Clinical Characteristics, Treatment Patterns, and Outcomes of Patients With Locally Advanced/Metastatic Hepatocellular Carcinoma Treated at the Veterans Health Administration

Abstract

Purpose

This study retrospectively reviewed the outcomes of patients with advanced hepatocellular carcinoma (HCC) receiving atezolizumab with bevacizumab (A + B) therapy at the Veterans Health Administration (VHA).

Patients and Methods

Patients with advanced HCC who received first-line systemic therapy with A + B at the VHA between December 1, 2019, and March 1, 2022, were selected from electronic medical records (EMR) using ICD-9 and ICD-10 codes. Abstractors reviewed the EMR of the patients from their index date of A + B initiation until death or their last VHA visit, with the study period ending on January 31, 2023. The chi-square test was used to compare rates, and the Mann-Whitney test was used to compare medians.

Results

A total of 332 patients met the study criteria. The median age was 67 years; 99% were male, 63% were non-Hispanic Whites, 26% were Black, and 66% had an Eastern Cooperative Oncology Group performance status of ≥1. 84% had child Pugh score (CPS) class A, 16% had CPS classes B and C, 62% had a grade 2 albumin-bilirubin score, 56% had HCC caused by viral hepatitis, 80% had cirrhosis, and 67% had received prior local therapies. The 6-month progression-free survival (PFS) was 59%, while the 1-year PFS rate was 36%. Overall survival (OS) at 1-year was 52% in our study.

Conclusion

In real world, despite having similar PFS as the phase III IMbrave 150 trial, our OS at 12 months was lower (52% vs. 67%) because our study included a higher proportion of elderly patients with moderate liver dysfunction and a 40% non-White. This study provided real-world outcomes that differed from the study population in a pivotal trial.

This study retrospectively reviewed the outcomes of patients with advanced hepatocellular carcinoma receiving atezolizumab with bevacizumab (A + B) therapy at the Veterans Health Administration.

Implications for Practice.

Real-world evidence is limited regarding the utilization of atezolizumab plus bevacizumab in advanced hepatocellular carcinoma and patient’s outcome in real-world data. This study aimed to evaluate if treatment outcome in White and non-White patients, elderly frail patients, and patients with advanced liver disease including child Pugh score (CPS) classes B and C, and albumin-bilirubin index (ALBI) of grades 2 and 3. A total of 332 patients were included in this study, making this one of the largest studies of its kind in Veteran Health administration. AFP, CPS class B, and ALBI > 1 negatively impacted the overall survival and progression-free survival.

Introduction

Hepatocellular carcinoma (HCC) is the most common primary liver malignancy. As most patients present with advanced disease, curative therapeutic modalities, such as surgical resection, liver transplantation, and ablation, are not feasible, and the focus is primarily on systemic treatments. Recent investigations have focused on optimal management of this disease, especially for the development of novel systemic therapies. Multiple options exist for first-line treatment of metastatic HCC. Before 2008, there were few options for the treatment of HCC; however, in 2008, the first tyrosine kinase inhibitor (TKI), sorafenib, was introduced and approved by the Food and Drug Administration (FDA) based on the SHARP trial, which demonstrated survival benefits.¹

Until recently, no new treatment options have been established for advanced inoperable or metastatic HCC that necessitates systemic therapy, with the introduction of new therapeutic drugs. Newer agents include TKIs (ie, lenvatinib,² regorafenib,³ and cabozantinib⁴), immunotherapy (ie, nivolumab⁵ pembrolizumab,⁶ durvalumab,⁷ tremelimumab plus durvalumab,⁷ nivolumab plus ipilimumab,⁸ and vascular endothelial growth factor inhibitor [VEGF] (ie, ramucirumab).⁹

According to the guidelines of the National Comprehensive Cancer Network (NCCN)¹⁰ and the American Society of Clinical Oncology (ASCO)¹¹ a combination of atezolizumab plus bevacizumab (A + B) is recommended as the first-line treatment for advanced or metastatic HCC based on a randomized clinical trial published in 2020, which was the first treatment to demonstrate a survival benefit over sorafenib.¹² The imbrave 150 trial included a highly selected patient population (such as those with Child-Pugh A cirrhosis, no history of recent gastrointestinal [GI] bleeding, and those who required baseline endoscopic gastroduodenoscopy [EGD]), which limits the generalization of the results to the broader real-world population.

