Atezolizumab plus bevacizumab as first-line systemic therapy for hepatocellular carcinoma: a multi-institutional cohort study

Background: Atezolizumab plus bevacizumab is the standard of care for advanced hepatocellular carcinoma (HCC) in the first-line setting, although was only evaluated in patients with Child-Pugh (CP) A liver function in the IMbrave150 trial. We sought to determine the outcomes of these patients based on CP score and ALBI grade in the US population. Methods: This multicenter cohort study included patients with HCC who received atezolizumab with bevacizumab as first-line systemic therapy between March 2018 and November 2023. Overall survival (OS) was determined using the Kaplan-Meier method and multivariate analyses were performed using Cox proportional hazard regression method. Results: Among 322 patients, 226, 86, and 10 patients had CP-A, CP-B, and CP-C liver function, respectively. Median age was 66.5 years, 78.6% were male, and 82.6% were White. Median OS (mOS) was 21.6 months for those with CP-A, 9.1 months for those with CP-B7, and 4.7 months for those with CP-B8-C12 ( P < .0001). Among patients with CP-A, those with ALBI grade 1 had an mOS of 34.9 months versus 14.2 months in those with grade 2. In multivariate analyses, CP score, ALBI grade, hepatitis B, performance status, and macrovascular invasion were significantly associated with survival. Conclusions: CP score is an important prognostic tool for patients with HCC receiving atezolizumab plus bevacizumab, and this regimen remains a viable option for patients with CP-B7 with no additional safety concern, although the benefit is significantly less than those with CP-A. ALBI score has independent predictive value in patients with CP-A liver function.


Introduction
Approximately 90% of patients diagnosed with hepatocellular carcinoma (HCC) have some form of chronic liver disease. 1 The Child-Turcotte score was first developed in 1964 as a means to predict mortality of patients undergoing portocaval shunt surgery, and was computed using subjective factors including encephalopathy, nutritional status, and ascites, as well as objective factors including serum albumin and bilirubin. 2This score was later modified by Pugh to include the international normalized ratio (INR) in place of nutritional status, hence why the score is now commonly referred to as the Child-Pugh (CP) score. 3,4CP score has been incorporated in the Barcelona Clinic Liver Cancer staging system, which guides therapeutic options for patients with HCC based on tumor spread and liver function. 5ystemic therapy for advanced HCC has been primarily evaluated in patients with CP class A liver function due to concurrent mortality risk in patients with higher liver dysfunction.Currently, the combination of atezolizumab and bevacizumab is a standard of care option in the first-line setting for eligible patients with advanced HCC based on the IMbrave150 clinical trial, which demonstrated survival benefit compared to sorafenib. 6However, this trial was limited to patients with CP class A liver function, with 72% having CP-A5 and 28% having CP-A6 liver function.At this time, there is little evidence to support the use of this combination in patients with CP-B or CP-C liver function.Moreover, there is limited data regarding outcomes of treatment of HCC with atezolizumab plus bevacizumab in Western populations.In the IMbrave150 trial, 57% of patients were Asian and 35% of patients were White.Further, only 10 Black/African American patients were enrolled in this trial, limiting the generalizability of results in this patient population.Other FDA-approved first-line systemic therapeutic options include tremelimumab plus durvalumab, lenvatinib, and sorafenib, but these are less frequently used compared to atezolizumab plus bevacizumab.
More recently, the ALBI score has gained favor as a simplified model for liver function analysis.This score is calculated using serum albumin and bilirubin levels, and excludes subjective measures that are required to calculate a CP score. 7LBI score has been demonstrated to have prognostic value in patients with cirrhosis.ALBI has been shown to have a good correlation with CP and the model of the end-stage liver disease (MELD) scores. 8The predictive value of ALBI in assessing patient outcomes when undergoing treatment with atezolizumab plus bevacizumab for HCC is not well known at this time.
In this study, we conducted a multi-institution retrospective cohort study of patients with advanced HCC who received atezolizumab plus bevacizumab in the first-line setting in the United States and evaluated patients' outcomes based on multiple patient and tumor factors, with a focus on baseline liver function.

