PDF
Citation
Print
Introduction
The systemic treatment of unresectable hepatocellular carcinoma (HCC) has evolved with an understanding of the molecular pathogenesis of the disease and the development of various therapeutic agents [1-6]. The recent emergence of immune checkpoint inhibitors (ICIs) has dramatically changed the treatment landscape for unresectable HCC and opened up the possibility of long-term survival in certain patients [7-10]. In particular, atezolizumab plus bevacizumab (Ate/Bev), a combination of an ICI with a monoclonal antibody against vascular endothelial growth factor, has demonstrated significantly improved survival outcomes compared to those of sorafenib [11]. The Food and Drug Administration in May 2020 approved the use of Ate/Bev for patients with unresectable HCC who have not received prior systemic therapy, making Ate/Bev the standard of care in the first-line setting [12].
Since the approval of Ate/Bev, numerous studies have been conducted to identify real-world data and optimal biomarkers or clinical characteristics for predicting the treatment response [13-18]. However, studies on patients who receive long-term Ate/Bev treatment are lacking. Therefore, we aimed to evaluate the clinical characteristics of patients treated with Ate/Bev for more than 1 year, and the clinical impact, such as liver function decline, associated with long-term Ate/Bev treatment.
Materials and Methods
1. Study population
This retrospective study included patients with unresectable HCC who received Ate/Bev treatment at three tertiary cancer centers (CHA Bundang Medical Center, Ulsan University Hospital, and Haeundae Paik Hospital) in Korea between May 2020 and April 2022. Patients who received Ate/Bev in later lines, had no baseline clinical data, and had no target lesion according to Response Evaluation Criteria in Solid Tumors (RECIST) ver. 1.1, Child-Pugh class C, or Barcelona Clinical Liver Cancer (BCLC) stage A were excluded. Patients were subsequently divided into two groups based on the treatment duration: the long-term treatment group (≥ 1 year) and the short-term treatment group (< 1 year). This study was conducted in accordance with the ethical guidelines of the Declaration of Helsinki and approved by the participating hospitals’ institutional review boards (CHA Bundang Medical Center, CHA-2020-12-030; Ulsan University Hospital, 2020-12-006; Haeundae Paik Hospital, 2020-12-019-001). The need for informed consent in this study was waived, as Korean regulations do not require consent for retrospective analyses. Data were anonymized and deidentified prior to analysis.
2. Treatment and assessment
Patients were treated with Ate/Bev following the IMbrave150 trial (1,200 mg atezolizumab plus 15 mg/kg bevacizumab intravenously every 3 weeks) [11]. Dose interruptions or reductions were made at the discretion of the attending physicians. Ate/Bev was continued until disease progression or occurrence of unacceptable toxicity. Tumor response was assessed every 6 or 9 weeks according to RECIST ver. 1.1. The objective response rate (ORR) was defined as the combined proportion of patients with a complete response (CR) or partial response (PR). Overall survival (OS) was defined as the time from the date of Ate/Bev treatment initiation to death. Progression-free survival (PFS) was defined as the time from the date of Ate/Bev treatment initiation to disease progression or death from any cause, whichever occurred first. Data from patients free of disease progression, death, or loss to follow-up were excluded at the last follow-up visit.
Blood samples were collected at every visit, and the Child-Pugh score and albumin-bilirubin (ALBI) grade were monitored. Treatment-related adverse events (AEs) were evaluated using the National Cancer Institute Common Terminology Criteria for Adverse Events (CTCAE) ver. 5.0. The intrahepatic tumor burden, defined as the volume of hepatic tumors occupying the remaining liver, was evaluated using dynamic contrast-enhanced computed tomography images obtained as a baseline study. Two board-certified radiologists, S.J. and C.A., each with over 10 years of experience in liver cancer imaging, conducted the assessment collaboratively and reached a consensus (S1 Fig.).
3. Statistical analysis
Categorical variables were analyzed using frequencies and percentages and compared using Fisher’s exact test or Pearson’s chi-square test, as appropriate. Continuous variables were analyzed using the median and range and compared using a Student’s t-test. Survival analysis was performed using the Kaplan-Meier method, and the log-rank test was used to compare the groups. Statistical analyses were performed using SPSS software ver. 22 (IBM Corp., Armonk, NY). Statistical significance was set at p < 0.05. GraphPad Prism 8.0 software (GraphPad Software Inc., San Diego, CA) was used to create the graphs.