Few real-world retrospective cohort studies have demonstrated the efficacy and tolerability of A + B.^13-17 However, there is limited published data on U.S. patients and no published data on the veteran population, which typically includes patients who are older, frail, and have multiple comorbidities and advanced liver disease. Therefore, we conducted a retrospective study to explore real-world data regarding the use of A + B in veterans. Additionally, we assessed the most common drugs used as second- or third-line therapy in patients who experienced disease progression after receiving A + B. We reported patient outcomes, including the objective response rate (ORR), overall survival (OS), progression-free survival (PFS), adverse events (AEs) in the elderly population, and hospice enrollment.

Patients and Methods

Data Source

The Veterans Health Administration (VHA) operates health care facilities nationwide, including Washington, DC and the surrounding regions. They maintain an electronic health record (EHR) that encompasses administrative, clinical, laboratory, and pharmacy data repositories from both hospital and outpatient settings. Automated EHR queries to identify patients who met the initial inclusion and exclusion criteria were performed using the VHA Corporate Data Warehouse (CDW). The Joint Longitudinal Viewer (JLV) was subsequently used by data abstractors to manually search for the EHR for each selected patient to build the study dataset.

Study Design

This retrospective study started from a set of patients with HCC treated at the VHA identified by an automated query of the VHA CDW outpatient and inpatient records between March 15, 2005, and December 31, 2121 with ICD-9 or ICD-10 diagnosis codes for HCC. The list was further refined using an automated query of VHA pharmacy records for patients who received at least one dose of A + B in a VHA clinical or inpatient setting, with a search period of January 1, 2017 through September 1, 2022. The automated query dates were purposefully wide, to ensure eligible participants were not inadvertently excluded. Next, clinician abstractors reviewed the VHA medical records of each patient on the list using JLV. Patients were included in the study only if they had a confirmed HCC diagnosis, and received A + B as first-line systemic treatment within the VHA health care system, between December 2019 and March 2022.Medical records were reviewed from the initiation of A + B treatment until either death or their last VHA visit, with the study period ending on January 31, 2023. In addition to demographic data, the study collected information on pathology, etiology, cancer stage within 4 weeks of starting A + B therapy, prior local treatments, outcomes of the first-line therapy determined by scans, toxicities associated with atezolizumab, data on second-line or subsequent therapies, if applicable, and relevant information on palliative or hospice care.

Study Population

The study population consisted of adult patients diagnosed with locally advanced or metastatic HCC, who initially received A + B treatment. The EHR inclusion criteria were as follows: (1) age ≥ 18 years; (2) inpatient or outpatient HCC diagnosis (ICD-9 and ICD-10 codes) between March 15, 2005, and December 31, 2021; and (3) an order for A + B (drug name, HCPCS C9492 or J9173, or NDC 0310-4500-12 or 0310-4611-50) between January 1, 2017, and September 1, 2022. The following chart review criteria were used to confirm patient inclusion derived from the EHR data: (1) diagnosis of locally advanced inoperable HCC or metastatic HCC via pathology or imaging reports during the cohort inclusion period, (2) patients who experienced either initial disease regression or stable disease after at least one dose of A + B and then experienced tumor regrowth or disease progression, or patients who experienced initial disease progression after at least one dose of A + B. Patients were excluded from the study if they were diagnosed with HCC but were not treated within the VHA system.

