Esophagogastric Variceal Bleeding in Cirrhotic Patients with Hepatocellular Carcinoma Receiving Systemic Therapy: A Retrospective Cohort Study

Introduction

Liver cancer is the sixth most common cancer globally, ranking third among cancer-related deaths.¹ Hepatocellular carcinoma (HCC) accounts for the majority of liver cancers, and approximately 80–90% of cases develop in patients with cirrhosis.^2,3 Recent studies have highlighted the molecular heterogeneity and complex pathogenesis of HCC, from novel molecular subtyping approaches to potential biomarkers linked to tumor microenvironment.^4,5 Portal hypertension (PHT) is a common complication of cirrhosis, resulting from structural distortion and increased intrahepatic resistance. Advanced HCC further exacerbates PHT through vascular invasion and hepatic architectural changes.^6,7 Among PHT-related complications, esophagogastric variceal (EGV) bleeding independently predicts mortality in HCC patients regardless of liver disease severity or tumor stage.⁸ Moreover, hepatic decompensation is associated with systemic treatment options and discontinuation, further worsening prognosis.⁹ Therefore, assessing EGV bleeding risk is critical in cirrhotic HCC patients receiving systemic therapy.

Tyrosine kinase inhibitors (TKIs) and immune checkpoint inhibitors (ICIs) have transformed treatment for advanced HCC over the past decade.¹⁰ However, TKIs may worsen PHT and compromise liver regeneration.¹¹ A meta-analysis confirmed increased EGV bleeding risk with sorafenib, attributable primarily to drug effects rather than disease progression.¹² The first-line landscape has evolved from sorafenib monotherapy to new standards such as atezolizumab plus bevacizumab and lenvatinib.¹³ Anti-programmed cell death-1 antibody (anti-PD-1) combined with TKIs have shown promising efficacy but may increase bleeding risk. Atezolizumab plus bevacizumab as a first-line regimen also has been associated with increased EGV bleeding.¹⁴ The incidence of gastrointestinal bleeding events due to increased PHT was also higher with this combination than with sorafenib.¹⁵ Thus, the EGV bleeding risk during systemic therapy for advanced HCC remains a critical concern.

This study evaluated the EGV bleeding risk and mortality in cirrhotic HCC patients receiving TKIs, with or without anti-PD-1, and identified independent risk factors for both outcomes. We also compared the efficacy of TKIs monotherapy or combination anti-PD-1 therapy and assessed virological response in patients with hepatitis B virus (HBV)-related HCC.

Materials and Methods

Patients

Cirrhotic patients with intermediate to advanced HCC who received systemic therapy between January 2022 and January 2024 at the Beijing You’an Hospital were included. The inclusion criteria were as follows: (1) Cirrhotic HCC patients who were inoperable for surgery or experienced failure of prior surgical or locoregional treatments, with HCC diagnosis confirmed by histology or radiological findings.¹⁶ All patients had diagnosed cirrhosis prior to or at the time of HCC diagnosis. (2) Continuously receiving systemic therapy (TKIs or TKIs in combination with anti-PD-1) for more than one month and at least 1-year follow-up. (3) No prior systemic therapy was administered. (4) Patients underwent oesophagogastroduodenoscopy before treatment to determine the presence of an EGV. (5) Available demographic, clinical, and laboratory data available at the initiation of systemic therapy and during follow-up. The exclusion criteria were as follows: (1) age <18 years or > 80 years, as elderly patients often have limited life expectancy, multiple comorbidities, and poor tolerance to systemic therapy, which may confound outcome assessment. (2) Lack of baseline or follow-up data. (3) Comorbidities with other advanced cancers.

This retrospective study was approved by the Ethics Committee of Beijing Youan Hospital, Capital Medical University (Approval No. LL-2023-027-K). The requirement for informed consent was waived by the Ethics Committee due to the retrospective nature of the study. All de-identified patient data were collected from routine clinical care. This study was conducted in accordance with the Declaration of Helsinki and all applicable regulations.