Results
1. Patient characteristics
A total of 246 patients who received first-line Ate/Bev treatment between May 2020 and April 2022 were included in this study (S2 Fig.). The median age at diagnosis was 61 years (interquartile range, 54 to 68 years), and 209 patients (85.0%) were male. Hepatitis B virus infection (n=166, 67.5%)was the most common etiology of HCC, and 207 patients (84.1%) had liver cirrhosis. Baseline α-fetoprotein (AFP) levels at diagnosis were < 400 ng/mL in 150 patients (61.0%), and 241 patients had an Eastern Cooperative Oncology Group (ECOG) performance status of < 2. Portal vein tumor thrombosis (PVTT) was observed in 102 patients (41.5%). More than half of the patients had extrahepatic spread (n=155, 63.0%) and had previously received local therapy (n=165, 67.1%) for HCC. Regarding liver function, 128 patients (52.0%) had Child-Pugh class A5 and 126 patients (51.2%) had ALBI grade 1. Most patients had BCLC stage C (n=206, 83.7%). Additionally, 134 patients (54.5%) had an intrahepatic tumor burden of < 25% (Table 1).
Table 1.
Demographic characteristics of the patients at baseline
| Variable | Total (n=246) | Short-term treatment groupa) (n=177) | Long-term treatment groupb) (n=69) | p-valuec) |
|---|---|---|---|---|
| Age (yr) | 61 (54-68) | 62 (53-68) | 61 (55-68) | 0.810d) |
| Sex | ||||
| Male | 209 (85.0) | 152 (85.9) | 57 (82.6) | 0.520 |
| Female | 37 (15.0) | 25 (14.1) | 12 (17.4) | |
| Etiology of HCC | ||||
| Hepatitis B | 166 (67.5) | 115 (65.0) | 51 (73.9) | 0.130e) |
| Hepatitis C | 18 (7.3) | 17 (9.6) | 1 (1.5) | |
| Alcohol | 31 (12.6) | 22 (12.4) | 9 (13.0) | |
| Others | 31 (12.6) | 23 (13.0) | 8 (11.6) | |
| Cirrhosis | ||||
| Yes | 207 (84.1) | 151 (85.3) | 56 (81.2) | 0.423 |
| No | 39 (15.9) | 26 (14.7) | 13 (18.8) | |
| AFP (ng/mL) | ||||
| < 400 | 150 (61.0) | 103 (58.2) | 47 (68.1) | 0.152 |
| ≥ 400 | 96 (39.0) | 74 (41.8) | 22 (31.9) | |
| ECOG performance status | ||||
| 0 | 102 (41.5) | 58 (32.8) | 44 (63.8) | < 0.001e) |
| 1 | 139 (56.5) | 115 (65.0) | 24 (34.8) | |
| 2 | 5 (2.0) | 4 (2.3) | 1 (1.5) | |
| Portal vein tumor thrombosis | ||||
| No | 144 (58.5) | 95 (53.7) | 49 (71.0) | 0.013 |
| Yes | 102 (41.5) | 82 (46.3) | 20 (29.0) | |
| Extrahepatic spread | ||||
| No | 91 (37.0) | 62 (35.0) | 29 (42.0) | 0.467 |
| Yes | 155 (63.0) | 115 (65.0) | 40 (58.0) | |
| Prior local treatment | ||||
| No | 81 (32.9) | 63 (35.6) | 18 (26.1) | 0.154 |
| Yes | 165 (67.1) | 114 (64.4) | 51 (73.9) | |
| Child-Pugh classification | ||||
| A5 | 128 (52.0) | 82 (46.3) | 46 (66.7) | 0.006e) |
| A6 | 62 (25.2) | 47 (26.6) | 15 (21.7) | |
| B7 | 36 (14.6) | 33 (18.6) | 3 (4.4) | |
| B8-9 | 20 (8.1) | 15 (8.5) | 5 (7.3) | |
| ALBI grade | ||||
| 1 | 126 (51.2) | 79 (44.6) | 47 (68.1) | 0.002e) |
| 2 | 115 (46.7) | 95 (53.7) | 20 (29.0) | |
| 3 | 5 (2.0) | 3 (1.7) | 2 (2.9) | |
| BCLC stage | ||||
| B | 41 (16.7) | 26 (14.7) | 15 (21.7) | 0.183 |
| C | 502 (83.3) | 151 (85.3) | 54 (78.3) | |
| Intrahepatic tumor burden (%) | ||||
| < 25 | 134 (54.5) | 83 (46.9) | 51(73.9) | 0.002e) |
| 25-50 | 57 (23.2) | 47 (26.6) | 10 (14.5) | |
