Five years after IMbrave150, atezolizumab plus bevacizumab (ate/bev) remains the first-line standard for unresectable hepatocellular carcinoma (uHCC). High response rates allow downstaging and curative conversion, enabling drug-free remission in some patients. Durable responses suggest potential for de-escalation, though discontinuation criteria need validation. In hepatitis B virus–related HCC, ate/bev appears safe across baseline viral loads when administered with concomitant antiviral therapy. Bleeding risk, particularly esophagogastric variceal hemorrhage, warrants baseline endoscopy and proactive management. Hypertension and proteinuria require close surveillance and early control to maintain bevacizumab use. Late bevacizumab interruption is acceptable, whereas early discontinuation is associated with poorer outcomes. After progression on ate/bev, evidence across cohorts favors tyrosine kinase inhibitor strategies such as lenvatinib over sorafenib. Cabozantinib and ramucirumab (for alpha-fetoprotein ≥400 ng/mL) are options and selected patients may benefit from immune checkpoint inhibitor–based combinations. Combining ate/bev with locoregional therapy (LRT) is promising. Phase III data with transarterial chemoembolization show improved progression-related endpoints. Early series suggest potential benefit with Yttrium-90 radioembolization, hepatic arterial infusion chemotherapy and radiotherapy, although high-level evidence is still lacking. Translational markers such as pre-existing immunity and transcriptomic models are promising but unvalidated. Adjuvant ate/bev has not demonstrated durable benefit. The neoadjuvant role and additional combination strategies require further exploration. Although ate/bev has improved outcomes in uHCC, key gaps remain, including biomarkers for regimen selection, de-escalation criteria, optimal sequencing, and integration with LRT or perioperative care. Prospective studies are needed to guide practice.
Hepatocellular carcinoma (HCC), the most common primary liver cancer, remains a major global health challenge, with the incidence expected to exceed one million cases by 2025. While hepatitis B and C remain the predominant etiologies, nonalcoholic steatohepatitis (NASH), linked to metabolic syndrome and diabetes, is increasingly shaping the epidemiology, particularly in Western countries. NASH-related HCC exhibits distinct molecular features, and although approximately one-quarter of HCCs harbor potentially actionable mutations, these findings have yet to influence routine clinical practice [1].
In 2020, the IMbrave150 trial established atezolizumab plus bevacizumab (ate/bev) as the first immunotherapy-based regimen to demonstrate superiority over sorafenib, marking a paradigm shift in the management of unresectable HCC (uHCC) [2]. Its significant survival benefit and higher response rates have reshaped treatment expectations. Systemic therapy, once limited to Barcelona Clinic Liver Cancer (BCLC) stage C or transarterial chemoembolization (TACE)-refractory BCLC stage B, is now considered within a redefined framework. The 2022 BCLC update underscores the heterogeneity of intermediate-stage disease (BCLC-B) and advocates individualized strategies—such as extended liver transplantation criteria, selective TACE, or systemic therapy—based on tumor burden, liver function, and biomarkers like alpha-fetoprotein (AFP) [3].
The higher response rate with ate/bev enables downstaging and durable responses, creating opportunities for integration with other treatment modalities. This highlights the central role of multidisciplinary team (MDT) management, with treatment strategies tailored to tumor stage, liver function, and patient factors. Recent guidance further emphasizes minimizing bias in MDT discussions through clear role definitions and structured feedback to ensure evidence-based, patient-centered care [4].
Five years after IMbrave150, accumulating clinical experience underscores not only the efficacy and safety of ate/bev but also its pivotal role in advancing individualized, evidence-based, patient-centered care in HCC. This review highlights the clinical application of ate/bev, focusing on therapeutic outcomes, adverse events (AEs) management, and evolving strategies for optimal integration into practice. The current evidence for the clinical integration of ate/bev in practice is summarized in an algorithm (Fig. 1).
Clinical flow chart for atezolizumab plus bevacizumab in HCC.
The algorithm summarizes pragmatic decision steps across stages.
This schematic reflects current evidence and expert consensus and should be applied within MDT decision-making and institutional standards.
*High-risk varices/ulcer refers to untreated or incompletely treated varices with active bleeding or at high risk of bleeding, as well as active ulcers.
Abbreviations: ABC, atezolizumab–bevacizumab–curative conversion; AFP/AFP-L3/DCP, tumor markers; CEUS, contrast-enhanced ultrasound; CR/PR/SD/PD, complete/partial/stable/progressive disease; EBRT, external-beam radiotherapy; EVL, endoscopic variceal ligation; HAIC, hepatic arterial infusion chemotherapy; HBV, hepatitis B virus; HTN, hypertension; ICI/IO, immune checkpoint inhibitor/immuno-oncology; LRT, locoregional therapy; LT, liver transplantation; MAFLD, metabolic dysfunction–associated fatty liver disease; MDT, multidisciplinary team; PPI, proton-pump inhibitor; RECIST/mRECIST, response criteria; RT, radiotherapy; TACE, transarterial chemoembolization; TKI, tyrosine-kinase inhibitor; Y-90, yttrium-90 radioembolization.
Atezolizumab is a monoclonal antibody targeting programmed death ligand 1 (PD-L1), which blocks its interaction with programmed death-1 (PD-1) and B7–1, thereby restoring cytotoxic T-cell activity and reinvigorating antitumor immune responses [5]. Bevacizumab, an antibody against vascular endothelial growth factor A (VEGF-A), was initially developed as an anti-angiogenic therapy but is now recognized to exert profound immunomodulatory effects [6]. VEGF-A contributes to tumor immune evasion by impairing dendritic cell maturation, expanding regulatory T cells and myeloid-derived suppressor cells, and promoting T-cell exhaustion while simultaneously fostering abnormal tumor vasculature [7,8]. Inhibition of VEGF-A not only normalizes the tumor vasculature to enhance immune cell infiltration but also alleviates these immunosuppressive mechanisms, shifting the tumor microenvironment toward a more immune-permissive state [9]. The combination of atezolizumab and bevacizumab therefore acts synergistically, coupling direct activation of antitumor immunity with reversal of VEGF-mediated immunosuppression, and represents a rational therapeutic strategy in HCC.
3Ate/bev as first-line systemic therapy for unresectable HCCIMbrave150 established ate/bev as the first regimen to exceed sorafenib in uHCC, including patients with Vp4, reducing the risk of death [hazard ratio (HR) 0.58] and improving 12-month overall survival (OS; 67.2% vs. 54.6%); progression-free survival (PFS) was also prolonged (6.8 vs. 4.3 months; HR 0.59) with a manageable safety profile [2].
