Optimal candidates and surrogate endpoints for HAIC versus Sorafenib in

hepatocellular carcinoma: an updated systematic review and meta-analysis

Tengfei Si1, Qing Shao2, Wayel Jassem2, Yun Ma*2, Nigel Heaton*2

1. Department of Hepatobiliary and Pancreatic Surgery, The First Affiliated Hospital,

School of Medicine, Zhejiang University, 310003, Hangzhou, Zhejiang, China.

2.Institute of Liver Studies, King's College Hospital NHS Foundation Trust, Denmark

Hill, SE5 9RS, London, UK.

Short title: HAIC vs Sorafenib in HCC: systematic review and meta-analysis

* Contributed equally to the work

Correspondence: Institute of Liver Studies, King’s College Hospital NHS

Foundation Trust, Denmark Hill, London, SE5 9RS, United Kingdom. Tel.:

+44(0)2032993369 E-mail address: nigel.heaton@nhs.net or yun.ma@kcl.ac.uk

Acknowledgements statement:

Assistance with the study: none

Financial support and sponsorship: none

Conflicts of interest: none

Presentation: none

Copyright © 2024 The Author(s). Published by Wolters Kluwer Health, Inc. This is an open access article distributed

under the Creative Commons Attribution License 4.0 (CCBY), which permits unrestricted use, distribution, and

reproduction in any medium, provided the original work is properly cited.

International Journal of Surgery Publish Ahead of Print

DOI:10.1097/JS9.0000000000001889

ACCEPTED Downloaded from http://journals.lww.com/international-journal-of-surgery by BhDMf5ePHKav1zEoum1tQfN4 a+kJLhEZgbsIHo4XMi0hCywCX1AWnYQp/IlQrHD3i3D0OdRyi7TvSFl4Cf3VC4/OAVpDDa8K2+Ya6H515kE= on 08/05/2024

Author Contribution: Study Design: TFS, NH; manuscript writing: TFS, QS; study

screening and Data analysis: TFS, QS, YM; result discussion: TFS, QS, YM, NH;

proof reading and editing: YM, WJ, NH

Data Availability Statement

All data generated or analysed during this study are included in this manuscript.

Further enquiries can be directed to the corresponding author.

Highlights

1. This updated meta-analysis was performed with 6,456 HCC patients from 26

studies.

2. The potential risk and protective factors of OS for HAIC versus Sorafenib in HCC

are unknown from prior meta-analyses and firstly explored in the present study.

3. HAIC was consistently associated with higher OS in HCC compared to Sorafenib.

However, for patients who were refractory to TACE, Sorafenib demonstrated

better OS compared to HAIC.

4. Higher response rates (ORR/DCR) after HAIC does not necessarily translate into

survival improvement compared to Sorafenib monotherapy, but the intermediate

milestone endpoints (1y-OS/1y-PFS) strongly correlated with patients’ OS.

5. Higher ECOG score, larger tumour size(>5cm), heavier tumour burden (>50%),

the existence of MVI or EHS and high AFP level (>400ng/ml) were independent

risk factors for patients’ OS. Conversely, HAIC treatment and lower BCLC stage

were potentially protective factors.

ACCEPTED Downloaded from http://journals.lww.com/international-journal-of-surgery by BhDMf5ePHKav1zEoum1tQfN4 a+kJLhEZgbsIHo4XMi0hCywCX1AWnYQp/IlQrHD3i3D0OdRyi7TvSFl4Cf3VC4/OAVpDDa8K2+Ya6H515kE= on 08/05/2024

Data statement

All data generated and analysed during this study are included in this article. The data

supporting the findings of this study are available from the corresponding author upon

reasonable request.

Abstract

Background and Aims:

Hepatic artery infusion chemotherapy (HAIC) has been a long-standing intervention

for hepatocellular carcinoma (HCC). Despite positive clinical outcomes, its inclusion

in guidelines remains limited due to a lack of evidence-based support. This study aims

to identify optimal target populations for HAIC and validate associations between

intermediate endpoints with overall survival (OS).

Methods:

Following PRISMA guidelines, a comprehensive search was conducted in PubMed,

Embase, Cochrane Library, and Web of Science. The primary search strategy was

based on medical subject headings terms (MeSH) using “Hepatic arterial infusion

chemotherapy”, “HAIC”, “Sorafenib”, “Nexavar”, “hepatocellular carcinoma”,

“HCC”, ‘‘Liver cancer’’, combined with free text words. Data extraction, quality

assessment, and analysis were performed according to pre-registered protocol.

Results:

A total of 26 studies, 6456 HCC patients were included for analysis (HAIC, n=2648;

Sorafenib, n=3808). Pooled outcomes revealed that Sorafenib demonstrated better OS

only in patients who were refractory to trans-arterial chemoembolization (TACE)

(HR=1.32,95%CI [1.01-1.73]), in other subgroups or overall HCC population HAIC

consistently outperformed Sorafenib in patients’ survival. Radiologically, higher

response rates in the HAIC group does not necessarily translate into survival

ACCEPTED Downloaded from http://journals.lww.com/international-journal-of-surgery by BhDMf5ePHKav1zEoum1tQfN4 a+kJLhEZgbsIHo4XMi0hCywCX1AWnYQp/IlQrHD3i3D0OdRyi7TvSFl4Cf3VC4/OAVpDDa8K2+Ya6H515kE= on 08/05/2024

improvement, but the hazard ratios (HRs) of 1y-OS (R2=0.41, p=0.0044) and 1yprogression free survival (1y-PFS) (R2=0.77, p=0.0002) strongly correlated with the

patients OS. Meanwhile, larger tumour size (HR=1.86,95%CI [1.12-3.1, 95%),

heavier tumour burden (HR=2.32, 95%CI [1.33-4.02), existence of MVI or EHS

(HR=1.65,95%CI[1.36-2]; HR=1.60,95%CI[1.19-2.14]), and AFP >400ng/ml

(HR=1.52, 95%CI [1.20-1.92]) were identified as independent risk factors for OS,

while HAIC treatment (HR= 0.54, 95%CI[0.35-0.82]) and lower BCLC stage

(HR=0.44, 95%CI[0.28-0.69]) were potential protective factors for HCC patients.