Data and Statistical Analysis

Patient demographics, disease characteristics (including staging), A + B treatment patterns and toxicities, second- and third-line treatments, and treatment outcomes were evaluated descriptively. Categorical demographic and staging variables were reported as proportions and compared against treatment characteristics using the chi-squared test. Treatment outcomes included the ORR (proportion of patients with complete response [CR] or partial response [PR] in the first scan after initiating A + B treatment), disease control rate (proportion of patients with CR, PR, or stable disease at the first scan), duration of treatment (time from A + B initiation until discontinuation), PFS (time from A + B initiation until progression, death, or end of the study period), OS (time from A + B initiation until death or end of the study period), and duration of response (time from the first scan to progression/death/end of the study period for those with CR, PR, or SD). Hazard ratios (HRs) were calculated from the survival data using the Cox proportional hazards model. Additionally, OS and PFS were analyzed in terms of the proportion of patients who survived for at least 6 and 12 months, respectively. Palliative and hospice care for deceased patients were compared based on demographic characteristics using the chi-square test.

Results

A total of 418 medical records were reviewed. A total of 332 charts met the inclusion criteria and 86 were excluded. Fifty-five charts were excluded because A + B was not the first-line systemic therapy, 13 did not have sufficient information in the VHA medical records, 9 were not treated in the VHA system, 2 had no HCC diagnosis in the charts, and 7 for other reasons. Other reasons included that A + B was planned, but the patient died before starting A + B treatment; A + B was started, but serious complications from other diseases precluded continuation (sepsis, bladder cancer, or sudden decline); or non-compliant patients. We observed that 81% of the deceased patients received palliative care consultations and 70% of them were enrolled in hospice at the time of their death. Age was identified as the only significant determinant of hospice utilization, with 79% of veterans aged ≥ 70 years enrolled in hospice before death compared with 62% of veterans aged < 65 years (P = .027). The median time from enrollment in hospice care until death varied among age groups:13 days for veterans aged < 65 years, 31 days for veterans aged 65-69 years, and 18 days for veterans aged ≥ 70 years (P = .35).

We also conducted a subgroup analysis of the 107 patients who received second-line treatment. A total of 87% started TKIs with lenvatinib (48%), sorafenib (27%), cabozantinib (11%), and regorafenib (1%). Two second-line cohorts were selected: Group A (52 patients treated with lenvatinib) and Group B (29 patients treated with sorafenib). The median ages were 66 and 65 years in cohorts A and B, respectively. Both cohorts included 60% non-Hispanic Whites and 90% of patients with Eastern Cooperative Oncology Group (ECOG) scores ≤ 1. No statistically significant differences were observed between the 2 cohorts in terms of OS or PFS.

Patient Characteristics

The patients’ baseline characteristics are listed in Table 1. The median patient age was 67 years. patients were predominantly male (99%) and non-Hispanic White (63%), with 26% Black and 8% Hispanic individuals. 88% of patients had Eastern Cooperative Oncology Group (ECOG) scores of 0-1, 84% had CPS class A, and 62% had grade 2 Albumin-Bilirubin Index score (ALBI).

Table 1.

Patient baseline characteristics.

Characteristic	Cohort results
	(N = 332 pts)
Age at diagnosis (median [IQR])	67 (64-71)
Males (%)	329 (99%)
Race/ethnicity (%)
Hispanic	27 (8%)
Non-Hispanic Black	86 (26%)
Asian or indigenous	4 (1%)
Non-Hispanic White	208 (63%)
Unknown	7 (2%)
ECOG performance status (%)
0	112 (34%)
1	179 (54%)
2	40 (12%)
3	1 (0%)
Child–Pugh classification (%)
A	278 (84%)
B	52 (16%)
C	2 (1%)
Barcelona clinic liver cancer stage (%)
A	11 (3%)
B	96 (29%)
C	224 (67%)
D	1 (0%)
ALBI grade (%)
1	97 (29%)
2	207 (62%)
3	28 (8%)
Alpha-fetoprotein
≥ 400 ng/mL (%)	102 (31%)
Portal vein thrombosis (PVT) present (%)	120 (36%)
Extrahepatic spread present (%)	167 (50%)
Cause of HCC (%)
Viral (hepatitis)	184 (56%)
Non-viral	79 (24%)
No Cirrhosis present	68 (20%)
Prior local therapy for HCC (%)
0 treatment	110 (33%)
1 treatment	66 (20%)
2 treatments	59 (18%)
3 treatments	97 (29%)