Real-world evidence corroborates these outcomes: in a 332-patient multicenter cohort, 1-year OS was 52% and 1-year PFS 36%; Child–Pugh B, albumin–bilirubin (ALBI) grade >1, and AFP ≥400 ng/mL were adverse factors [10], and another real-world study also echoed the findings of IMbrave150 [11]. A meta-analysis of 2179 patients (12 cohorts) showed a pooled 12-month OS of 65%, median OS in Child–Pugh A comparable to IMbrave150 (20.9 vs 19.2 months), and longer PFS in practice (11.8 vs 6.9 months) [12]. Studies of ate/bev in the first-line setting are summarized in Table 1.
Key clinical outcomes of first-line atezolizumab plus bevacizumab.
| Study (year) | Design (numbers) | Populations | CR/ORR/DCR (%) | PFS/OS (months) | Key findings | ≥ grade 3 AEs | Ref. |
|---|---|---|---|---|---|---|---|
| IMbrave150 (2020) | RCT(n = 336) | CP-ABCLC B/CECOG 0–1 | 5.5/27.3/73.6 | 6.8/19.2 | Ate/bev significantly improved OS (HR 0.58) and PFS (HR 0.59) over sorafenib; established new first-line standard. | TRAEs ∼23%; HTN 15%; bleeding 6%; discontinuation due to AEs 15.5% | [2] |
| Shao et al. (2022) | RCTPost-hoc(n = 40) | IMbrave150/GO30140 Taiwanese subgroup;CP-A(B7 allowed in GO30140) | 7.5/37.5/85 | 8.6/24.9 | Efficacy/safety consistent with global cohorts despite more advanced disease; skipping bevacizumab common yet no clear PFS/OS harm in ≥6 m subgroup. | TRAEs ∼50%; HTN 27.5%; ALT elevation and proteinuria 10%; discontinuation due to bevacizumab 10%; discontinuation due to atezolizumab 2.5% | [53] |
| Cheon et al. (2022) | RWE(n = 138) | CP-ABCLC B/C | 1.7/24/76 | 6.5/NR | Real-world efficacy and safety of ate/bev in Korean patients with advanced HCC; baseline neutrophil-to-lymphocyte ratio (NLR) ≥5 predicted worse PFS and OS in advanced HCC patients receiving ate/bev. | AEs 28.9%; AST elevation 10.7%; HTN 6.6%; thrombocytopenia 4.9%; discontinuation due to AEs 10.7% | [34] |
| Vithayathil et al. (2022) | RWE(n = 191) | CP-A/B | 0/24.5/73 | ≥65y: 7.1/14.9; <65y: 5.5/15.1 | Comparable efficacy/tolerability in ≥65 vs <65; CP-B associated with reduced OS/PFS. | AEs 20.4%; HTN 23%; proteinuria 19.9%; fatigue 16.2%; discontinuation 6.3% | [33] |
| Fulgenzi et al. (2022) | RWE(n = 296) | CP-ABCLC B/CECOG 0–1 | 2.9/30.8/77.7 | 6.9/15.7 | Confirms reproducible safety/efficacy in routine practice; ALBI grade and PVTT independently tied to OS; prior locoregional treatment associated with higher ORR. | AEs 23.6%; hepatotoxicity and proteinuria 5.4%; HTN and bleeding 3.7%; discontinuation due to AEs 8.4% | [35] |
| Fukushima et al. (2023) | RWE(n = 150) | CP-ABCLC B/CECOG 0–1 | irAEs group: 3.1/37.5/84.4non-irAEs:0/24.6/73.7 | irAEs group: 9/NR;non-irAEs group: 6/15.2 | Grade 1/2 irAEs linked to longer PFS/OS | irAEs 6.0%; irAEs group: permanently discontinued due to irAEs 40.6% | [57] |
| Song et al. (2024) | RWE(n = 111) | CP-ABCLC B/CECOG 0–1 | 4.5/27/63 | 6.5/NR | Real-world outcomes align with trials; peritoneal seeding associated with higher CR/ORR; radiotherapy combination improved responses | NA | [11] |
| Suzuki et al. (2024) | RWE(n = 130) | CP-ABCLC B/CECOG ≤2 | 3.1/20.8/68.5 | irAEs group 8.9/NA;non-irAEs group 4.6/20.5 | Low-grade irAEs and hypothyroidism associate with higher DCR/ORR and longer PFS; high-grade irAEs often led to discontinuation. | irAEs 8.5%; dermatological irAEs 3.8%; elevated liver enzymes 1.5%; discontinuation due to irAEs 11.5% | [58] |
| Nam et al. (2024) | RWE(n = 150) | CP-A/BBCLC B/CECOG ≤2 | 4.5/35.6/71.2 | 5.7/13.6 | Mild irAEs independently predict better OS/PFS; severe irAEs tend toward worse outcomes; time-to-treatment discontinuation correlates with OS/PFS. | irAEs 19.3%; hepatitis 8%; colitis 3.3%; pneumonitis and fatigue 2.0%; discontinuation due to AEs 25.3% | [56] |
| Alkadimi et al. (2024) | RWE (n = 332) | CP-A/B/C BCLC A-D ECOG ≤3 | NA/30/NA | 6-m PFS 59%; 12-m PFS/OS 36%/52% | Real-world outcomes differed from those of the IMbrave150 trial because of a higher proportion of elderly patients | NA | [10] |
| Celsa et al. (2025) | RWE (n = 571) | CP-ABCLC A-CECOG PS 0–1 | NA | 3/6/12-m progression 23%/34%/48%; OS 17.4 | Hepatic decompensation is a major early event driving mortality; ALBI grade 2/3 and >3 nodules predict worse OS; successful viral therapy lowers decompensation risk. | NA (focus on decompensation/ progression as competing risks) | [37] |
| Persano et al. (2024) | RWE(n = 823) | CP-ABCLC B/C | 4.2/27.3/79.3 | 7.6/15.9 | AEs correlate with outcomes; any-grade proteinuria/diarrhea or ≥ grade 2 HTN linked to longer PFS; decreased appetite < grade 2 and irAEs < grade 2 linked to longer OS. | AEs 17.3%; proteinuria 6.9%; HTN 6.3% | [59] |
| Gairing et al. (2025) | RWE(n = 683) | CP-A/B ECOG ≤2 | Training cohort: NA/31.8/71.7; validation cohort: NA/31/69.7 | Training cohort: NA/13.7; validation cohort: NA/16 | CABLE score (CRP, albumin, bilirubin, lymphocytes, ECOG, EHS) stratifies prognosis for HCC patients treated with first line ate/bev | NA (study focused on prognostic modeling) | [36] |
| Manfredi et al. (2025) | Meta-analysis (n = 2179) | NA | RECIST 1.1: NA/30/78mRECIST: NA/36/77 | Pooled PFS/OS 6-m 57/82%; 12-m 35/65%; 18-m 26/52%; 24-m 25/39% | Real-world effectiveness of ate/bev in unresectable HCC closely mirrors the outcomes observed in clinical trials | Pooled TRAEs 26%; pooled GI bleeding 8%; pooled HTN and proteinuria 5%; discontinuation due to TRAEs 12% | [12] |
ALBI grade: albumin-bilirubin grade; AEs: adverse events; ALT: alanine aminotransferase; AST: aspartate aminotransferase; BCLC: barcelona clinic liver cancer; CP: Child-Pugh score; CR: complete response; DCR: disease control rate; HTN: hypertension; irAEs: immune-related adverse events; NA: non-available; ORR: overall response rate; OS: overall survival; PFS: progression-free survival; PVTT: portal vein tumor thrombosis; RCT: randomized controlled trial; RWE: real-world evidence; TRAEs: treatment-related adverse events.
