Abstract
Introduction
Optimal treatment for recurrent hepatocellular carcinoma (HCC) after curative resection remains unclear, particularly in patients with microvascular invasion (MVI). This study evaluated post-recurrence treatment patterns and survival outcomes in this population.
Methods
We retrospectively included patients with MVI-positive HCC who developed recurrence after curative hepatectomy at a single center. Post-recurrence progression-free survival (PFS) was the primary endpoint and overall survival (OS) was the secondary endpoint. Kaplan–Meier analysis was used to compare survival across post-recurrence treatment groups. Cox regression, subgroup analysis, propensity score matching, and correlation analysis between recurrence-free survival (RFS) and post-recurrence PFS were also performed.
Results
A total of 150 patients were included, and 132 received post-recurrence treatment. Treatment strategies were heterogeneous, with locoregional therapy plus tyrosine kinase inhibitor (TKI) and immune checkpoint inhibitor (ICI), surgery-based therapy, locoregional therapy plus TKI, and locoregional therapy alone being the main approaches. Median PFS and OS were 10.4 and 26.7 months, respectively. Survival outcomes differed across treatment groups in the unadjusted analysis. Patients receiving surgery-based therapy showed more favorable PFS and OS than those receiving locoregional therapy alone, although this finding should be interpreted cautiously given the non-randomized treatment allocation and potential baseline imbalances among treatment groups. In exploratory multivariable analysis, locoregional therapy plus TKI, locoregional therapy plus TKI and ICI, and surgery-based therapy were associated with longer PFS, whereas no treatment regimen was independently associated with OS. Among patients with potentially resectable recurrence, surgery-based therapy showed numerically more favorable outcomes before matching. However, the differences were not statistically significant and became less apparent after propensity score matching.
Conclusion
Post-recurrence management in MVI-positive HCC after curative resection was highly heterogeneous. Surgery-based therapy was associated with favorable outcomes in selected patients, but this association may have been influenced by recurrence pattern and patient selection. Locoregional therapy combined with systemic treatment may be considered a clinically relevant option for non-surgical candidates, although its comparative effectiveness requires further validation.
Keywords: hepatocellular carcinoma, microvascular invasion, recurrent hepatocellular carcinoma, post-recurrence treatment, survival
Introduction
According to the latest data from the National Cancer Center of China based on cancer registry and follow-up surveillance, liver cancer ranks fourth in incidence and second in cancer-related mortality among all malignant tumors in China, posing a major threat to public health. Despite continuous advances in diagnostic and therapeutic techniques in recent years, the overall prognosis of hepatocellular carcinoma (HCC) remains unsatisfactory, with a 5-year survival rate of only approximately 18% and a 5-year recurrence rate of up to 70% after curative treatment.1–3 Recurrence and metastasis are the major determinants of long-term survival in patients with HCC. For patients undergoing curative resection, whether postoperative management should rely on active surveillance alone or incorporate adjuvant therapy to reduce the risk of recurrence has long remained controversial. To improve outcomes, an increasing number of studies have explored effective postoperative adjuvant treatment strategies. The international multicenter Phase III IMbrave050 trial, reported in 2023, showed that among patients with HCC at high risk of recurrence after curative resection or ablation, atezolizumab plus bevacizumab improved recurrence-free survival (RFS) compared with active surveillance. However, subsequent updated results suggested that this early RFS benefit was not sustained over time,4 and therefore the optimal choice of postoperative adjuvant therapy remains inconclusive.
Among the pathological factors associated with postoperative recurrence, microvascular invasion (MVI) is particularly important because it reflects microscopic vascular dissemination and an invasive tumor phenotype. Patients with MVI-positive HCC have a substantially higher risk of early recurrence after curative resection, and recurrence in this subgroup may represent biologically aggressive disease rather than simple local tumor regrowth.5 Therefore, recurrent HCC arising from an MVI-positive primary tumor may differ from MVI-negative recurrent disease in recurrence timing, recurrence pattern, treatment sensitivity, and long-term prognosis. These features provide a rationale for evaluating post-recurrence treatment strategies specifically in patients with MVI-positive HCC.
In comparison with the controversy surrounding postoperative adjuvant therapy, treatment decision-making for patients who develop recurrence after curative resection presents an even more complex and practical challenge in clinical practice. Recurrent HCC is highly heterogeneous, and factors such as time to recurrence, location and number of recurrent lesions, the presence of vascular invasion or extrahepatic metastasis, residual liver function reserve, and patient performance status may all substantially influence treatment selection and prognosis.6 Existing studies have shown that a wide range of treatment options are available for these patients, including repeat hepatectomy, ablation, transarterial chemoembolization (TACE), and systemic approaches such as targeted therapy, immunotherapy, or combined locoregional and systemic treatment.7,8 For carefully selected patients, repeat curative-intent treatments such as re-resection or ablation may still provide favorable long-term survival benefits, whereas for those who are not candidates for repeat curative treatment, TACE, radiotherapy, and systemic therapy represent important alternative options.
Nevertheless, it should be noted that most current domestic and international guidelines continue to focus primarily on treatment-naive HCC, and specific management pathways for recurrence after curative resection are still not supported by high-level evidence. As a result, no unified or standardized recommendation has yet been established. Previous reviews have clearly pointed out that there is still no universally accepted standard treatment guideline for recurrent HCC, and that clinical decision-making generally continues to follow the treatment principles for primary HCC while taking into account recurrence pattern and individual patient characteristics. In other words, whether patients with postoperative recurrence should preferentially undergo repeat surgery, locoregional treatment, or combined systemic therapy remains uncertain, and substantial variation in treatment strategies persists across centers.7
For this reason, real-world studies focusing on retreatment strategies for patients with recurrent HCC after curative resection are of important clinical value. Given the distinct biological behavior and high recurrence risk associated with MVI, dedicated evidence in the MVI-positive population is needed to better understand post-recurrence treatment selection and outcomes. In this single-center retrospective study, we enrolled 150 patients with HCC who developed recurrence after curative resection and had concomitant MVI. We compared the outcomes associated with different treatment modalities administered after recurrence. In addition, we further explored clinical factors that may predict survival benefit from post-recurrence treatment in this patient population.
Methods
Patients and Study Design
This study retrospectively enrolled patients who underwent curative-intent hepatectomy for HCC at Peking Union Medical College Hospital (PUMCH) between January 1, 2017 and March 31, 2024. Eligible patients met all of the following criteria: (1) curative resection was achieved, as defined by both intraoperative and postoperative assessments. Intraoperative criteria included: (i) no macroscopic tumor thrombus in the hepatic veins, portal veins, bile ducts, or inferior vena cava; (ii) no invasion of adjacent organs, and no hilar lymph node involvement or distant metastasis; and (iii) a resection margin of ≥1 cm from the tumor boundary, or a negative resection margin when the margin was <1 cm. Postoperative criteria included: (i) no evidence of residual tumor on at least two imaging modalities, including ultrasonography, computed tomography (CT), or magnetic resonance imaging (MRI), performed 1–2 months after surgery; and (ii) normalization of serum tumor markers within 2 months after surgery in patients with elevated preoperative alpha-fetoprotein (AFP), des-gamma-carboxy prothrombin (DCP), or other tumor markers. Additional inclusion criteria were: (2) age 18–75 years; and (3) pathologically confirmed HCC with MVI. MVI was defined and graded according to the Chinese guidelines for the standardized pathological evaluation of primary liver cancer. Briefly, M1 was defined as ≤5 MVI foci within 1 cm from the tumor margin, whereas M2 was defined as >5 MVI foci or any MVI focus located more than 1 cm away from the tumor margin. Only patients with M1 or M2 disease were included in this study. Patients were excluded if they: (1) had another active malignancy, including malignancies in sustained remission; (2) died of causes unrelated to HCC during follow-up; or (3) had missing key clinical data.