Data Acquisition

Baseline and follow-up data were extracted from the electronic medical record system, including demographics (age and sex), medical history, cirrhosis-related complications (ascites, hepatic encephalopathy, and portal vein thrombosis [PVT]), laboratory parameters (white blood cell [WBC] count, hemoglobin [Hb], platelet [PLT] count, alanine aminotransferase [ALT], aspartate aminotransferase [AST], total bilirubin [TBIL], albumin [ALB], prothrombin time activity [PTA], and international normalized ratio [INR]), HBV DNA levels, hepatitis B surface antigen (HBsAg) levels, alpha-fetoprotein (AFP) levels, and tumor characteristics (etiology, tumor size, number of nodules, portal vein invasion, and extrahepatic metastasis). Liver function and tumor stage were assessed using the Child-Pugh score, Model for End-stage Liver Disease (MELD) score, and Barcelona Clinic Liver Cancer (BCLC) staging system.¹⁷ For patients with HBV-related HCC, concurrent nucleos(t)ide analogues therapy status at baseline and during follow-up was documented.

The presence and severity of EGV were assessed by oesophagogastroduodenoscopy performed before systemic therapy initiation. Varices were classified by location as esophageal varices (EV), gastric varices (GV), or both. The severity of varices was graded as mild, moderate, or severe according to the criteria of the Chinese Society of Gastroenterology and the Chinese Society of Digestive Endoscopy.¹⁸ Mild varices were defined as straight or slightly tortuous veins without red color signs; moderate varices as straight or slightly tortuous veins with red color signs, or serpentine-tortuous protrusions without red color signs; and severe varices as serpentine-tortuous protrusions with red color signs, or beaded, nodular, or tumor-like varices regardless of red color signs. The moderate and severe EGV was regarded as high bleeding risk varices.

Portal vein tumor thrombus (PVTT) was defined at baseline CT scan or MRI as a filling defect, partially or completely occluding the vessel in the portal venous phase, with clear evidence of enhancement during the arterial phase of dynamic imaging. Portal vein emboli without this clear-cut evidence were considered as PVT. PVTT was classified according to the Cheng classification: type I, involving segmental or sectoral branches; type II, involving the right or left portal vein; type III, involving the main portal vein; and type IV, involving the superior mesenteric vein.¹⁹

Treatment Schedule

Systemic therapy consisted of TKIs monotherapy or TKIs combined with anti-PD-1 antibody. TKIs included lenvatinib (8–12 mg/day based on body weight), sorafenib (400 mg twice daily), and regorafenib (160 mg daily for 3 weeks per 4-week cycle) as second-line therapies. Anti-PD-1 agents included camrelizumab, sintilimab, and tislelizumab (200 mg intravenously every 3 weeks). Dosing and modifications followed the respective drug labels.

Nearly all patients with HBV-related HCC received nucleos(t)ide analogs during systemic treatment. Most patients had been on long-term antiviral therapy before starting TKIs, while a few initiated antiviral therapy at the same time as the systemic therapy.

Follow-Up and Endpoints Definitions

The cut-off date for follow-up was January 31, 2025, and all data was obtained from patient medical records. Endpoints were followed from the date of systemic therapy initiation to the date of bleeding/death or the date of the last visit.

The primary outcome was the rate of EGV bleeding, and the definition of EGV bleeding was based on one or more of the following signs on oesophagogastroduodenoscopy: 1) active bleeding or blood oozing of the varicose veins, 2) varicose veins with a white thrombus head or an overlying blood clot, and 3) no other potential causes of bleeding besides variceal bleeding.

The secondary outcomes were the 1-year mortality and objective response rates (ORR). Response to systemic therapy was evaluated according to the modified Response Evaluation Criteria in Solid Tumors standard system based on abdominal enhanced computed tomography or magnetic resonance imaging.²⁰ Briefly, complete response (CR) of HCC patients after treatment was defined as the disappearance of intratumoral arterial enhancement in all target lesions; partial response (PR) was defined as at least a 30% decrease in the total diameter of the target lesion; progressive disease (PD) was defined as an increase in the total diameter of the target lesion by at least 20% or the appearance of new lesions; and reduction or increase without reaching PR or PD was defined as stable disease (SD). Thus, ORR was defined as the number of (CR+PR) cases divided by the total number of cases.