As hepatitis B virus (HBV)-related HCC declines with vaccination and antivirals, the share attributable to metabolic dysfunction–associated fatty liver disease [MAFLD; formerly NAFLD (nonalcoholic fatty liver disease)/NASH (nonalcoholic steatohepatitis)] is rising [13,14]. Emerging data suggest that NASH-driven HCC may be less responsive to immunotherapy, as dysfunctional CD8+PD-1 + T cells impair antitumor surveillance, and both preclinical studies and meta-analyses indicate reduced benefit from PD-1/PD-L1 inhibitors in non-viral—particularly NASH-related—HCC, although these subgroup findings may be influenced by selection bias [15]. A large retrospective multicenter study showed no overall survival difference between ate/bev and lenvatinib overall, but greater benefit of ate/bev in viral HCC and a relative advantage of lenvatinib in NASH/NAFLD [16]. Conversely, a meta-analysis of nine phase III trials found that the survival advantage of immune checkpoint inhibitors (ICIs) over tyrosine kinase inhibitors (TKIs) was consistent across HBV, hepatitis C virus (HCV), and non-viral etiologies (HR 0.74, 0.77, and 0.86, respectively), indicating that etiology alone should not determine treatment selection in current practice [17].
5Hepatic safety and viral kinetics in HBV-Related HCCIn virus-related HCC, hepatic events and viral reactivation remain key concerns with ICIs. In an exploratory analysis of IMbrave150, Hsu et al. assessed hepatic safety and viral kinetics in patients with HBV (31%) or HCV (12%). Hepatic serious adverse events occurred in 11% with ate/bev and 8% with sorafenib, and reactivation rates were low and similar (HBV: 2% vs 7%; HCV: 16% vs 14%); notably, no hepatitis flares occurred with ate/bev [18]. These data support a consistent hepatic safety profile of ate/bev across viral statuses.
Because IMbrave150 excluded patients with high HBV DNA, we conducted a Taiwan Cooperative Oncology Group study (NCT04180072) in uHCC with HBV DNA >2000 IU/mL. In the 30-patient expansion cohort receiving concomitant antiviral therapy with ate/bev, no HBV reactivation occurred, and 17 patients achieved a rapid virological response. Liver adverse events were manageable, with one case of immune-related hepatitis; efficacy was encouraging (5 partial responses, 15 stable diseases; median PFS 6.3 months; OS 19.7 months) [19]. Therefore, in HBV-related HCC, ate/bev with concomitant antivirals demonstrates low, manageable rates of hepatic events and viral reactivation across baseline HBV DNA levels.
6Impact of radiological responseDirect comparisons of radiological response to ate/bev by Response Evaluation Criteria in Solid Tumors (RECIST) 1.1 versus modified RECIST (mRECIST) have been lacking. In a 125-patient cohort, both criteria showed substantial interobserver agreement (k = 0.79); mRECIST identified more responders (20.8%vs 9.6%), while both stratified overall survivals similarly. Discrepancies between blinded central review and multidisciplinary board decisions—particularly for intrahepatic progression—support refined assessment approaches, including progression-pattern analysis and circulating biomarkers [20].
7Durable response and discontinuationUnlike the TKI era, ate/bev induces complete response (CR) or durable partial response (PR)/stable disease (SD) in a subset of patients, making de-escalation or discontinuation a reasonable consideration.
In a post hoc analysis of IMbrave150, 37 patients (13%) achieved RECIST CR and had superior outcomes (2-year PFS: 58% vs. 7%; 2-year OS: 81% vs. 21%); CR was strongly associated with survival (HR for OS, 0.22; 95% CI, 0.07–0.70) [21]. In a multicenter real-world cohort of 3933 advanced HCC patients (60% received ate/bev), CR rates were 2.5% by RECIST and 2.0% by mRECIST; 89% achieved pathological CR after curative conversion. Recurrence occurred in 34% (38% by RECIST vs 29% by mRECIST). Continuing ICI therapy for ≥6 months beyond CR correlated with improved recurrence-free survival (RFS), suggesting response consolidation before treatment holidays [22]. Durable radiologic responses do not always signal complete tumor eradication. Among patients with RECIST PR or SD lasting >6 months, pathological CR was more frequent after PR than SD (57.7%vs 16.7%) and increased with longer treatment intervals before resection; deeper histologic responses translated into better RFS and OS [23].
A preliminary framework proposes considering discontinuation when three criteria are met: CR by mRECIST (± locoregional therapy); sustained normalization of AFP, AFP-L3, and des-gamma-carboxy prothrombin for 12–24 weeks; and either disappearance of tumor vascularity on contrast-enhanced ultrasonography or pathological cure after resection. These criteria require validation in larger, prospective datasets [24].
8Curative conversion after ate/bevThe atezolizumab–bevacizumab–curative conversion (ABC-conversion) concept—ate/bev followed by curative conversion—proposed by Kudo highlights how systemic therapy can enable curative interventions in TACE-unsuitable intermediate-stage HCC [25]. In IMbrave150, ate/bev outperformed sorafenib and showed particularly strong efficacy in intermediate-stage disease [OS: 25.8 vs. 17.5 months; PFS: 12.6 vs. 6.5 months; objective response rate (ORR) 44% vs. 27%], supporting downstaging and conversion. ABC conversion may follow tumor shrinkage that permits resection or ablation, or an ABC-TACE “sandwich” approach using selective TACE between cycles of ate/bev therapy.