Data Collection
Patients were routinely followed up beginning 4 weeks after surgery. Follow-up assessments included serum tumor marker testing, abdominal ultrasonography, and contrast-enhanced abdominal CT or MRI. When distant metastasis was clinically suspected, additional examinations such as chest CT, bone scintigraphy, or positron emission tomography-computed tomography (PET-CT) were performed as appropriate. Follow-up continued until death or loss to follow-up. The data cutoff date was April 1, 2026. The following clinicopathological variables were collected: age, sex, hepatitis B surface antigen (HBsAg), hepatitis B virus (HBV) DNA, hepatitis C virus antibody (HCV Ab), modified albumin-bilirubin (mALBI) grade, Child–Pugh class, AFP, China Liver Cancer (CNLC) stage, Barcelona Clinic Liver Cancer (BCLC) stage, tumor differentiation, resection margin status, tumor number, tumor size, satellite lesions, tumor embolus, MVI status, cirrhosis, postoperative treatment, number of intrahepatic recurrent tumors, presence of portal vein tumor thrombus at recurrence, metastatic organs, post-recurrence treatment strategies, treatment response, and time to post-recurrence progression. This study was reviewed and approved by the Ethics Committee of PUMCH (approval ID: I-26PJ0238).
Study Outcomes
The primary endpoint was post-recurrence progression-free survival (PFS), and overall survival (OS) was defined as the secondary endpoint. Recurrence-free survival (RFS) was defined as the interval from initial curative resection to first documented recurrence. Post-recurrence PFS was calculated from the date of first documented recurrence to radiological disease progression after recurrence, death, or last follow-up, whichever occurred first. OS was calculated from the date of first documented recurrence to death or last follow-up. Tumor response and disease progression were evaluated based on contrast-enhanced computed tomography or magnetic resonance imaging according to the Response Evaluation Criteria in Solid Tumors (RECIST), version 1.1, together with modified RECIST (mRECIST) criteria for HCC when applicable. After recurrence, imaging evaluation was generally performed every 2–3 months, or earlier when clinically indicated.
Post-Recurrence Treatment Decision-Making and Treatment Classification
Post-recurrence treatment was determined in routine clinical practice through multidisciplinary evaluation by hepatobiliary surgeons, interventional radiologists, oncologists, and radiologists. Treatment decisions were based on recurrence pattern, number and location of recurrent lesions, technical resectability, liver functional reserve, performance status, presence of portal vein tumor thrombus or extrahepatic spread, prior treatment history, patient preference, and the availability of targeted and immune-based therapies during the study period. Repeat surgery was generally considered for patients with technically resectable intrahepatic recurrence, adequate future liver remnant, preserved liver function, and acceptable operative risk. Locoregional therapy was selected for patients with intrahepatic recurrence who were not optimal candidates for repeat surgery or who required tumor control by transarterial or ablative approaches. Systemic therapy, alone or in combination with locoregional therapy, was considered for patients with more extensive tumor burden, high-risk recurrence features, unresectable disease, or anticipated high risk of further progression. Radiotherapy-based treatment was mainly used for patients with portal vein tumor thrombus or lesions suitable for local radiation control. Treatment groups were mutually exclusive and were classified according to the initial post-recurrence treatment strategy. Subsequent treatment modification or crossover after disease progression was recorded separately but did not alter the initial treatment classification.
Statistical Analysis
Continuous variables are presented as mean ± standard deviation (SD) or median with interquartile range (IQR), as appropriate. For comparisons between two groups, Student’s t test or the Mann–Whitney U-test was used according to data distribution. For comparisons across multiple treatment groups, one-way analysis of variance or the Kruskal–Wallis test was used, as appropriate. Categorical variables are presented as frequencies and percentages and were compared using the Pearson chi-square test or Fisher’s exact test, as appropriate. RFS, PFS, and OS were estimated using the Kaplan–Meier method and compared with the Log rank test. Cox proportional hazards regression analyses were performed to identify independent prognostic factors. Variables with P < 0.05 in univariable analysis, together with clinically relevant covariates, were entered into the multivariable Cox regression model.
To further address potential selection bias related to recurrence patterns, an additional subgroup analysis was performed among patients with potentially resectable recurrence, defined as intrahepatic-only recurrence without portal vein invasion or extrahepatic spread. In this subgroup, patients were classified into surgery-based and non-surgery-based treatment groups. Propensity score matching was performed using nearest-neighbor matching at a 1:2 ratio with a caliper of 0.2, with exact matching for the number of intrahepatic recurrent tumors. The propensity score model included RFS, primary tumor size, age, log-transformed AFP, and MVI grade. Covariate balance before and after matching was assessed using standardized mean differences (SMDs). Survival outcomes before and after matching were evaluated using Kaplan–Meier analysis and Cox regression, with robust standard errors used for matched data. Because treatment allocation was highly heterogeneous and several treatment groups had small sample sizes, fully balanced propensity score-based comparisons across all post-recurrence treatment categories were not statistically robust. Therefore, propensity score matching was restricted to the clinically relevant comparison between surgery-based and non-surgery-based therapy among patients with potentially resectable recurrence. Other multivariable and subgroup analyses were considered exploratory.
All statistical analyses were performed using R software (version 4.3.1), and a two-sided P < 0.05 was considered statistically significant.
Results
Patient Characteristics
Between January 2017 and March 2024, a total of 1223 patients with HCC who underwent curative-intent hepatectomy were initially screened. Of these, 785 patients did not have MVI, 75 were pathologically diagnosed with combined hepatocellular-cholangiocarcinoma (cHCC-CCA), and 14 experienced recurrence within 2 months after surgery. Among the remaining 349 patients with MVI-positive HCC, 199 had no recurrence by the follow-up cutoff. Ultimately, 150 patients with postoperative recurrence were included in the final analysis. Among these 150 patients, 43 were lost to follow-up during the post-recurrence course.
Post-Recurrence Treatment Regimens and Outcomes
The median post-recurrence follow-up for the entire cohort was 24.6 months (95% CI: 17.0–29.6). Among patients who received treatment after recurrence, the median PFS and OS were 10.4 months (95% CI: 8.2–13.2) and 26.7 months (95% CI: 23.0–44.5), respectively.
Post-recurrence treatment strategies were heterogeneous (Figure 1). Treatment categories were mutually exclusive and were defined according to the initial post-recurrence treatment strategy. Of the 150 patients, 50 received locoregional therapy combined with tyrosine kinase inhibitors and immune checkpoint inhibitors (Locoregional-TKI-ICI), 23 received surgery-based therapy (defined as repeat curative-intent surgery as the core treatment, either alone or combined with other locoregional or systemic treatments), 21 received locoregional therapy combined with TKI (Locoregional-TKI), 18 received locoregional therapy alone, 15 received radiotherapy-based therapy (defined as treatment regimens containing radiotherapy as a major component), and 5 received lenvatinib alone; 7 patients received no treatment and treatment information was unavailable for 11 patients. Locoregional therapy included transarterial chemoembolization, radiofrequency ablation, and/or hepatic arterial infusion chemotherapy. Subsequent treatment changes or crossover after disease progression were recorded separately and did not alter the initial post-recurrence treatment classification. Among the 132 patients who received any post-recurrence treatment, 127 subsequently developed disease progression. After progression, 64 patients underwent further treatment, most of whom received ICI-based therapy (n = 40, 62.5%). Baseline clinicopathological characteristics stratified by post-recurrence treatment regimen are shown in Table 1. The specific targeted agents and immune checkpoint inhibitors used in systemic-therapy-containing groups are summarized in Supplementary Table S1.
Figure 1.
Treatment patterns across the post-resection, post-recurrence, and post-progression stages in patients with MVI-positive HCC after curative resection. Post-recurrence treatment categories were defined according to the first treatment strategy used after recurrence. Surgery-based therapy was centered on repeat curative-intent surgery, with or without additional locoregional or systemic treatment. Locoregional therapy consisted of TACE, RFA, and/or HAIC. Combination groups were defined as locoregional therapy plus TKI, or locoregional therapy plus both TKI and ICI.
Table 1.
