HCC Recurrence After Curative Intent Treatment: The Need for New High-Risk Criteria in the Context of Adjuvant Therapy

Abstract

Background and Aim: Adjuvant therapy after curative intent treatment for hepatocellular carcinoma (HCC) is a significant unmet need. The IMbrave050 study demonstrated improved recurrence-free survival (RFS) in patients with high-risk HCC receiving adjuvant atezolizumab and bevacizumab post-curative treatment compared to active surveillance. However, the IMbrave050 cohort was predominantly Asian, largely underwent surgical resection, and had chronic liver disease (CLD) mainly due to hepatitis B features that differ markedly from the Australian setting, where microwave ablation (MWA) is more common and hepatitis B-related CLD is less prevalent. Given these differences, this study aimed to explore the performance of the IMbrave050 risk criteria in an Australian population of patients with early-stage HCC undergoing curative treatment to determine if the criteria identified patients with a high risk of recurrence who may benefit from adjuvant treatment. Method: We performed a retrospective 5-year study of 50 patients with early-stage HCC undergoing MWA with curative intent or liver resection. Patients were stratified into high- and low-risk groups using the IMbrave050 criteria. The primary outcomes were RFS and overall survival (OS) in the high- and low-risk cohorts. Results: For patients who underwent liver resection, the 1-year RFS was 77.8% and 100% in high- and low-risk patients respectively (p = NS). In those who underwent MWA, the 1-year RFS was 89.5% in the high-risk cohort and 73.3% in the low-risk cohort (p = NS). OS at 1-year was 100% in all cohorts (p = NS). Conclusions: In this Western cohort receiving predominantly ablation as curative therapy the current high-risk criteria do not reliably distinguish between those with increased risk of early recurrence and those without. Criteria defining high-risk may need to be refined to better identify patients who may benefit from adjuvant therapy in this setting.

1. Introduction

Hepatocellular carcinoma (HCC) was the seventh most common cause of cancer-related death in Australia in 2021 and is the third leading cause of death from cancer worldwide [1]. The incidence of HCC is rising in Australia and has doubled from 1999 to 2020 [2]. The management of HCC is uniquely challenging as it is almost always diagnosed in patients with liver cirrhosis. The stage of disease, performance status and underlying liver function are incorporated in the Barcelona Clinic Liver Cancer (BCLC) staging system which guides treatment options for HCC [3]. Curative treatment is recommended for patients with early-stage disease who have preserved liver function and good performance status (BCLC 0/A) [3]. The curative treatment options include ablation, resection or liver transplant depending on the stage, presence of portal hypertension and number of tumours. HCC carries a poor prognosis with a 5-year survival of 24.2% and as such new therapeutic strategies are urgently needed [2].

Akin to other cancers, curative treatment for HCC is compromised by a risk of recurrent disease which can be high as 70%, highlighting the need for adjuvant therapy [4]. A number of agents have been investigated as adjuvant options in HCC. The STORM trial which reviewed adjuvant sorafenib was negative [5]. Adjuvant interferon did not improve recurrence-free survival (RFS) in patients with viral hepatitis [6,7]. Improved RFS and overall survival (OS), however, was observed in a Korean study evaluating the use of adjuvant activated cytokine-induced cells in patients with curative HCC, suggesting a possible role for immunotherapy in this setting [8].

The use of adjuvant immunotherapy has revolutionised modern cancer therapy and has led to improved outcomes in other solid organ malignancies compared to traditional chemotherapy. For example, the addition of neoadjuvant and adjuvant pembrolizumab compared to standard care improved OS in patients with triple-negative breast cancer [9]. Neoadjuvant and adjuvant durvalumab in addition to standard therapy led to significantly improved event-free survival and OS in patients with muscle-invasive bladder cancer [10]. The goal of treatment with immunotherapy in HCC is twofold, firstly to prevent early disease recurrence by eliminating any microscopic residual disease but also to prevent late recurrence of HCC through secondary prevention. Previously the use of immunotherapy for HCC has been limited to the treatment of advanced disease where the IMbrave150 trial established atezolizumab and bevacizumab as the standard of care for unresectable HCC, replacing sorafenib as the preferred first-line treatment [11].