Abbreviations: IQR, interquartile range; ECOG, Eastern Cooperative Oncology Group; ALBI, albumin-bilirubin; HCC, hepatocellular carcinoma.

Efficacy

In total, 332 patients received A + B as first-line treatment. The 6-month PFS was 59%, while the 1-year PFS was 36%, and the 1-year OS in our study was 52%. The 6-month PFS in non-White patients was 60% compared to 58% in non-Hispanic Whites (P = .91), the 1-year PFS was 41% in non-Whites compared to 33% in non-Hispanic Whites (P = .21). The 6-month OS was 75% among non-Whites compared to 72% in Whites (P = .49), and 1-year OS was 59% in non-Whites compared to 47% in White patients (P = .06), with a median OS of 17.7 months in non-Whites vs. 11.4 months in White patients (P = .07) (Fig. 1). Patients who were ≥ 70-years old had modest 6-month PFS of 57% and 1-year PFS of 35% (P = .82 and .77, respectively, when compared to the other age groups), as well as a 6-month OS of 71% and 1-year OS of 49% (P = .09 and .34, respectively, when compared to the other age groups) (Fig. 2). Patients who received at least one line of local therapy prior to A + B had a 6-month OS of 76% compared to 68% in patients who did not receive any prior local therapy (P = .15) and a better 1-year OS of 55% compared to 47% (P = .22) (Fig. 3). In patients with non-viral hepatitis, 45% of the patients had at least 1-year OS, clinically inferior to the 1-year OS of 55% in patients with viral hepatitis (P = .13). Patients with ALBI grade 1exhibited improved 1-year OS compared to those with ALBI grades 2 and 3; 71% vs. 48% vs. 16% (P < .0001) (Fig. 4).

Figure 1.

Overall survival by race.

Figure 2.

Overall survival by age group.

Figure 3.

Overall survival by prior local therapy.

Figure 4.

Overall survival according to the albumin bilirubin (ALBI) grades.

Safety

Approximately 30% of the patients receiving A + B did not report any toxicities, and 70% of the patients experienced a wide range of toxicities that resulted in treatment discontinuation. A total of 23% of patients discontinued therapy due to treatment-related toxicities. The median number of A + B doses administered was 6. Hepatic toxicity was the most frequently reported toxicity leading to treatment discontinuation (13%), followed by renal toxicity (9%), and GI toxicity (6%). The other toxicities associated with discontinuation are listed in Table 2. Immune-related AEs that led to discontinuation included hepatitis, colitis, nephritis, thyroiditis, pneumonitis, and encephalitis. After adjusting for AEs in the 3 age groups (<65, 65-69, and > 70 years), no significant differences were observed in the median A + B doses, therapy cessation due to toxicities, or median reported toxicities of any grade (Table 2).

Table 2.

Toxicities to atezolizumab plus bevacizumab treatment by age group.