Patient Characteristics According to Post-Recurrence Treatment Regimens
| Variable | Total (n = 150) |
Locoregional-TKI-ICI (n = 50) |
Locoregional-TKI (n = 21) |
Surgery-Based Therapy (n = 23) |
No More Treatment (n = 7) |
Locoregional Therapy (n = 18) |
RT-Based Therapy (n = 15) |
Unknown (n = 11) |
Lenvatinib (n = 5) |
P value |
|---|---|---|---|---|---|---|---|---|---|---|
| Median age (IQR) | 58 (51–66) | 58 (51–65) | 50 (45–65) | 56 (53–61) | 60 (54–72) | 64 (54–67) | 58 (54–62) | 61 (56–67) | 53 (52–58) | 0.426 |
| Age, n (%) | 0.211 | |||||||||
| <65 years | 107 (71.3%) | 37 (74.0%) | 15 (71.4%) | 19 (82.6%) | 4 (57.1%) | 9 (50.0%) | 13 (86.7%) | 6 (54.5%) | 4 (80.0%) | |
| ≥65 years | 43 (28.7%) | 13 (26.0%) | 6 (28.6%) | 4 (17.4%) | 3 (42.9%) | 9 (50.0%) | 2 (13.3%) | 5 (45.5%) | 1 (20.0%) | |
| Sex, n (%) | 0.594 | |||||||||
| Male | 118 (78.7%) | 38 (76.0%) | 16 (76.2%) | 21 (91.3%) | 7 (100.0%) | 13 (72.2%) | 11 (73.3%) | 8 (72.7%) | 4 (80.0%) | |
| Female | 32 (21.3%) | 12 (24.0%) | 5 (23.8%) | 2 (8.7%) | 0 (0.0%) | 5 (27.8%) | 4 (26.7%) | 3 (27.3%) | 1 (20.0%) | |
| HBsAg, n (%) | 0.792 | |||||||||
| Positive | 116 (77.3%) | 41 (82.0%) | 16 (76.2%) | 15 (65.2%) | 5 (71.4%) | 13 (72.2%) | 13 (86.7%) | 9 (81.8%) | 4 (80.0%) | |
| Negative | 34 (22.7%) | 9 (18.0%) | 5 (23.8%) | 8 (34.8%) | 2 (28.6%) | 5 (27.8%) | 2 (13.3%) | 2 (18.2%) | 1 (20.0%) | |
| HBV DNA, n (%) | 0.742 | |||||||||
| ≤20 IU/mL | 100 (66.7%) | 32 (64.0%) | 14 (66.7%) | 16 (69.6%) | 6 (85.7%) | 13 (72.2%) | 7 (46.7%) | 8 (72.7%) | 4 (80.0%) | |
| >20 IU/mL | 50 (33.3%) | 18 (36.0%) | 7 (33.3%) | 7 (30.4%) | 1 (14.3%) | 5 (27.8%) | 8 (53.3%) | 3 (27.3%) | 1 (20.0%) | |
| HCV-Ab, n (%) | 0.454 | |||||||||
| Positive | 9 (6.0%) | 1 (2.0%) | 2 (9.5%) | 3 (13.0%) | 0 (0.0%) | 2 (11.1%) | 1 (6.7%) | 0 (0.0%) | 0 (0.0%) | |
| Negative | 141 (94.0%) | 49 (98.0%) | 19 (90.5%) | 20 (87.0%) | 7 (100.0%) | 16 (88.9%) | 14 (93.3%) | 11 (100.0%) | 5 (100.0%) | |
| mALBI grade, n (%) | 0.310 | |||||||||
| 1 | 105 (70.0%) | 37 (74.0%) | 14 (66.7%) | 18 (78.3%) | 4 (57.1%) | 14 (77.8%) | 8 (53.3%) | 7 (63.6%) | 3 (60.0%) | |
| 2a | 31 (20.7%) | 7 (14.0%) | 7 (33.3%) | 5 (21.7%) | 2 (28.6%) | 3 (16.7%) | 4 (26.7%) | 1 (9.1%) | 2 (40.0%) | |
| 2b | 12 (8.0%) | 5 (10.0%) | 0 (0.0%) | 0 (0.0%) | 1 (14.3%) | 1 (5.6%) | 2 (13.3%) | 3 (27.3%) | 0 (0.0%) | |
| 3 | 2 (1.3%) | 1 (2.0%) | 0 (0.0%) | 0 (0.0%) | 0 (0.0%) | 0 (0.0%) | 1 (6.7%) | 0 (0.0%) | 0 (0.0%) | |
| Child-Pugh, n (%) | 0.632 | |||||||||
| A | 145 (96.7%) | 48 (96.0%) | 21 (100.0%) | 23 (100.0%) | 7 (100.0%) | 17 (94.4%) | 14 (93.3%) | 10 (90.9%) | 5 (100.0%) | |
| B | 5 (3.3%) | 2 (4.0%) | 0 (0.0%) | 0 (0.0%) | 0 (0.0%) | 1 (5.6%) | 1 (6.7%) | 1 (9.1%) | 0 (0.0%) | |
| AFP, n (%) | 0.729 | |||||||||
| ≤400 ng/mL | 100 (66.7%) | 32 (64.0%) | 12 (57.1%) | 17 (73.9%) | 4 (57.1%) | 15 (83.3%) | 10 (66.7%) | 7 (63.6%) | 3 (60.0%) | |
| >400 ng/mL | 50 (33.3%) | 18 (36.0%) | 9 (42.9%) | 6 (26.1%) | 3 (42.9%) | 3 (16.7%) | 5 (33.3%) | 4 (36.4%) | 2 (40.0%) | |
| Post-surgery treatment, n (%) | 0.017 | |||||||||
| TACE | 55 (36.7%) | 17 (34.0%) | 7 (33.3%) | 12 (52.2%) | 1 (14.3%) | 7 (38.9%) | 6 (40.0%) | 3 (27.3%) | 2 (40.0%) | |
| No adjuvant treatment | 29 (19.3%) | 8 (16.0%) | 3 (14.3%) | 5 (21.7%) | 5 (71.4%) | 3 (16.7%) | 0 (0.0%) | 4 (36.4%) | 1 (20.0%) | |
| TKI-based therapy | 31 (20.7%) | 8 (16.0%) | 8 (38.1%) | 1 (4.3%) | 1 (14.3%) | 4 (22.2%) | 3 (20.0%) | 4 (36.4%) | 2 (40.0%) | |
| IC- based therapy | 35 (23.3%) | 17 (34.0%) | 3 (14.3%) | 5 (21.7%) | 0 (0.0%) | 4 (22.2%) | 6 (40.0%) | 0 (0.0%) | 0 (0.0%) | |
| CNLC stage, n (%) | 0.609 | |||||||||
| Ia | 53 (35.3%) | 17 (34.0%) | 8 (38.1%) | 9 (39.1%) | 4 (57.1%) | 7 (38.9%) | 3 (20.0%) | 2 (18.2%) | 3 (60.0%) | |
| Ib | 69 (46.0%) | 24 (48.0%) | 9 (42.9%) | 9 (39.1%) | 2 (28.6%) | 6 (33.3%) | 10 (66.7%) | 7 (63.6%) | 2 (40.0%) | |
| IIa | 25 (16.7%) | 9 (18.0%) | 4 (19.0%) | 3 (13.0%) | 1 (14.3%) | 5 (27.8%) | 1 (6.7%) | 2 (18.2%) | 0 (0.0%) | |
| IIb | 3 (2.0%) | 0 (0.0%) | 0 (0.0%) | 2 (8.7%) | 0 (0.0%) | 0 (0.0%) | 1 (6.7%) | 0 (0.0%) | 0 (0.0%) | |
| BCLC stage, n (%) | 0.273 | |||||||||
| 0 | 5 (3.3%) | 1 (2.0%) | 0 (0.0%) | 2 (8.7%) | 0 (0.0%) | 1 (5.6%) | 0 (0.0%) | 0 (0.0%) | 1 (20.0%) | |
| A | 26 (17.3%) | 7 (14.0%) | 6 (28.6%) | 6 (26.1%) | 1 (14.3%) | 2 (11.1%) | 4 (26.7%) | 0 (0.0%) | 0 (0.0%) | |
| B | 119 (79.3%) | 42 (84.0%) | 15 (71.4%) | 15 (65.2%) | 6 (85.7%) | 15 (83.3%) | 11 (73.3%) | 11 (100.0%) | 4 (80.0%) | |
| Differentiation status, n (%) | 0.496 | |||||||||
| Well-moderately | 109 (72.7%) | 36 (72.0%) | 16 (76.2%) | 18 (78.3%) | 7 (100.0%) | 11 (61.1%) | 9 (60.0%) | 9 (81.8%) | 3 (60.0%) | |
| Poorly | 41 (27.3%) | 14 (28.0%) | 5 (23.8%) | 5 (21.7%) | 0 (0.0%) | 7 (38.9%) | 6 (40.0%) | 2 (18.2%) | 2 (40.0%) | |
| Margin, n (%) | 0.291 | |||||||||
