Atezolizumab/bevacizumab and lenvatinib for hepatocellular carcinoma: A comparative analysis in a European real-world cohort Authors: Tiago de Castro, Sabrina Welland, Leonie Jochheim, Cathrine Leyh, Kateryna Shmanko, Fabian Finkelmeier, Petia Jeliazkova, Andre Jefremow, Maria A Gonzalez-Carmona, Arne Kandulski, Daniel Roessler, Najib Ben Khaled, Stefan Enssle, Marino Venerito, Thorben W Fründt, Michael Schultheiß, Angela Djanani, Maria Pangerl, Andreas Maieron, Thomas C Wirth, Jens U Marquardt, Richard Greil, Christina Fricke, Rainer Günther, Andreas Schmiderer, Dominik Bettinger, Henning Wege, Bernhard Scheiner, Martina Müller, Christian P Strassburg, Jürgen Siebler, Ursula Ehmer, Oliver Waidmann, Arndt Weinmann, Matthias Pinter, Christian M Lange, Anna Saborowski, Arndt Vogel Published: 2024 ABSTRACT AbstractBackground:Immunotherapy-based combinations are currently the standard of care in the systemic treatment of patients with HCC. Recent studies have reported unexpectedly long survival with lenvatinib (LEN), supporting its use in first-line treatment for HCC. This study aims to compare the real-world effectiveness of LEN to atezolizumab/bevacizumab (AZ/BV).Methods:A retrospective analysis was conducted to evaluate the effectiveness and safety of frontline AZ/BV or LEN therapy in patients with advanced HCC across 18 university hospitals in Europe.Results:The study included 412 patients (AZ/BV: n=207; LEN: n=205). Baseline characteristics were comparable between the 2 treatment groups. However, patients treated with AZ/BV had a significantly longer median progression-free survival compared to those receiving LEN. The risk of hepatic decompensation was significantly higher in patients with impaired baseline liver function (albumin-bilirubin [ALBI] grade 2) treated with AZ/BV compared to those with preserved liver function. Patients with alcohol-associated liver disease had poorer baseline liver function compared to other etiologies and exhibited a worse outcome under AZ/BV.Conclusions:In this real-world cohort, survival rates were similar between patients treated with LEN and those treated with AZ/BV, confirming that both are viable first-line options for HCC. The increased risk of hepatic decompensation in patients treated with AZ/BV who have impaired baseline liver function underscores the need for careful monitoring. Future trials should aim to distinguish more clearly between metabolic dysfunction–associated steatotic liver disease and alcohol-associated liver disease.Keywords: albumin-bilirubin grade, hepatic dysfunction, immunotherapy, liver cancer, tyrosine kinase inhibitor INTRODUCTION HCC is the most prevalent malignant liver tumor, with an annual incidence of 750,000. 1 For over a decade, the tyrosine kinase inhibitor (TKI) sorafenib (SOR) was the only approved drug for HCC, and numerous randomized trials investigating other TKIs and combinations with SOR failed to show significant survival improvements. 2 In 2018, the multikinase inhibitor lenvatinib (LEN) demonstrated noninferiority to SOR in the REFLECT trial, leading to its approval as an alternative first-line option. 3 Notably, the median overall survival (mOS) in the SOR and LEN control arms of recent phase III HCC trials, including LEAP-002 and CHECKMATE 9DW, exceeded the mOS by up to 6 months reported in the respective pivotal trials, likely due to improved patient management and access to effective second-line therapies.4–7 The implementation of sequential systemic therapies in advanced HCC has also translated into a survival benefit in real-world (RW) cohorts.8,9 The introduction of immunotherapy-based strategies marked the next breakthrough in the treatment of advanced HCC. Despite very promising phase-I/II data, the initial phase III trials with single-agent immunotherapy failed in both the first-line and second-line settings.4,5 However, subsequent combinations of immune checkpoint inhibitors (ICIs) with anti-VEGF therapies significantly improved survival compared to SOR. The IMbrave150 study first reported significantly longer mOS with the combination of anti-programmed death-ligand 1 atezolizumab (AZ) plus anti-VEGF bevacizumab (BV) compared to SOR (19.2 vs. 13.4 mo), 10 which was subsequently endorsed by national and international guidelines as the new standard of care (SOC) for the first-line HCC. 11 Recently, the combination of camrelizumab, a programmed cell death-1 inhibitor, and rivoceranib, a TKI with VEGFR-2 directed activity, also significantly improved survival compared to SOR, reinforcing the effectiveness of ICI/VEGF-targeted combinations in HCC. 12 Additionally, the HIMALAYA trial expanded the immunotherapeutic options for HCC by demonstrating significant benefit for dual checkpoint inhibition with anti-programmed death-ligand 1 durvalumab and anti-cytotoxic T-lymphocyte–associated protein 4 tremelimumab, establishing it as a valid first-line option in HCC.13,14 Furthermore, the phase-3 CHECKMATE 9DW trial confirmed the benefit of combined programmed cell death-1 and cytotoxic T-lymphocyte–associated protein 4 targeting in the first-line treatment of HCC, showing a mOS of 23.7 months with nivolumab and ipilimumab. 7 The pivotal trials evaluating the efficacy of LEN and AZ/BV were both conducted with SOR in the control arm, and it is unlikely that prospective randomized head-to-head trials comparing the efficacy of AZ/BV to LEN will be conducted. Indirect cross-trial comparisons between recent phase III studies favored AZ/BV over LEN in OS analysis.15,16 Considering the recent efficacy data of LEN in the LEAP-002 and CHECKMATE 9DW studies, with an mOS similar to AZ/BV in IMbrave150, there is speculation that both treatments may have comparable effectiveness in the RW setting. It is important to note that RW patients often have reduced performance status and more impaired liver function compared to those eligible for clinical trials. Recently, several groups have reported real-world data (RWD) for both treatment options. Most of these data were derived from Asian patients, who often have well-preserved liver function (Child-Pugh score A) and HBV infection. While some studies reported similar effectiveness,17–20 others observed superior outcomes with AZ/BV compared to LEN.21,22 In contrast, a recent RW study including both Asian and Caucasian patients with excellent liver function found higher effectiveness of LEN compared to AZ/BV, particularly in patients with metabolic dysfunction–associated steatotic liver disease (MASLD). 