Conclusion:

HAIC monotherapy appears noninferior to Sorafenib in HCC treatment, with potential

benefits in specific subgroups. The robust correlation between 1y-OS/1y-PFS and OS,

alongside identified risk and protective factors from the present study, offers valuable

insights for designing future large prospective studies in this field.

Key Words: HAIC; Sorafenib; hepatocellular carcinoma; surrogate endpoint;

systematic review; meta-analysis

ACCEPTED Downloaded from http://journals.lww.com/international-journal-of-surgery by BhDMf5ePHKav1zEoum1tQfN4 a+kJLhEZgbsIHo4XMi0hCywCX1AWnYQp/IlQrHD3i3D0OdRyi7TvSFl4Cf3VC4/OAVpDDa8K2+Ya6H515kE= on 08/05/2024

Introduction:

Hepatocellular carcinoma (HCC) poses a formidable challenge for global heath, often

manifesting at an advanced stage with limited therapeutic avenues. As a traditional

intervention treatment, hepatic artery infusion chemotherapy (HAIC) has regained

new vitality in recent clinical studies. Many centres reported its application in the

treatment of advanced HCCs, and most studies demonstrated promising results.

Particularly when combined with Sorafenib, HAIC has demonstrated remarkable

efficacy [1, 2]. However, despite the positive clinical outcomes reported, the absence

of evidence-based medical study support has hindered the inclusion of HAIC as a

standard HCC treatment option in most guidelines. In the Barcelona Liver Cancer

Staging (BCLC), HAIC is not included in the treatment plan for primary liver cancer

[3]. The American Association for the Study of Liver Diseases (AASLD) also refrains

from recommending HAIC as routine locoregional therapy [4]. The European

Association of Liver Diseases (EASL) pointed out the FOLFOX chemotherapy

regimen commonly used in HAIC lacks evidence of a survival advantage [5]. In

contrast, Sorafenib is widely adopted as a standard of care for advanced HCC in many

countries. Since approved by the U.S. Food and Drug Administration (FDA) in 2007,

it has received subsequent endorsement in various clinical practice guidelines for

HCC management.

So far, the clinical application of HAIC is mainly limited to Asia. The Pan-Asian–

adapted European Society for Medical Oncology Guidelines recommended HAIC as

one of the first-line options for advanced, nonmetastatic HCC with macrovascular

invasion (MVI) [6]. The widespread acceptance of HAIC as one of the treatments for

HCC stems primarily its significant locoregional control effect and low systemic

toxicity [7]. Our prior investigation showed that HAIC yielded more favourable

outcomes in advanced HCC compared to conventional trans-arterial

chemoembolization (TACE) [8]. Ongoing research endeavours are dedicated to

identifying patients who may benefit from HAIC is required. With additional studies

ACCEPTED Downloaded from http://journals.lww.com/international-journal-of-surgery by BhDMf5ePHKav1zEoum1tQfN4 a+kJLhEZgbsIHo4XMi0hCywCX1AWnYQp/IlQrHD3i3D0OdRyi7TvSFl4Cf3VC4/OAVpDDa8K2+Ya6H515kE= on 08/05/2024

published comparing HAIC with Sorafenib, which suggest that HAIC alone may also

achieve similar or even superior therapeutic effects in comparison to Sorafenib

monotherapy [9].

Nevertheless, advanced HCC constitutes a highly heterogeneous group, influenced by

factors such as macrovascular invasion and/or metastatic disease, the degree of

underlying cirrhosis, and patient's performance status, all of which can impact

patients’ treatment outcome [10]. Consequently, further subclassification is warranted

to validate the reliability of conclusion drawn from overall study population. The

present study aims to systematically evaluate the clinical outcomes of both therapies

in HCC thus to identify: (a) the optimal target populations for HAIC or Sorafenib, and

(b) alternative surrogate endpoints which may enhance the design of future clinical

trials, along with potential risk/protection factors of OS in patients receiving these

treatments.

Methods

We performed this study according to the Preferred Reporting Items for Systematic

Reviews and Meta-Analyses (PRISMA) statement [11] and assessed the

methodological quality by the AMSTAR guidelines[12]. The protocol was registered

in PROSPERO international prospective register of systematic reviews.

Search strategy and selection criteria

Databases including PubMed, Embase, Cochrane Library and Web of Science were

searched to collect all available studies about using HAIC and Sorafenib mono

therapy in the treatment of advanced HCC. The primary search strategy was based on

medical subject headings terms (MeSH), combined with free text words. The

following key words were used as MeSH: “Hepatic arterial infusion chemotherapy”,

“Hepatic artery infusion”, “HAIC”, “HAI”, “Sorafenib”, “Nexavar”, “SORA”

ACCEPTED Downloaded from http://journals.lww.com/international-journal-of-surgery by BhDMf5ePHKav1zEoum1tQfN4 a+kJLhEZgbsIHo4XMi0hCywCX1AWnYQp/IlQrHD3i3D0OdRyi7TvSFl4Cf3VC4/OAVpDDa8K2+Ya6H515kE= on 08/05/2024

“hepatocellular carcinoma”, “HCC”, ‘‘Liver cancer’’. The original searching cutoff

date was 05th August 2023 with an extra updated search was conducted on 15th

December 2023. We also checked the reference lists of all identified studies for

additional eligible data. Detailed searching strategy was shown in Supplementary

Table 3, Supplemental Digital Content 1, http://links.lww.com/JS9/D228

Inclusion criteria: (1) Patients: individuals diagnosed with HCC were included; (2)

Type of studies: primary research, clinical trials (randomised controlled trials, nonrandomised trials), prospective or retrospective studies (single centre study, cohort

study), and case-controlled studies were considered. Both prospective and

retrospective studies were eligible; (3) Outcomes: included studies must report at least

one of the following outcomes: overall survival (OS), progression free survival (PFS),

radiological response (CR, complete response; PR, partial response; SD, stable

disease; PD, progression disease), adverse events and mortality data; (4) The language

of the published literature was limited to English only.