Study design and patient selection
We conducted a multisite, multi-institution retrospective analysis of patients seen within the Mayo Clinic Health system (with sites in Minnesota, Wisconsin, Iowa, Florida, and Arizona) and University Hospitals Siedman Cancer Center and Case Western Reserve University (with sites in Ohio).Patients were included who were ≥18 years of age, had a prior diagnosis of HCC and planned to receive atezolizumab plus bevacizumab as first-line systemic therapy from March 1, 2018, to November 1, 2023.Pathologic confirmation of HCC was not required if the diagnosis was based on the Liver Imaging Reporting and Data System. 9The keywords used to retrieve the data were International Classification of Diseases-10-CM code of "C20.0"and/or "hepatocellular cancer."Patients were excluded if they had mixed hepatocellular cancer-cholangiocarcinoma or had more than one primary malignancy requiring concurrent treatment.Exclusion criteria included patients who had received systemic therapy for HCC prior to receiving atezolizumab plus bevacizumab.All cases were reviewed at multidisciplinary tumor boards at respective sites and were considered having advanced disease not amenable to locoregional liver-directed therapy.
Eastern Cooperative Oncology Group (ECOG) performance score, laboratory data at the time of diagnosis including albumin, bilirubin, and INR, prior treatment history, and patient demographics were collected by retrospective review of electronic medical records.Response was assessed using Response Evaluation Criteria in Solid Tumors v 1.1.Response assessment was performed by 2 authors (M.H.S. and A.M.).This research was conducted ethically in accordance with the World Medical Association Declaration of Helsinki.The study protocol was reviewed and approved by Case Western Reserve University and Mayo Clinic Institutional Review Board.

Analysis
Patient demographics were compared across CP score levels.For categorical variables, frequencies (percentages) were calculated overall and within CP score levels.For continuous variables, mean and standard deviation (SD) were calculated overall and within CP score levels.Analysis of both overall survival (OS) and progression-free survival (PFS) was performed using the Kaplan-Meier method.OS was defined as the time from the date of first treatment with atezolizumab plus bevacizumab till death.If patients were alive at the last follow-up, they were censored on that date.PFS was defined as the time from the date of treatment to the time of progression or death from any cause.If a patient was alive and progression-free, they were censored at the date of last follow-up.Both OS and PFS were compared across CP liver status using the log-rank P-value.Additionally, a subgroup analysis of Child-Pugh A patients was performed to compare both OS and PFS based on ALBI scores.A univariate analysis was performed for both PFS and OS to identify prognostic factors.Factors to be included in the univariable analysis were determined a priori and the covariates that had a significance level of <0.2 were included in the candidate factors to be used in the multivariable analysis.A backward selection model (using P-value threshold for remaining in a model of .05) was performed to determine the final multivariable model for both OS and PFS.All statistical analysis was performed using SAS v9.4.

Cohort characteristics
We identified 322 patients with HCC who planned to receive atezolizumab plus bevacizumab as first-line systemic therapy, with calculable CP liver function.Fourteen patients were excluded from efficacy analysis due to the inability to calculate CP score based on available data, and 5 patients were excluded due to a synchronous, non-HCC malignancy.In our cohort, 22 (6.8%) patients received atezolizumab alone without bevacizumab primarily due to large varices that could not be treated in time to initiate systemic therapy.Median age was 66.5 years, 78.6% were male, and 82.6% were White.The study population included 28 (8.7%)Black/African American patients.Risk factors for HCC included hepatitis

Response and adverse events by baseline liver function
A complete response (CR) was seen in 3.4% of patients, including 4.4% with CP-A and 1.2% with CP-B liver function.A partial response (PR) was seen in 29.2%, 19.8%, and 10.0% with CP-A, CP-B, and CP-C, respectively, with stable disease (SD) in 32.7%, 33.7%, and 40.0%, respectively.A disease control rate of 66.4%, 54.7%, and 50% was seen in CP-A, CP-B, and CP-C, respectively.Responses are listed in Table 2.
Mean time on treatment was longest in patients with CP-A (8.0 months), followed by CP-B (4.5 months) and CP-C (3.5 months).Adverse events (AEs) experienced by patients included fatigue in 59.9% and hypertension in 19.6%, with grade 3 or greater hypertension in 3.7%.Sixteen percent of patients experienced an immune-related AE, with 4.7% experiencing a grade 3 or greater event.Bleeding occurred in 12.1% of the patients.Of these, 15 patients had a gastrointestinal bleed that was not clearly variceal, which included 4 with grade 1/2 bleeding, 10 with grade 3/4 bleeding, and 1 patient with a grade 5 bleed (of note, 3 patients with a grade 3 gastrointestinal bleed were not on bevacizumab at the time of bleed).Seven patients experienced variceal bleeding, all of which were grade 3. None of the patients with variceal bleed underwent prior embolization and 5 patients were receiving nonselective beta blockers at the time of bleeding event.Twelve patients experienced grade 1 epistaxis, 2 patient had bleeding from a tumor (one grade 2 prior to starting therapy and the other grade 3), and 3 patients had an intracranial bleed (two grade 3 and one grade 5).Thirty-seven percent of the patients received second-line systemic therapy, with most common among these including lenvatinib (22.8%), cabozantinib (3.4%), and ipilimumab plus nivolumab (2.5%).A higher proportion of patients with CP-A received second-line therapy compared to those with CP-B liver function (41% vs 30%).No patients with CP-C liver function received second-line systemic therapy.