For HBV-related outcomes, HBV DNA negative conversion was defined as a change from detectable to undetectable HBV DNA levels during follow-up. HBsAg decline was expressed as a percentage change from the baseline.

Statistical Analysis

Statistical analyses were performed using R version 4.4.1. Continuous variables were expressed as median with interquartile range and compared using Wilcoxon-Mann–Whitney test. Categorical variables were presented as numbers (percentages) and compared using the chi-square test or Fisher’s exact test. Paired comparisons between baseline and follow-up data were analyzed using the Wilcoxon signed-rank test.

Variables with P<0.10 in univariable analysis were assessed for multicollinearity before entry into multivariable logistic regression. A backward stepwise selection method with a removal criterion of P>0.10 was then applied to identify independent risk factors. The cumulative incidence of EGV bleeding and mortality was estimated using the Kaplan-Meier method, and between-group comparisons were performed using the log-rank test. All tests were two-sided, and statistical significance was set at P<0.05.

Results

Patients Characteristics

Total 263 patients with cirrhosis with intermediate or advanced HCC who received systemic therapy were enrolled. 85 (32.3%) of them were confirmed by liver pathology. The baseline characteristics of the cohort are presented in Table 1. The cohort was predominantly male (85.9%) with a median age of 59 years. The median CTP score, MELD score, and AFP level were 5 (5–6), 7 (6–10) and 9.71 (3.06–141.75), respectively. The predominant etiology was HBV infection (79.5%), followed by hepatitis C virus (HCV) (6.1%), and other causes (14.4%).

Table 1 Baseline Characteristics of 263 HCC Patients with Systemic Therapy

At baseline, 94 patients (35.7%) had no EGV, and 169 (64.3%) had EGV. Among patients with EGV, 106 (40.3%) had esophageal varices (EV) alone, 22 (8.4%) had gastric varices (GV) alone, and 41 (15.6%) had both. Variceal severity was classified as mild in 83 patients (31.6%), moderate in 36 (13.7%), and severe in 50 (19.0%). According to the BCLC staging system, 156 (59.3%) patients were classified as stage B and 107 (40.7%) were classified as stage C. 44 patients (16.7%) had PVT, 89 (33.8%) had extrahepatic spread and 39 (14.8%) had PVTT. Among patients with PVTT, 25 (64.1%) had type I–II PVTT, 14 (35.9%) had type III-IV PVTT.

Risk Factors of EGV Bleeding

EGV bleeding occurred in 43 (16.4%) patients after 150 (60–365) days of systemic therapy. Kaplan-Meier analysis estimated 1-year and 2-year cumulative incidence rates of 16.7% and 21.6%, respectively. In univariable analysis, EGV bleeding was associated with PTA (P=0.038), CTP scores (P=0.032), presence of PVT (P<0.001), type of PVTT (P=0.002), degree of varices (P=0.005), site of varices (P=0.012), and degree of ascites (P=0.002) (Supplementary Table 1). In the final multivariable analysis, the presence of PVT (odds ratio [OR]=3.84, 95% CI: 1.61–9.20, P=0.003), degree of varices (P=0.007, mild, OR=2.34, 95% CI: 0.73–7.49; moderate, OR=3.19, 95% CI: 0.86–11.75; severe, OR=6.77, 95% CI: 2.14–21.38), and degree of ascites (P=0.033, mild, OR=1.76, 95% CI:0.73–4.23; moderate-severe, OR=6.95, 95% CI:1.53–31.45) were independent risk factors for EGV bleeding (Table 2), exhibiting a higher cumulative incidence of EGV bleeding (Figure 1).

Table 2 Risk Factors of EGV Bleeding in Cirrhotic HCC Patients with Systemic Therapy

Figure 1 Kaplan–Meier survival estimated curves in patients with independent risk factors of 1-year EGV bleeding. (a) PVT; (b) The severity of varices; (c) The severity of ascites. Abbreviations: EGV, Esophagogastric varices; PVT, Portal vein thrombosis.