Exclusion criteria: (1) Patients without clear diagnosis of HCC or those with

metastatic liver cancer; (2) Patients who received combination therapy of HAIC and

Sorafenib; (3) In the case of single‐centre series with repeated publication or

overlapping cases, the research manuscript with more comprehensive data was

retained, and (4) letters, editorials, expert opinions and reviews were excluded to

ensure only original data were used.

Data extraction

Data extraction from each study was performed by two authors (TFS and QS)

independently. Patients’ basic characteristics, OS, PFS, details of interventions used,

objective response rate (ORR)/disease control rate (DCR), adverse events (Grade3/4)

and mortality data were collected from each study using a pre-designed data

extraction form (Supplementary Table 1 , Supplemental Digital Content 1,

http://links.lww.com/JS9/D228 & Table 2, Supplemental Digital Content 1,

ACCEPTED Downloaded from http://journals.lww.com/international-journal-of-surgery by BhDMf5ePHKav1zEoum1tQfN4 a+kJLhEZgbsIHo4XMi0hCywCX1AWnYQp/IlQrHD3i3D0OdRyi7TvSFl4Cf3VC4/OAVpDDa8K2+Ya6H515kE= on 08/05/2024

http://links.lww.com/JS9/D228 ).). In cases of missing information, attempts were

made to contact the authors of original articles. Any disagreements during the data

extraction process were resolved through discussion or, if necessary, with the

involvement of a third reviewer (YM).

Quality assessment

Two independent reviewers conducted the study quality assessment and risk of bias

analysis. The Newcastle-Ottawa Scale (NOS) was utilized to assess the quality of

included studies. Regarding the NOS assessment, a maximum of one star could be

assigned for each numbered item within the Selection and Exposure categories, while

a maximum of two stars would be given for Comparability. Each study was rated

from 0 to 9 stars, with 0–3, 4–6, and 7–9 considered low, moderate, and high

qualities, respectively, on the NOS scale. The Grading Recommendations Assessment,

Development and Evaluation (GRADE) approach was applied to evaluate the

certainty of evidence for meta-results [13]. The Guideline Development Tool was

accessed from https://www.gradepro.org to create the Summary of Findings table.

During the process of quality assessment, disagreement was resolved by discussion or

with a third reviewer if necessary (YM).

Statistical analysis

Review Manager (version 5.3) recommended by Cochrane Collaboration was used to

perform the meta-analysis. Hazard Ratio (HR) with a 95% confidence interval was

selected as effect measure for OS and PFS. Dichotomous variables were tested by

Risk Ratio (RR) with a 95% confidence interval. RRs were calculated using inverse

variance method. HRs were calculated using the method suggested by Tierney et

al[14] for incorporating summary time-to-event data into meta-analysis. A randomeffect model was used for all calculations. Heterogeneity between studies was tested

ACCEPTED Downloaded from http://journals.lww.com/international-journal-of-surgery by BhDMf5ePHKav1zEoum1tQfN4 a+kJLhEZgbsIHo4XMi0hCywCX1AWnYQp/IlQrHD3i3D0OdRyi7TvSFl4Cf3VC4/OAVpDDa8K2+Ya6H515kE= on 08/05/2024

by χ² test (with significance set at p>0.1) and I² test. Funnel plot was used to

investigate publication bias if sufficient studies were available (Supplementary File

14, Supplemental Digital Content 1, http://links.lww.com/JS9/D228 ). Statistical

analysis of patients' background information was conducted using Fisher’s exact test

and chi‐squared test with GraphPad Prism 8.0 (GraphPad Software, Inc.). A p-value

<0.05 was considered statistically significant. Pooling HRs from multivariate Cox

regression were also extracted from each study to identify potential risk or protective

factors for OS. Subgroup analysis was performed according to tumour stage, liver

function, patient’s treatment history and other confounding factors which may affect

patients’ prognosis.

Results:

Patients

The study screening process is provided in Figure 1. A total of 26 studies [9, 15-39]

involving 6456 patients were included for the final quantitative analysis. Among

whom, 2648 received HAIC treatment, while 3808 patients received Sorafenib

monotherapy simultaneously. The participants were mainly from Japan, South Korea,

and China. Patients’ tumour characteristics varied across these studies, but the

majority exhibited stable liver function (Child-Pugh A-B) (Table 1). To address

baseline differences between the HAIC and Sorafenib groups, we summarized 8

studies [9, 19, 25, 29-31, 36, 37] that employed propensity scoring matching (PSM)

design (Supplementary Table 4, Supplemental Digital Content 1,

http://links.lww.com/JS9/D228 ). However, even after PSM, pooled data indicated

that in comparison to the Sorafenib group (n=562), the HAIC group (n=560)

continued to have a tendency towards a higher proportion of patients with MVI

(46.6% vs. 40.3%, p=0.035).

ACCEPTED Downloaded from http://journals.lww.com/international-journal-of-surgery by BhDMf5ePHKav1zEoum1tQfN4 a+kJLhEZgbsIHo4XMi0hCywCX1AWnYQp/IlQrHD3i3D0OdRyi7TvSFl4Cf3VC4/OAVpDDa8K2+Ya6H515kE= on 08/05/2024

Survival

In general, both OS (HR=0.72, 95%CI=0.61-0.86, I2=80%) and PFS (HR=0.57,

95%CI=0.46-0.71, I2

=71%) favoured HAIC compared to Sorafenib (Figure 2),

especially for those with tumours larger than 5cm, who were chemotherapy naïve or

had MVI but without extrahepatic spread (EHS) (Table 2). In patients with ChildPugh A grade liver function, no significant difference was found in the OS between

the two groups, but the HAIC group seemed to have a longer PFS (HR=0.50,

95%CI=0.28-0.91). However, for patients who were refractory to TACE treatment,

Sorafenib may offer better OS compared to HAIC (HR=1.32, 95%CI=1.01-1.73).