23 Given the ambiguous results to date and the underrepresentation of Caucasian patients in RW cohorts, this study aimed to retrospectively compare the effectiveness and tolerability of AZ/BV and LEN as first-line therapy in a large, European multicenter cohort. METHODS This study included patients with advanced HCC from 18 different European tertiary hospitals. These patients were considered unsuitable for resection or locoregional therapies and received either first-line AZ/BV or LEN. A subset of the cohorts has been previously published.24,25 Treatment allocation was discussed at the respective tumor boards. Ethical approval for this study was obtained from the Ethical Committees of each participating institution as follows: Hannover (ID 940-2011); Essen (ID 21-71009-BO); Mainz (ID 837.199.10); Frankfurt and Oncology Bethanien (ID SG1-3-2018 [Amendment 12019]); Munich, Klinikum rechts der Isar (ID 2022-605-S-KH); Erlangen-Nuremberg (ID 277_20_Bc); Bonn (ID 341/17-2); Regensburg (ID 20-2010-104); LMU Munich (ID 18-604 and 20-439); Vienna (ID 2033/2017); Magdeburg (ID 70/22); Hamburg (ID PV-3578); Freiburg (ID EK 62/14); Innsbruck (ID 1195/2021); Schleswig-Holstein, Campus Kiel (BIC UKSH); St. Pölten (ID EK1092/2021); Schleswig-Holstein, Campus Lübeck (ID AZ-21-423); Salzburg (ID 1099/2020). All research was conducted in accordance with the Declaration of Helsinki. The cutoff date for follow-up data was April 30, 2024. Tumor response was assessed by CT and/or MRI every 8–12 weeks by local radiology teams, following the RECIST 1.1 criteria. Serious adverse events, graded as 3 or higher according to the National Cancer Institute—Common Terminology Criteria for Adverse Events, version 5.0, were included in the safety analysis. Statistics Statistical analysis was performed using IBM SPSS statistics version 26 (SPSS Inc., Chicago IL) and R software (version 4.2.2, SPSS Inc., Chicago IL). Survival was defined as the time from the start of treatment until death or last follow-up, while progression-free survival (PFS) was measured from the start of therapy to either radiological progression or death, whichever occurred first. Continuous variables were analyzed using the Student t test if sample means were normally distributed. If population variances were unequal, Welch t test was used. Categorial variables were analyzed using the Chi-Square test or Fisher exact test, as appropriate. mOS and median progression-free survival (mPFS) were estimated using the Kaplan-Meier method and compared with log-rank tests. Adjusted and unadjusted HRs for clinical variables were analyzed using the Cox proportional hazards model and reported with a 95% CI. To minimize potential imbalances in baseline parameters between the treatment groups, a propensity score matching was performed using a logistic regression model. Covariates were included in a 1:1 nearest neighbor matching algorithm with a caliper of 0.15, as described. 26 A significance level of p<0.05 was considered significant for all analyses. RESULTS Baseline characteristics Baseline patient characteristics are shown in Table 1. Data from 412 patients were available for effectiveness and safety analysis, of which 207 (50.2%) received AZ/BV and 205 (49.8%) were treated with LEN. TABLE 1Baseline characteristics by treatment cohortOverall (n=412) n (%)Lenvatinib first-line (n=205) n (%)Atezolizumab plus bevacizumab first-line (n=207) n (%) p Age (years) Mean (range) (years)67.9 (27-96)67.7 (31-89)68.1 (27-96)0.43 Median (years)686967Sex0.74 Male333 (80.8)167 (81.5)167 (81.5)ECOG performance status at baseline0.10 0174 (42.2)85 (41.5)89 (43.0)0.75 1185 (44.9)91 (44.4)94 (45.4)0.84 ≥247 (11.4)23 (11.2)24 (11.6)0.90 Missing6 (1.5)6 (2.9)0 (0)0.01BCLC status at baseline0.02 BCLC A8 (1.9)6 (2.9)2 (1.0)0.17 BCLC B97 (23.5)58 (28.3)39 (18.8)0.02 BCLC C305 (74.0)141 (68.8)164 (79.2)0.02 BCLC D2 (0.5)0 (0)2 (1.0)0.50Child-Pugh score at baseline0.40 CP A316 (76.7)157 (76.5)159 (76.8)0.96 CP B90 (21.8)45 (22.0)45 (21.7)0.96 B747 (11.4)26 (12.7)21 (10.1)0.42 B828 (6.8)15 (7.3)13 (6.3)0.70 B915 (3.6)4 (2.0)11 (5.3)0.11 Missing6 (1.5)3 (1.5)3 (1.4)>0.99ALBI grade at baseline0.77 ALBI 1145 (35.2)69 (33.7)76 (36.7)0.52 ALBI 2232 (56.3)119 (58.0)113 (54.6)0.48 ALBI 329 (7.0)14 (6.8)15 (7.2)>0.99 Missing6 (1.5)3 (1.5)3 (1.4)>0.99Etiology0.01 ALD112 (27.2)55 (26.8)57 (27.5)0.87 Hepatitis B49 (11.9)28 (13.7)21 (10.1)0.27 Hepatitis C78 (18.9)27 (13.2)51 (24.6)0.003 MASLD86 (20.9)42 (20.5)44 (21.3)0.85 Unknown/others87 (21.1)53 (25.8)34 (16.4)0.02Presence of PVI/ extrahepatic spread Main portal vein invasion156 (38.0)86 (42.0)70 (34.0)0.09 Extrahepatic spread163 (39.8)80 (39.0)83 (40.1)0.89Hepatic dysfunction a Ascites (moderate)69 (16.7)38 (18.5)31 (15.0)0.36 Ascites (severe)8 (1.9)4 (2.0)4 (1.9)>0.99 Hepatic encephalopathy (moderate)5 (1.2)0 (0)5 (2.4)0.03Esophageal varices Screening prior to bevacizumab–n.a.175 (84.5)– No varices–n.a.87 (49.7)– Varices grade I-II–n.a.84 (48.0)– Varices grade III–n.a.4 (2.3)–AFP at baseline AFP ≥200 ng/mL158 (38.3)87 (42.4)71 (35.0)0.09 AFP ≥400 ng/mL137 (33.3)75 (36.6)62 (30.5)0.16 Missing4 (1.0)0 (0)4 (1.9)0.12 Median AFP (ng/ml)50.36648.00.08Previous treatment Any previous anti-cancer treatment244 (59.2)148 (72.2)96 (46.4)<0.001 Resection90 (21.8)53 (25.9)37 (17.9)0.05 Ablation68 (16.5)46 (22.4)22 (10.6)0.001 Liver transplantation4 (1.0)4 (1.9)0 (0)0.06 TACE139 (33.8)87 (42.4)52 (25.2)<0.001 SIRT56 (13.6)31 (15.1)25 (12.1)0.37 Radiation11 (2.7)6 (2.9)5 (2.4)0.75 a Grading of ascites and hepatic encephalopathy were performed according to child-pugh classification.Baseline characteristics of patients treated with first-line