	Age group
Characteristic	<65	65-69	≥70
	(n = 98)	(n = 111)	(n = 123)
Had no toxicities to atezolizumab plus bevacizumab	30 (31%)	26 (23%)	43 (35%)
Median doses of atezolizumab plus bevacizumab	5.5 (IQR 3-12, range 1-32)	6 (IQR 3-14, range 1-34)	5 (IQR 3-13, range 1-34)
Stopped treatment due to toxicity	23 (23%)	29 (26%)	26 (21%)
Median # of toxicities (for those with ≥ 1)	2 (1-3)	2 (1-3)	2 (1-3)
(Min, Max) toxicities	(1,8)	(1,6)	(1,7)
Most common discontinuation toxicities	5	9	6
Proteinuria	3	7	6
Aspartate aminotransferase increase	3	5	6
Alanine aminotransferase increase	1	3	4
Blood bilirubin increase	2	3	3
Diarrhea	3	1	3
Rash	0	2	5
Colitis	1	2	3
Kidney infection/toxicity	3	2	1
GI bleed	1	5	0
Hepatitis	3	3	0
Fatigue	1	2	1
Hypertension	6	10	6
Most common anytime toxicities (whether it causes discontinuation or not)	29	44	37
Fatigue	14	12	17
Decreased appetite	11	16	14
Proteinuria	9	11	14
Aspartate aminotransferase increase	9	9	13
Alanine aminotransferase increase	9	9	15
Weight decrease	9	14	10
Diarrhea	6	10	10
Hypertension	10	5	10
Rash	4	8	7
Nausea	3	10	3
Pruritis	3	4	8
Blood bilirubin increase	4	5	4
Abdominal pain	3	4	4
Constipation	4	3	3
Hypothyroidism	3	2	3
Platelet count decrease	1	4	3
Cough	4	2	1
GI bleeding	1	5	1
Hepatitis	0	2	5
Colitis	1	2	3
Kidney infection/toxicity	27	11	10

Abbreviation: GI: gastrointestinal.

Prognostic Factors Associated With Survival

High AFP, CPS class B, and ALBI grade > 1 were associated with adverse effects on OS, whereas other variables did not significantly affect survival time. When adjusting for all factors, AFP ≥ 400 leads to a 64% higher chance of death treatment initiation compared to lower AFP levels. CPS class B corresponds to a 114% higher probability of death than class A. ALBI grade 2 or 3 associated with a 68% higher probability of death than grade 1.

Disease progression is similarly influenced by high AFP levels, CPS class B, and ALBI grade > 1. Adjusted for all variables, AFP ≥ 400 results in a 33% higher likelihood of disease progression (including death) post-treatment initiation compared with lower AFP. CPS class B corresponded to a 68% higher probability of death than class A. While limited evidence suggests the effect of ALBI grade on disease progression (P = .076) after adjusting for other variables, its impact requires further investigation.

Discussion

Immunotherapy-based regimens have become the new standard of care for first-line treatment of advanced HCC. We report the first real-world experience of using A + B in a VHA system in the US. Our cohort represents the largest analysis of patients with advanced HCC who received A + B as first-line treatment. The 6-month PFS was 59% and the 1-year PFS was 36%, comparable to results from the phase III IMbrave 150 trial.¹² However, the 1 year OS in this study was lower (52% vs. 67%) because of the higher representation of Hispanic and Black patients and a higher median age compared to the pivotal trial. Non-White patients had a 1-year OS of 59% compared to 47% in White patients, with a median OS of 15 months in non-White and 10 months in White.

Seventy percent of the study population were > 65 years old. The efficacy of immunotherapy has been scarcely studied in geriatric patients compared to younger cohorts, secondary to its underrepresentation in clinical trials. Challenges arise in the treatment of geriatric patients undergoing immunotherapy owing to frailty, multimorbidity, metabolic changes, and immune senescence. However, previous studies have challenged these statements by demonstrating the efficacy of immunotherapy in both older and younger adults.^18,19 A subgroup analysis of the IMbrave 150 trial demonstrated increased OS and PFS with A + B therapy in patients over 65 years of age.^20,21 In this study, patients aged > 70 years had lower 6-month and 1-year OS than those in the IMbrave 150 trial, which may be attributed to the higher median age of the participants (67 years vs. 64 years).

To our knowledge, this is the first study to investigate the toxicities in geriatric patients with VHA. Among veterans aged > 70 years, there was a significantly lower incidence of cirrhosis attributed to viral hepatitis (P < .0001) and lack of prior local therapy (P = .002). There was no significant difference in the number of AEs between the age groups. Thus, the A + B combination can be safely used in geriatric patients with HCC.