Pairwise log-rank tests showed that the cumulative incidence of EGV bleeding was significantly higher in the severe varices group compared with the absent (P<0.001) and mild groups (P=0.013) (Supplementary Figure 1). For ascites, both mild (P=0.003) and moderate-severe (P<0.001) ascites groups had significantly higher cumulative incidence of EGV bleeding than the absent group (Supplementary Figure 2). Furthermore, the cumulative incidence of EGV bleeding increased progressively with the number of independent risk factors present, from 3.1% with no risk factors to 12.1%, 35.3%, and 46.7% with 1, 2, and 3 factors, respectively (P for trend <0.001) (Figure 2a).

Figure 2 Kaplan–Meier survival estimated curves of EGV bleeding and overall survival between the number of independent risk factors present. (a) 1-year EGV bleeding; (b) 1-year overall survival. Panel (b) included 250 patients with evaluable tumor response data. Abbreviation: EGV, Esophagogastric varices.

Risk Factors of 1-Year Mortality

After one year of follow-up, 16 patients (6.1%) died, of whom 4 (25.0%) were attributed to EGV bleeding. Univariable logistic regression analysis indicated that PTA (P=0.019), PVTT (P<0.001), type of PVTT (P=0.002), AFP (P=0.004), AFP level >400 ng/mL (P<0.001), EGV bleeding during therapy (P=0.001), and therapy response (P=0.023) were significantly associated with 1-year mortality in HCC patients receiving systemic therapy (Supplementary Table 2). Multivariable logistic regression analysis with backward stepwise selection revealed that AFP >400 ng/mL (OR=9.18, 95% CI: 2.21–38.15, P=0.001), EGV bleeding (OR=6.66, 95% CI: 1.40–31.68, P=0.014), type of PVTT (P=0.022, I-II, OR=8.43, 95% CI: 1.60–44.43; III-IV, OR=4.04, 95% CI: 0.62–26.30), and tumor response during systemic therapy (P=0.025, SD, OR=3.11, 95% CI: 0.28–34.5; PD, OR=13.35, 95% CI: 1.19–150.18) were independent risk factors for composite outcomes of 1-year mortality (Table 3), exhibiting a shorter survival time (Figure 3).

Table 3 Risk Factors Related to 1-year Mortality in HCC Patients with Systemic Therapy

Figure 3 Kaplan–Meier survival estimated curves in patients with independent risk factors of 1-year mortality. (a) AFP>400 ng/mL; (b) EGV bleeding; (c) Tumor response; (d) Type of PVTT. Panel (c) included 250 patients with evaluable tumor response data. Abbreviations: AFP, Alpha-fetoprotein; EGV, Esophagogastric varices; CR, Complete response; PD, Progressive disease; PR, Partial response; SD, Stable disease; PVTT, Portal vein tumor thrombus.

Pairwise comparisons for tumor response showed that patients with PD had significantly lower survival than those with CR/PR (P=0.010) (Supplementary Figure 3). For PVTT type, patients with either type I–II (P<0.001) or type III–IV (P=0.002) PVTT had significantly lower survival than those without PVTT (Supplementary Figure 4). Like EGV bleeding, 1-year mortality increased progressively with the number of independent risk factors, from 0.9% with no risk factors to 2.5%, 15.9%, and 45.5% with 1, 2, and ≥3 factors, respectively (P for trend <0.001) (Figure 2b).

Efficacy of Systemic Therapy

Of the 263 patients, 166 (63.1%) received TKIs monotherapy (lenvatinib, n=128; sorafenib, n=27; regorafenib, n=11) and 97 (36.9%) received TKIs plus anti-PD-1 combination therapy. The ORR was 32.0% (80/250) among the 250 patients with evaluable imaging findings. TKIs plus anti-PD-1 showed higher ORR than TKIs alone (39.5% vs. 27.7%, P=0.017), while the EGV bleeding rate and 1-year mortality were comparable between the groups (Table 4). Among the TKIs monotherapy group, no significant differences were observed in ORR, EGV bleeding, or 1-year mortality among the three TKIs agents (Supplementary Table 3). Kaplan-Meier analysis also confirmed no significant difference in the cumulative incidence of EGV bleeding (log-rank P=0.991, Figure 4a) or overall survival (log-rank P=0.549, Figure 4b) between the two treatment groups. Extended follow-up analysis showed that the 2-year cumulative incidence of EGV bleeding remained comparable between the TKIs alone and TKIs plus anti-PD-1 groups (log-rank P=0.860, Figure 4c). Importantly, AFP levels were elevated in PD patients, suggesting a potential predictive value for tumor response (Figure 5).