In a multicenter cohort of 110 consecutive Child–Pugh A patients with TACE-unsuitable intermediate-stage HCC treated with ate/bev, clinical or pathological CR was achieved in 35%, 23% attained durable drug-free status without recurrence, and conversion modalities included resection (7), ablation (13), and superselective TACE (15). Among patients achieving CR, no deaths were observed and PFS was not reached, whereas non-conversion patients had median OS 18.5 months and PFS 7.9 months; pathological CR was confirmed in 3 of 7 resected cases [26]. A separate multicenter retrospective study of 188 patients reported an 8% conversion rate after ate/bev; BCLC stage A or B was the only independent predictor. Cancer-free status was most often achieved in those with fewer intrahepatic lesions, earlier conversion, and use of resection or ablation [27].
In the phase III TALENTOP study, presented in the 2025 ESMO Congress, HCC with macrovascular invasion (MVI) received induction therapy with ate/bev (three cycles) followed by one additional cycle of atezolizumab; those deemed “technically resectable” with PR or SD were randomized to resection followed by maintenance ate/bev versus continued ate/bev. Surgery arm improved time to failure of treatment strategy, while OS remains immature. Although ABC conversion is associated with favorable outcomes, the unexpectedly high conversion-to-resection rate (∼40%), the inclusion of patients without tumor shrinkage before randomization (∼30%), and center-dependent definitions of “technical resectability” in MVI-positive HCC limit generalizability and complicate attribution of benefit to conversion surgery in this study [28].
Key limitations include difficulties in prospectively identifying ideal candidates and uncertainty about long-term durability. Although prospective validation and standardized definitions of cancer-free and drug-free status are needed, ABC conversion offers encouraging evidence that selected, early responding patients can be transitioned from previously incurable to potentially curable disease.
9Liver transplantation after ate/bevThe safety and outcomes of liver transplantation (LT) after downstaging with ate/bev in patients initially beyond Milan criteria remain uncertain, particularly regarding rejection and recurrence. Two case reports illustrate feasibility: a 57-year-old man with multinodular HCC achieved pathological CR with no recurrence or rejection [29], and a 54-year-old man with HBV-related HCC and Vp4 portal vein tumor thrombosis (PVTT) attained radiologic CR, underwent ABO-incompatible living-donor transplantation six weeks after stopping immunotherapy, and remained recurrence-free with preserved graft function at 24 months [30].
Case series provide additional support. In a prospective multicenter cohort of 17 patients beyond Milan criteria, ate/bev produced a 94% ORR, 59% CR, and 88% pathological response at explant, with excellent post-transplantation survival and no severe rejection [31]. Similarly, in a single-center series of 115 patients, 12 were downstage and five underwent liver transplantation, almost all with pathological CR and without recurrence or graft rejection [32].
Overall, ate/bev appears promising as a bridge to liver transplantation for carefully selected patients beyond Milan criteria; prospective studies are needed to refine selection, timing, and peri‑transplant strategies.
10Prognostic factors in real-world evidenceReal-world studies have expanded our understanding of prognostic determinants. Age does not appear to limit benefit. Vithayathil et al. reported that patients ≥65 (and even ≥75) achieved survival, response, and tolerability comparable to younger patients despite differing baseline features [33].
By contrast, systemic inflammation and liver function are key. A Korean multicenter study (n = 121) identified AFP increase, baseline neutrophil-to-lymphocyte ratio (NLR) ≥5, and non-response as predictors of shorter PFS, while macrovascular invasion and NLR ≥5 independently predicted worse OS [34]. A global cohort (n = 296) similarly showed higher ALBI grade and PVTT predicted inferior OS, and AFP ≥400 ng/mL, worse ALBI, and extrahepatic spread were linked to shorter PFS [35]. To refine risk, the CABLE score (n = 683)—C-reactive protein (CRP), albumin, bilirubin, lymphocytes, Eastern Cooperative Oncology Group (ECOG) performance status, and extrahepatic spread—outperformed ALBI, mALBI, CRP- and AFP-based immunotherapy score (CRAFITY score), NLR, and Glasgow prognostic score, with a practical web-based calculator for individualized prognostication [36]. The AB-real cohort (n = 571) further showed hepatic decompensation is a stronger mortality driver than tumor progression; risk was associated with baseline ALBI grade 2/3 and multiple nodules, whereas viral etiology and effective antiviral therapy were protective. Preserving liver function and controlling viral hepatitis are therefore critical to optimize outcomes on ate/bev [37].
In summary, real-world evidence highlights liver function (ALBI grade, decompensation), systemic inflammation (NLR, CRP), tumor burden (PVTT, extrahepatic spread, nodule count), and viral control as the principal prognostic factors for patients receiving ate/bev.
11Safety profile of ate/bevIn IMbrave150, the incidence of immune-related adverse events (irAEs) with ate/bev mirrored prior anti–PD-L1 monotherapy; the most frequent irAEs were hepatitis (13.1%), hypothyroidism (10.9%), and pneumonitis (1.2%). With a median 15.6-month follow-up, bevacizumab-related toxicities were mainly proteinuria (29%) and hypertension (28%). Bleeding events were slightly higher with ate/bev vs. sorafenib (25%vs. 17%), predominantly grade 1–2 epistaxis; grade 3–4 bleeding occurred in 6% in both arms [2].
A systematic review of 30 real-world studies (n = 3867) reported adverse events in 79% of patients (56% grade 1–2; 30% grade ≥3), with the overall spectrum largely paralleling that observed in IMbrave150. The most common adverse events were hypertension, proteinuria, and fatigue. Reporting of bleeding events and rare irAEs was heterogeneous, underscoring the need for standardized AE documentation [38].
12Risk of bleeding with bevacizumabEarly phase II bevacizumab monotherapy in HCC showed activity but raised safety concerns, with gastrointestinal bleeding in 5–11% and fatal variceal hemorrhage in up to 11% of patients; baseline endoscopy and prophylactic ligation were recommended, though they could delay therapy [6,39].
In the ate/bev era, bleeding remains clinically relevant. Reported variceal bleeding rates range from 3% to 14%, typically within the first three months. Risk is highest with prior varices bleeding, untreated high-risk varices, poor liver function (Child–Pugh B, ALBI 2–3), portal vein invasion, infiltrative tumors, or thrombocytopenia. Prospective and real-world studies consistently emphasize baseline endoscopy and prophylactic varices ligation to mitigate risk [40–44].