Results from PSM subgroup analysis showed that PFS favouring HAIC over

Sorafenib (HR=0.68, 95%CI=0.48-0.98), while no difference existed in OS between

the two groups.

The intermediate milestone survival at 1 year (1y-OS and 1y-PFS) was also explored.

The pooled results are generally consistent, HAIC outperformed Sorafenib in terms of

1y-PFS. However, regarding 1y-OS, a statistical difference was found in the subgroup

analysis that was not evident in the previous comparison. 1y-OS of patients with EHS

but no MVI favoured the HAIC group compared to Sorafenib (HR=0.69,

95%CI=0.51-0.95) (Table 2).

Radiological Response

Regarding radiological response, the HAIC group demonstrated overall advantages

compared to Sorafenib in both ORR (RR=4.29, 95%CI=3.12-5.89) and DCR

(RR=1.33, 95%CI=1.16-1.52) (Figure 2). Across subgroup analysis, the objective

response relative risk (ORRRR) and disease control relative risk (DCRRR) of the HAIC

group were higher (indicating more patients with ORR /DCR in the HAIC group) than

those in the Sorafenib group. Although no difference in DCR was observed between

the two groups for patients who were refractory to TACE (RR=1.06, 95%CI=0.77-

1.45), the ORR in the HAIC group was still significantly higher compared to the

Sorafenib group (RR=4.05, 95%CI [1.42-11.52]). Pooled data from PSM analysis

ACCEPTED Downloaded from http://journals.lww.com/international-journal-of-surgery by BhDMf5ePHKav1zEoum1tQfN4 a+kJLhEZgbsIHo4XMi0hCywCX1AWnYQp/IlQrHD3i3D0OdRyi7TvSFl4Cf3VC4/OAVpDDa8K2+Ya6H515kE= on 08/05/2024

showed similar findings: ORR favoured the HAIC group compared to the Sorafenib

group (RR=3.80, 95%CI=2.00-7.21) (Table 2).

Surrogacy analysis of intermediate endpoints

Correlation analysis revealed a strong association between PFSHR and OSHR, with 1yPFSHR exhibiting excellent compatibility with PFSHR (R2

=0.975, 95%Cl=0.959-

0.996). Both OSHR (R2

=0.771, 95%CI=0.613-0.965) and 1y- OSHR (R2=0.722,

95%CI=0.561-0.954) demonstrated a significant correlation with 1y-PFSHR (Figure

3A). Although OSHR and PFSHR had a downward trend (indicating longer OS/PFS)

with the increasing values of ORRRR and DCRRR, no significant correlations were

found between ORRRR or DCRRR with OSHR/PFSHR. Similar results were also

observed for PFS analysis (Figure 3B), while both 1y-OSHR (R2

=0.3080, P=0.026) and

1y-PFSHR (R2=0.3163, P=0.036) had a statistical association with DCRRR: with

increase in DCRRR corresponded to lower HRs of OS and PFS (Figure 3C).

Seven pre-defined subgroup analyses, including tumour characteristics, liver function,

treatment history, and study design were performed to validate the robustness of

correlation analysis results. Overall, the findings remained consistent across

subgroups: ORR/DCR showed low association with OS while 1y-OS and 1y-PFS had

moderate to high association with OS (Table 3).

Risk/Protective Factors of OS

Independent risk and protective factors of OS between the HAIC group and the

Sorafenib group were identified in each study. Pooled HRs of multivariate Cox

regression indicated that higher ECOG score (HR=2.06, 95%Cl=1.26-3.37), larger

tumour size(>5cm) (HR=1.86, 95%CI=1.12-3.10), heavier tumour burden (>50%)

(HR=2.32, 95%CI=1.33-4.02), the existence of MVI (HR=1.60, 95%CI[1.19-2.14]) or

EHS(HR=1.65, 95%CI[1.36-2.00]) and AFP >400ng/ml (HR=1.52, 95%CI=1.20-

ACCEPTED Downloaded from http://journals.lww.com/international-journal-of-surgery by BhDMf5ePHKav1zEoum1tQfN4 a+kJLhEZgbsIHo4XMi0hCywCX1AWnYQp/IlQrHD3i3D0OdRyi7TvSFl4Cf3VC4/OAVpDDa8K2+Ya6H515kE= on 08/05/2024

1.92) were independent risk factors for patients’ OS. Conversely, HAIC treatment

(HR=0.54, 95%CI=0.35-0.82) and lower BCLC stage (HR=0.44, 95%CI=0.28-0.69)

were potentially protective factors (Figure 4).

Adverse Events

The safety profiles of the two treatments in HCC patients were also compared through

meta-analysis. No treatment-related mortality was reported in any of the included

studies. Data about grade 3/4 adverse events were documented. Patients in the HAIC

group exhibited a higher incidence of elevated ALT (RR=1.29, 95%CI=1.03-1.60),

whereas patients receiving Sorafenib monotherapy reported elevated risks for skin

complications (RR=0.03, 95%CI=0.01-0.08), diarrhoea (RR=0.37, 95%CI=0.22-

0.64), ascites (RR=0.62, 95%CI=0.45-0.85), hepatic encephalopathy (RR=0.31,

95%CI=0.11-0.85) and total serum bilirubin elevation (RR=0.65, 95%CI=0.43-0.97).