| 50-75 | 43 (17.5) | 36 (20.3) | 7 (10.1) | |
| > 75 | 12 (4.9) | 11 (6.2) | 1 (1.5) | |
| PIVKA-II (mAU/mL) | ||||
| < 200 | 120 (48.8) | 78 (44.1) | 42 (60.9) | 0.018 |
| ≥ 200 | 126 (51.2) | 99 (55.9) | 27 (39.1) | |
| NLR (n=244) | 2.78 (1.81-4.27) | 2.87 (1.98-4.43) | 2.12 (1.47-3.93) | 0.061d) |
| PLR (n=244) | 0.13 (0.09-0.20) | 0.14 (0.10-0.20) | 0.10 (0.08-0.17) | 0.323d) |
| CRP (mg/dL) (n=198) | 0.52 (0.16-1.66) | 0.81 (0.18-2.00) | 0.25 (0.09-0.99) | 0.172d) |
Values are presented as median (interquartile range) or number (%). AFP, α-fetoprotein; ALBI, albumin-bilirubin grade; BCLC, Barcelona Clinical Liver Cancer stage; CRP, C-reactive protein; ECOG, Eastern Cooperative Oncology Group; HCC, hepatocellular carcinoma; NLR, neutrophil-to-lymphocyte ratio; PIVKA-II, protein induced by vitamin K absence or antagonist-II; PLR, platelet-to-lymphocyte ratio.
2. Characteristics of the long-term treatment group
In the long-term treatment group, more patients had intrahepatic tumor burdens of < 25% compared to the short-term group (73.9% vs. 46.9%, p=0.002). PVTT occurred less frequently in the long-term treatment group (29.0% vs. 46.3%, p=0.013). In addition, more patients had an ECOG performance status of 0 (63.8% vs. 32.8%, p < 0.001) and were classified as Child-Pugh class A5 (66.7% vs. 46.3%, p=0.006) or ALBI grade 1 (68.1% vs. 44.6%, p=0.002) than did those in the short-term group. Other baseline characteristics, including known prognostic factors [19-22] such as C-reactive protein, neutrophil-to-lymphocyte ratio, platelet-to-lymphocyte ratio, or protein induced by vitamin K absence or antagonistII demonstrated no significant differences between the longand short-term groups (Table 1).
3. Survival outcomes
At the data cutoff of May 10, 2023, the median follow-up duration was 15.5 months (95% confidence interval [CI], 12.5 to 17). Overall, the median PFS and OS were 6.9 months (95% CI, 4.4 to 8.6) and 15.8 months (95% CI, 12.5 to 19.9), respectively. The ORR was 32.1% (79/246).
Considering the survival outcomes according to the treatment duration, the long-term treatment group demonstrated better PFS (25.0 months vs. 3.5 months, p < 0.001) and OS (not reached vs. 9.8 months, p < 0.001) than the short-term group (S3 Fig.). Moreover, the ORR was significantly higher in the long-term treatment group than in the short-term treatment group (75.4% vs. 15.3%, p < 0.001) (S4 Table).
4. Survival outcomes according to the intrahepatic tumor burden
To further evaluate the impact of the intrahepatic tumor burden on the survival outcomes, we evaluated PFS and OS according to the intrahepatic tumor burden. PFS was significantly different according to the intrahepatic tumor burden. The median PFS of patients with burdens of < 25%, 25%-50%, 50%-75%, and > 75% were 11.3, 4.0, 3.5, and 1.5 months, respectively (p < 0.001) (Fig. 1A). The OS demonstrated a similar pattern. The median OS of patients with a burden of < 25%, 25%-50%, 50%-75%, and > 75% was not reached at 11.1, 4.5, and 3.0 months, respectively (p < 0.001) (Fig. 1B).
Fig. 1.