A meta-analysis of 28 studies (n = 3895) reported pooled bleeding prevalence of 8.4%, grade ≥3 bleeding of 4.4%, and fatal bleeding of 2.1%. Variceal hemorrhage was most common (5.5%) and occurred at more than twice the rate seen with TKIs; high body mass index and ALBI grade 3 were additional predictors. Thus, despite strong efficacy, bleeding—particularly variceal hemorrhage—remains a safety concern, and careful patient selection, baseline endoscopy, and proactive management of high-risk varices are critical to optimize safety without compromising benefit [45].
13Bevacizumab-related hypertensionIn IMbrave150, hypertension was more frequent with ate/bev than with sorafenib, with any-grade events in 29.8% vs. 15.2% and grade 3–4 in 24.4% vs. 12.2%. Bevacizumab (anti-VEGF) promotes hypertension by reducing endothelial nitric oxide–mediated vasodilation, increasing endothelin-1–driven vasoconstriction, promoting microvascular rarefaction and vascular stiffness, and impairing renal function [46,47].
A meta-analysis of 72 studies (n = 21,902) reported pooled incidences of bevacizumab-induced all-grade and high-grade hypertension of 25.3% and 8.2%, with relative risks of 3.6 and 5.2 vs. controls [48]. Another meta-analysis of 20 studies (n = 12,656) confirmed a high-grade incidence of 7.9% (RR 5.28), consistent across tumor types and low- (2.5 mg/kg/week) vs. high-dose (5 mg/kg/week) regimens, with grade 4 crises in 0.5% [49]. These findings underscore the importance of close blood-pressure monitoring and timely antihypertensive therapy during bevacizumab treatment.
14Bevacizumab-related proteinuriaIn IMbrave150, proteinuria was more frequent with ate/bev than with sorafenib for both any grade (20.1%vs. 7.1%) and grade 3–4 events (7.1%vs. 0.6%). Bevacizumab shows a dose-dependent risk, with early pooled data reporting proteinuria in 21–63% and grade ≥3 in up to 1.8% [50].
A 2017 meta-analysis found pooled incidences of 18% for all-grade and 2.4% for grade ≥3 proteinuria, with higher doses increasing severe events [48]. A single-center study in gynecologic cancers reported proteinuria in 38%, reaching a plateau around 35% (3% grade 3–4), suggesting cumulative exposure contributes to risk [51]. Collectively, proteinuria is a clinically relevant, dose-dependent toxicity of bevacizumab, warranting early detection and appropriate management.
15Impact of bevacizumab interruption on outcomesIn an exploratory IMbrave150 analysis of patients treated ≥6 months, withholding bevacizumab for VEGF-related adverse events did not worsen outcomes (OS: HR 1.04; PFS: HR 1.07–1.10) [52]. A Taiwanese cohort from IMbrave150 and GO30140 likewise showed that 50% required bevacizumab skipping (median 175 days) without significant differences in PFS or OS [53].
Conversely, early interruption correlated with poorer results in Japanese real-world cohorts. In a 239-patient study (landmark 9 weeks), no-interruption vs. interruption yielded higher ORR (34.5%vs. 17.3%), longer PFS (9.0 vs. 6.5 months, p = 0.021), and higher 12-month OS (82.2%vs. 49.4%, p = 0.004); later-line therapy and ALBI 2b predicted interruption [54]. Another multicenter series (n = 123) found that interruption—most often for proteinuria and more frequent with hypertension and/or diabetes—within 24 weeks was associated with inferior PFS [55].
Overall, late bevacizumab interruption after prolonged exposure appears acceptable, whereas early interruption is unfavorable. Close monitoring and proactive management of VEGF-related toxicities are critical to maintaining treatment continuity and preserving clinical outcomes.
16Immune-related adverse events and clinical impactAcross retrospective series, irAEs carry prognostic significance. In two multicenter cohorts (∼150 patients each), grade 1–2 irAEs were independently associated with longer survival (>20 months) compared with markedly shorter OS in patients without irAEs or with severe irAEs [56,57]. A single-center study (n = 130) likewise linked irAEs to prolonged PFS (8.9 vs. 4.6 months) and higher disease control, with hypothyroidism particularly predictive of benefit [58]. Beyond irAEs, a multicountry real-world analysis (n = 823) found that low-grade irAE, decreased appetite, diarrhea, hypertension, and proteinuria independently associated with improved PFS and OS, suggesting treatment-emergent toxicities may serve as efficacy surrogates [59].
Collectively, mild irAEs appear to be positive prognostic markers in patients receiving ate/bev, whereas severe irAEs often necessitate discontinuation and portend poorer outcomes. Careful evaluation and early management of irAEs, with the goal of keeping events low grade, are essential to maximize efficacy while maintaining safety.
17Post-ate/bev therapeutic strategiesAfter progression on ate/bev, prognosis remains poor and no high-level evidence exists to guide the subsequent systemic therapy. Real-world data showed that patients with preserved Child–Pugh A reserve derive markedly longer survival from second-line therapy than those with impaired liver function [60].
Across cohorts, lenvatinib generally outperforms sorafenib post-ate/bev, with higher disease control and longer PFS and a trend toward longer OS [61–63]. A phase II study reported ORR 14%, disease control rate (DCR) 82%, PFS 5.4 months, and OS 9.8 months with second-line lenvatinib [64], concordant with real-world series (PFS: ∼4–5 months; OS: ∼16 months) [65,66]. Regarding cabozantinib, a phase II study showed activity after ate/bev (PFS: 4.1 months; OS: 9.9 months) and performs best in Child–Pugh A/ECOG 0–1, aligning with large Asia-Pacific data and CELESTIAL benchmarks [61,67–69]. For AFP ≥400 ng/mL, ramucirumab remains a viable option with disease control but frequent grade ≥3 toxicities and liver function decline in some series [70–72]. ICI-based strategies may benefit selected patients after ate/bev. Multicenter data show that TKI–ICI combinations are associated with longer survival compared with TKI monotherapy [61,73]. Notably, in a retrospective study of 60 patients, ipilimumab plus nivolumab achieved response rates of 42.9% in ICI-naïve and 19.4% in patients pretreated with ate/bev, with some responses lasting up to 19.5 months. These findings support the rationale for ICI-based combination approaches as viable options beyond first-line ate/bev treatment [74]. All studies are summarized in Table 2.