lenvatinib and atezolizumab plus bevacizumab.Abbreviations: AFP, alfa-feto protein; ALBI, Albumin-bilirubin grade; ALD, alcoholic liver disease; BCLC, Barcelona Clinic Liver Cancer; CPS, Child-Pugh score; ECOG, European Cooperative Oncology Group performance status; MASLD, metabolic dysfunction-associated steatotic liver disease; n.a., not available; SIRT, selective internal radiotherapy; TACE, transarterial chemoembolization. Baseline characteristics by treatment cohort Grading of ascites and hepatic encephalopathy were performed according to child-pugh classification. Baseline characteristics of patients treated with first-line lenvatinib and atezolizumab plus bevacizumab. Abbreviations: AFP, alfa-feto protein; ALBI, Albumin-bilirubin grade; ALD, alcoholic liver disease; BCLC, Barcelona Clinic Liver Cancer; CPS, Child-Pugh score; ECOG, European Cooperative Oncology Group performance status; MASLD, metabolic dysfunction-associated steatotic liver disease; n.a., not available; SIRT, selective internal radiotherapy; TACE, transarterial chemoembolization. The LEN cohort mainly included patients who initiated systemic therapy between 2017 and 2020. Following the IMbrave150 trial, AZ/BV replaced LEN as the SOC in the first-line setting. Consequently, most patients starting systemic therapy after 2020 received AZ/BV. As a result, no active decision against either drug was generally made, and baseline characteristics were mostly well balanced between the cohorts (Table 1). However, the rate of Barcelona Clinic Liver Cancer stage A–B was higher in the LEN cohort (30.6% vs. 19.8%, p=0.02), as was the rate of prior locoregional therapies (72.2% vs. 46.4%, p<0.001). Additional baseline characteristics by liver disease etiology and albumin-bilirubin (ALBI) grade are detailed in Supplemental Tables S1A, B, http://links.lww.com/HC9/B68. More than half of the patients had an Eastern Cooperative Oncology Group (ECOG) performance status ≥1 and impaired liver function (ALBI grade ≥2). By the data cutoff date, 29 (16.3%) patients treated with AZ/BV and 12 (6.7%) patients treated with LEN were still alive. Of these, 5 (2.4%) in the AZ/BV group and 1 (0.5%) in the LEN group remained on first-line treatment. The median treatment duration was comparable between the cohorts (AZ/BV: 4.9 mo; LEN: 4.0 mo, p=0.28). The median follow-up time was 12.0 months (range 0.4–47.8) in the AZ/BV group and 10.3 months (range 0.6–74.4) in the LEN group (Table 2). TABLE 2Treatment characteristics by treatment cohortOverall (n=412) n (%)Lenvatinib first-line (n=205) n (%)Atezolizumab plus bevacizumab first-line (n=207) n (%) p Radiologic response Overall response104 (25.2)49 (23.9)55 (26.6)0.53 Complete response13 (3.2)1 (0.5)12 (5.8)0.0015 Partial response91 (22.1)48 (23.4)43 (20.8)0.52 Disease control246 (59.7)119 (58.0)127 (61.4)0.49 Stable disease142 (34.5)70 (34.1)72 (34.8)0.89 Progressive disease94 (22.8)43 (21.0)51 (24.6)0.38 EOT before first imaging36 (17.6)36 (17.6)26 (12.6)0.16 No data available10 (2.4)7 (3.4)3 (1.4)0.22Treatment status Median treatment duration (months, range)4.1 (0–40.5)4.0 (0.2–40.3)4.9 (0–40.5)0.28 Median follow-up duration11.1 (0.4–74.4)10.3 (0.6–74.4)12.0 (0.4–47.8)0.88 Treatment ongoing6 (1.5)1 (0.5)5 (2.4)0.21 End of treatment406 (98.5)204 (99.5)202 (97.6)0.21 Complete response12 (3.0)1 (0.5)11 (5.4)0.003 Progression and/or death234 (57.6)114 (55.9)120 (59.4)0.47 Adverse events116 (28.6)77 (37.7)39 (19.3)<0.0001 Patient’s choice9 (2.2)4 (2.0)6 (2.9)0.54 Lost to follow-up35 (8.6)8 (3.9)26 (13.0)0.002Subsequent therapy Any subsequent therapy183 (44.4)104 (50.7)79 (38.2)0.04 Locoregional therapy47 (11.4)16 (7.8)31 (15.0)0.07 Surgical resection4 (1.0)0 (0)4 (1.9)0.13 Ablation7 (3.2)4 (2.0)3 (1.4)0.90 Liver transplantation3 (0.7)1 (0.5)2 (1.0)0.82 TACE17 (4.1)1 (0.5)16 (7.7)<0.001 SIRT8 (1.9)5 (2.4)3 (1.4)0.74 Radiation14 (3.4)5 (2.4)9 (4.3)0.54 Systemic second-line therapy155 (37.6)95 (46.3)60 (29.0)0.001 ICI therapy14 (3.4)9 (4.4)5 (2.4)0.29 Sorafenib94 (22.8)61 (29.8)33 (15.9)0.0008 Lenvatinib15 (3.6)0 (0)15 (7.2)<0.0001 Cabozantinib21 (5.1)12 (5.9)9 (4.3)0.51 Regorafenib3 (0.7)3 (1.5)0 (0)0.12 Ramucirumab9 (2.2)9 (4.4)0 (0)0.0017 Best supportive care163 (39.6)81 (39.5)82 (39.6)>0.99 Note: Treatment characteristics according to1 first-line lenvatinib and atezolizumab plus bevacizumab.Abbreviations: EOT, end of treatment; ICI, immune checkpoint inhibition; SIRT, selective internal radiotherapy; TACE, transarterial chemoembolization. Treatment characteristics by treatment cohort Note: Treatment characteristics according to1 first-line lenvatinib and atezolizumab plus bevacizumab. Abbreviations: EOT, end of treatment; ICI, immune checkpoint inhibition; SIRT, selective internal radiotherapy; TACE, transarterial chemoembolization. Therapy effectiveness Best radiologic responses by treatment cohort and etiology are presented in Table 2 and Supplemental Table S4, http://links.lww.com/HC9/B68, respectively. Complete responses were generally rare but more frequent in the AZ/BV group compared to the LEN group (5.8% vs. 0.5%, p=0.0015). The overall response rate (ORR) and disease control rate were similar between the AZ/BV and LEN cohorts (ORR: 26.6% vs. 23.9%; disease control rate: 61.4% vs. 58.0%). Among patients still alive at data cutoff, 3 patients treated with AZ/BV and 1 patient in the LEN cohort had an objective response as their best radiologic outcome. Treatment discontinuation prior to the first radiological assessment occurred at similar rates (AZ/BV: 12.6% vs. LEN: 17.6%, p=0.16). The primary reasons for stopping treatment were clinical progression or death (AZ/BV: 59.4% vs. LEN: 55.9%, p=0.47) and adverse events (AZ/BV: 19.3% vs. LEN: 37.7%, p<0.0001). Loss to follow-up before the end of first-line treatment was more frequent in the AZ/BV cohort than in the LEN group (13.0% vs. 3.9%, p=0.002). Safety analysis Serious adverse events classified as Common Terminology Criteria for Adverse Events grade ≥3 were reported in 147 (71.0%) patients treated with AZ/BV and 120 (58.5%) patients treated with LEN (p=0.008) (Table 3). Overall, bleeding complications were more frequent in the AZ/BV group compared to the LEN group (17.9% vs. 2.9%, p<0.0001), including variceal bleeding (3.9% vs. 1.0%, p=0.11), nonvariceal gastrointestinal bleeding (5.3% vs. 1.0%, p=0.02), and non-gastrointestinal bleeding events (8.7% vs. 1.0%, p=0.0003). Upper gastrointestinal-endoscopy dates prior to treatment initiation were available for 175 patients treated with AZ/BV (84.5%). Among these, 4 (2.3%) were diagnosed with grade III esophageal varices, 84 (48.0%) had grade I–II varices, and 87 (49.7%), varices were ruled out. These data were not available for the LEN cohort. Varices were managed according to local SOC. Thromboembolic events were more common in the AZ/BV group than in the LEN group (5.3% vs. 1.0%, p=0.02). Immune-related adverse events Common Terminology Criteria for Adverse Events grade ≥3 occurred in 34 patients (16.4%) in the AZ/BV cohort. In the LEN cohort, most adverse events were related to common TKI side effects, including hand-foot syndrome (3.4% vs. 0.5%, p=0.04), diarrhea (4.9% vs. 1.0%, p=0.02), and worsening general condition/fatigue (10.2% vs. 3.4%, p=0.006). Severe ascitic decompensation and severe HE requiring hospitalization occurred at similar rates in both treatment groups (ascites: 11.6% [AZ/BV] vs. 7.3% [LEN], p=0.14; HE: 1.4% [AZ/BV] vs. 2.0% [LEN], p=0.72). TABLE 3Frequency of adverse events CTCAE grade ≥3 by treatment cohortOverall (n=412) n (%)Lenvatinib first-line (n=205) n (%)Atezolizumab plus bevacizumab first-line (n=207) n (%) p Any CTCAE grade ≥3267 (64.8)120 (58.5)147 (71.0)0.008Bleeding complications43 (10.4)6 (2.9)37 (17.9)<0.0001 Variceal upper GI bleeding10 (2.4)2 (1.0)8 (3.9)0.11 Nonvariceal GI bleeding13 (3.2)2 (1.0)11 (5.3)0.02 Non-GI bleeding20 (4.9)2 (1.0)18 (8.7)0.0003Thromboembolism12 (2.9)2 (1.0)10 (4.8)0.04 Pulmonary embolism5 (1.2)2 (1.0)3 (1.4)>0.99 Others7 (1.7)0 (0)7 (3.4)0.01Immune-related AEs Hepatitis—NA6 (2.9)— Arthritis and myositis—NA5 (2.4)— Hypophysis-adrenal-axis—NA1 (0.5)— Colitis—NA2 (1.0)— Nephritis—NA4 (1.9)— Vasculitis—NA2 (1.0)— Pneumonitis—NA4 (1.9)— Skin and mucosa disorders—NA4 (1.9)— Thyroiditis—NA5 (2.4)— Neuropathy—NA1 (0.5)—Severe ascites39 (9.5)15 (7.3)24 (11.6)0.14Severe HE7 (1.7)4 (2.0)3 (1.4)0.72Worsening general condition/fatigue28 (6.8)21 (10.2)7 (3.4)0.006Others75 (18.2)49 (23.9)26 (12.6)0.003Diarrhea12 (2.9)10 (4.9)2 (1.0)0.02HFS8 (1.9)7 (3.4)1 (0.5)0.04Hypertension a 9 (2.2)6 (2.9)3 (1.4)0.34 Note: Frequency of adverse events of CTCAE grade ≥3 in patients treated with first-line lenvatinib and atezolizumab plus bevacizumab a Only Hypertension Grade ≥4.Abbreviations: AEs, adverse events; CTCAE, Common Terminology Criteria for Adverse Events; GI, gastrointestinal; HE, hepatic encephalopathy; HFS, hand-foot syndrome; NA, not applicable. Frequency of adverse events CTCAE grade ≥3 by treatment cohort Note: Frequency of adverse events of CTCAE grade ≥3 in patients treated with first-line lenvatinib and atezolizumab plus bevacizumab Only Hypertension Grade ≥4. Abbreviations: AEs, adverse events; CTCAE, Common Terminology Criteria for Adverse Events; GI, gastrointestinal; HE, hepatic encephalopathy; HFS, hand-foot syndrome; NA, not applicable. Survival analysis The mOS was similar between the treatment groups (AZ/BV: 13.0 months [95% CI: 9.8–16.2]; LEN: 12.5 months [95% CI: 10.5–14.4]; HR 0.85 [95% CI: 0.68–1.07], p=0.16) (Figure 1A). However, mPFS was significantly longer in the AZ/BV group (6.6 mo [95% CI: 5.0–8.2] vs. 5.8 mo [95% CI: 4.9–6.7]; HR 0.73 [95% CI: 0.58–0.90], p=0.004) (Figure 1B). Univariate analysis of prognostic factors for mOS and mPFS are presented in Figure 2A, B, respectively. Patients treated with AZ/BV had a lower risk of progression regardless of prior local therapies (no prior local therapy: HR 0.70 [95% CI: 0.49–0.99], p=0.045; prior local therapy: HR 0.72 [95% CI: 0.54–0.96], p=0.03). Kaplan-Meier analysis of (A) overall survival and (B) progression-free survival for patients treated with first-line LEN and AZ/BV. Abbreviations: AZ/BV, atezolizumab plus bevacizumab; LEN, lenvatinib; mOS, median overall survival; mPFS, median progression-free survival; No., number. Forrest plot of unadjusted HRs for (A) overall survival and (B) progression-free survival in patients treated with first-line LEN and AZ/BV. Abbreviations: AFP, alfa-feto protein; ALBI, albumin-bilirubin grade; ALD, alcohol-associated liver disease; AZ/BV, atezolizumab plus bevacizumab; CPS, Child-Pugh score; ECOG, European Cooperative Oncology Group performance status; EHS, extrahepatic spread; MASLD, metabolic dysfunction–associated steatotic liver disease; n, number of events reached/total of patients; PVI, portal vein invasion. A longer PFS was also observed for AZ/BV in patients with ECOG 0 (HR 0.59 [95% CI: 0.42–0.83], p=0.003), viral etiology (HR 0.61 (95% CI: 0.41–0.92], p=0.019), absence of extrahepatic spread (HR 0.63 [95% CI: 0.47–0.84], p=0.011), presence of portal vein invasion (HR 0.54 [95% CI: 0.38–0.79], p=0.001), higher levels of alpha-fetoprotein (HR 0.66 (95% CI: 0.47–0.94], p=0.017), and younger age (HR 0.74 [95% CI: 0.57–0.95], p=0.017]. Multivariate analysis confirmed that impaired baseline liver function and poor ECOG performance status were independent prognostic factors for both OS and PFS. In contrast, portal vein invasion, extrahepatic spread, and alcohol-related liver disease (ALD) were only associated with shorter OS. For frontline treatment, AZ/BV therapy was an independent prognostic factor for longer PFS but not for OS (Supplemental Tables S2A, B, http://links.lww.com/HC9/B68). Further analysis examined the impact of etiology of the underlying liver disease and baseline liver function according to ALBI grade on OS and PFS (Supplemental Figures S1A–F, http://links.lww.com/HC9/B68, S2A–D, http://links.lww.com/HC9/B68). Patients with viral etiology showed a significantly lower risk of death under AZ/BV (HR 0.74 [95% CI: 0.56–0.96], p=0.0026). Additionally, patients with viral-HCC and those with impaired liver function (ALBI