The IMbrave050 study investigated the use of atezolizumab and bevacizumab in the adjuvant setting, demonstrating a significant improvement in RFS compared to active surveillance [12]. However, this benefit diminished over time, as the interim analysis revealed a convergence of the RFS curves [13]. Defined high-risk criteria for patients undergoing liver resection included poor tumour differentiation, tumours greater than 5 cm, four or more tumours, or vascular invasion (microvascular invasion or segmental portal vein invasion). For patients who underwent ablation, the high-risk criteria were tumours greater than 2 cm or multiple tumours [12].

Given the potential benefit of adjuvant treatment for high-risk patients undergoing curative treatment for HCC, our aim was to confirm that the IMbrave050 risk stratification criteria could be applied to an Australian population to accurately identify a cohort of patients who may benefit from adjuvant therapy.

2. Materials and Methods

The study design was a retrospective cohort study over a 5-year period between 2019 and 2024 in a tertiary hepatology centre in Queensland, Australia. Data was extracted from the statewide electronic medical record (Cerner Millennium, Kansas City, MO, USA). The inclusion criteria were persons aged 18 years or older, first diagnosis of HCC, and curative intent treatment with either microwave ablation (MWA) or liver resection. No patients were excluded that met these criteria. Patients were risk-stratified into high-risk and low-risk groups according to the IMbrave050 criteria. Biochemistry, Child–Pugh and Model of End-Stage Liver Disease (MELD-Na) scores, Eastern Cooperative Oncology Group (ECOG) status and BCLC stage were obtained at time of HCC diagnosis.

All patients underwent HCC screening and were discussed at a hepatology multidisciplinary meeting (MDT) with the interval of follow up determined by the findings on imaging. The follow up interval and choice of imaging modality for HCC surveillance was determined by the MDT group and included liver ultrasound, multiphase computed tomography (CT) of the liver, and multiphase magnetic imaging resonance (MRI). HCC was diagnosed radiologically using the CT/MRI Liver Imaging Reporting and Data System (LI-RADS) and MDT consensus opinion with histopathology confirming the diagnosis in some ablation patients and all resection patients [14]. The Edmondson–Steiner grading system was used for grading of liver biopsies [15]. A pairwise deletion approach was used for missing data and some patients had missing data for tumour differentiation, microvascular invasion, Child–Pugh and MELD-Na status which has been indicated in

Table 1. Patient demographics.

Sex	N (%)
	Male: 41 (82%) Female: 9 (17.3%)
Age (years)	Median	Interquartile range
	69	61–75
Aetiology of liver disease	N (%)
Hepatitis C virus-related cirrhosis	17 (34%)
Alcohol-related cirrhosis	16 (32%)
Metabolic dysfunction-associated steatotic liver disease	9 (18%)
Idiopathic	4 (8%)
Hepatitis B virus-related cirrhosis	3 (6%)
Haemochromatosis	1 (2%)
Model for End-Stage Liver Disease (MELD-Na) (n = 42)	Median	Interquartile range
	10	7–13
Childs Pugh score	N (%)
	A: 35 B: 6 C: 1 Not specified: 8
Barcelona Clinic Liver Cancer (BCLC)	N (%)
	BCLC 0: 13 (26%) BCLC A: 35 (70%) BCLC B: 2 (4%)
Eastern Cooperative Oncology Group (ECOG)	N (%)
	ECOG 0: 49 (98%) ECOG 1: 1 (2%)

,

Table 2. Patient demographics according to high- and low-risk groups.