Post-atezolizumab/bevacizumab second-line options.
| Study (year) | Regimen | Design (numbers) | Populations | CR/ORR/DCR (%) | PFS/OS (months) | Key findings | ≥ Grade 3 AEs | Ref. |
|---|---|---|---|---|---|---|---|---|
| Kuzuya et al. (2022) | Ramucirumab | RWE(n = 13) | Advanced HCC, AFP ≥400 ng/mL; CPS A (69%), ≥3 line (85%) | 0/15.4/69.2 | 3.0/4.8 | AFP decreased in 85% at 2 weeks; liver function deterioration noted; caution in ≥3 line | Ascites 15.4%; proteinuria 7.7%; bleeding 7.7% | [71] |
| Shimose et al. (2023) | Ramucirumab | RWE(n = 46; 12 post-ate/bev) | Advanced HCC, unresectable | 0/33.3/83.3 | 3.9/7.6 | RAM more effective after ate/bev failure; splenomegaly and low BMI predicted ascites | Ascites 23.9%; proteinuria 10.8%; HTN 2.0% | [72] |
| Chan et al. (2024) | Cabozantinib | Phase II, single-arm, multicenter(n = 47) | Advanced HCC, prior ICI (1–2 lines), CPS A | 0/6.4/83 | 4.1/9.9 | Demonstrated efficacy post-ICI; longer OS in 2 line vs ≥3 line; number of prior lines prognostic | Thrombocytopenia 6.4%; HTN 4.3%; proteinuria 2.1% | [67] |
| Kuzuya et al. (2023) | Cabozantinib | RWE(n = 19) | Advanced HCC | 0/5.3/52.6 | 3.9/13.4CPS A +ECOG 0/1 | Benefit limited to patients with preserved liver function and ECOG 0/1 | Proteinuria 31.6%; hand–foot syndrome 5.3%; bleeding 5.3% | [68] |
| Kim et al. (2025) | Lenvatinib | Prospective, phase II(n = 50) | Advanced HCC, CPS A | 0/14/82 | 5.4/9.8 | First prospective second line LEN study; ORR correlated with survival | ALT elevation 10%; proteinuria 8%; HTN 8%; bilirubin elevation 4%; fatigue 4% | [64] |
| Mohri et al. (2024) | Lenvatinib and other TKI | RWE(LEN=20) | Unresectable HCC | NA/45/75 | 3.5/24.2 | LEN effective post ate/bev; FGF-19 change may predict benefit | Not specified | [65] |
| Hiraoka et al. (2023) | Lenvatinib and other TKI | RWE(LEN=101) | Unresectable HCC; CPS A (81%) | NA/15.4/66.2 | 4.4/15.7 | LEN effective post ate/bev comparable to first line; liver reserve is a key prognostic factor | Proteinuria 12.9%; anorexia 9.9%; fatigue 5.9%; HTN 5.0% | [66] |
| Chen et al. (2022) | SorafenibLenvatinib | RWE(SOR=19, LEN=9) | Advanced HCC, CPS B: 15%, CPS C:17% | SOR: 0/0/47LEN: NA/11/22 | SOR: 2.6/8.3LEN: 2/3.8 | No difference between LEN and SOR | NA | [60] |
| Lee et al. (2024) | SorafenibLenvatinibRegorafenibCabozantinibTKI+ICI | RWE;(n = 629,SOR=339LEN=161REG=37CAB=12TKI+ICI=50) | Unresectable HCC | NA | SOR:2.3/6.3LEN:4.0/8.0REG:3.6/9.7CAB:5.4/11.2TKI+ICI:5.4/12.6 | LEN superior to SOR; preserved liver function and lower AFP predicted OS; TKI+ICI promising for continued ICI benefit | NA | [61] |
| Wu et al. (2025) | SorafenibLenvatinibICI | RWE(n = 406,TKI=155,ICI=45,BSC=184) | Advanced HCC CPS A: 79% | NA | SOR: NA/7.0LEN: NA/10.2ICI: NA/14.9 | Active 2 L improved PPS; ICI >1y PPS suggests ICI continuation benefit; LEN numerically > SOR (trend) | NA | [73] |
| Kim et al. (2025) | Ipilimumab plus Nivolumab | RWE(n = 60) | Advanced HCC | 6.7/30.5/42.4 | 1.4/7.8 | Ipi/Nivo is a meaningful option in after prior exposure to ate/bev | NA | [74] |
CR: complete response; ORR: overall response rate; DCR: disease control rate; PFS: progression-free survival; OS: overall survival; AEs: adverse events; RWE: real-world evidence; AFP: alpha-fetoprotein; CPS: Child-Pugh score; RAM: ramucirumab; BMI: body mass index; HTN: hypertension; ICI: immune checkpoint inhibitor; LEN: lenvatinib; ALT: alanine aminotransferase; TKI: tyrosine kinase inhibitor; FGF-19: fibroblast growth factor 19; SOR: sorafenib; REG: regorafenib; CAB: cabozantinib; BSC: best supportive care; PPS: post-progression survival.
Integrated analyses from GO30140 and IMbrave150 identified “pre-existing immunity” (high PD-L1, T-effector signatures, intratumoral CD8+ density) as markers of improved outcomes on ate/bev, whereas high Treg/Teff ratios and oncofetal gene expression (e.g., glypican-3, AFP) correlated with reduced benefit; the advantage of ate/bev over atezolizumab monotherapy was enriched in tumors with VEGFR2 expression, Treg signatures, and myeloid inflammation, underscoring the immunomodulatory role of VEGF blockade [75]. Single-cell RNA sequencing further delineated responder phenotypes (CD8+ effector T cells, CXCL10+ macrophages or angiogenesis-driven, low neuropilin-1) and resistance states (immunosuppressive myeloid subsets, Notch/TGF-β activation) [76].
Although CTNNB1 mutations typically mark non–T-cell-inflamed, ICI-refractory HCC, a small ate/bev cohort (n = 33) reported similar outcomes in CTNNB1-mutant and wild-type tumors, suggesting VEGF blockade may mitigate this resistance [77]. Peripheral immunophenotyping linked elevated baseline and early-expanded CTLA4+ lymphocytes to resistance and inferior survival, whereas circulating PD-L1 was non-predictive [78]. Transcriptome-derived models (Immune signature score, ISS) showed ISS10-high tumors achieving better survival and response on ate/bev versus sorafenib [79].
Despite these advances, no clinically validated biomarker currently guides patient selection, emphasizing the need for prospective validation before integration into routine practice.
19Investigational combinations with ate/bevAnti-T cell immunoreceptor with Ig and ITIM domains (TIGIT) antibodies target the inhibitory receptor TIGIT to restore antitumor T-cell activity [80]. In MORPHEUS-Liver (phase Ib/II), adding tiragolumab to ate/bev improved outcomes, with ORR 43% vs. 11% and median PFS 12.3 vs. 4.2 months, without new safety signals [81]. However, the confirmatory IMbrave152/SKYSCRAPER-14 phase III trial did not demonstrate benefit in PFS and showed no favorable OS trend, according to recent 2025 ESMO Congress [82].