grade 2) had a significantly longer PFS with AZ/BV compared to LEN (HR 0.63 [95% CI: 0.42–0.94], p=0.017; and HR 0.70 [95% CI: 0.52–0.93], p=0.016), respectively. In the AZ/BV cohort, mOS was longer for patients with viral and MASLD-related HCC compared to those with ALD (HR 0.45 [95% CI: 0.30–0.67], p<0.001; and HR 0.57 [95% CI: 0.37–0.90], p=0.016), respectively (Figure 3S). Similarly, both cohorts showed longer mPFS compared to those with ALD [HR 0.52 [95% CI: 0.35–0.77], p=0.001; and HR 0.70 [95% CI: 0.45–1.09], p=0.115), respectively (Figure 3B). In contrast, no etiology-associated differences in mOS or mPFS were observed in LEN-treated patients (Figure 3C, D). Kaplan-Meier analysis of (A, C) overall survival and (B, D) progression-free survival for patients treated with first-line LEN and AZ/BV, stratified by etiology. Abbreviations: ALD, alcohol-associated liver disease; AZ/BV, atezolizumab plus bevacizumab; LEN, lenvatinib; MASLD, metabolic dysfunction–associated steatotic liver disease; mOS, median overall survival; mPFS, median progression-free survival; No., number. Propensity score matching analysis To mitigate the influence of baseline characteristics, a propensity score matching analysis was conducted, matching 155 patients based on age, gender, HCC etiology, prior locoregional therapy, ALBI grade, alfa-feto protein level, ECOG performance status, Barcelona Clinic Liver Cancer stage, portal vein invasion, and extrahepatic spread (Supplemental Table S3, http://links.lww.com/HC9/B68). The overall balance test yielded a p-value of p=0.959, indicating no remaining unbalanced covariates (Cohen d cutoff 0.25). After matching, there was no significant difference in mOS between the treatment groups (AZ/BV: 13.7 [95% CI: 10.5–16.9] months vs. LEN: 12.5 [95% CI: 10.5–14.5] months; HR 0.82 [95% CI: 0.64–1.07], p=0.15). However, mPFS was notably longer in the AZ/BV group (7.6 [95% CI: 5.8–9.3] months vs. 5.8 [95% CI: 4.9–6.7] months; HR 0.68 [95% CI: 0.53–0.87], p=0.002), further confirming the results from the unmatched population (Supplemental Figures S3A, B, http://links.lww.com/HC9/B68). Impact on hepatic dysfunction ALBI scores at the end of first-line treatment (EOT) were available for 157 (77.7%) patients in the AZ/BV group and 166 (81.4%) patients in the LEN group. An increase in ALBI scores was observed in both treatment groups (AZ/BV: ALBI baseline : −2.40±0.58 and ALBI EOT : −1.97±0.74, p<0.001; LEN: ALBI baseline : −2.29±0.58 and ALBI EOT : −1.99±0.72, p<0.001). However, the worsening in liver function (ΔALBI baseline - EOT ) was more pronounced in the AZ/BV group compared to the LEN group (−0.44±0.53 vs. −0.30±0.53, p=0.008) (Figure 4A). Deterioration in liver function was associated with a shorter mOS in both groups (AZ/BV: HR 1.36 [95% CI: 10.3–17.0], p=0.027; LEN: HR 1.23 [95% CI: 1.00–1.50], p=0.047) (Figure 4B). To understand whether the worsening of ALBI scores was driven by changes in albumin or bilirubin levels, both parameters were analyzed individually. The difference in albumin from baseline to EOT (Δalbumin baseline - EOT ) was significantly more pronounced in the AZ/BV group than in the LEN group (−3.91 g/L±4.96 vs. −2.63 g/L±5.65, p=0.030). Changes in bilirubin levels, although more pronounced in the AZ/BV group, did not reach statistical significance (−17.0 µmol/L±50.5 vs. −10.4 µmol/L±29.3, p=0.14). Mean ALBI score at baseline and EOT and respective change (Δ) according to (A) treatment cohort, (C) ALBI grade, and (D) etiology. The gray area indicates the range of ALBI grade 2 (>−2.60 to ≤−1.39). Kaplan-Meier analysis of overall survival for patients with (B) worsening ALBI score, (E) worsening ascites, and (F) worsening HE at EOT. Abbreviations: ALBI, albumin-bilirubin; ALD, alcohol-associated liver disease; AZ/BV, atezolizumab/bevacizumab; EOT, end of treatment; LEN, lenvatinib; MASLD, metabolic dysfunction–associated steatotic liver disease; mOS, median overall survival; No., number; w/o, without; w/, with; wors., worsening. Changes in ALBI score at EOT (ΔALBI baseline - EOT ) were independent of the baseline ALBI grade (ALBI 1: −0.40±0.51 vs. ALBI 2: −0.36±0.51, p=0.50) (Figure 4C) and were largely independent of etiology, except for a significant difference between viral hepatitis and ALD (−0.28±0.58 vs. −0.48±0.51, p=0.01) (Figure 4D). Data on ascites and HE at EOT was available for 174 (86.1%) patients treated with AZ/BV and 174 (85.3%) patients treated with LEN and classified as absent, moderate or severe whenever hospitalization for paracentesis was necessary. Overall, aggravation of ascites at EOT was similar between both treatment cohorts (AZ/BV: 24.3% vs. LEN: 21.1%, p=0.47) (Supplemental Figure S4A, http://links.lww.com/HC9/B68), including severe ascitic decompensation (11.6% vs. 7.3%, p=0.14) (Table 3). Notably, within the AZ/BV group, patients with baseline ALBI grade 2 were significantly more likely to experience worsening of ascites at EOT compared to those with ALBI grade 1 (31.3% vs. 12.5%, p=0.002), with only a trend observed in the LEN cohort (ALBI 1: 17.6% vs. ALBI 2: 21.8%, p=0.57) (Supplemental Figure S4B, http://links.lww.com/HC9/B68). Subgroup survival analysis revealed that mOS was significantly shorter in patients treated with AZ/BV with worsening ascites (17.3 [95% CI: 13.5–21.2] months vs. 8.6 [95% CI: 5.9–11.3] months; HR 1.55 [95% CI: 1.29–1.89], p<0.001), but not in the LEN group (12.6 [95% CI: 10.4–14.9] months vs. 9.4 [95% CI: 4.4–14.4] months; HR 1.04 [95% 0.86–1.27], p=0.65) (Figure 4E). Overall, the occurrence of worsening HE at EOT was similar in both treatment cohorts (Supplemental Figure S4C, http://links.lww.com/HC9/B68). However, within the AZ/BV group, only patients with impaired baseline liver function were at risk of HE under treatment (ALBI 1: 0% vs. ALBI 2: 13.4%, p<0.001), whereas in the LEN group, worsening of HE occurred independently of baseline ALBI grade (ALBI 1: 7.8% vs. ALBI 2: 10.9%, p=0.30) (Supplemental Figure S4D, http://links.lww.com/HC9/B68). The mOS was significantly shorter for patients with worsening