Survival outcomes according to the intrahepatic tumor burden: PFS (A), OS (B), objec tive response rate (C). CI, confidence interval; CR, complete response; NR, not reached; OS, overall survival; PD, progressive disease; PFS, progression-free survival; PR, partial response; SD, stable disease.
Subsequently, the ORR based on the intrahepatic tumor burden was tested. ORR decreased as the intrahepatic tumor burden increased (p=0.002) (Fig. 1C). The ORRs of patients with tumor burdens of < 25%, 25%-50%, 50%-75%, and > 75% were 41.1%, 26.3%, 18.6%, and 8.3%, respectively. CR was achieved in 7.5% (10/134) and 1.8% (1/57) of patients with tumor burdens of < 25% and 25%-50%, respectively. However, no CR was observed in patients with a tumor burden of 50%-75% and > 75%.
5. Safety of long-term treatment with Ate/Bev
In the long-term treatment group, atezolizumab-related thyroid dysfunction (10.2% vs. 31.9%, p < 0.001), dermatological toxicity (14.7% vs. 29.0%, p=0.010), and arthritis (1.7% vs. 7.2%, p=0.041) occurred more frequently in the long- than in the short-term treatment group. Moreover, the incidence of bevacizumab-related AEs, such as hypertension (22.6% vs. 44.9%, p=0.001) and proteinuria (38.4% vs. 69.6%, p < 0.001), was higher in the long-term treatment group. Although no patient discontinued atezolizumab due to AEs, 34 patients (13.8%) discontinued bevacizumab due to AEs. The rate of bevacizumab discontinuation was higher in the long-term treatment group than in the short-term treatment group (8.5% vs. 27.5%, p < 0.001) (Table 2).
Table 2.
Treatment-related AEs
| Short-term treatment groupa) (n=177) | Long-term treatment groupb) (n=69) | p-valuec) | |
|---|---|---|---|
| Atezolizumab-related AEs (any grade) | |||
| Diabetes mellitus | 6 (3.4) | 3 (4.3) | 0.713 |
| Adrenal insufficiency | 4 (2.3) | 3 (4.3) | 0.404d) |
| Thyroid toxicity | 18 (10.2) | 22 (31.9) | < 0.001 |
| Dermatologic toxicity | 26 (14.7) | 20 (29.0) | 0.010 |
| Colitis | 14 (7.9) | 10 (14.5) | 0.118 |
| Fatigue | 45 (25.4) | 13 (18.8) | 0.275 |
| Liver toxicity | 60 (33.9) | 30 (43.5) | 0.161 |
| Pituitary toxicity | 2 (1.1) | 4 (5.8) | 0.054d) |
| Arthritis | 3 (1.7) | 5 (7.3) | 0.041d) |
| Pneumonitis | 1 (0.6) | 1 (1.5) | 0.483d) |
| Bevacizumab-related AEs (any grade) | |||
| Hypertension | 40 (22.6) | 31 (44.9) | 0.001 |
| Proteinuria | 68 (38.4) | 48 (69.6) | < 0.001 |
| Palmo-plantar erythrodysesthesia | 1 (0.6) | 0 | > 0.99d) |
| Bleeding | 14 (7.9) | 10 (14.5) | 0.118 |
| Thrombosis | 2 (1.1) | 0 | > 0.99d) |
| AEs leading to treatment discontinuation | |||
| Atezolizumab | 0 | 0 | |
| Bevacizumab | 15 (8.5) | 19 (27.5) | < 0.001 |
| Grade 5 Aese) | 7 (4.0) | 1 (1.5) | 0.448d) |
e) Grade 5 adverse events in the short-term group included gastrointestinal hemorrhage (in 3 patients), multi-organ dysfunction syndrome, intracranial hemorrhage, duodenal ulcer perforation, and mesenteric vein thrombosis (in 1 patient each); grade 5 events in the long-term treatment group included gastrointestinal hemorrhage (in 1 patient).
The median time to onset (mTTO) of atezolizumab-related thyroid dysfunction, dermatologic toxicity, and arthritis was 2.8 months (interquartile range [IQR], 1.8 to 4.3), 3.3 (IQR, 1.4 to 8.4), and 5.5 (IQR, 4.4 to 9.2), respectively. On the other hand, the mTTO of bevacizumab-related hypertension and proteinuria was 4.2 (IQR, 2.1 to 8.3) and 6.8 months (IQR, 3.7 to 10.5), respectively.