| ≥0.5 cm | 32 (21.3%) | 13 (26.0%) | 5 (23.8%) | 7 (30.4%) | 1 (14.3%) | 3 (16.7%) | 0 (0.0%) | 3 (27.3%) | 0 (0.0%) | |
| <0.5 cm | 118 (78.7%) | 37 (74.0%) | 16 (76.2%) | 16 (69.6%) | 6 (85.7%) | 15 (83.3%) | 15 (100.0%) | 8 (72.7%) | 5 (100.0%) | |
| Tumor number, n (%) | 0.607 | |||||||||
| 1 | 114 (76.0%) | 40 (80.0%) | 15 (71.4%) | 15 (65.2%) | 6 (85.7%) | 13 (72.2%) | 11 (73.3%) | 9 (81.8%) | 5 (100.0%) | |
| 2 | 26 (17.3%) | 8 (16.0%) | 5 (23.8%) | 5 (21.7%) | 1 (14.3%) | 5 (27.8%) | 1 (6.7%) | 1 (9.1%) | 0 (0.0%) | |
| ≥3 | 10 (6.7%) | 2 (4.0%) | 1 (4.8%) | 3 (13.0%) | 0 (0.0%) | 0 (0.0%) | 3 (20.0%) | 1 (9.1%) | 0 (0.0%) | |
| Tumor size, median (IQR) | 5.6 (3.7–8) | 6 (4.5–8) | 5 (3–7.9) | 4.5 (2.6–6) | 4 (3.9–8.2) | 5.2 (4–6.7) | 6.5 (3.4–8.7) | 8.5 (5.3–10.2) | 4.8 (4.5–9) | 0.149 |
| Satellite, n (%) | 0.238 | |||||||||
| Absent | 97 (64.7%) | 35 (70.0%) | 12 (57.1%) | 17 (73.9%) | 5 (71.4%) | 11 (61.1%) | 6 (40.0%) | 6 (54.5%) | 5 (100.0%) | |
| Present | 53 (35.3%) | 15 (30.0%) | 9 (42.9%) | 6 (26.1%) | 2 (28.6%) | 7 (38.9%) | 9 (60.0%) | 5 (45.5%) | 0 (0.0%) | |
| Tumor embolus, n (%) | 0.128 | |||||||||
| Absent | 125 (83.3%) | 42 (84.0%) | 20 (95.2%) | 18 (78.3%) | 5 (71.4%) | 15 (83.3%) | 14 (93.3%) | 6 (54.5%) | 5 (100.0%) | |
| Present | 25 (16.7%) | 8 (16.0%) | 1 (4.8%) | 5 (21.7%) | 2 (28.6%) | 3 (16.7%) | 1 (6.7%) | 5 (45.5%) | 0 (0.0%) | |
| MVI, n (%) | 0.008 | |||||||||
| MVI 1 | 84 (56.0%) | 28 (56.0%) | 9 (42.9%) | 14 (60.9%) | 3 (42.9%) | 14 (77.8%) | 11 (73.3%) | 1 (9.1%) | 4 (80.0%) | |
| MVI 2 | 66 (44.0%) | 22 (44.0%) | 12 (57.1%) | 9 (39.1%) | 4 (57.1%) | 4 (22.2%) | 4 (26.7%) | 10 (90.9%) | 1 (20.0%) | |
| Cirrhosis, n (%) | 0.449 | |||||||||
| Absent | 23 (15.3%) | 8 (16.0%) | 4 (19.0%) | 1 (4.3%) | 2 (28.6%) | 3 (16.7%) | 2 (13.3%) | 1 (9.1%) | 2 (40.0%) | |
| Present | 127 (84.7%) | 42 (84.0%) | 17 (81.0%) | 22 (95.7%) | 5 (71.4%) | 15 (83.3%) | 13 (86.7%) | 10 (90.9%) | 3 (60.0%) | |
| Recurrence intrahepatic tumors, n (%) | <0.001 | |||||||||
| None | 19 (12.7%) | 10 (20.0%) | 0 (0.0%) | 3 (13.0%) | 1 (14.3%) | 0 (0.0%) | 0 (0.0%) | 5 (45.5%) | 0 (0.0%) | |
| 1 | 44 (29.3%) | 8 (16.0%) | 7 (33.3%) | 14 (60.9%) | 1 (14.3%) | 10 (55.6%) | 3 (20.0%) | 0 (0.0%) | 1 (20.0%) | |
| ≥2 | 80 (53.3%) | 31 (62.0%) | 12 (57.1%) | 6 (26.1%) | 4 (57.1%) | 8 (44.4%) | 12 (80.0%) | 4 (36.4%) | 3 (60.0%) | |
| Unknown | 7 (4.7%) | 1 (2.0%) | 2 (9.5%) | 0 (0.0%) | 1 (14.3%) | 0 (0.0%) | 0 (0.0%) | 2 (18.2%) | 1 (20.0%) | |
| Portal vein invasion, n (%) | <0.001 | |||||||||
| Absent | 124 (82.7%) | 50 (100.0%) | 21 (100.0%) | 23 (100.0%) | 7 (100.0%) | 18 (100.0%) | 0 (0.0%) | 0 (0.0%) | 5 (100.0%) | |
| Present | 15 (10.0%) | 0 (0.0%) | 0 (0.0%) | 0 (0.0%) | 0 (0.0%) | 0 (0.0%) | 15 (100.0%) | 0 (0.0%) | 0 (0.0%) | |
| Unknown | 11 (7.3%) | 0 (0.0%) | 0 (0.0%) | 0 (0.0%) | 0 (0.0%) | 0 (0.0%) | 0 (0.0%) | 11 (100.0%) | 0 (0.0%) | |
| Extrahepatic spread, n (%) | <0.001 | |||||||||
| Absent | 116 (77.3%) | 37 (74.0%) | 19 (90.5%) | 20 (87.0%) | 5 (71.4%) | 18 (100.0%) | 10 (66.7%) | 4 (36.4%) | 3 (60.0%) | |
| Present | 27 (18.0%) | 12 (24.0%) | 0 (0.0%) | 3 (13.0%) | 1 (14.3%) | 0 (0.0%) | 5 (33.3%) | 5 (45.5%) | 1 (20.0%) | |
| Unknown | 7 (4.7%) | 1 (2.0%) | 2 (9.5%) | 0 (0.0%) | 1 (14.3%) | 0 (0.0%) | 0 (0.0%) | 2 (18.2%) | 1 (20.0%) | |
| Metastatic organs, n (%) | 0.006 | |||||||||
| Liver | 116 (77.3%) | 37 (74.0%) | 19 (90.5%) | 20 (87.0%) | 5 (71.4%) | 18 (100.0%) | 10 (66.7%) | 4 (36.4%) | 3 (60.0%) | |
| Bone | 10 (6.7%) | 3 (6.0%) | 0 (0.0%) | 1 (4.3%) | 0 (0.0%) | 0 (0.0%) | 2 (13.3%) | 4 (36.4%) | 0 (0.0%) | |
| Lung | 7 (4.7%) | 5 (10.0%) | 0 (0.0%) | 0 (0.0%) | 1 (14.3%) | 0 (0.0%) | 0 (0.0%) | 1 (9.1%) | 0 (0.0%) | |
| Peritoneum | 4 (2.7%) | 2 (4.0%) | 0 (0.0%) | 0 (0.0%) | 0 (0.0%) | 0 (0.0%) | 1 (6.7%) | 0 (0.0%) | 1 (20.0%) | |
| Lymph node | 4 (2.7%) | 1 (2.0%) | 0 (0.0%) | 1 (4.3%) | 0 (0.0%) | 0 (0.0%) | 2 (13.3%) | 0 (0.0%) | 0 (0.0%) | |
| Spleen | 1 (0.7%) | 1 (2.0%) | 0 (0.0%) | 0 (0.0%) | 0 (0.0%) | 0 (0.0%) | 0 (0.0%) | 0 (0.0%) | 0 (0.0%) | |
| Brain | 1 (0.7%) | 0 (0.0%) | 0 (0.0%) | 1 (4.3%) | 0 (0.0%) | 0 (0.0%) | 0 (0.0%) | 0 (0.0%) | 0 (0.0%) | |
| Unknown | 7 (4.7%) | 1 (2.0%) | 2 (9.5%) | 0 (0.0%) | 1 (14.3%) | 0 (0.0%) | 0 (0.0%) | 2 (18.2%) | 1 (20.0%) |
Notes: Treatment groups were assigned based on the initial treatment strategy after recurrence. Surgery-based therapy referred to repeat curative-intent surgery as the core treatment, either alone or in combination with other modalities. Locoregional therapy included TACE, radiofrequency ablation, and/or hepatic arterial infusion chemotherapy. Locoregional-TKI and Locoregional-TKI-ICI indicated locoregional therapy combined with tyrosine kinase inhibitors, without or with immune checkpoint inhibitors, respectively.