HE in the AZ/BV cohort (15.0 mo [95% CI: 11.8–18.2] vs. 3.1 mo [95% CI: 0.0–9.9]; HR 1.82 [95% CI: 1.37–2.42], p<0.001), but this was not observed in LEN-treated patients (12.5 mo [95% CI: 10.2–14.7) vs. 8.4 mo [95% CI: 3.4–13.6]; HR 1.09 [95% CI: 0.84–1.40], p=0.52) (Figure 4F). Therapeutic sequencing At data cutoff date, 5 (2.4%) patients treated with AZ/BV and 1 (0.5%) patient treated with LEN remained on first-line therapy. Among those who discontinued first-line therapy (excluding patients with complete response or those lost to follow-up), patients treated with LEN stopped treatment more frequently due to adverse events (37.3% vs. 19.3%, p<0.0001), but were more likely to receive subsequent systemic therapies (46.3% vs. 29.0%, p=0.001) (Table 2). There was a trend for patients treated with AZ/BV to receive more often local therapies (5.0% vs. 7.8%, p=0.07), particularly transarterial chemoembolization (7.7% vs. 0.5%, p<0.001) (Table 2). In line with the approval and reimbursement policy in Germany, SOR was the most frequently used second-line therapy in both cohorts (LEN: 29.8% vs. AZ/BV: 15.9%, p=0.0008). Patients receiving at least 2 lines of therapy had longer mOS compared to those who received only the best supportive care, although there was no significant difference between both cohorts (Figure 5A, B). The mOS for patients treated with at least 2 lines of therapy was 17.3 (95% CI: 14.6–20.1) months for AZ/BV vs. 15.7 (95% CI: 12.2–19.2) months for LEN (HR 0.87 [95% CI: 0.60–1.26], p=0.45). For patients who received best supportive care after first-line therapy, mOS was 7.2 (95% CI: 5.0–9.3) months for AZ/BV and 6.2 (95% CI: 5.0–7.3) months for LEN (HR 1.26 [95% CI: 0.90–1.75], p=0.18) (Figure 5B). Kaplan-Meier analysis of (A) overall survival in patients receiving 2L and (B) patients receiving best supportive care (BSC) after first-line LEN and AZ/BV. Abbreviations: AZ/BV, atezolizumab/bevacizumab; 2L, second-line systemic therapy; LEN, lenvatinib; mOS, median overall survival; No., number. DISCUSSION Following the favorable outcomes observed in the IMbrave150 trial, 27 the combination of AZ/BV has been established as the SOC in first-line systemic therapy for advanced HCC, alongside durvalumab and tremelimumab.13,28,29 LEN is nowadays primarily considered for patients who are ineligible for ICI therapy or for use in subsequent lines of treatment. However, the notable survival of 19 and 20.7 months in the LEN control arms of the recent LEAP-002 and CHECKMATE 9DW trials, paired with longer mPFS in the REFLECT, LEAP-002, and CHECKMATE 9DW studies compared to AZ/BV in IMbrave150, have reinforced the value of LEN as an option in first-line HCC therapy.3,6,7 A direct comparison between AZ/BV and LEN through randomized trials is currently unavailable, and the likelihood of a prospective head-to-head trial is extremely low. While recommendations in national and international guidelines primarily rely on evidence from pivotal trials, RWD that mirrors clinical scenarios can provide valuable insights and potentially influence treatment decisions. The discrepancy between trial populations and patients encountered in everyday clinical practice is illustrated by factors such as the proportion of patients with well-maintained liver function. In phase III trials, over 50% of patients typically present with ALBI grade 1, whereas in our large Caucasian RW cohort, this proportion drops to 34%.30,31 Additionally, the proportion of HBV-positive patients is usually larger in global trials compared to European multicenter cohorts (eg,12.9% in our cohort vs. 48% in IMbrave150). 27 Our retrospective RWD also encompasses patients diagnosed with CP-B liver cirrhosis and an ECOG status of 2, which are commonly excluded from clinical trials in HCC, confirming the effectiveness and manageable toxicity of frontline treatment even in this more frail patient cohort. 32 Our study aligns with recent RW findings from Asia,21,22 which also reported longer PFS with AZ/BV compared to LEN. However, this PFS benefit did not translate into a survival benefit. As expected, multivariate analysis confirmed established negative prognostic factors such as poor ECOG performance status, the presence of extrahepatic metastases, and impaired liver function. Regarding etiology, we observed poor outcomes in patients with ALD-related HCC. While inclusion criteria remain pivotal for assessing the efficacy of therapies, RWD shape expectations in routine clinical practice. Our analysis not only underscores the prognostic significance of the ALBI grade but also suggests that the onset of clinical decompensation during therapy may affect survival differently based on the treatment approach. A longitudinal decline in the ALBI score was associated with worse OS in both AZ/BV and LEN-treated patients, whereas ascitic decompensation or onset of encephalopathy led to a decline in OS only in the AZ/BV cohort. It remains speculative whether this observation is related to AZ or BV, but it highlights the importance of longitudinal assessment not only of biochemical liver function but also clinical signs of hepatic deterioration. The underlying cause is most likely multifactorial and potentially related to liver disease, extent of fibrosis/cirrhosis, side effects and mechanisms of action of the respective drugs, or due to tumor progression. Our results raise the question whether a decline in liver function, including clinical signs of early hepatic decompensation under treatment, should trigger prompt adjustment of therapy in patients treated with AZ/BV. The observed bleeding events in our AZ/BV cohort were consistent with other RW cohorts, supporting the need for a diagnostic upper endoscopy before initiating anti-VEGF therapy.32,33 Based on observations in preclinical models of MASLD-induced HCC, it has been postulated that these patients may be less responsive to immunotherapy, 34 which was supported by a significant survival benefit for LEN compared to AZ/BV in a large RW study focused on patients with MASLD from Europe and Asia (Korea, Japan) with excellent liver function (95% ALBI 1). 