In terms of other common systematic drug-related side effects such as neutropenia,

fatigue, leukopenia, thrombocytopenia, reduced haemoglobin, and hypoalbuminemia,

no statistical differences were observed between the two groups (Supplementary Table

5, Supplemental Digital Content 1, http://links.lww.com/JS9/D228 ).

Quality Assessment and GRADE Summary of Findings

Most studies were assessed to be of medium- (>3 stars, n=15) or high quality (>6

stars, n=11). No low-quality studies (≤ 3 stars) were found to be included. The

certainty of evidence after GRADE assessment showed that most findings were of

low/very low quality, findings from comparisons of ORR and OS in patients who are

refractory to TACE and comparison of OS in patients with EHS but no MVI presented

with moderate quality (Supplementary Table 6, Supplemental Digital Content 1,

http://links.lww.com/JS9/D228 & Table 7, Supplemental Digital Content 1,

http://links.lww.com/JS9/D228 ).

ACCEPTED Downloaded from http://journals.lww.com/international-journal-of-surgery by BhDMf5ePHKav1zEoum1tQfN4 a+kJLhEZgbsIHo4XMi0hCywCX1AWnYQp/IlQrHD3i3D0OdRyi7TvSFl4Cf3VC4/OAVpDDa8K2+Ya6H515kE= on 08/05/2024

Discussion

Based on 6,456 patients from 26 studies, our study showed that HAIC as

monotherapy is noninferior to Sorafenib. In specific subgroups, it demonstrates better

clinical outcomes, with anticipated longer OS and PFS for individuals with large

tumour size (>5 cm), chemotherapy naïve patients and those with MVI but no EHS.

Furthermore, HAIC exhibits advantages over Sorafenib in terms of radiological

response. Though previous meta-analyses reported similar findings [40] [41],

limitations including small sample sizes and high degree of heterogeneity of patient

groups which restricted further subgroup analyses based on liver function and

treatment history.

Encompassing 26 studies and over 6,000 HCC patients, our study stands as the most

extensive meta-analysis to date comparing the outcomes of HAIC versus Sorafenib

(Supplementary File 8-13, Supplemental Digital Content 1,

http://links.lww.com/JS9/D228 & File15-16, Supplemental Digital Content 1,

http://links.lww.com/JS9/D228 ). The substantial sample size allows for a

comprehensive subgroup-analysis, confirming that patients with Child-Pugh A liver

function have better PFS following HAIC compared to those receiving Sorafenib.

However, HAIC therapy did not benefit all patients: for those who were refractory to

TACE, Sorafenib demonstrated better OS compared to HAIC. This finding contrasts

with the outcomes of previous meta-analyses (Liu et al[40], n=417; Zhang et al[42],

n=672; Ni et al[43], n=1264; Zhuang et al[44], n=1779); these smaller sized studies

generally supported better OS with HAIC compared to Sorafenib in HCC. We believe

that in TACE-refractory HCCs, vascular injury due to repeated catheterisation and

reduced sensitivity of tumour cells to chemotherapy drugs may provide explanation

for the diminished efficacy of HAIC in these patients. Similar findings were also

reported in our previous publication [45], where repeated TACE procedures correlated

with a reduced progression-free survival after liver transplant and higher rate of

vascular complications.

ACCEPTED Downloaded from http://journals.lww.com/international-journal-of-surgery by BhDMf5ePHKav1zEoum1tQfN4 a+kJLhEZgbsIHo4XMi0hCywCX1AWnYQp/IlQrHD3i3D0OdRyi7TvSFl4Cf3VC4/OAVpDDa8K2+Ya6H515kE= on 08/05/2024

The advantage of HAIC is the high first pass in the liver allowing higher

concentration of chemotherapy drug exposure in the tumour [46]. But there are

concerns that chemotherapy from HAIC may not reach effective concentrations in

extrahepatic tissues which may also explain the findings from our study: HAIC did

not provide greater benefit compared to Sorafenib in the presence of EHS. In such

cases, a rational approach might involve adding a systemic chemotherapy agent to

HAIC to provide effective systemic control in distant metastatic disease [47, 48]. In

recent years, numerous studies have reported on the use of combined treatment of

HAIC plus Sorafenib for advanced HCC [1, 49-51]. While the results of different

studies vary, a recent meta-analysis [52] showed that the anti-tumour effect of

Sorafenib combined with HAIC was better than Sorafenib alone in advanced HCCs

with acceptable safety concerns of chemotherapy toxicity.

One novelty of this study was the exploration of surrogate endpoints for effective

comparisons of HAIC and Sorafenib. In phase III clinical trials of advanced HCC, OS

is usually used as the primary endpoint to verify whether the treatment regimen can

truly prolong patient survival. However, in recent years, many studies have included

surrogate endpoints such as ORR/PFS in their evaluation. In our study, we found a

relatively weak correlation between ORR/DCR and patients’ ultimate survival. A high

DCR/ORR does not necessarily translate into clinical benefit. Although radiological

response is the most direct indicator of anti-tumour activity, it is essentially an

imaging assessment and may not predict continuing response of patient survival. This

potentially explains why, despite overall ORR and DCR in the HAIC group being

superior to that from the Sorafenib group, many patients did not exhibit significant

improvement in OS after HAIC.

It is noteworthy that during the efficacy comparison between HAIC and Sorafenib,

general PFS is significantly correlated with patients’ OS, and the value of 1-year

PFSHR was closely related to overall PFSHR. A recent meta-analysis [53] assessing the

ability of PFS to predict OS in phase III clinical trials of systemic treatment for

advanced HCC, based on 21 randomised controlled trials (RCTs), identified a

ACCEPTED Downloaded from http://journals.lww.com/international-journal-of-surgery by BhDMf5ePHKav1zEoum1tQfN4 a+kJLhEZgbsIHo4XMi0hCywCX1AWnYQp/IlQrHD3i3D0OdRyi7TvSFl4Cf3VC4/OAVpDDa8K2+Ya6H515kE= on 08/05/2024

moderate correlation between PFS and OS. Consequently, the authors proposed a

conservative alternative threshold, a PFSHR <0.6, which could be used to predict

clinically relevant OS improvement. According to our study, in most comparisons

which favour the HAIC group had a PFSHR value below 0.6 and each PFSHR matched

OSHR also demonstrated favourable results for HAIC. The robust correlation between

OSHR and PFSHR indirectly substantiated the finding that a PFSHR <0.6 could predict

clinically relevant OS improvement.