Another study just presented in 2025 ESMO Congress, phase II portion of PRODIGE 81/FFCD 2101 (TRIPLET-HCC), adding low-dose ipilimumab (1 mg/kg for four cycles) to ate/bev failed to improve efficacy over the doublet. The study did not meet its primary endpoint, with ORR of 30.1% versus 27.4%, and no meaningful differences in PFS, OS, time to progression, duration of response, or time to response. Ate/bev also showed slightly better tolerability. The OS data remain immature, but if CTLA-4 blockade is to add benefit in this setting, higher ipilimumab dosing (e.g., 3 mg/kg for four cycles as in CheckMate-040/9DW) may be necessary [83]. These findings underscore both the promise and the current limitations of extending checkpoint blockade beyond PD-(L)1 in HCC, highlighting the need for refined targets and biomarker-driven strategies.
20Ate/bev combined locoregional therapyAlthough the integration of ate/bev with locoregional therapies (LRT) is biologically rational, the overall level of evidence remains limited. Most data are supported by small retrospective cohorts, single-arm studies, or exploratory analyses with immature overall survival results and potential selection bias. Therefore, while preliminary findings suggest synergy, prospective validation in adequately powered randomized trials is essential before routine adoption into standard practice. Studies of ate/bev combined with locoregional therapy are summarized in Table 3.
Atezolizumab plus bevacizumab combined with locoregional therapy.
| Study (year) | Combined LRT | Design (numbers) | Populations | CR/ORR/DCR (%) | PFS/OS (months) | Key findings | ≥ Grade 3 AEs | Ref. |
|---|---|---|---|---|---|---|---|---|
| TALENTACE (2025) | TACE (on-demand) | RCT(n = 342) | BCLC A/B/CTACE eligibleTumor score ≥6, CP-A | 3.5/49.1/84.2(RECIST 1.1) | 10.3/34.5 | Improvement of TACE-PFS (HR, 0.71); OS remains immature; showed a safety profile | TRAE: 60.8%, SAE: 25.9%, AST elevation: 33.1%, post-embolization syndrome: 45.8% | [84] |
| Xin et al. (2022) | HAIC-FOLFOX | Multicenter retrospective(n = 52) | Advanced HCC, treatment-naïve | ORR: 67.3% (mRECIST), ORR: 44.2% (RECIST 1.1) | 10.6/NR. | Triple therapy effective, extrahepatic metastasis is poor risk factor | HTN (7.7%), AST elevation (5.8%), and thrombocytopenia (3.8%). HAIC discontinued due to AEs (3.8%), HAIC dose reduction (9.6%) | [98] |
| He et al. (2025) | HAIC-FOLFOX | Real-world PSM(n = 80) | Unresectable HCC | 0/48.4/96.8% (RECIST 1.1)8.6/57/96.8 (mRECIST) | 8.8/NR. | Ate/bev + HAIC is effective and tolerable | NA | [99] |
| Yu et al. (2024) | HAIC-FOLFOX ± TACE | Meta-analysis (10 studies, n = 405) | Advanced HCC | 10.8/ 57.2/86 (mRECIST) | 10.9/NA. | Demonstrated pooled efficacy and tolerance of Ate/bev + HAIC/TACE | Any-grade AEs: 91.0%; grade ≥3 AEs: 24.8%. | [100] |
| Villalobos et al. (2023) | Y90-TARE | RWE(n = 10) | BCLC B-C | 75/100/100(mRECIST) | 6 m PFS: 78.8%; 12 m PFS: 66.7%; 6 m OS: 90.0%; 12 m OS: 77.1% | Demonstrated feasibility and safety; 3/4 intermediate-stage patients downstaged to Milan criteria | Grade ≥3 AST elevation & hypoglobulinemia (1 pt); ICI-related myositis (1 pt) | [89] |
| Kim et al. (2023) | RT (IMRT or PBT) | RWE(n = 7) | Advanced HCC | 14.3/28.6/85.7(RECIST 1.1) | 4/14.8 | Additional RT during ate/bev feasible; showed lesion control and acceptable safety | None | [92] |
| Wang et al. (2023) | IMRT | Multicenter prospective (n = 30) | HCC with extrahepatic PVTT | 6.7/76.6/96.7(RECIST 1.1) | 8/9.8 | Ate/bev + IMRT achieved high ORR and disease control in extrahepatic PVTT | HTN 10%; AST elevation 10%; neutropenia 6.7%; no treatment-related deaths | [93] |
| Su et al. (2024) | High-dose EBRT (PBT/SBRT) | RWE(n = 14) | Highly advanced HCC with Vp4 PVTT or >50% liver volume | 7.1/50/78.6(mRECIST) | 5.2/NR. | Ate/bev + high-dose external beam RT improved ORR and OS without increasing severe AEs | AEs 14.3%: GI bleeding, dermatitis; similar to ate/bev alone | [95] |
LRT: locoregional therapy; CR: complete response; ORR: overall response rate; DCR: disease control rate; PFS: progression-free survival; OS: overall survival; AEs: adverse events; TRAE: treatment-related adverse events; SAE: serious adverse events; TACE: transarterial chemoembolization; RCT: randomized controlled trial; BCLC: Barcelona Clinic Liver Cancer staging; CP/CP-A: Child–Pugh (A = class A); HR: hazard ratio; AST: aspartate aminotransferase; HAIC: hepatic arterial infusion chemotherapy; FOLFOX: 5-fluorouracil, leucovorin, oxaliplatin; mRECIST: modified Response Evaluation Criteria in Solid Tumors; RECIST 1.1: Response Evaluation Criteria in Solid Tumors, version 1.1; NR: not reached; NA: not available/not applicable; HTN: hypertension; Y90-TARE: yttrium-90 transarterial radioembolization; RWE: real-world evidence; RT: radiotherapy; IMRT: intensity-modulated radiotherapy; PBT: proton beam therapy; EBRT: external-beam radiotherapy; SBRT: stereotactic body radiotherapy; PVTT: portal vein tumor thrombus; Vp4: main trunk portal vein invasion classification; GI: gastrointestinal; ICI: immune checkpoint inhibitor.
The phase III TALENTACE trial (open-label, randomized) compared ate/bev plus on-demand TACE vs. TACE alone in unresectable, systemic-therapy–naïve HCC. The combination met its primary endpoint, prolonging TACE-PFS (11.3 vs. 7.0 months; HR 0.71; p = 0.009) and improving RECIST 1.1 PFS (10.3 vs. 6.4 months; HR 0.64), with higher response rates in the combination arm [84]. Although OS data are immature, early findings suggest ate/bev plus TACE may be a promising option for intermediate- to high-tumor-burden HCC; however, the currently available results are interim and non–peer-reviewed, and longer follow-up and full publication are required to define survival impact and practice implications [Chugai press release, May 21, 2025].
