–0.90)	0.016	NS	0.100

Abbreviations: CI Confidence interval, HR Hazard ratio, aHR adjustment Hazard ratio, MV multivariate, UV univariate, NS not significant

Table 4.

Cox regression model and competing risk regression model analyses of risk factors for recurrence

		Cox regression model				Fine and Gray’s competing risk model
Variables	HR Comparison	UV HR (95% CI)	UV P	MV HR (95% CI)	MV P	UV sHR (95% CI)	UV P	MV sHR (95% CI)	MV P
Age	> 60 vs. ≤ 60 years	1.22 (0.84–1.76)	0.301			0.91 (0.77–1.09)	0.300
Sex	Male vs. Female	1.25 (0.72–2.19)	0.432			1.32 (1.33–1.69)	0.027	1.47 (1.13–1.91)	0.004
ASA score	> 2 vs. ≤ 2	0.89 (0.59–1.35)	0.589			0.73 (0.58–0.93)	0.012	NS	0.310
Body mass index	> 24 vs. ≤ 24 kg/m2	0.96 (0.67–1.39)	0.846			0.81 (0.69–0.95)	0.009	NS	0.430
Cirrhosis	Presence vs. Absence	1.68 (1.14–2.45)	0.008	1.55 (1.05–2.30)	0.028	1.52 (1.25–1.85)	< 0.001	1.44 (1.18–1.77)	< 0.001
Portal hypertension	Presence vs. Absence	0.89 (0.50–1.59)	0.704			1.01 (0.83–1.22)	0.940
Child–Pugh grade	B vs. A	1.38 (0.61–3.14)	0.441			1.91 (1.47–2.47)	< 0.001	1.40 (1.08–1.81)	0.012
Preoperative AFP level	> 400 vs. ≤ 400 μg/L	1.54 (0.98–2.44)	0.064			1.62 (1.35–1.95)	< 0.001	NS	0.150
Preoperative HBV-DNA level	≥ 104 vs. < 104 IU/ml	1.38 (0.92–2.07)	0.120			1.52 (1.28–1.80)	< 0.001	NS	0.190
Largest tumor diameter	> 5 vs. ≤ 5 cm	1.63 (1.13–2.36)	0.009	1.62 (1.11–2.38)	0.013	1.97 (1.67–2.31)	< 0.001	1.48 (1.21–1.79)	< 0.001
Tumor number	Multiple vs. Solitary	1.62 (0.96–2.75)	0.073			1.89 (1.52–2.37)	< 0.001	1.79 (1.42–2.56)	< 0.001
Macrovascular invasion	Yes vs. no	8.86 (3.51–22.3)	< 0.001	5.85 (1.99–17.2)	0.001	5.20 (3.74–7.21)	< 0.001	2.69 (1.95–3.71)	< 0.001
Microvascular invasion	Yes vs. no	1.89 (1.31–2.73)	0.001	1.65 (1.13–2.44)	0.010	1.82 (1.52–2.37)	< 0.001	1.49 (1.25–1.78)	< 0.001
Satellites	Yes vs. no	2.65 (1.54–4.58)	< 0.001	2.30 (1.32–4.04)	0.004	3.16 (2.50–3.99)	< 0.001	2.02 (1.59–2.56)	< 0.001
Poor tumor differentiation	Yes vs. no	1.24 (0.63–2.45)	0.537			1.43 (1.21–1.69)	< 0.001	NS	0.510
Incomplete tumor encapsulation	Yes vs. no	1.23 (0.82–1.84)	0.327			1.34 (1.12–1.61)	0.001	NS	0.430
Intraoperative blood loss	> 400 vs. ≤ 400 ml	1.86 (1.02–3.38)	0.043	NS	0.458	2.09 (1.63–2.68)	< 0.001	1.42 (1.05–1.91)	0.024
Intraoperative transfusion	Yes vs. no	1.41 (0.65–3.02)	0.384			1.83 (1.36–2.45)	< 0.001	NS	0.730
Extent of liver resection	Major vs. ≤ Minor	1.21 (0.75–1.94)	0.439			1.74 (1.41–2.16)	< 0.001	NS	0.880
Non-anatomical liver resection	Yes vs. no	1.08 (0.63–1.86)	0.781			0.85 (0.67–1.07)	0.170
Resection margin	< 1 vs. ≥ 1 cm	1.34 (0.90–1.99)	0.150			1.80 (1.48–2.17)	< 0.001	1.34 (1.09–1.65)	< 0.001
Preoperative antiviral therapy	Yes vs. no	0.83 (0.52–1.32)	0.426			0.72 (0.61–0.86)	< 0.001	0.82 (0.68–1.01)	0.052
HBsAg seroclearance	Yes vs. no	0.56 (0.37–0.85)	0.006	0.47 (0.31–0.73)	< 0.001	0.54 (0.38–0.76)	< 0.001	0.59 (0.40–0.86)	0.006