One major limitation in this study was that majority of collected data originated from

observational studies and only two RCTs were included. This unavoidably increased

the risk of publication bias during the analysis. Future large scale RCTs comparing

HAIC with Sorafenib are essential to offer more robust evidence on the effectiveness

of HAIC. Such trials should aim to minimize biases and provide comprehensive data

that can enhance the reliability of comparative analyses. Additionally, all included

studies were from Aisa, the results generated may have regional limitations. HCC in

Asia is mostly caused by hepatitis B virus (HBV), with many patients having

established liver cirrhosis [10]. At the time of diagnosis, they often present with large

tumour size or accompanied by portal vein tumour thrombus, resulting in a generally

high-risk classification. While in western countries, HCC is mainly caused by chronic

hepatitis C virus (HCV) infection and alcohol related liver disease followed by fatty

liver disease and diabetes [10]. The diagnostic rate of early-stage HCC is relatively

higher with tumours had median or average diameter less than 5 cm when diagnosed,

classifying most lesions as low risk [54]. The IMbrave150 study found that the

median OS of atezolizumab combined with bevacizumab in low-risk HCC (BCLC B,

no MVI and /or EHS) was more than 2 years, while the median OS of high-risk HCC

(VP4) was only 7.6 months [55]. Differences in epidemiology and tumour burden

may result in significant differences in the efficacy of the same treatment regimen.

Thus, multiple confounding factors need to be carefully considered in clinical and

practical applications of HAIC.

ACCEPTED Downloaded from http://journals.lww.com/international-journal-of-surgery by BhDMf5ePHKav1zEoum1tQfN4 a+kJLhEZgbsIHo4XMi0hCywCX1AWnYQp/IlQrHD3i3D0OdRyi7TvSFl4Cf3VC4/OAVpDDa8K2+Ya6H515kE= on 08/05/2024

This study is different from previous meta-analysis in that it examines independent

risk and protective factors for patients’ OS after HAIC or Sorafenib. Factors such as

larger tumour size (>5cm), the presence of MVI or EHS were found to be independent

risk factors for patients’ OS, while the choice of HAIC treatment emerged as a

protective factor. However, although pooled data indicated the benefit of HAIC for

high-risk patients (tumour size >5cm or with MVI but no EHS), it is crucial to assess

the adverse effects of tumour size and MVI on OS when considering HAIC as a

treatment option. Tumour size and MVI often reflect the status of tumour blood

supply which are closely tied to the efficacy of HAIC. But both factors would also

upgrade the BCLC stage of tumours, and pooled HRs suggested that lower BCLC

stage may serve as potential protective factor for patients’ OS. Therefore, striking a

balance between risk factors and protective factors is pivotal in the consideration of

HAIC as the optimal treatment for HCC.

To our knowledge, the present study is the most extensive meta-analysis of HAIC

versus Sorafenib in treating HCC. With over 6,400 patients across 26 studies, we

conducted comprehensive subgroup analysis yielding results generally consistent with

those from overall study population. But certain limitations are acknowledged. Firstly,

the assessment criteria for radiological response varied among individual studies,

potentially introducing bias in the collection and reporting of ORR/DCR. Secondly,

considerable heterogeneity exists among different comparisons, partially explainable

through the establishment of various subgroups. This diversity may stem from

inherent patient differences or variations in HAIC treatment protocols, which

currently cannot be definitively attributed due to insufficient data. Thirdly, for singlecentre studies covering the same patient population we selected the ones with most

comprehensive data, but the inclusion of multicentre studies may lead to some overlap

among patients. Finally, most findings from our study generally present with low

quality after GRADE assessment, with a small portion attaining moderate quality.

Consequently, high-quality evidence is still lacking.

ACCEPTED Downloaded from http://journals.lww.com/international-journal-of-surgery by BhDMf5ePHKav1zEoum1tQfN4 a+kJLhEZgbsIHo4XMi0hCywCX1AWnYQp/IlQrHD3i3D0OdRyi7TvSFl4Cf3VC4/OAVpDDa8K2+Ya6H515kE= on 08/05/2024

In conclusion, our findings indicate that some patients may derive benefit from HAIC

compared to Sorafenib monotherapy. The strong correlation between 1y-OS/1y-PFS

with OS, along with the potential risk and protection factors screened in this study,

may inform the design of future large prospective studies focusing on this subject.

Provenance and peer review

Not commissioned, externally peer-reviewed

ACCEPTED Downloaded from http://journals.lww.com/international-journal-of-surgery by BhDMf5ePHKav1zEoum1tQfN4 a+kJLhEZgbsIHo4XMi0hCywCX1AWnYQp/IlQrHD3i3D0OdRyi7TvSFl4Cf3VC4/OAVpDDa8K2+Ya6H515kE= on 08/05/2024

References

1. Zheng, K., et al., Sorafenib Plus Hepatic Arterial Infusion Chemotherapy

versus Sorafenib for Hepatocellular Carcinoma with Major Portal Vein Tumor

Thrombosis: A Randomized Trial. Radiology, 2022. 303(2): p. 455-464.

2. Ueshima, K., et al., Phase I/II Study of Sorafenib in Combination with Hepatic

Arterial Infusion Chemotherapy Using Low-Dose Cisplatin and 5-

Fluorouracil. Liver Cancer, 2015. 4(4): p. 263-73.