Abbreviations: TKI, tyrosine kinase inhibitor; ICI, immune checkpoint inhibitor; IQR, interquartile range; MVI, microvascular invasion; HBsAg, hepatitis B surface antigen; HBV, hepatitis B virus; HCV-Ab, hepatitis C virus antibody; mALBI, modified albumin-bilirubin grade; AFP, α-fetoprotein; TACE, transarterial chemoembolization; RT, radiotherapy; CNLC, China Liver Cancer staging system; BCLC, Barcelona Clinic Liver Cancer staging system; IU, international units.
Survival analysis showed significant differences among post-recurrence treatment groups in both PFS (p < 0.0001, Figure 2a) and OS (p < 0.0001, Figure 2b). For subsequent comparative analyses, we focused on the four principal treatment categories with sufficient sample size and relative clinical comparability: Locoregional-TKI-ICI, surgery-based therapy, Locoregional-TKI, and locoregional therapy alone. The radiotherapy-based group was not included in pairwise comparison because all 15 patients had portal vein tumor thrombus, indicating a distinct clinical context, whereas the lenvatinib-only, untreated, and unknown-treatment groups were too small for reliable estimation. Among the four major groups, patients receiving surgery-based therapy had the most favorable unadjusted survival outcomes. The median PFS and OS in the surgery-based therapy group were 26.4 months (95% CI: 12.6–43.2) and 46.3 months (95% CI: 25.8–NR), respectively. In the locoregional therapy alone group, the median PFS and OS were 3.6 months (95% CI: 2.8–NR) and 15.6 months (95% CI: 9.7–NR), respectively. In the unadjusted comparison, surgery-based therapy was associated with longer PFS and OS than locoregional therapy alone (PFS: HR = 0.35, 95% CI: 0.17–0.73, p = 0.004; OS: HR = 0.31, 95% CI: 0.12–0.79, p = 0.010). However, these findings were not based on a PSM comparison and should therefore be interpreted with caution.
Figure 2.
PFS and OS in all included patients. Kaplan–Meier curves for PFS (a) and OS (b), stratified by initial post-recurrence treatment strategy.
Abbreviations: PFS, progression-free survival; OS, overall survival; TKI, tyrosine kinase inhibitor; ICI, immune checkpoint inhibitor; RT, radiotherapy.
A substantial proportion of patients received systemic therapy in addition to locoregional treatment after recurrence. In the Locoregional-TKI group, the median PFS was 13.2 months (95% CI: 6.5–NR) and the median OS was 26.4 months (95% CI: 16.4–NR). In the Locoregional-TKI-ICI group, the median PFS was 12.1 months (95% CI: 9.3–19.3) and the median OS was 29.6 months (95% CI: 23.0–NR). Compared with locoregional therapy alone, the addition of targeted therapy or combined targeted-immunotherapy appeared to improve both PFS and OS. However, these differences did not reach statistical significance, which may be attributable to limited sample size, treatment heterogeneity, and potential confounding related to patient selection in this retrospective cohort.
To further explore whether specific post-recurrence treatment strategies might overcome limited sensitivity to postoperative adjuvant therapy, we analyzed the association between RFS after initial surgery and post-recurrence PFS according to subsequent treatment pattern (Figure 3). Overall, RFS was positively correlated with post-recurrence PFS (r = 0.200, p = 0.015, Figure 4), suggesting that patients with longer disease control after initial curative resection tended to have longer PFS after post-recurrence treatment.
Figure 3.
Swimmer plot showing recurrence-free survival after initial surgery and post-recurrence survival outcomes according to the initial post-recurrence treatment strategy. Patients were grouped by treatment regimen, and treatment group labels are shown adjacent to the corresponding patient clusters. The left side of the plot represents post-surgery recurrence-free survival, and the right side represents post-recurrence progression-free survival. Black circles indicate disease progression, red squares indicate death, and gray lines indicate post-progression survival.
Abbreviations: RFS, recurrence-free survival; PFS, progression-free survival; TKI, tyrosine kinase inhibitor; ICI, immune checkpoint inhibitor; RT, radiotherapy.
Figure 4.
Spearman correlation between recurrence-free survival and progression-free survival.
Subgroup Analysis Restricted to Patients with Potentially Resectable Recurrence
To further evaluate the potential influence of selection bias arising from differences in recurrence patterns, we performed a subgroup analysis restricted to patients with potentially resectable recurrence, defined as intrahepatic-only recurrence without portal vein invasion or extrahepatic spread. A total of 101 patients met these criteria, including 20 patients who received surgery-based therapy and 81 patients who received non-surgery-based therapy. Before propensity score matching, surgery-based therapy showed a favorable trend for post-recurrence PFS (HR = 0.63, 95% CI: 0.37–1.09, P = 0.101; Supplementary Figure S1a) and OS (HR = 0.47, 95% CI: 0.22–1.00, P = 0.050; Supplementary Figure S1b). After 1:2 propensity score matching with exact matching for the number of intrahepatic recurrent tumors, 15 patients in the surgery-based group and 23 patients in the non-surgery-based group were retained. Baseline characteristics before and after matching are shown in Supplementary Table S2. After matching, surgery-based therapy was not significantly associated with improved post-recurrence PFS (HR = 0.95, 95% CI: 0.48–1.90, P = 0.891; Supplementary Figure S1c) or OS (HR = 0.85, 95% CI: 0.34–2.09, P = 0.720; Supplementary Figure S1d). These findings suggest that the apparent survival advantage of surgery-based therapy observed before matching may be partly influenced by differences in recurrence patterns and patient selection.
Subgroup and Multivariable Analyses
Subgroup analyses were performed among the 132 patients who received post-recurrence treatment (Table 2). As expected, patients with portal vein invasion had shorter post-recurrence PFS and OS. Although patients with AFP >400 ng/mL, multiple intrahepatic recurrent tumors, or extrahepatic metastasis tended to have worse PFS and OS, these associations did not reach statistical significance, possibly due to the limited sample size of this retrospective study.
Table 2.