Abbreviations: CI Confidence interval, HR Hazard ratio, aHR adjustment Hazard ratio, MV multivariate, UV univariate, NS not significant

Discussion

Chronic HBV infection represents a substantial global health burden, affecting an estimated 296 million individuals and contributing significantly to the development of cirrhosis and HCC [6, 28, 29]. Nearly half of all HCC cases are diagnosed in China, a region with a high prevalence of HBV infection [30]. In the entire cohort, it can be observed that, except for AFP, there were no significantly statistical differences in other variables related to the tumor, indicating a certain degree of similarity in the malignancy of the tumor between the two groups, a finding also consistently described in previous study [31]. However, patients in the HBsAg-seroclearance group have a lower proportion of cirrhosis and portal hypertension and better viral control. Kaplan–Meier curves showed better OS (p = 0.022) and lower TTR (p < 0.001). To further balance the baseline differences between the two groups, PSM method was used. However, the Kaplan–Meier curve showed that the 5-year OS rates were comparable between the HBsAg-seroclearance and the HBsAg positive groups in the PSM cohort (p = 0.134). Multivariable Cox regression and competing risk regression analyses evaluated that HBsAg seroclearance was independently associated with lower TTR (p < 0.05), but not with better OS and CSS (p > 0.05).

Previous studies had shown that effectively controlling the viral load and even achieving serological clearance of HBsAg can effectively reduce the risk of the development of HCC, especially for patients who experienced HBsAg seroclearance at age < 50 years [16, 32]. In CHB patients with cirrhosis achieving HBsAg seroclearance, liver function can improve or remain stable in the absence of viral superinfection, and occurrences of hepatic decompensation are extremely infrequent. The long-term outcomes for CHB patients without pre-existing cirrhosis or viral superinfection are excellent following HBsAg seroclearance [33]. In the present study, both spontaneous clearance and drug-induced seroconversion of HBsAg had effectively suppressed the replication of HBV. Despite not being able to completely prevent the occurrence of liver cancer, HCC patients with low preoperative HBV-DNA load have a better long-term prognosis after liver resection than patients with high HBV-DNA load [34, 35]. Therefore, whether as secondary prevention or tertiary prevention, antiviral treatment is indispensable for CHB patients. Despite all patients with positive HBsAg undergoing postoperative antiviral therapy, this study did not investigate which antiviral regimen was superior. Recent studies indicated that, in patients undergoing curative liver resection for HBV-related HCC, tenofovir disoproxil offered superior long-term OS and RFS rates compared to entecavir [36, 37].