Table 4 Efficacy and Outcomes of Cirrhotic HCC Patients with TKIs and TKIs with Anti-PD-1

Figure 4 Kaplan–Meier survival estimated curves of EGV bleeding and overall survival between TKIs alone and TKIs plus anti-PD-1 groups. (a) 1-year EGV bleeding; (b) 1-year overall survival; (c) 2-year EGV bleeding. Abbreviations: Anti-PD-1, Anti-programmed cell death-1 antibody; EGV, Esophagogastric varices; TKIs, Tyrosine kinase inhibitors.

Figure 5 Baseline characteristics and trends of some clinical parameters with different tumor responses. (a) WBC; (b) PLT; (c) Hb; (d) AST; (e) ALT; (f) TBIL; (g) ALB; (h) PTA; (i) AFP. Blue boxes represent baseline values and purple boxes represent follow-up values. ns: P>0.05; **: ≤0.01; ***: <0.001. Abbreviations: AFP, Alpha-fetoprotein; ALB, Albumin; ALT, Alanine aminotransferase; AST, Aspartate aminotransferase; CR, Complete response; TBIL, Total bilirubin; Hb, Hemoglobin; PD, Progressive disease; PLT, Platelet count; PR, Partial response; PTA, Prothrombin time activity; SD, Stable disease; WBC, White blood cell.

HBV Biomarkers Changes

Of 263 patients included in this study, 209 (79.5%) had HBV-related HCC. Among these patients, 162 (77.5%) had received long-term antiviral therapy before systemic treatment, 45 (21.5%) had initiated antiviral therapy at the start of systemic therapy, and 2 (1.0%) did not receive antiviral therapy. Virological responses were analyzed in 162 patients with pre-existing long-term antiviral therapy to minimize confounding from treatment duration. At the 1-year follow-up, HBV DNA negative conversion was achieved in 29.6% of patients in the TKIs alone group and in 37.5% in the TKIs plus anti-PD-1 group (P=0.739, Figure 6a). The median percentage decline in HBsAg levels from baseline was -14.3% versus -15.1% (P=0.883, Figure 6b).

Figure 6 The correlation between TKIs plus anti-PD-1 therapy and the changes of HBsAg and HBV DNA in HBV-related HCC patients with pre-existing long-term antiviral therapy. (a) showed HBV DNA negative conversion before and after treatment and (b) showed the median percentage decline in HBsAg. Abbreviations: Anti-PD-1, Anti-programmed cell death-1 antibody; HBsAg, Hepatitis B surface antigen; HBV, Hepatitis B virus; TKIs, Tyrosine kinase inhibitors.

Discussion

This retrospective cohort study evaluated 263 cirrhotic HCC patients receiving systemic therapy. PVT, ascites, and high bleeding risk of EGV were independent predictors of EGV bleeding, while AFP >400 ng/mL, EGV bleeding, PVTT type, and tumor progression independently predicted mortality. Moreover, the risk of both EGV bleeding and mortality increased progressively with the number of concurrent risk factors. Importantly, TKIs with anti-PD-1 combination therapy achieved superior tumor response rates without increasing EGV bleeding risk or mortality. These findings suggest that effective systemic treatment can be delivered safely in this high-risk population with careful baseline assessment of EGV bleeding risk.