Subgroup Analyses of PFS and OS in Patients Receiving Post-Recurrence Treatment (n = 132), Stratified by Baseline Demographic and Disease Characteristics
| Subgroup | PFS | OS | ||||||
|---|---|---|---|---|---|---|---|---|
| Events/patients | Median (95% CI) | HR (95% CI) | P value | Events/patients | Median (95% CI) | HR (95% CI) | P value | |
| Age (years) | ||||||||
| <65 | 78/97 | 11.5 (9.1–15.0) | Reference | 43/97 | 29.1 (24.6-NR) | Reference | ||
| ≥65 | 31/35 | 8.8 (6.5–15.5) | 1.16 (0.76–1.76) | 0.4916 | 22/35 | 20.1 (15.6-NR) | 1.41 (0.84–2.35) | 0.1934 |
| Sex | ||||||||
| Male | 87/103 | 11.1 (9.0–15.0) | Reference | 47/103 | 29.6 (25.8-NR) | Reference | ||
| Female | 22/29 | 7.4 (4.1–15.5) | 1.56 (0.97–2.52) | 0.0691 | 18/29 | 12.8 (10.4–27.0) | 2.60 (1.48–4.57) | 0.0009 |
| HBsAg | ||||||||
| Negative | 26/30 | 9.0 (7.9–24.6) | Reference | 15/30 | 30.1 (25.8-NR) | Reference | ||
| Positive | 83/102 | 10.7 (8.1–13.3) | 0.99 (0.64–1.54) | 0.9715 | 50/102 | 25.0 (20.1–44.5) | 1.19 (0.67–2.13) | 0.5527 |
| HCV-Ab | ||||||||
| Negative | 101/123 | 10.4 (8.8–13.3) | Reference | 60/123 | 25.9 (20.3–44.5) | Reference | ||
| Positive | 8/9 | 6.5 (2.7-NR) | 0.96 (0.47–2.00) | 0.9222 | 5/9 | 32.0 (9.2-NR) | 0.81 (0.32–2.02) | 0.6480 |
| mALBI grade | ||||||||
| 1 | 77/94 | 11.5 (8.1–15.5) | Reference | 47/94 | 27.0 (23.0–45.7) | Reference | ||
| 2a | 25/28 | 8.2 (4.3–16.4) | 1.29 (0.82–2.04) | 0.2659 | 15/28 | 24.6 (11.1-NR) | 1.32 (0.74–2.37) | 0.3480 |
| 2b-3 | 7/10 | 10.8 (8.2-NR) | 0.88 (0.41–1.92) | 0.7507 | 3/10 | NR (12.8-NR) | 0.76 (0.24–2.44) | 0.6432 |
| Child-Pugh | ||||||||
| A | 105/128 | 10.7 (8.8–13.3) | Reference | 63/128 | 26.7 (23.0–44.5) | Reference | ||
| B | 4/4 | 8.1 (3.6-NR) | 1.25 (0.46–3.41) | 0.6603 | 2/4 | 17.5 (9.7-NR) | 0.96 (0.23–3.96) | 0.9559 |
| AFP (ng/mL) | ||||||||
| ≤400 | 71/89 | 12.1 (9.3–16.5) | Reference | 41/89 | 29.3 (25.0-NR) | Reference | ||
| >400 | 38/43 | 8.1 (4.1–13.0) | 1.46 (0.98–2.17) | 0.0605 | 24/43 | 20.1 (13.1-NR) | 1.35 (0.81–2.23) | 0.2472 |
| MVI | ||||||||
| M1 | 66/80 | 10.4 (7.4–13.3) | Reference | 39/80 | 26.4 (20.1-NR) | Reference | ||
| M2 | 43/52 | 12.1 (8.8–19.3) | 0.78 (0.53–1.16) | 0.2163 | 26/52 | 27.0 (20.3-NR) | 0.88 (0.53–1.45) | 0.6174 |
| Recurrence intrahepatic tumors | ||||||||
| None | 11/13 | 5.3 (3.2-NR) | Reference | 5/13 | 44.5 (10.3-NR) | Reference | ||
| 1 | 34/43 | 16.0 (12.6–26.9) | 0.63 (0.31–1.26) | 0.1934 | 18/43 | 45.7 (26.7-NR) | 0.66 (0.24–1.77) | 0.4043 |
| Multiple | 62/72 | 8.2 (6.5–11.5) | 1.00 (0.52–1.92) | 0.9881 | 40/72 | 20.3 (15.6–29.3) | 1.28 (0.50–3.26) | 0.6042 |
| Unknown | 2/4 | 2.8 (2.2-NR) | 0.88 (0.19–4.02) | 0.8684 | 2/4 | 16.4 (4.9-NR) | 1.81 (0.35–9.48) | 0.4801 |
| Portal vein invasion | ||||||||
| No | 95/117 | 11.5 (9.2–15.0) | Reference | 58/117 | 27.0 (24.1–45.7) | Reference | ||
| Yes | 14/15 | 4.4 (1.7–13.2) | 3.09 (1.73–5.51) | 0.0001 | 7/15 | 13.0 (6.3-NR) | 2.63 (1.19–5.83) | 0.0171 |
| Extrahepatic spread | ||||||||
| No | 89/107 | 12.1 (9.2–15.5) | Reference | 54/107 | 27.0 (24.1–45.7) | Reference | ||
| Yes | 18/21 | 5.1 (3.2–15.0) | 1.41 (0.84–2.36) | 0.1934 | 9/21 | 13.1 (10.3-NR) | 1.36 (0.67–2.77) | 0.3933 |
| Metastatic organs | ||||||||
| Liver | 89/107 | 12.1 (9.2–15.5) | Reference | 54/107 | 27.0 (24.1–45.7) | Reference | ||
| Other | 20/25 | 5.1 (3.2–15.0) | 1.36 (0.83–2.22) | 0.2244 | 11/25 | 14.6 (10.3-NR) | 1.43 (0.75–2.75) | 0.2798 |
| Treatment regimen | ||||||||
| Lenvatinib | 5/5 | 3.6 (3.0-NR) | Reference | 4/5 | 12.4 (7.8-NR) | Reference | ||
| RT-based therapy | 14/15 | 4.4 (1.7–13.2) | 0.82 (0.29–2.29) | 0.7050 | 7/15 | 13.0 (6.3-NR) | 0.95 (0.28–3.26) | 0.9339 |
| Locoregional therapy | 15/18 | 3.6 (2.8-NR) | 0.45 (0.16–1.27) | 0.1317 | 11/18 | 15.6 (9.7-NR) | 0.67 (0.21–2.10) | 0.4885 |
| Locoregional-TKI | 17/21 | 13.2 (6.5-NR) | 0.23 (0.08–0.64) | 0.0051 | 12/21 | 26.4 (16.4-NR) | 0.36 (0.11–1.13) | 0.0810 |
| Locoregional-TKI-ICI | 39/50 | 12.1 (9.3–19.3) | 0.25 (0.10–0.66) | 0.0048 | 21/50 | 29.6 (23.0-NR) | 0.33 (0.11–0.96) | 0.0420 |
| Surgery-based therapy | 19/23 | 26.4 (12.6–43.2) | 0.17 (0.06–0.47) | 0.0006 | 10/23 | 46.3 (25.8-NR) | 0.21 (0.06–0.67) | 0.0088 |
Abbreviations: PFS, progression-free survival; OS, overall survival; CI, confidence interval; HR, hazard ratio; HBsAg, hepatitis B surface antigen; HCV-Ab, hepatitis C virus antibody; mALBI, modified albumin-bilirubin; AFP, alpha-fetoprotein; MVI, microvascular invasion; RT, radiotherapy; TKI, tyrosine kinase inhibitor; ICI, immune checkpoint inhibitor; NR, not reached.
Multivariable Cox proportional hazards regression analyses were subsequently performed by incorporating variables that were significant in univariable analyses (Table 3), together with clinically relevant prognostic factors previously reported in the literature. After adjustment for AFP, MVI, extrahepatic metastasis, and portal vein invasion, Locoregional-TKI, Locoregional-TKI-ICI, and surgery-based therapy remained significant factors associated with PFS when compared with lenvatinib alone. For OS, no variable remained independently significant in the multivariable model, which may also be related to the relatively limited number of patients and events. In addition, OS may have been influenced by post-progression treatment heterogeneity, including subsequent-line therapies and crossover between treatment strategies, whereas PFS more directly reflected the effect of the initial post-recurrence treatment.
Table 3.
Multivariable Cox Regression Analyses for Post-Recurrence PFS and OS
| Variable | PFS | OS | ||
|---|---|---|---|---|
| HR (95% CI) | p | HR (95% CI) | p | |
| AFP | ||||
| ≤400 ng/mL | Reference | – | Reference | – |
| >400 ng/mL | 1.48 (0.97–2.26) | 0.0705 | 1.31 (0.75–2.29) | 0.3354 |
| MVI | ||||
| M 1 | Reference | – | Reference | – |
| M 2 | 0.83 (0.55–1.27) | 0.3886 | 0.99 (0.58–1.68) | 0.9576 |
| Extrahepatic spread | ||||
| Absent | Reference | – | Reference | – |
| Present | 1.41 (0.81–2.45) | 0.2308 | 1.51 (0.71–3.22) | 0.2895 |
| Portal vein invasion | ||||
| Absent | Reference | – | Reference | – |
| Present | 0.91 (0.30–2.77) | 0.8645 | 1.15 (0.29–4.48) | 0.8437 |
| Treatment regimen | ||||
| Lenvatinib | Reference | – | Reference | – |
| Locoregional therapy | 0.63 (0.20–1.96) | 0.4219 | 1.00 (0.27–3.76) | 0.9972 |
| Locoregional-TKI | 0.29 (0.09–0.89) | 0.0308 | 0.47 (0.13–1.73) | 0.2582 |
| Locoregional-TKI-ICI | 0.33 (0.11–0.94) | 0.0388 | 0.43 (0.13–1.46) | 0.1744 |
| Surgery-based therapy | 0.20 (0.07–0.62) | 0.0052 | 0.27 (0.07–1.01) | 0.0511 |
Abbreviations: PFS, progression-free survival; OS, overall survival; HR, hazard ratio; CI, confidence interval; AFP, α-fetoprotein; MVI, microvascular invasion; TKI, tyrosine kinase inhibitor; ICI, immune checkpoint inhibitor.