In our cohort, PFS and OS were better in patients with viral hepatitis than in those without viral hepatitis, which was also observed in a pivotal study.¹² An analysis of Chinese patients treated in the IMbrave150 trial also supported the high efficacy of A + B compared to sorafenib. In this group, over 90% had a history of viral hepatitis infection, mainly hepatitis B, and 77% were alive 12 months after treatment initiation.

Patients who received local therapies prior to initiating A + B demonstrated improved PFS and OS.

The ORR in our real-world cohort was 30%, which is comparable to the treatment response rate of 27% reported in a phase III trial.¹² Univariate and multivariate analyses revealed that a high AFP level (>400), CPS stage B, and ALBI grade had a negative impact on OS and PFS. 1. In this cohort, the ALBI grade was not correlated with disease progression (P = .076) after adjusting for other variables.

Emerging data suggest that immunotherapy may be less effective in patients with non-alcoholic steatohepatitis (NASH)/non-alcoholic fatty liver disease (NAFLD) and HCC, possibly because of an altered immune environment.^22,23 Most patients in these studies had received prior small-molecule treatment, hampering direct comparisons. Currently, there is insufficient data demonstrating the outcomes of small-molecule treatment following immunotherapy in patients without steatohepatitis or fatty liver cirrhosis. However, the HIMALAYA study using dual immune checkpoint inhibitors demonstrated efficacy in a non-viral hepatitis subgroup.⁷

In our real-world data, we used ALBI grade as a surrogate for liver function rather than the CPS system, considering the known limitations of the latter. We found an obvious increase in PFS and OS in patients with ALBI grade 1 compared with those with ALBI grades 2 and 3. We also observed modest PFS and OS in patients with ALBI grade 2, thereby making this a useful tool for selecting patients for systemic treatment with immunotherapy and prognostication.²⁴

Early integration of palliative care for assistance with symptom burden and hospice utilization plays an important role in quality of life, but not end-of-life care.²⁵ The utilization of hospice care in our study was comparable to that observed in a similar nationwide VA study of patients with HCC.²⁵ Our study shows potential gaps in improving the quality of end-of-life care for patients with poor prognosis. Furthermore, quality improvement projects should be implemented to increase the number and timing of hospice care utilization in patients with advanced HCC.

Treatment with VEGF carries the risk of variceal bleeding and requires EGD before the initiation of systemic therapy with A + B.¹⁵ Therefore, VEGF-free treatment, such as tremelimumab plus durvalumab or durvalumab monotherapy, which was recently approved as the first-line systemic therapy for HCC, may be an option.⁷ A notable advantage of this VEGF-free treatment approach is that it does not require EGD, and this therapy is associated with less risk of variceal bleeding and associated B. Furthermore, as newer therapies are approved for HCC, real-world data in the VHA cohort will be important to study.

Our study has several limitations. First, because the study design was retrospective, selection bias may have been introduced. Second, we did not have enough time to follow up patients who underwent salvage therapy after A + B. Nevertheless, long-term follow-up studies with data on various salvage therapies after disease progression are required to support real-world evidence for treatment guidance strategies.

Conclusions

The A + B combination demonstrated efficacy and safety in this real-world VA cohort of predominantly older patients and ALBI was identified as a predictor of response. As a therapeutic approach for HCC development, it is important to explore real-world data regarding the use of these new treatments for advanced HCC. Real-world data provide insights into how treatments are performed outside clinical trials and can provide valuable information on their impact on patients’ quality of life.

Conflict of Interest

The authors indicated no financial relationships.

Author Contributions

All authors contributed to the conception/design, provision of study material or patients, collection and/or assembly of data, data analysis and interpretation, manuscript writing, and final approval of manuscript.

Data Availability

No new data were generated or analyzed in support of this research.