3. Reig, M., et al., BCLC strategy for prognosis prediction and treatment

recommendation: The 2022 update. J Hepatol, 2022. 76(3): p. 681-693.

4. Heimbach, J.K., et al., AASLD guidelines for the treatment of hepatocellular

carcinoma. Hepatology, 2018. 67(1): p. 358-380.

5. European Association for the Study of the Liver. Electronic address, e.e.e. and

L. European Association for the Study of the, EASL Clinical Practice

Guidelines: Management of hepatocellular carcinoma. J Hepatol, 2018. 69(1):

p. 182-236.

6. Chen, L.T., et al., Pan-Asian adapted ESMO Clinical Practice Guidelines for

the management of patients with intermediate and advanced/relapsed

hepatocellular carcinoma: a TOS-ESMO initiative endorsed by CSCO,

ISMPO, JSMO, KSMO, MOS and SSO. Ann Oncol, 2020. 31(3): p. 334-351.

7. Kudo, M., et al., Management of Hepatocellular Carcinoma in Japan: JSH

Consensus Statements and Recommendations 2021 Update. Liver Cancer,

2021. 10(3): p. 181-223.

8. Si, T., et al., Hepatic arterial infusion chemotherapy versus transarterial

chemoembolization for unresectable hepatocellular carcinoma: A systematic

review with meta-analysis. Front Bioeng Biotechnol, 2022. 10: p. 1010824.

9. Lyu, N., et al., Arterial Chemotherapy of Oxaliplatin Plus Fluorouracil Versus

Sorafenib in Advanced Hepatocellular Carcinoma: A Biomolecular

Exploratory, Randomized, Phase III Trial (FOHAIC-1). J Clin Oncol, 2022.

40(5): p. 468-480.

10. Llovet, J.M., et al., Hepatocellular carcinoma. Nat Rev Dis Primers, 2021.

7(1): p. 6.

11. Page, M.J., et al., The PRISMA 2020 statement: an updated guideline for

reporting systematic reviews. BMJ, 2021. 372: p. n71.

12. Shea, B.J., et al., AMSTAR 2: a critical appraisal tool for systematic reviews

that include randomised or non-randomised studies of healthcare

interventions, or both. BMJ, 2017. 358: p. j4008.

13. Guyatt, G.H., et al., GRADE: an emerging consensus on rating quality of

evidence and strength of recommendations. BMJ, 2008. 336(7650): p. 924-6.

14. Tierney, J.F., et al., Practical methods for incorporating summary time-toevent data into meta-analysis. Trials, 2007. 8: p. 16.

15. Kang, M.K., J.G. Park, and H.J. Lee, Comparison of clinical outcomes

between sorafenib and hepatic artery infusion chemotherapy in advanced

ACCEPTED Downloaded from http://journals.lww.com/international-journal-of-surgery by BhDMf5ePHKav1zEoum1tQfN4 a+kJLhEZgbsIHo4XMi0hCywCX1AWnYQp/IlQrHD3i3D0OdRyi7TvSFl4Cf3VC4/OAVpDDa8K2+Ya6H515kE= on 08/05/2024

hepatocellular carcinoma: A STROBE-compliant article. Medicine

(Baltimore), 2018. 97(17): p. e0611.

16. Kodama, K., et al., Comparison of clinical outcome of hepatic arterial

infusion chemotherapy and sorafenib for advanced hepatocellular carcinoma

according to macrovascular invasion and transcatheter arterial

chemoembolization refractory status. J Gastroenterol Hepatol, 2018. 33(10): p.

1780-1786.

17. Ahn, Y.E., et al., Comparison of Sorafenib versus Hepatic Arterial Infusion

Chemotherapy-Based Treatment for Advanced Hepatocellular Carcinoma with

Portal Vein Tumor Thrombosis. Gut Liver, 2021. 15(2): p. 284-294.

18. Abdelmaksoud, A.H.K., et al., Hepatic arterial infusion chemotherapy in the

treatment of advanced hepatocellular carcinoma with portal vein thrombosis:

a case-control study. Clin Radiol, 2021. 76(9): p. 709 e1-709 e6.

19. Choi, J.H., et al., Randomized, prospective, comparative study on the effects

and safety of sorafenib vs. hepatic arterial infusion chemotherapy in patients

with advanced hepatocellular carcinoma with portal vein tumor thrombosis.

Cancer Chemother Pharmacol, 2018. 82(3): p. 469-478.

20. Moriguchi, M., et al., Sorafenib versus Hepatic Arterial Infusion

Chemotherapy as Initial Treatment for Hepatocellular Carcinoma with

Advanced Portal Vein Tumor Thrombosis. Liver Cancer, 2017. 6(4): p. 275-

286.

21. Song, D.S., et al., A comparative study between sorafenib and hepatic arterial

infusion chemotherapy for advanced hepatocellular carcinoma with portal

vein tumor thrombosis. J Gastroenterol, 2015. 50(4): p. 445-54.

22. Yang, H., et al., A comparative study of sorafenib and metronomic

chemotherapy for Barcelona Clinic Liver Cancer-stage C hepatocellular

carcinoma with poor liver function. Clin Mol Hepatol, 2017. 23(2): p. 128-

137.

23. Nakano, M., et al., Clinical effects and safety of intra-arterial infusion therapy

of cisplatin suspension in lipiodol combined with 5-fluorouracil versus

sorafenib, for advanced hepatocellular carcinoma with macroscopic vascular

invasion without extra-hepatic spread: A prospective cohort study. Mol Clin

Oncol, 2017. 7(6): p. 1013-1020.

24. Kawaoka, T., et al., Comparison of hepatic arterial infusion chemotherapy

versus sorafenib monotherapy in patients with advanced hepatocellular

carcinoma. J Dig Dis, 2015. 16(9): p. 505-12.