Discussion
Management of recurrent HCC after curative-intent resection remains highly heterogeneous in real-world practice. Although repeat resection is generally considered the preferred curative option for carefully selected patients with intrahepatic recurrence and preserved liver function, the optimal strategy after recurrence is still not fully defined, particularly in the era of expanding targeted and immune-based therapies. Recent reviews and guidelines consistently emphasize that treatment selection after recurrence should be individualized according to recurrence pattern, liver functional reserve, tumor burden, vascular invasion, and prior treatment exposure, rather than determined by tumor status alone.7,9
Several studies have shown that, among patients with limited recurrent disease, repeat hepatectomy and ablation can provide comparable long-term survival, particularly for solitary and small recurrent tumors. A randomized clinical trial by Xia et al reported no significant difference in overall survival between repeat hepatectomy and percutaneous radiofrequency ablation for early-stage recurrent HCC, although repeat hepatectomy provided better local disease control and appeared to confer greater benefit in patients with tumors larger than 3 cm or AFP levels above 200 ng/mL.10 Similarly, subsequent meta-analyses and retrospective comparative studies have suggested that the survival gap between repeat resection and ablation narrows considerably in highly selected “ideal candidates”, whereas the advantage of surgery becomes more apparent in patients with larger tumor burden or biologically more aggressive disease.11–13
Against this background, our study provides a real-world view of treatment allocation and outcomes in patients with MVI-positive HCC who developed recurrence after curative resection. Among the 150 recurrent cases, 132 patients underwent further treatment, indicating that most postoperative recurrences remained clinically actionable. Notably, post-recurrence management was diverse, with locoregional therapy plus TKI and ICI being the most common initial strategy, followed by surgery-based therapy, locoregional therapy plus TKI, and locoregional therapy alone. These findings reflect contemporary practice, in which repeated surgery is feasible only in a subset of patients, while many others are directed toward locoregional or combined systemic approaches because of impaired liver reserve, multifocal recurrence, unfavorable tumor location, or the presence of cirrhosis. This treatment heterogeneity is also in line with current recommendations highlighting multidisciplinary, risk-adapted decision-making for recurrent HCC.
A key finding of our study is that surgery-based therapy was associated with significantly longer post-recurrence PFS and OS than locoregional therapy alone. This result is biologically plausible and clinically consistent with prior literature.10 Surgery can achieve complete macroscopic tumor clearance and may provide superior local control, especially in patients with limited intrahepatic recurrence and sufficient remnant liver function. By contrast, locoregional therapy alone, although less invasive and more widely applicable, may be more vulnerable to incomplete tumor eradication and early intrahepatic re-progression, particularly in patients with underlying aggressive tumor biology. However, this finding should be interpreted cautiously because patients undergoing surgery-based therapy were highly selected and tended to have more favorable recurrence characteristics. Therefore, the observed survival advantage of surgery-based therapy may reflect, at least in part, differences in recurrence pattern and patient selection rather than treatment efficacy alone. Accordingly, our data further support the view that whenever recurrent disease remains technically resectable and liver function permits, surgery-based treatment should still be actively considered as a potential option for carefully selected patients, rather than as a universally superior post-recurrence strategy. Importantly, treatment allocation in this cohort was not random. Patients selected for surgery-based therapy generally had more favorable clinical conditions, including technically resectable disease, preserved liver function, and acceptable operative risk. These factors are themselves associated with better prognosis. Therefore, the favorable outcomes observed in the surgery-based group should be interpreted as associations in selected patients rather than definitive evidence of treatment superiority.
To further address this issue, we performed an additional subgroup analysis restricted to patients with potentially resectable recurrence, defined as intrahepatic-only recurrence without portal vein invasion or extrahepatic spread. In this subgroup, surgery-based therapy showed a favorable trend for post-recurrence PFS and OS before propensity score matching. However, after 1:2 propensity score matching with exact matching for the number of intrahepatic recurrent tumors, surgery-based therapy was no longer significantly associated with improved post-recurrence PFS or OS. These results suggest that recurrence-pattern-related selection bias may substantially contribute to the apparent survival benefit of surgery-based therapy. Thus, surgery-based treatment should be interpreted as a feasible strategy for selected patients with potentially resectable recurrence, but its comparative survival advantage requires further validation in prospective studies.
At the same time, our results also suggest that adding targeted therapy or targeted-immunotherapy to locoregional treatment may prolong both PFS and OS compared with locoregional therapy alone, even though these differences did not reach statistical significance. This trend is clinically meaningful. Increasing evidence suggests that locoregional treatment may synergize with systemic agents by debulking tumor burden, altering the tumor microenvironment, and potentially enhancing antitumor immune responses,14 while TKIs and ICIs may help suppress residual microscopic disease and delay further dissemination. Emerging studies of TACE combined with lenvatinib and PD-1 blockade have reported improved tumor response and longer PFS with manageable safety profiles in advanced or unresectable HCC,15,16 supporting the biological rationale for combination treatment.
Another notable finding is the positive correlation between postoperative RFS and post-recurrence PFS. This suggests that patients who derived greater benefit from their initial curative and adjuvant phase were also more likely to benefit from treatment administered after recurrence. One possible explanation is that longer RFS may reflect a less aggressive underlying tumor biology, including lower metastatic potential, slower clonal evolution, and reduced resistance to therapy. In contrast, early or rapidly recurring tumors are often considered manifestations of occult dissemination or biologically aggressive disease, which may also be less responsive to subsequent treatment. Previous studies have similarly shown that recurrence timing is strongly associated with prognosis after resection, with early recurrence generally indicating more aggressive disease behavior and poorer post-recurrence outcomes.17,18 From this perspective, RFS may partly reflect underlying tumor biology and may provide limited information regarding subsequent treatment responsiveness after recurrence. However, given the weak correlation observed in this study, this association should be interpreted cautiously.
Our subgroup analyses are also clinically informative. As expected, patients with portal vein invasion had worse post-recurrence PFS and OS. This is consistent with the established role of portal vein tumor thrombus as a marker of aggressive disease, impaired intrahepatic hemodynamics, and limited treatment tolerance.19,20 We also observed trends toward poorer outcomes among patients with AFP >400 ng/mL, multiple intrahepatic recurrent tumors, and extrahepatic disease, although these associations did not reach statistical significance. These findings remain directionally consistent with prior studies and likely reflect the limited statistical power of our sample rather than the absence of prognostic relevance.
An additional point worth noting is that no variable remained independently associated with OS in multivariable analysis, despite several factors being associated with PFS. This discrepancy is not unexpected in recurrent HCC. OS is influenced not only by the efficacy of first post-recurrence therapy, but also by subsequent-line treatments, crossover between regimens, supportive care, and changes in liver function over time. In our cohort, half of the patients who progressed received additional therapy, and ICI-based regimens became the dominant subsequent-line option. Such post-progression treatment heterogeneity may dilute the effect of the initial post-recurrence regimen on OS, making OS a less treatment-specific endpoint than PFS in this setting. Recent studies in advanced HCC have similarly suggested that the expansion of sequential systemic options can substantially affect survival independently of first-line treatment choice.21
This study has several limitations. First, this was a retrospective single-center study, and treatment allocation was inherently subject to selection bias and treatment allocation bias. Although we clarified the real-world decision-making process and performed a subgroup analysis with propensity score matching among patients with potentially resectable recurrence, residual confounding could not be fully eliminated. In particular, surgical candidacy depended on factors that were difficult to quantify retrospectively, including tumor location, technical resectability, future liver remnant, performance status, patient preference, and multidisciplinary judgement. Therefore, the observed favorable outcomes of surgery-based therapy in the overall cohort should be interpreted as associations in selected patients rather than causal treatment effects. Second, the sample sizes of several treatment subgroups were limited, which reduced the statistical power of comparative and subgroup analyses. The matched subgroup analysis also involved a relatively small number of patients, and therefore should be considered exploratory. Third, the treatment landscape evolved during the study period, particularly with the increasing use of targeted agents and immune checkpoint inhibitors, introducing temporal heterogeneity. Finally, post-progression management was nonuniform, which may have diluted the impact of initial post-recurrence treatment on OS. Despite these limitations, our study provides clinically relevant real-world evidence in patients with MVI-positive HCC after curative resection. The findings suggest that surgery-based therapy remains a feasible option for selected patients with resectable recurrence, while locoregional therapy combined with systemic treatment may represent a promising alternative for those who are not surgical candidates. Future multicenter prospective studies with larger sample sizes are needed to validate these findings and refine individualized treatment strategies for recurrent HCC.