Sex	N in High-Risk Group	N in Low-Risk Group
	Male: 22 Female: 8 Total: 30	Male: 19 Female: 1 Total: 20
Age (years)	Median in high-risk group	Median in low-risk group
	Median: 68 years (IQR 61–75 years)	Median: 71 years (IQR 59–75 years)
Aetiology of liver disease	N in high-risk group	N in low-risk group
Hepatitis C virus-related cirrhosis	11	6
Alcohol-related cirrhosis	7	9
Metabolic dysfunction-associated steatotic liver disease	8	1
Idiopathic	3	1
Hepatitis B virus-related cirrhosis	1	2
Haemochromatosis	0	1
	Total: 30	Total: 20
Model for End-Stage Liver Disease (MELD-Na) (n = 42)	Median in high-risk group	Median in low-risk group
	Median: 10 (IQR 7–12)	Median: 10 (IQR 8–13)
Childs Pugh score	N in high-risk group	N in low-risk group
	A: 22 B: 3 C: 0 Not specified: 5	A: 13 B: 3 C: 1 Not specified: 3
Barcelona Clinic Liver Cancer (BCLC)	N in high-risk group	N in low-risk group
	BCLC 0: 2 BCLC A: 26 BCLC B: 2 Total: 30	BCLC 0: 11 BCLC A: 9 BCLC B: 0 Total: 20
Eastern Cooperative Oncology Group (ECOG)	N in high-risk group	N in low-risk group
	ECOG 0: 30 ECOG 1: 0 Total: 30	ECOG 0: 19 ECOG 1: 1 Total: 20

,

Table 3. Tumour characteristics.

Tumour Grade (Edmondson–Steiner Grading System)	N (%)
	Well differentiated: 7 (21.9%) Moderately differentiated: 16 (50%) Poorly differentiated: 6 (18.8%) Biopsy non-diagnostic for HCC: 2 (6.3%) Not specified: 1 (3.1%) Total: 32
Number of intrahepatic tumours at diagnosis	N (%)
	1: 43 (86%) 2: 6 (12%) 3: 1 (2%)
Presence of any intrahepatic tumours > 5 cm	N (%)
	Yes: 6 (12%) No: 44 (88%)
Presence of microvascular invasion (in patients who underwent resection)	N (%)
	Yes: 6 No: 5
Diameter of the largest intrahepatic tumour at diagnosis (cm)	Median 2.3	Interquartile range 1.8–2.9
Alpha-fetoprotein (AFP) (ng/mL)	Median 5	Interquartile range 3.2–9.4

and

Table 4. Tumour characteristics according to curative treatment.

Number of Intrahepatic Tumours at Diagnosis	Ablation	Resection	Total
1	34	9	43
2	5	1	6
3	0	1	1
Total	39	11	50
Tumour grade (Edmondson–Steiner grading system)	Ablation	Resection	Total
Well differentiated	5	2	7
Moderately differentiated	101	6	16
Poorly differentiated	3	3	6
Biopsy not diagnostic of HCC	2	0	2
Not specified	0	1	1
Total	21	11	32
Diameter of the largest intrahepatic tumour at diagnosis (cm)
Ablation (n = 39)	Median: 2.2 cm (IQR 1.7–2.5 cm)
Resection (n = 11)	Median: 5.2 cm (IQR 2.7–7 cm)
Alpha-fetoprotein (AFP) (ng/mL)
Ablation (n = 39)	Median: 4.6 ng/mL (IQR 1–12 ng/mL)
Resection (n = 11)	Median: 5 ng/mL (IQR 3.4–8.9 ng/mL)

.

The primary outcomes were RFS and OS in the high- and low-risk cohorts and outcomes were compared to explore whether the IMbrave050 risk stratification appropriately selects patients who may benefit from adjuvant immunotherapy. The Kaplan–Meier method was used for survival analysis, and the log rank test was used to compare the cumulative probability of disease recurrence and death in high- and low-risk groups. A p value of <0.05 was considered significant. Statistical analysis was done using SPSS version 15.0 (SPSS, Chicago, IL, USA). This retrospective observational study was reviewed and approved by the Townsville ethical review board (HREC/2024/QTHS/113109), and the requirement for informed consent was waived.

3. Results

Fifty patients met the inclusion criteria. The median age was 69 (IQR 61–75). 82% of patients (n = 41) were male. The most common cause of underlying chronic liver disease (CLD) was Hepatitis C (34%), followed by alcohol-related liver disease (ARLD) (32%) and metabolic dysfunction-associated steatotic liver disease (MASLD) (18%) (Table 1 and Table 2). The majority of patients had a single lesion (86%) and had Barcelona clinic liver cancer (BCLC) grade 0/A HCC at time of diagnosis. Baseline tumour characteristics including the grade, number, size, presence of microvascular invasion and baseline alpha-fetoprotein (AFP) are detailed in Table 3 and Table 4. The initial treatment was MWA in 78% (n = 39) of patients. Two patients who had a liver resection later received MWA (with one patient receiving intra-operative MWA and later MWA of a different lesion). Eight patients received transarterial chemoembolisation (TACE). Of these patients, two underwent initial TACE as a downstaging procedure to facilitate curative treatment. Three were bridging to liver transplant and three had TACE after disease progression.