Univariate and multivariate analysis for factors predictive of survival
In a univariate analysis for factors associated with OS, factors of negative prognostic value included higher CP score, higher ALBI score, non-White race, history of hepatitis B, ECOG performance status of 1 or 2, and macrovascular invasion (Table 3).Factors predictive of improved OS included prior surgery and prior embolization.In a univariate analysis for factors predictive of PFS, negative prognostic factors  S1).Factors predictive of improved PFS include prior embolization and prior radiotherapy.
In multivariate analyses, higher CP score and ALBI grade, hepatitis B, higher ECOG PS, and macrovascular invasion were significantly associated with poorer survival.There was a significant statistical interaction between the CP score and ALBI grade (Table 4).In particular, in patients with CP-A, there was significant prognostic value of ALBI grade vs grade 2. Evaluating factors predictive of PFS using multivariate analyses, factors associated with poorer PFS include higher CP score and female sex, while the only factor predictive of more favorable PFS was prior radiotherapy (Supplementary Table S2).

CP-A (N = 226) CP-B (N = 86) CP-C (N = 10) Total (N = 322)
Yes We also evaluated outcomes separately in Black/African American patients (n = 28).Median PFS and mOS were 6.2 and 12.8 months, respectively.Among 19 patients with CP-A liver function, mPFS and mOS were 6.2 months and 16.1 months, respectively.