EGV Bleeding Risk Assessment and Management

The pathophysiological basis of PHT is sinusoidal microvascular dysfunction. Capillarised liver sinusoidal endothelial cells and activated hepatic stellate cells reduce intrahepatic nitric oxide bioavailability and increase vascular resistance. The resultant splanchnic angiogenesis and increased portal pressure lead to EGV formation. These collateral vessels have thin and immature walls that are prone to rupture.²¹ Vascular endothelial growth factor-A (VEGF-A) is a key driver of this angiogenic process and is essential for both endothelial integrity and tumor angiogenesis.²² In HCC, PVT formation is further potentiated by tumor-derived IL-6 and TNF-α that injure portal venous endothelium, HIF-1α and VEGF upregulation that promotes blood stasis, and tissue factor overexpression that activates the coagulation cascade.²³

Bevacizumab, a monoclonal antibody that targets VEGF-A, has been linked to an increased risk of gastrointestinal bleeding, limiting its use in PHT.²² Although none of the patients in our study received bevacizumab, all patients received TKIs. These therapies also inhibit VEGF receptors among multiple kinases and may affect PHT pathophysiology. A prospective study suggested that lenvatinib may aggravate PHT,²⁴ and the specific relationship between ICIs and EGV bleeding remains unclear.²⁵

In our study, the prevalence of EGV in patients with cirrhotic HCC was approximately 60%, which is consistent with previous reports of 40%-60%.^26,27 EGV bleeding occurred in 16.35% of patients after a median of 150 days of systemic therapy, and high bleeding risk EGV emerged as independent predictors of bleeding regardless of tumor response. Notably, eight patients with red color signs on baseline oesophagogastroduodenoscopy developed bleeding during treatment. Consistent with prior evidence that PVT increases the risk of high-risk varices and variceal bleeding,²⁸ our analysis identified PVT and moderate-to-severe ascites as additional independent risk factors for EGV bleeding.

Acute EGV bleeding is independently associated with poor survival outcome in HCC patients.²⁹ EGV bleeding can be managed with vasoactive drugs, endoscopic therapy, or transjugular intrahepatic portosystemic shunt, which has been reported to improve outcomes in HCC patients receiving systemic therapy.^30–32 In our study, most patients with high bleeding risk EGV at baseline did not receive prophylactic endoscopic treatment before systemic therapy. However, most patients with a history of bleeding have undergone endoscopic intervention, which may explain the lack of an association between prior bleeding and on-treatment bleeding risk. This suggests that identifying bleeding risk and providing prophylactic treatment for high bleeding risk EGV can reduce bleeding events in patients with cirrhotic HCC receiving systemic therapy.

Efficacy and Safety of Systemic Therapy

In our study, combinations of TKIs with anti-PD-1 demonstrated superiority over TKIs alone in tumor response. This may be partly attributed to the immunomodulatory effects of TKIs. Lenvatinib has been shown to modulate the tumor immune microenvironment and enhance anti-PD-1 efficacy.³³ Moreover, this combination did not increase EGV bleeding rates. Among TKIs monotherapy subgroups, no significant differences in EGV bleeding or mortality were observed among lenvatinib, sorafenib, and regorafenib.

Beyond EGV bleeding, AFP >400 ng/mL, PVTT type, and tumor progression were independently associated with mortality. Elevated AFP may reflect an aggressive HCC subtype with increased VEGF pathway activity and is associated with poorer TKIs response and worse prognosis.^34,35 Consistent with these findings, elevated AFP was associated with tumor progression in our cohort. These findings emphasize the importance of baseline and on-treatment AFP monitoring in guiding clinical decision-making. PVTT type was also independently associated with mortality, consistent with prior evidence that survival differs substantially according to the extent of thrombus involvement.³⁶

Health-related quality of life (HRQOL) is influenced by demographic, clinical, psychological, and symptom severity.³⁷ The independent mortality predictors identified in our study are also likely to impair HRQOL through recurrent hospitalization, endoscopic intervention, and treatment discontinuation. Recent evidence indicates that atezolizumab/bevacizumab and lenvatinib delay HRQOL deterioration compared to sorafenib, while anti-PD-1 therapy has minimal impact on HRQOL.³⁸ Noting that TKIs plus anti-PD-1 achieved superior tumor response without increasing bleeding risk suggests a potential HRQOL advantage of combination therapy.