There were no procedure-related deaths. Four patients experienced complications related to the curative procedure which included pneumothorax (n = 2), surgical site infection (n = 1) and perihepatic collection (n = 1).

The majority of patients fulfilled high-risk criteria as defined by IMbrave050 with 53.8% undergoing MWA (n = 21) and 81.8% (n = 9) undergoing liver resection. One-year RFS in patients who underwent liver resection was 77.8% in high-risk and 100% in low-risk patients (p = NS). Median follow up time was 35 months (95% CI 25–45 months) in the high-risk resection group and 34 months (95% CI NE) in the low-risk resection group. In patients who underwent MWA, 1-year RFS was 89.5% in the high-risk cohort and 73.3% in the low-risk cohort (p = NS) (Figure 1). Median follow up time was 30 months (95% CI 25–35 months) in the high-risk ablation group and 26 months (95% CI 11–41 months) in the low-risk ablation group. One-year RFS for the entire cohort was 82.2%. OS at 1-year was 100% in all cohorts. There was no significant difference in RFS when accounting for gender, age, aetiology or MELD-Na score (Figure 2, Figure 3, Figure 4, Figure 5 and Figure 6).

4. Discussion

The major finding of this study is that accepted criteria to define patients with HCC who may benefit from adjuvant therapy do not appear to apply to Australian patients. We found no significant difference in 1-year RFS or OS in our ablation cohort after stratification into high- and low-risk groups per the IMbrave050 protocol. The high-risk group in fact had improved RFS compared to the low-risk group although this did not reach statistical significance. These results are likely due to key epidemiological differences in Australian patients diagnosed with HCC compared to the IMbrave050 cohort. Liver resection was the predominant curative treatment in the IMbrave050 cohort, and the majority of patients were of Asian background with CLD secondary to Hepatitis B infection. This differs from the Australian context, where the majority of patients undergo MWA and the prevalence of Hepatitis B-associated CLD is much lower. The proportion of patients undergoing MWA as the curative treatment was 78% in our cohort compared to only 12.4% in the IMbrave050 cohort. The most common causes of CLD in our cohort were Hepatitis C, ARLD and MASLD which are typical for a Western population. There is some evidence that patients with non-viral HCC, particularly MASLD, may not respond as well to immune therapy compared to viral-related HCC. A meta-analysis of 1656 patients with advanced HCC undergoing PDL1 immunotherapy found that patients with non-viral HCC had significantly worse overall survival compared to patients with viral HCC [16]. Additionally, patients with MASLD were found to have significantly worse overall survival compared to patients without [16].

Early and late recurrence of HCC are mechanistically distinct. Early recurrence typically refers to disease recurrence occurring prior to 2 years which is secondary to metastases and is thought to be influenced by tumour characteristics such as size and number of lesions whereas late recurrence occurs after 2 years and is attributed to de novo HCC [17]. Ablation is associated with a higher risk of recurrence compared to resection, noting that patients undergoing ablation tend to be older and have more advanced liver disease which are both predictors of recurrence in their own right [18]. Despite our population being older and predominantly undergoing ablation, OS was excellent with all patients in the study alive at 1-year and a 1-year RFS of 89.5% and 73.3% in the high- and low-risk ablation groups respectively (p = NS). RFS was numerically improved in the high-risk ablation group compared to the low-risk ablation group across the study period though this was not statistically significant (HR 1.38, 95% CI, 0.56–3.42, p = NS). In the IMbrave050 study, the ablation cohort was much smaller than the resection group (n = 83) and there was no significant difference in RFS in the subgroup analysis of these ablation patients (HR 0.61, 95% CI 0.26–1.41) [12].