Discussion
To our knowledge, this is the largest study to report real-world outcomes of patients with advanced HCC receiving atezolizumab with bevacizumab in the first-line setting.Further, this is one of the first studies to evaluate outcomes in a Western population.While only studied in patients with CP-A liver function, atezolizumab with bevacizumab is frequently used in patients with CP-B7 liver function and beyond in clinical practice, although until this point, data regarding patient outcomes in this population have been lacking.Additionally, ALBI liver function has been more recently developed as an objective method to quantify liver function, although its use in prognostication of patients with HCC receiving atezolizumab with bevacizumab has not been widely evaluated.In this study, we were able to provide real-world data regarding patient outcomes based on these metrics.
In the present study, those with CP-A had an mOS of 21.6 months, which is comparable to the 19.2 months seen in an updated analysis from the IMbrave150 clinical trial. 102][13][14][15] In the present study, further stratification of those with CP-B liver function revealed an mOS of 9.1, 5.5, and 4.0 months in those with CP-B7, CP-B8, and CP-B9, liver function, respectively, showing a significant decline in survival benefit between those with CP-B7, and those with CP-B8 and CP-B9 liver function.In a univariate analysis, we also show CP-B7 and CP-B8-C12 to be predictive of worse survival which is consistent with prior multivariate analyses. 16hile patients with CP-B8 and beyond may be candidates for atezolizumab plus bevacizumab, it is also important that they recognize that the survival benefit seen in IMbrave150 Downloaded from https://academic.oup.com/oncolo/advance-article/doi/10.1093/oncolo/oyae142/7709605 by guest on 29 July 2024 may be significantly reduced by baseline liver function.The same holds true for those with CP-B7 liver function, although this group still receives greater benefit than those with CP-B8 and beyond.It is important to note that we did not observe additional safety concerns in terms of AEs, immune-related AEs, or bleeding events in patients with CP-B7 liver function.
In addition to CP, we evaluated the role of ALBI liver score in assessing patient outcomes and found that those with grade 1 had the best survival of 34.9 months.Among patients with CP-A liver function, those with ALBI grade 1 had an mOS of 34.9 months compared to 14.2 months in those with ALBI grade 2, which shows that ALBI may serve a role for further stratification of prognosis in patients with CP-A liver function receiving atezolizumab with bevacizumab.When first developed, ALBI score was able to predict 2 prognostic subpopulations among patients with HCC and CP-A liver function, and we show that this holds true among patients receiving atezolizumab with bevacizumab. 7In multivariate analyses, we also show that ALBI grade has the ability to further predict outcomes among patients with CP-B7, which has previously been shown in a real-world analysis of patients with CP-B liver function receiving first-line atezolizumab with bevacizumab. 12n addition to baseline liver function, non-White race was predictive of poorer survival in univariate analysis.While the majority of patients in this study were White, which may limit statistical analysis to some extent, this is a concerning finding that merits further exploration.When looking at survival of 19 Black patients with CP-A liver function, mOS was 16.1 months, which was shorter than that observed in the overall population (21.6 months).In the 9 Black patients with CP-B  17 The reasons for these disparities in outcomes may include genetic variations, socioeconomic status, varying access to screening, and disparities in care.One study demonstrated that Black and Hispanic patients may be more likely to be diagnosed at an advanced stage of disease, which may reflect disparities in screening. 18o our knowledge, this study includes the largest series of Black patients treated with atezolizumab plus bevacizumab.Future trials should investigate outcomes specifically in minority population.In our study, response to atezolizumab plus bevacizumab was highly correlated with OS.One-year survival rates for patients achieving CR/PR, SD, and PD were 86.5%, 57.6%, and 31.7%,respectively.This study suggests that response can be considered as a surrogate measure for OS in HCC with immunotherapy.Patients who have progressive disease on first-line therapy should be considered for clinical trials using novel therapeutics as they are otherwise expected to have poor outcomes.
Overall, this study demonstrates variable outcomes in patients with HCC receiving atezolizumab plus bevacizumab based on baseline liver function.A key limitation of this study rests within the intrinsic limitations of the CP score, in that 2 of its components, ascites and encephalopathy, are subjective in nature.Within our review, we attempted to determine the volume of ascites based on imaging and physical findings, and encephalopathy based on clinical documentation, which is dependent upon the assessment of the provider documenting their visit with the patient.There can be variability between providers, and these factors may also be transient.On the other hand, the ALBI score is an objective measure, which can be calculated with certainty near the time of treatment.In our study, the concordance between provider documentation and investigator-assessed CP score was only 80% suggesting high rates of miscalculation of CP score in the clinical practice.Further, the CP score was missing from the clinical notes in one-third of the patients reflecting the need for standardizing oncology notes for patients with HCC.In our study cohort, patients with CP-B and CP-C liver function who received atezolizumab with bevacizumab are likely a selected population, who were deemed fit to undergo systemic therapy despite poor liver function and may lead to overestimation of the benefit of treatment in this population.Other limitations of the study include its retrospective nature and inclusion primarily of patients at high-volume academic centers, questioning the generalizability of its conclusions.However, in this study, patients were included from the sites at 6 different states including Minnesota, Wisconsin, Iowa, Ohio, Arizona, and Florida, broadening the applicability of the study results.Patients were treated at both academic and community sites, although most of the cases were included from tertiary care centers.

Conclusions
CP and ALBI liver function are both valuable methods to predict outcomes of patients receiving atezolizumab plus bevacizumab as first-line systemic therapy for advanced HCC.This regimen remains a viable option for patients with CP-B7 liver function with no additional safety concern, although the benefit is significantly less than those with CP-A liver function.ALBI has additional prognostic value, with 2 subpopulations with significantly different outcomes identified among patients with CP-A liver function and should be routinely calculated in clinical practice for patients receiving atezolizumab plus bevacizumab.Future trials should focus on underserved populations, especially Black patients, who may have differential outcomes.

Table 2 .
Response rate, adverse events, and second-line therapy by baseline liver function.

Table 3 .
Univariate analysis for factors predictive of overall survival.
liver function, mOS was 12.5 months.Prior data reporting outcomes of atezolizumab plus bevacizumab in this population has been highly limited to date.Non-White patients with HCC may experience poorer survival.