At the 2025 ESMO Congress, the phase IIIb IKF-035/ABCHCC trial reported interim results showing that ate/bev outperformed TACE in intermediate-stage HCC by the composite endpoint of time-to-failure of treatment strategy (TTFS). However, because TTFS includes subjective components (e.g., “loss of clinical benefit,” treatment inapplicability), “unTACEable” status was not clearly defined, and tumor burden was not prespecified—introducing potential selection and assessment bias—these data are insufficient to claim definitive superiority over TACE in the absence of mature survival outcomes [85].
20.2Yttrium-90 radioembolizationY90 radioembolization is used for intermediate- to advanced-stage HCC—particularly with portal vein thrombosis and no distant metastasis—and can serve for downstaging or bridging to transplantation [86]. Preclinical data support synergy among VEGF blockade, immune checkpoint inhibition, and radioembolization [87]. Early reports suggest feasibility and clinical activity in 10 patients treated with Y90 plus ate/bev. Local control was 100%, with 6- and 12-month PFS of 78.8% and 66.7%, respectively, and OS of 90.0% and 77.1%. Three patients were successfully downstaged to within Milan criteria. Reported adverse events included grade 3 transaminitis and one case of immunotherapy-related myositis [88]. A multicenter cohort (n = 19) treated with Y90 plus ate/bev or nivolumab showed ORR 58%, CR 16%, and median OS 10.7 months in the ate/bev subgroup, with acceptable safety [89]. An ongoing feasibility trial (NCT07059494) is evaluating ate/bev plus Y90 as a bridging or downstaging strategy. Overall, retrospective data indicate feasibility, safety, and downstaging potential, pending prospective validation.
20.3RadiotherapyExternal-beam radiotherapy is an effective local modality for HCC, and combination with ate/bev is biologically plausible yet requires further clinical validation. In preclinical HCC models, low-dose irradiation enhanced dual PD-L1 and VEGFA blockade by activating CXCL10/CXCR3 signaling, recruiting stem-like CD8+ Tpex cells from tumor-draining lymph nodes and driving their differentiation into cytotoxic Tex cells, thereby augmenting antitumor immunity [90]. Abscopal responses have been described; for example, adding stereotactic body radiotherapy to an adrenal metastasis after suboptimal control on ate/bev was followed by complete response upon resuming therapy [91]. Small clinical cohorts similarly suggest meaningful local control and survival with acceptable safety, including in extrahepatic portal vein tumor thrombus [92,93].
Particle therapy (e.g., protons, heavy ions) has matured and offers higher biological effectiveness and improved normal-tissue sparing vs. photons [94]. In a retrospective study of highly advanced HCC (Vp4 or >50% liver involvement), concurrent proton therapy plus ate/bev vs. ate/bev alone yielded higher ORR (50.0%vs. 11.8%) and longer OS (HR 0.18; p < 0.01) without increased grade ≥ 3 adverse events [95]. Case reports also describe conversion to curative therapy with particle-based regimens [96].
Collectively, available data support feasibility and potential synergy. Prospective trials are needed to define patient selection, dose/fractionation, and optimal sequencing with ate/bev.
20.4Hepatic arterial infusion chemotherapyHepatic arterial infusion chemotherapy (HAIC) delivers high intratumoral drug exposure with reduced systemic toxicity. HAIC with 5-fluorouracil, leucovorin, and oxaliplatin (FOLFOX) is effective for advanced HCC with PVTT [97]. Combining HAIC with immunotherapy appears active. Ate/bev plus HAIC-FOLFOX achieved an ORR of 67.3% (mRECIST) and median PFS of 10.6 months in a multicenter series [98]. In propensity score–matched cohorts, ate/bev-HAIC and lenvatinib-based HAIC showed similar ORR, whereas lenvatinib-based HAIC yielded longer PFS and OS [99]. A meta-analysis of ten trials (n = 405) reported pooled ORR 57.2%, DCR 85.9%, PFS 10.9 months, and grade ≥ 3 AEs 24.8% [100].
Additionally, a small retrospective study found that HAIC as a subsequent therapy after ate/bev failure yielded an ORR of 35.3%, PFS of 8.2 months, and OS of 12.4 months [101]. These findings suggest HAIC may be a reasonable second-line consideration after ate/bev in selected patients with liver-dominant disease, though evidence remains limited and retrospective. Despite these encouraging results, broader adoption is constrained by the lack of large-scale randomized trials in Western populations, regional differences in HAIC availability and procedural standardization, and variability in technical expertise. Consequently, multicenter prospective studies across diverse practice settings are needed to determine generalizability and inform consensus recommendations.
21Ate/bev for early-stage HCCThe phase III IMbrave050 trial compared adjuvant ate/bev with active surveillance in patients at high risk of recurrence after curative resection or ablation (included tumor >5 cm, ≥3 tumors, micro/macrovascular invasion [Vp1–Vp2], or poor differentiation) [102]. Although an interim analysis met the primary endpoint of RFS by independent review, subsequent updates failed to confirm benefit and suggested no advantage vs. active surveillance, while adding treatment-related toxicity. At present, these data do not support adjuvant ate/bev after curative therapy; whether neoadjuvant strategies will confer benefit remains to be established.
22ConclusionsOver five years after IMbrave150, cumulative evidence indicates that ate/bev enables downstaging and curative conversion, achieving disease-free, drug-free status in a subset of patients. However, criteria for de-escalation and discontinuation remain to be prospectively defined. In HBV-related HCC, ate/bev can be administered safely across baseline viral loads when delivered with concomitant antiviral therapy, with low and manageable rates of hepatic events. Rigorous pre-treatment assessment of bleeding and other VEGF-related toxicities, coupled with continued monitoring and timely management, is essential to maintain bevacizumab exposure and clinical benefit. Integration with locoregional therapies is promising and warrants prospective validation to refine patient selection, sequencing, and perioperative strategies.
FundingThis study is supported by Taiwan Clinical Oncology Research Foundation and Taipei Veterans General Hospital to San-Chi Chen.
Authors’ contributionsConceptualization and project administration: S-CC
Visualization, writing original draft, reviewing, and editing: K-CL, S-CC
All authors had full access to all the data and responsibility for the decision to submit for publication.
None.
We sincerely appreciate Dr. Caleb for his valuable assistance in reviewing and refining the English language of this manuscript.