Conclusions
In this real-world cohort of patients with MVI-positive HCC who developed recurrence after curative resection, post-recurrence treatment strategies were highly heterogeneous. Surgery-based therapy was associated with longer PFS and OS than locoregional therapy alone in the overall analysis. However, this association was attenuated after restricting the analysis to patients with potentially resectable recurrence and balancing recurrence-pattern-related differences by PSM. These findings suggest that the favorable outcomes observed in the surgery-based group were likely influenced by patient selection, and surgery-based therapy should be viewed as a potential option for carefully selected patients rather than a universally superior strategy. RFS was weakly but significantly correlated with post-recurrence PFS, suggesting that recurrence timing may partly reflect tumor biology and subsequent treatment sensitivity. For patients who are not candidates for repeat surgery, locoregional therapy combined with targeted therapy with or without immunotherapy may be considered a clinically relevant option, but its comparative effectiveness requires further validation in prospective studies.
Acknowledgments
Xiao Liu, Xiaokun Chen, Ziyue Huang, and Taifeng Zhu contributed equally to this study. We thank all the patients and the medical staff.
Funding Statement
This work was supported by the National Natural Science Foundation of China (81972698); the CAMS Innovation Fund for Medical Sciences (CIFMS 2021-I2M-01-014); the Central High-Level Hospital Clinical Research Special Key Cultivation Project (2025-PUMCH-C-028); and the Peking Union Medical Foundation.
Data Sharing Statement
All data supporting the results of this study are included in the article and its Supplementary Materials. Further inquiries can be directed to the corresponding author.
Statement of Ethics
This study was conducted in accordance with the Declaration of Helsinki and was approved by the Ethics Committee of Peking Union Medical College Hospital (Approval ID: I-26PJ0238). Given the retrospective nature of the study and the use of anonymized clinical data, the requirement for informed consent was waived.
Author Contributions
All authors made a significant contribution to the work reported, whether that is in the conception, study design, execution, acquisition of data, analysis and interpretation, or in all these areas; took part in drafting, revising or critically reviewing the article; gave final approval of the version to be published; have agreed on the journal to which the article has been submitted; and agree to be accountable for all aspects of the work.
Disclosure
The authors have no conflicts of interest to declare for this work.
References
- 1.Han B, Zheng R, Zeng H, et al. Cancer incidence and mortality in China, 2022. J Natl Cancer Cent. 2024;4(1):47–17. doi: 10.1016/j.jncc.2024.01.006 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Rumgay H, Arnold M, Ferlay J, et al. Global burden of primary liver cancer in 2020 and predictions to 2040. J Hepatol. 2022;77(6):1598–1606. doi: 10.1016/j.jhep.2022.08.021 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Li J, Liang Y-B, Chen X-M, et al. Prognostic value of lymphatic vessel density in the capsule of early-stage hepatocellular carcinoma: implications for postoperative recurrence risk. Front Immunol. 2026;17:1714314. doi: 10.3389/fimmu.2026.1714314 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Yopp A, Chen M, Cheng A-L, et al. Updated data from IMbrave050: adjuvant atezolizumab plus bevacizumab for high-risk hepatocellular carcinoma. J Hepatol. 2026;84(6):1102–1111. doi: 10.1016/j.jhep.2026.01.006 [DOI] [PubMed] [Google Scholar]
- 5.Li K, Zhang R, Wen F, et al. Single-cell dissection of the multicellular ecosystem and molecular features underlying microvascular invasion in HCC. Hepatology. 2024;79(6):1293–1309. doi: 10.1097/HEP.0000000000000673 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Wen T, Jin C, Facciorusso A, et al. Multidisciplinary management of recurrent and metastatic hepatocellular carcinoma after resection: an international expert consensus. Hepatobiliary Surg Nutr. 2018;7(5):353–371. doi: 10.21037/hbsn.2018.08.01 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Herrero A, Toubert C, Bedoya JU, et al. Management of hepatocellular carcinoma recurrence after liver surgery and thermal ablations: state of the art and future perspectives. Hepatobiliary Surg Nutr. 2024;13(1):71–88. doi: 10.21037/hbsn-22-579 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Aquina CT, Eskander MF, Pawlik TM. Liver-directed treatment options following liver tumor recurrence: a review of the literature. Front Oncol. 2022;12:832405. doi: 10.3389/fonc.2022.832405 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Singal AG, Llovet JM, Yarchoan M, et al. AASLD practice guidance on prevention, diagnosis, and treatment of hepatocellular carcinoma. Hepatology. 2023;78(6):1922–1965. doi: 10.1097/HEP.0000000000000466 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Xia Y, Li J, Liu G, et al. Long-term effects of repeat hepatectomy vs percutaneous radiofrequency ablation among patients with recurrent hepatocellular carcinoma: a randomized clinical trial. JAMA Oncol. 2020;6(2):255–263. doi: 10.1001/jamaoncol.2019.4477 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Yeow M, Zhao JJ, Fong KY, et al. Radiofrequency ablation versus repeat hepatectomy for recurrent hepatocellular carcinoma: a systematic review and meta-analysis. World J Surg. 2022;46(11):2778–2787. doi: 10.1007/s00268-022-06691-x [DOI] [PubMed] [Google Scholar]
- 12.Hsiao CY, Hu R-H, Liang P-C, et al. Comparison of long-term outcomes between repeated hepatic resection and radiofrequency ablation in patients with small recurrent hepatocellular carcinoma after initial curative resection: a propensity score matched study. J Hepatocell Carcinoma. 2025;12:1587–1598. doi: 10.2147/JHC.S496470 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Chua DW, Koh Y-X, Syn NL, et al. Repeat hepatectomy versus radiofrequency ablation in management of recurrent hepatocellular carcinoma: an average treatment effect analysis. Ann Surg Oncol. 2021;28(12):7731–7740. doi: 10.1245/s10434-021-09948-2 [DOI] [PubMed] [Google Scholar]
- 14.Chen QF, Chen S, Zhao M. The synergistic mechanisms and prospects of transarterial chemoembolization combined with immunotherapy for hepatocellular carcinoma. J Hepatocell Carcinoma. 2025;12:841–854. doi: 10.2147/JHC.S514881 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Wang WJ, Liu Z-H, Wang K, et al. Efficacy and safety of TACE combined with lenvatinib and PD −1 inhibitors for unresectable recurrent HCC: a multicenter, retrospective study. Cancer Med. 2023;12(10):11513–11524. doi: 10.1002/cam4.5880 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Chen X, Huang M, Liu R, et al. Clinical efficacy of immunotherapy in combination of locoregional therapies for advanced hepatocellular carcinoma: a systematic review and meta-analysis. Front Immunol. 2026;17:1706375. doi: 10.3389/fimmu.2026.1706375 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Chan AWH, Zhong J, Berhane S, et al. Development of pre and post-operative models to predict early recurrence of hepatocellular carcinoma after surgical resection. J Hepatol. 2018;69(6):1284–1293. doi: 10.1016/j.jhep.2018.08.027 [DOI] [PubMed] [Google Scholar]
- 18.Abdelhamed W, El-Kassas M. Hepatocellular carcinoma recurrence: predictors and management. Liver Res. 2023;7(4):321–332. doi: 10.1016/j.livres.2023.11.004 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Papaconstantinou D, Tsilimigras DI, Pawlik TM. Recurrent hepatocellular carcinoma: patterns, detection, staging and treatment. J Hepatocell Carcinoma. 2022;9:947–957. doi: 10.2147/JHC.S342266 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Wang YC, Lee J-C, Wu T-H, et al. Improving outcomes of liver resection for hepatocellular carcinoma associated with portal vein tumor thrombosis over the evolving eras of treatment. World J Surg Oncol. 2021;19(1):313. doi: 10.1186/s12957-021-02425-w [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Nakamura Y, Hirooka M, Hiraoka A, et al. Survival improvements in advanced hepatocellular carcinoma with sequential therapy by era. Cancers. 2023;15(21):5298. doi: 10.3390/cancers15215298 [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Data Availability Statement
All data supporting the results of this study are included in the article and its Supplementary Materials. Further inquiries can be directed to the corresponding author.