23 In contrast, in our study, we observed a trend toward superior survival in patients with MASLD treated with AZ/BV compared to LEN. Thus, at this time, no conclusive evidence exists for a negative predictive value of steatotic liver disease for patients treated with checkpoint inhibitors. Given the complexities of post hoc analyses and recent data indicating that harmful drinking can be observed in up to 30% of obese patients, the distinction between MASLD and ALD is challenging. 35 Despite these limitations, our RW study aimed to identify patients with dominant MASLD versus ALD and specifically observed a poor outcome in the ALD group treated with AZ/BV. mOS was not only shorter in patients with ALD treated with AZ/BV compared to LEN-treated patients, but also, within the AZ/BV group, the risk of death was significantly higher in patients with ALD compared to those with MASLD. At this point, it is premature to conclude that patients with ALD benefit less from ICI-based therapies, but our data highlight the need for a more granular separation of patients with nonviral HCC in prospective trials (MASLD/metabolic dysfunction–associated steatotic liver disease with greater alcohol consumption/ALD). As expected, mOS of patients who received second and further lines of therapy was longer than in those who received only first-line treatment. Despite recent discussions on downstaging from systemic to local therapies, almost all patients remained on systemic treatments, suggesting that this concept has not yet been implemented in clinical practice. Ongoing prospective trials, such as ABC-HCC (NCT 04803994), may help define which patients should start on local versus systemic therapy. Additionally, with higher ORRs reported from trials combining local and systemic therapies (EMERALD-1; ORR>40%), as well as from the CheckMate9DW study for the dual ICI combination nivolumab/ipilimumab (ORR 36%), the number of patients able to receive subsequent curative therapies may be further increased.7,36 Our study has several limitations, including its retrospective nature and the lack of randomization. The analysis was limited in power due to the small sample size and the potential risk of type I error, and the time lag in treatment initiation could have impacted survival outcomes due to the greater availability of subsequent therapies. Our study lacks quality-of-life data and a comprehensive assessment of adverse events, including detailed radiologic assessment at specific time points of decompensation. Additional important clinical parameters, such as the presence of cirrhosis and tumor burden, were not available and could have substantially influenced the outcomes. Furthermore, the prospective assessment of alcohol consumption, which could provide better insight into the poor prognosis associated with ALD, was lacking. Finally, data on severe COVID-19 infections or missed treatment visits during the pandemic were not available. In conclusion, our findings support the clinical effectiveness of both AZ/BV and LEN as first-line therapies for advanced HCC. In the RW setting, PFS was significantly longer in the AZ/BV cohort, although no significant differences were observed in OS. The risk of hepatic decompensation increased in patients with impaired baseline liver function, particularly for those with ALBI grade 2 treated with AZ/BV. LEN continues to be a valid alternative to ICI-based combinations for first-line treatment of HCC. Given the growing number of first-line therapy options, identifying clinical or molecular biomarkers to guide treatment allocation remains an unmet need. Supplementary Material AUTHOR CONTRIBUTIONS Tiago de Castro and Sabrina Welland contributed to data curation, investigation, formal analysis, methodology, data analysis, visualization, writing—original draft; Arndt Vogel and Anna Saborowski contributed to conceptualization, methodology, supervision, writing—review and editing; Leonie Jochheim, Cathrine Leyh, Kateryna Shmanko, Fabian Finkelmeier, Petia Jeliazkova, Andre Jefremow, Maria A. Gonzalez-Carmona, Arne Kandulski, Daniel Rössler, Najib Ben Khaled, Stefan Enssle, Marino Venerito, Thorben W. Fründt, Michael Schultheiß, Andreas Schmiderer, Maria Pangerl, Andreas Maieron, Jens U. Marquardt, Thomas C. Wirth, Richard Greil, Christina Fricke, Rainer Günther, Angela Djanani, Dominik Bettinger, Henning Wege, Bernhard Scheiner, Martina Müller, Christian P. Strassburg, Jürgen Siebler, Ursula Ehmer, Oliver Waidmann, Arndt Weinmann, Christian M. Lange and Matthias Pinter contributed to data curation and writing—review and editing. CONFLICTS OF INTEREST Tiago de Castro received speaker honoraria from AstraZeneca and BMS, and travel support from MSD. Sabrina Welland received speaker honoraria from AstraZeneca. Leonie Jochheim received speaker honoraria from Falk Foundation, Boston scientific, Abbvie, AstraZeneca, and travel support from Biotest and Abbvie; she is consultant for Boston scientific and AstraZeneca. Fabian Finkelmeier received speaker honoraria from AbbVie, MSD, Ipsen, Astra, and travel support from Ipsen, Abbvie, AstraZeneca. Maria A. Gonzalez-Carmona received advisory honoraria from BMS, AZ, Lilly, Roche, MSD, Eisai, Amgen and travel support from AstraZeneca, Lilly, and BMS. Arne Kandulski received advisory honoraria and travel support from Roche, Eisai, Abbvie, Janssen-Cilag, MSD, Boston Scientific, Fujifilm, Micro-Tech Germany, and Bayer. Daniel Roessler received advisory honoraria from Bayer and Ipsen; he received grants from Ipsen. Najib Ben Khaled has received reimbursement of meeting attendance fees and travel support from EISAI and lecture honoraria from the Falk Foundation and AstraZeneca. Marino Venerito received speaker, consultancy, and advisory honoraria from Servier, Roche, BMS, MSD, EISAI, Bayer, Lilly, AstraZeneca, Merck Serono, Sirtex, Ipsen, Nordic Pharma, and Amgen. Michael Schultheiß received speaker honoraria from Falk Foundation, W. L. Gore & Associates, Roche, and Bentley InnoMed. He is a consultant for Bayer. Andreas