Spontaneous clearance of HBsAg is uncommon and there is no notable geographical variability in the incidence of this phenomenon. A 15-year-and-8-month longitudinal study of 448 CHB patients revealed that only 40 patients achieved HBsAg clearance, with an annual clearance rate of 0.7% [38]. As such, CHB patients typically require an extended duration before the potential for spontaneous clearance of HBsAg can be achieved. Antiviral therapy has a modest impact on the occurrence of HBsAg clearance in our study, only 21.0% CHB patients achieved HBsAg clearance with a history of antiviral treatment. A large cohort study involving 2,112 CHB patients conducted in Japan conducted a long-term follow-up of up to 15 years for spontaneous and post-antiviral treatment HBsAg seroclearance. The results showed no statistical difference in the annual incidence rates between the two groups (1.65% vs. 2.05%, p = 0.289) [39]. Seroclearance of HBsAg has been observed at a rate of 0–3% following 1 year of treatment with nucleos(t)ide analogues [15]. Under conditions of persistent viral suppression, the rate of HBsAg seroclearance increased to 6.4–21.5% after 9–10 years of treatment with lamivudine or entecavir [40–42]. Similarly, the duration of antiviral treatment is the key to achieving HBsAg clearance. Patients in the HBsAg-positive group were significantly younger than those in the HBsAg-seroclearance group, which is consistent with previous research findings that advanced age is a factor influencing HBsAg loss [33, 38].

A total of 433 patients died during follow-up in this study, of which 64 (14.8%) died from non-CSM causes. Failure to account for competing risks can lead to exaggerated or obscured correlations between research factors and diseases in the study of competitive risk events. To comprehensively assess the true survival benefit following liver resection in HCC patients with HBsAg clearance, we not only conducted the Cox proportional hazards model but also utilized the competing risk model to elucidate the impact of HBsAg seroclearance on OS, CSS and TTR. Although the Kaplan–Meier curve of the entire cohort suggested that HBsAg clearance is associated with better OS, which may be due to the better liver function and lower proportions of liver cirrhosis and portal hypertension in the HBsAg-seroclearance patient population. There was no statistical difference in OS between the two groups after balancing baseline characteristics using PSM. Similarly, after multivariable regression adjustment, HBsAg seroclearance was not a protective factor for OS and CSS. These findings suggest that clinicians should comprehensively evaluate liver status to improve personalized prediction of postoperative survival in HCC patients. Additionally, our study results showed that HBsAg seroclearance was associated with lower recurrence rates among HCC patients after surgery, which is consistent with reports from other studies [21, 31].

The study is subject to several limitations. First, its retrospective design may introduce inherent selection bias. Second, chronic HBV infection is the leading etiology of HCC in China and many Asian countries, and the pattern of disease differs from the United States and other Western countries. Therefore, a large sample size from these regions is required for further validation. In addition, in this study, if preoperative serological testing fails to detect HBsAg, HCC patients who have been confirmed to have previously been infected with HBV are included in the HBsAg-seroclearance group. However, in real-world studies, many patients are unaware of their previous HBV infection, and preoperative serological testing also fails to confirm whether they have been infected with HBV. This leads to a lower HBsAg clearance rate than the actual situation, resulting in bias in the research results.

Authors’ contributions

Si-Yu Liu, Lin Zhu and Mu-Gen Dai wrote the main manuscript text and studied concept and design. Gui-Lin Xie, Jun-Wei Liu, Wen-Feng Lu and Bin Ye critical revision of the manuscript for important intellectual content. Si-Yu Liu and Lei Liang prepared figures 1-3. Si-Yu Liu, Lei Liang and Mu-Gen Dai obtained funding. All authors reviewed the manuscript.

Funding

This study was supported by the Public Welfare Technology Application Research Projects from the Science and Technology Bureau of Lishui (No. 2022GYX50), the Natural Science Foundation of Zhejiang Province (No. LQ23H160049), and Health Science and Technology Plan of Zhejiang Province (No. 2022KY453).

Data availability

The datasets used and analyzed during the current study are available from the corresponding author upon reasonable request.

Declarations

Ethics approval and consent to participate

The study was approved by the Medical Ethics Committee of the Fifth Affiliated Hospital of Wenzhou Medical University, Zhejiang Provincial People’s Hospital, Eastern Hepatobiliary Surgery Hospital, and Shaoxing Municipal Hospital (NO.2024347). All procedures were in accordance with the ethical standards of the above four Clinical Research Committees and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards. As patient identities were anonymized, the ethics committees of the four centers all waived the requirement of informed consent.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.