Beyond clinical outcomes, the heterogeneous tumor responses observed in our cohort may also be partly explained by the underlying molecular diversity of HCC. HCC is divided into three major subtypes (proliferative, CTNNB1-mutated and metabolic disease-associated) with distinctive molecular and immunological features, and only 20–30% of HCC patients respond to immunotherapies.³⁹ Immune-active HCCs, characterized by CD4+ T and CD8+ T cell infiltration, are more likely to respond to ICIs. In contrast, immune-exhausted HCCs overlapping with the CTNNB1-mutated subclass are associated with T cell exhaustion and increased Treg infiltration.⁴⁰ Such genetic heterogeneity frequently underlies primary resistance to systemic therapy, which may be mitigated by molecular subtype-targeted therapies.⁴¹ However, the relationship between molecular subtypes and EGV bleeding risk remains to be elucidated.⁴²

HBV Management and Virological Responses

Chronic HBV infection promotes an immunosuppressive microenvironment via the PD-1/PD-L1 pathway, thereby facilitating hepatocarcinogenesis.^43,44 In our cohort, nearly all patients received antiviral therapy, and most had been receiving long-term antiviral therapy before systemic treatment. Recent evidence indicates that baseline HBV viral load does not significantly affect outcomes in patients with HCC receiving systemic therapy when antiviral treatment is maintained,⁴⁵ underscoring the protective role of continuous antiviral treatment. However, elevations in HBV DNA and HBsAg titers during systemic therapy may indicate poorer tumor response and shorter survival.⁴⁶

Antiviral therapy and PD-1 blockade may demonstrate synergistic effects in HBV-related HCC.⁴⁷ Antiviral therapy may enhance immunotherapy efficacy by reducing viral antigen load, alleviating T-cell exhaustion, and attenuating hepatic inflammation.⁴⁸ Conversely, PD-1/PD-L1 blockade restores anti-tumor immunity and reinvigorates antiviral T-cell responses, potentially enhancing viral suppression.⁴⁹ Emerging evidence suggests that ICI therapy may facilitate HBsAg decline in HBV-related HCC.^43,50 In our cohort, patients receiving TKIs plus anti-PD-1 showed a numerically higher rate of HBV DNA negative conversion and greater HBsAg decline than those receiving TKIs alone, although the differences did not reach statistical significance. The 1-year observation period may have been insufficient to detect differences in HBsAg levels, and extended follow-up with larger sample sizes is warranted.

Limitations

This study has several limitations. First, its retrospective design and sample size may limit the generalizability of our findings. Moreover, the substantially unbalanced distribution of TKIs subtypes precluded a meaningful comparison of efficacy and bleeding risk. Second, genomic profiling and HRQOL assessment were not routinely performed at our center during the study period, limiting our ability to explore associations between molecular subtypes and clinical outcomes or to evaluate the impact of systemic therapy. Third, most patients did not receive prophylactic endoscopic therapy or beta-blockers before systemic therapy, limiting our ability to assess the impact of these interventions on EGV bleeding risk. Finally, we could not definitively attribute EGV bleeding events to drug-related adverse effects, disease progression, or the natural history of cirrhosis, because these factors often coexist and may be difficult to disentangle in clinical practice. Future well-designed prospective multicenter studies with larger sample sizes are warranted to validate our findings and compare outcomes among different systemic regimens. Such studies should also integrate molecular profiling and HRQOL instruments and evaluate EGV bleeding risk in cirrhotic versus non-cirrhotic HCC patients.

Conclusion

In conclusion, PVT, ascites, and high bleeding risk of EGV were independent predictors of EGV bleeding, while AFP >400 ng/mL, EGV bleeding, the type of PVTT, and tumor progression independently predicted mortality in cirrhotic HCC patients receiving systemic therapy. The risk of both EGV bleeding and mortality increased progressively with the accumulation of independent risk factors. TKIs with anti-PD-1 combination therapy achieved superior ORR without increasing EGV bleeding risk or mortality, supporting its use in appropriately selected patients with careful assessment of EGV bleeding risk.