Maieron received speaker honoraria from Eisai, MSD, and Roche and travel support from Roche; he is a consultant for AstraZeneca, Eisai, Ipsen, MSD, and Roche. Jens U. Marquardt received speaker and advisory honoraria from Roche, MSD, and Eisai, and travel support from AstraZeneca and Ipsen. Thomas C. Wirth is a consultant for Ipsen, Roche, Novartis, BMS, MSD, Daiichi Sankyo, Servier, Merck, Astrazeneca, Pierre Fabre, and Viatris and received grants from Hookipa Biotech and BMS. Richard Greil owns stocks or other ownership from Novo Nordisk and Lilly; he also received honoraria from Celgene, Roche, Merck, Takeda, AstraZeneca, Novartis, Amgen, BMS, MSD, Sandoz, Abbvie, Gilead, Daiicho Sankyo, and Sanofi, and research funding from Celgene, Roche, Merck, Takeda, AstraZeneca, Novartis, Amgen, BMS, MSD, Sandoz, Abbvie, Gilead, and Daiichi Sankyo, and travel support from Roche, Amgen, Janssen, AstraZeneca, Novartis, MSD, Celgene, Gilead, BMS, Abbvie, and Daiichi Sankyo, and consulting honoraria from Celgene, Novartis, Roche, BMS, Takeda, Abbvie, AstraZeneca, Janssen, MSD, Amgen, Merck, Gilead, Daiichi Sankyo, and Sanofi. Richard Greil received advisory honoraria from Roche and Eisai. Dominik Bettinger received speaker honoraria from W. L. Gore & Associates and travel support from Gilead. Henning Wege received consulting and speaker honoraria from Roche, AstraZeneca, and Eisai. Bernhard Scheiner received grant support from AstraZeneca and Eisai, speaker honoraria from Eisai and travel support from AbbVie, AstraZeneca, Ipsen, and Gilead. Jürgen Siebler received speaker honoraria from Ipsen, advisory honoraria from BMS, Lilly, and MSD, and travel support from Celgene. Ursula Ehmer received speaker honoraria from AstraZeneca, Falk Foundation, Ipsen, MSD, and Novartis, and travel support from AstraZeneca; she is a consultant for AstraZeneca, Bayer, Eisai, and MSD. Oliver Waidmann is a consultant and received speaker and investigator honoraria from AstraZeneca, Bayer, Eisai, MSD, and Roche. Arndt Weinmann received speaker honoraria from Eisai, Ipsen and MSD; he is a consultant for Bayer, BMS, Sanofi, Roche, AstraZeneca, and Servier. Matthias Pinter received speaker honoraria from Bayer, BMS, Eisai, Lilly, MSD, and Roche and travel support from Bayer, BMS, Ipsen, and Roche; he is a consultant for AstraZeneca, Bayer, BMS, Eisai, Ipsen, Lilly, MSD, and Roche. Christian M. Lange has received advisory and speaker honoraria from AbbVie, AstraZeneca, Boston Scientific, CSL Behring, Eisai, Falk, Gilead, MSD, Norgine, Novartis, Roche, Shionogi, and Sobi. Anna Saborowski is a consultant and received speaker honoraria from Eisai, Roche, Servier, Ipsen, Lilly, AstraZeneca, MSD, and travel support from Ipsen, Servier, Pierre-Fabre, and MSD. Arndt Vogel received personal fees from Roche, Bayer, BMS, Lilly, EISAI, AstraZeneca, Ipsen, MSD, Sirtex, BTG, Servier, Terumo, and Imaging Equipment Ltd (AAA). Cathrine Leyh, Kateryna Shmanko, Petia Jeliazkova, Andre Jefremow, Thorben W. Fründt, Angela Djanani, Maria Pangerl, Christina Fricke, Andreas Schmiderer, Stefan Enssle, Martina Müller, and Christian P. Strassburg have no conflicts to report. Footnotes Tiago de Castro and Sabrina Welland contributed equally and shared the first authorship. Anna Saborowski and Arndt Vogel contributed equally and shared the last authorship. Abbreviations: ALBI, albumin-bilirubin; ALD, alcohol-associated liver disease; AZ/BV, atezolizumab/bevacizumab; ECOG, Eastern Cooperative Oncology Group; EOT, end of treatment; ICI, immune checkpoint inhibitors; LEN, lenvatinib; MASLD, metabolic dysfunction–associated steatotic liver disease; mOS, median overall survival; mPFS, median progression-free survival; ORR, overall response rate; RW, real-world; RWD, real-world data; SOC, standard of care; SOR, sorafenib; TKI, tyrosine kinase inhibitor. Supplemental Digital Content is available for this article. Direct URL citations are provided in the HTML and PDF versions of this article on the journal’s website, www.hepcommjournal.com. Contributor Information Tiago de Castro, Email: decastro.tiago@mh-hannover.de. Sabrina Welland, Email: tiagucastro@gmail.com. Leonie Jochheim, Email: leonie.jochheim@uk-essen.de. Cathrine Leyh, Email: Catherine.Leyh@uk-essen.de. Kateryna Shmanko, Email: Kateryna.Shmanko@unimedizin-mainz.de. Fabian Finkelmeier, Email: fabian.finkelmeier@kgu.de. Petia Jeliazkova, Email: petia.Jeliazkova@mri.tum.de. Andre Jefremow, Email: andre.jefremow@googlemail.com. Maria A. Gonzalez-Carmona, Email: maria.gonzalez-carmona@ukbonn.de. Arne Kandulski, Email: Arne.Kandulski@klinik.uni-regensburg.de. Daniel Roessler, Email: daniel.roessler@med.uni-muenchen.de. Najib Ben Khaled, Email: najib.benkhaled@med.uni-muenchen.de. Stefan Enssle, Email: stefan.enssle@med.uni-muenchen.de. Marino Venerito, Email: m.venerito@med.ovgu.de. Thorben W. Fründt, Email: t.fruendt@uke.de. Angela Djanani, Email: Angela.Djanani@i-med.ac.at. Maria Pangerl, Email: maria.pangerl@uksh.de. Andreas Maieron, Email: andreas.maieron@stpoelten.lknoe.at. Thomas C. Wirth, Email: Wirth.thomas@mh-hannover.de. Jens U. Marquardt, Email: jens.marquardt@uksh.de. Richard Greil, Email: r.greil@salk.at. Christina Fricke, Email: Christina.Fricke@stpoelten.lknoe.at. Rainer Günther, Email: rainer.guenther@uksh.de. Andreas Schmiderer, Email: Andreas.Schmiderer@i-med.ac.at. Dominik Bettinger, Email: dominik.bettinger@uniklinik-freiburg.de. Henning Wege, Email: h.wege@klinikum-esslingen.de. Bernhard Scheiner, Email: bernhard.scheiner@meduniwien.ac.at. Martina Müller, Email: martina.mueller-schilling@ukr.de. Christian P. Strassburg, Email: christian.strassburg@ukbonn.de. Jürgen Siebler, Email: Juergen.Siebler@uk-erlangen.de. Ursula Ehmer, Email: ursula.ehmer@tum.de. Oliver Waidmann, Email: oliver.waidmann@chop-studien.de. Arndt Weinmann, Email: arndt.weinmann@unimedizin-mainz.de. Matthias Pinter, Email: matthias.pinter@meduniwien.ac.at. Christian M. Lange, Email: Christian.Lange@med.uni-muenchen.de. Anna Saborowski, Email: Saborowski.anna@mh-hannover.de. Arndt Vogel, Email: vogela@me.com. REFERENCES