6. Changes in liver function in the long-term treatment group
In the long-term treatment group, changes in liver function were tracked using the Child-Pugh score and ALBI grade over a 1-year period. As treatment continued, a fraction of the patients experienced deteriorating liver function. Specifically, 87.0% of the patients were classified as Child-Pugh class A before Ate/Bev treatment; however, 75.4% remained in Child-Pugh class A after 1 year of treatment. While the proportion of patients with Child-Pugh class A5 showed no change from 60.9%, the proportion of patients with Child-Pugh class A6 notably decreased from 26.1% to 14.5% (Fig. 2A). Moreover, the proportion of patients classified as having ALBI grade 1 decreased from 68.1% to 50.7% after 1 year of treatment (p=0.025) (Fig. 2B).
7. Proportions of patients maintained on Ate/Bev treatment over 2 years
To track the proportion of patients who maintained Ate/Bev treatment over a 2-year period, we separately evaluated 182 patients who started Ate/Bev treatment by August 2021 (S2 Fig.). Of them, 125 patients (68.7%) maintained treatment for more than 3 months, 87 (47.8%) for more than 6 months, 59 (32.4%) for more than 12 months, 41 (22.5%) for more than 18 months, and 31 (17.0%) for 24 months or more (Fig. 3A). Meanwhile, when the same analysis was conducted with only responders who showed CR or PR, all 61 responders maintained Ate/Bev treatment for more than 3 months, 55 (90.2%) for more than 6 months, 44 (72.1%) for more than 12 months, 36 (59.0%) for more than 18 months, and 28 (45.9%) for 24 months or more (Fig. 3B).
The proportion of patients completing 24 months of treatment was 24.8%, 35.6%, 52.5%, and 75.6% for those who received 3, 6, 12, and 18 months of Ate/Bev treatment, respectively; these rates were elevated among responders at 45.9%, 50.9%, 63.6%, and 77.8%, respectively.
Regarding patients with PFS of < 3 months, they had a significantly larger proportion of elevated AFP level (≥ 400 ng/mL) (56.4% vs. 29.7%, p=0.001) and PVTT (50.0% vs. 31.6%, p=0.013) and BCLC stage C (95.6% vs. 76.1%, p < 0.001); and a smaller proportion of ECOG performance status 0 (22.1% vs. 55.3%, p < 0.001), Child-Pugh A5 (38.8% vs. 65.2%, p=0.001) at baseline than those with PFS of 3 or more months (S5 Table).
Discussion
With the advent of immunotherapy, long-term survival has become possible in patients with advanced HCC. However, few studies have reported on how many patients benefit from Ate/Bev in the long term; the characteristics of these patients remain unclear. Moreover, few studies exist on the clinical changes, such as liver function decline, that may occur during long-term exposure to Ate/Bev treatment. In this study, one-third of the patients were treated with Ate/Bev for > 1 year (i.e., the long-term treatment group). This group had a better performance status (ECOG 0), better liver function (Child-Pugh class A5 or ALBI grade 1), lower PVTT prevalence, and a lower intrahepatic tumor burden (< 25%) than the short-term group. More patients in the long-term treatment group experienced atezolizumab- and bevacizumab-related AEs. In addition, some patients in the long-term treatment group experienced liver function deterioration during Ate/Bev treatment.
The long-term treatment group was characterized by a low intrahepatic tumor burden (< 25%). Further analysis of survival outcomes according to the intrahepatic tumor burden of < 25%, 25%-50%, 50%-75%, and > 75% revealed that the PFS, OS, and ORR of Ate/Bev-treated patients were well stratified according to the intrahepatic tumor burden. A recent study [23] reported that Ate/Bev, unlike anti–programmed death-1 monotherapy [24], demonstrated favorable responses upon treatment of intrahepatic lesions, comparable to those of extrahepatic lesions. However, larger intrahepatic tumors (i.e., ≥ 5 cm) were associated with inferior ORR and OS compared to smaller intrahepatic tumors, highlighting the need to verify the therapeutic efficacy of Ate/Bev according to the intrahepatic tumor burden. Our finding supports that increased intrahepatic tumor burden may prognosticate poor survival outcomes with Ate/Bev treatment. For patients with very large intrahepatic tumors, exploring strategies that incorporate pretreatment to control the large intrahepatic mass before Ate/Bev treatment may be necessary to enhance treatment efficacy.