Adverse effects of atezolizumab and bevacizumab are not insignificant. Atezolizumab like other immunotherapeutic agents can lead to a wide array of dermatological, gastrointestinal, endocrinological and other immunogenic effects [19]. Adverse effects of bevacizumab include hypertension and bleeding which underline the importance of variceal screening and prophylaxis prior to the commencement of therapy [19]. Treatment was required to be discontinued in 9% of patients in the IMbrave050 trial and almost half of all patients (41%) experienced a Grade 3 or 4 event [12]. Given the lack of evidence for adjuvant therapy in the ablation cohort and the risk of significant complications, it is unclear whether the benefits of adjuvant therapy outweigh the risks in this population.

In addition to atezolizumab and bevacizumab, there are a number of other potential adjuvant therapy options. Lee et al. (2015) administered adjuvant activated cytokine-induced cells and found significantly improved OS and RFS [8]. The patient cohort were Koreans with predominantly Hepatitis B-related HCC, and most underwent ablation as curative treatment. At 5 years follow up, the treatment arm continued to demonstrate improved RFS (HR 0.67, 95% CI 0.48–0.94) [20]. A more recent phase 2 trial looking at the use of sintilimab (a PD-1 inhibitor) in patients with high-risk HCC (defined as the presence of microvascular invasion (MVI)) demonstrated promising results with improved RFS compared to active surveillance [21].

Updated data from the IMbrave050 study published in 2025 revealed the improved RFS observed in the treatment arm was not sustained over time (median follow up 35.1 months, HR 0.90 (95% CI 0.72–1.12) [13]. Analysis of the OS end point is still immature. It is possible that the lack of sustained efficacy may be due to incomplete eradication of micrometastases with insignificant RFS after the 1-year time period. However, there has been some criticism of the risk stratification criteria utilised by IMbrave050 [22,23] and the use of different criteria may better select patients who would benefit from adjuvant immunotherapy. Tada et al. (2024) proposed the AD-ON score to assist in determining whether patients at high-risk of recurrence undergoing ablation should receive adjuvant immunotherapy [23]. Hepatitis C, alpha-fetoprotein (AFP)-L3 and des-gamma-carboxy prothrombin (DCP) were considered as predictors of recurrence and incorporated into the score [23]. The AFP-L3 isoform is produced by malignant liver cells, demonstrates good specificity for detecting HCC and is useful particularly for small tumours with a study by Li et al. demonstrating detection rates of approximately 35% in patients with small HCC [24]. DCP is a non-functional thrombin precursor produced by HCC cells [25]. Norman et al. demonstrated that the combination of AFP-L3 and DCP predicted 61% of HCC recurrences in patients who had undergone liver transplant [25]. The presence of MVI on histopathology is another independent negative prognostic marker of HCC recurrence [26].

Our exploratory findings raise several questions about the applicability of the IMbrave050 criteria to Australian and other Western populations. There are key differences in patient demographics which leads to variation in the types of curative treatment offered. Ablation is the most common treatment in our centre, and our patient cohort is older and frequently have co-morbid medical conditions that may limit their fitness to undergo surgery. In Australia there is universal access to specialist healthcare although despite this HCC screening rates are still well below target. MASLD is now the most common cause of CLD in many Western countries including Australia and there is some limited evidence suggesting that these patients do not respond as well to immunotherapy for HCC compared to patients with CLD secondary to viral hepatitis.

Our study is limited by its observational design and small sample size. There was no statistically significant difference in RFS between the high- and low-risk groups however the interpretation of these results is impacted by the small sample size. Treatment bias is a consideration with older patients more likely to undergo ablation. All cases were discussed at a hepatology MDT with input from hepatobiliary surgeons and consideration of resection in suitable patients. To our knowledge this is the first study to examine if criteria to select patients for adjuvant therapy are applicable in the Australian setting and further research into this area is needed with larger cohorts of patients to enable robust analysis of subgroups to explore our findings in more detail.

5. Conclusions

The role of adjuvant therapy for patients undergoing ablation or resection is unclear with limited evidence in this subgroup as well as in patients with non-viral HCC. The overrepresentation of Hepatitis B and resection in the IMbrave050 cohort affects the generalisability of findings to the Australian population. Further research is required to explore the practical application of the IMbrave050 model and to consider whether a new high-risk definition is required to identify which patients could benefit from adjuvant therapy.