In terms of AEs, the long-term treatment group experienced both atezolizumab- and bevacizumab-related AEs. However, the time-to-onset of AEs caused by the two agents showed different patterns. Atezolizumab-related AEs tended to occur early in the treatment course, whereas bevacizumab-related AEs tended to occur later, which suggest that atezolizumab-related AEs are immunological events that occur relatively early presumably caused by the bystander effects of ICI-induced T-cell activation events following atezolizumab treatment, rather than a consequence of chronic exposure to atezolizumab treatment [17,25]. As atezolizumab-related AEs, such as thyroid and dermatologic toxicities, are generally manageable, early detection and management efforts are needed to achieve long-term benefits without compromising the quality of life. However, later occurrences of bevacizumab-related AEs, such as hypertension and proteinuria, imply the consequences of cumulative stress on the vascular system resulting from prolonged use. Kudo et al. [26] recently reported that the efficacy of Ate/Bev in HCC was comparable between patients whose bevacizumab regimen was interrupted due to AEs and those who did not demonstrate AEs; therefore, an optimal bevacizumab application strategy should be considered in future studies to balance the potential risks and benefits.
Regarding changes in the liver function in the long-term treatment group in our cohort, some patients exhibited deterioration of liver function (i.e., an increase in the Child-Pugh score and ALBI grade). Intriguingly, the patients whose liver function worsened with long-term Ate/Bev treatment mainly including individuals who had already started to deteriorate, with a baseline Child-Pugh score of A6 or higher. This finding suggests that even in Child-Pugh class A patients, it is necessary to monitor A6 patients separately from A5 patients. The STRIDE regimen (tremelimumab plus durvalumab) [27,28] had little effect on liver function deterioration during long-term treatment, suggesting that, unlike dual immunotherapy, long-term immunotherapy combined with an anti-angiogenic agent may affect liver function.
In a separate cohort analysis of patients who started Ate/Bev treatment 2 years ago, 17% remained in treatment for 2 years or more; however, one-third discontinued treatment in the first 3 months, suggesting an urgent need to identify optimal biomarkers that can predict therapeutic response to Ate/Bev to select patients who can achieve a long-term clinical benefit [13,14]. On the other hand, when the same analysis was performed with only responders, 46% of patients continued treatment for 2 years or more, and all patients continued treatment in the first 3 months, suggesting that the combination of immunotherapy with an improved response rate is needed for patients to maintain long-term treatment. In this context, because the recent phase 1b/2 trial demonstrated that the addition of an anti-TIGIT (T-cell immunoglobulin and ITIM domain) antibody, tiragolumab, to Ate/Bev, improved the ORR (11.1% vs. 42.5%) [29], the results of a consecutive phase 3 trial are awaited (ClinicalTrials.gov Identifier: NCT05904886).
Our study has several limitations. First, it was susceptible to unintended biases owing to its retrospective nature, such as selection bias. Second, because all of the patients were of Asian ethnicity and since the majority of patients had chronic hepatitis B virus infection, the general applicability of the results warrants further investigation in diverse populations. Third, owing to the small number of patients who reached 2 years of treatment, liver function assessment according to treatment duration in the long-term treatment group was limited to 1 year. In addition, while adopting the concept of intrahepatic tumor burden in prognostication was a meaningful approach, the underlying mechanism needs to be elucidated in future translational research and warrants further validation in larger, prospective cohorts. Subsequently, constructing a predictive model with expanded cohorts of those who were treated with Ate/Bev with validation would be invaluable. Despite these limitations, our study is the first to demonstrate patterns of liver function deterioration in patients treated with long-term Ate/Bev treatment.
In conclusion, a good performance stage, good liver function, lower PVTT, and lower intrahepatic tumor burden are associated with long-term treatment in patients with unresectable HCC treated with first-line Ate/Bev. A higher proportion of patients experienced atezolizumab-related adverse immunological AEs early in the course of treatment. In addition, a fraction of patients, especially those with Child-Pugh A6, experienced deterioration of liver function during long-term treatment.
XML Download