25. Shiozawa, K., et al., Comparison of Sorafenib and Hepatic Arterial Infusion

Chemotherapy for Advanced Hepatocellular Carcinoma: A Propensity Score

Matching Study. Hepatogastroenterology, 2014. 61(132): p. 885-91.

26. Nemoto, T., et al., Comparison of hepatic arterial infusion chemotherapy and

sorafenib in elderly patients with advanced hepatocellular carcinoma: A case

series. Mol Clin Oncol, 2014. 2(6): p. 1028-1034.

ACCEPTED Downloaded from http://journals.lww.com/international-journal-of-surgery by BhDMf5ePHKav1zEoum1tQfN4 a+kJLhEZgbsIHo4XMi0hCywCX1AWnYQp/IlQrHD3i3D0OdRyi7TvSFl4Cf3VC4/OAVpDDa8K2+Ya6H515kE= on 08/05/2024

27. Jeong, S.W., et al., The efficacy of hepatic arterial infusion chemotherapy as

an alternative to sorafenib in advanced hepatocellular carcinoma. Asia Pac J

Clin Oncol, 2012. 8(2): p. 164-71.

28. Hiramine, Y., et al., Sorafenib and hepatic arterial infusion chemotherapy for

unresectable advanced hepatocellular carcinoma: A comparative study. Exp

Ther Med, 2011. 2(3): p. 433-441.

29. Lyu, N., et al., Hepatic arterial infusion of oxaliplatin plus

fluorouracil/leucovorin vs. sorafenib for advanced hepatocellular carcinoma.

J Hepatol, 2018. 69(1): p. 60-69.

30. Fukubayashi, K., et al., Evaluation of sorafenib treatment and hepatic arterial

infusion chemotherapy for advanced hepatocellular carcinoma: a comparative

study using the propensity score matching method. Cancer Med, 2015. 4(8): p.

1214-23.

31. Hatooka, M., et al., Comparison of Outcome of Hepatic Arterial Infusion

Chemotherapy and Sorafenib in Patients with Hepatocellular Carcinoma

Refractory to Transcatheter Arterial Chemoembolization. Anticancer Res,

2016. 36(7): p. 3523-9.

32. Ueshima, K., et al., Hepatic Arterial Infusion Chemotherapy versus Sorafenib

in Patients with Advanced Hepatocellular Carcinoma. Liver Cancer, 2020.

9(5): p. 583-595.

33. Terashima, T., et al., Beneficial Effect of Maintaining Hepatic Reserve during

Chemotherapy on the Outcomes of Patients with Hepatocellular Carcinoma.

Liver Cancer, 2017. 6(3): p. 236-249.

34. Moriya, K., et al., Efficacy of bi-monthly hepatic arterial infusion

chemotherapy for advanced hepatocellular carcinoma. J Gastrointest Oncol,

2018. 9(4): p. 741-749.

35. Kondo, M., et al., Hepatic arterial infusion chemotherapy with cisplatin and

sorafenib in hepatocellular carcinoma patients unresponsive to transarterial

chemoembolization: a propensity score-based weighting. J Dig Dis, 2015.

16(3): p. 143-51.

36. Zaizen, Y., et al., Hepatic Arterial Infusion Chemotherapy with Cisplatin

versus Sorafenib for Intrahepatic Advanced Hepatocellular Carcinoma: A

Propensity Score-Matched Analysis. Cancers (Basel), 2021. 13(21).

37. Han, S., et al., Treatment efficacy by hepatic arterial infusion chemotherapy

vs. sorafenib after liver-directed concurrent chemoradiotherapy for advanced

hepatocellular carcinoma. J Cancer Res Clin Oncol, 2021. 147(10): p. 3123-

3133.

38. Saeki, I., et al., Effect of body composition on survival benefit of hepatic

arterial infusion chemotherapy for advanced hepatocellular carcinoma: A

comparison with sorafenib therapy. PLoS One, 2019. 14(6): p. e0218136.

39. Iwamoto, H., et al., The Clinical Impact of Hepatic Arterial Infusion

Chemotherapy New-FP for Hepatocellular Carcinoma with Preserved Liver

Function. Cancers (Basel), 2022. 14(19).

ACCEPTED Downloaded from http://journals.lww.com/international-journal-of-surgery by BhDMf5ePHKav1zEoum1tQfN4 a+kJLhEZgbsIHo4XMi0hCywCX1AWnYQp/IlQrHD3i3D0OdRyi7TvSFl4Cf3VC4/OAVpDDa8K2+Ya6H515kE= on 08/05/2024

40. Liu, M., et al., Systematic review of hepatic arterial infusion chemotherapy

versus sorafenib in patients with hepatocellular carcinoma with portal vein

tumor thrombosis. J Gastroenterol Hepatol, 2020. 35(8): p. 1277-1287.

41. Yan, L., et al., A meta-analysis comparing hepatic arterial infusion

chemotherapy and sorafenib for advanced hepatocellular carcinoma. Transl

Cancer Res, 2022. 11(1): p. 99-112.

42. Zhang, W., et al., Hepatic arterial infusion chemotherapy versus sorafenib for

advanced hepatocellular carcinoma with portal vein tumor thrombus: An

updated meta-analysis and systematic review. Front Oncol, 2023. 13: p.

1085166.

43. Ni, J.Y., et al., Transcatheter hepatic arterial infusion chemotherapy vs

sorafenib in the treatment of patients with hepatocellular carcinoma of

Barcelona Clinic Liver Cancer stage C: a meta-analysis of Asian population.

Onco Targets Ther, 2018. 11: p. 7883-7894.

44. Zhuang, B.W., et al., Sorafenib versus hepatic arterial infusion chemotherapy

for advanced hepatocellular carcinoma: a systematic review and metaanalysis. Japanese Journal of Clinical Oncology, 2019. 49(9): p. 845-855.

45. Si, T., et al., Transarterial chemoembolization prior to liver transplantation

for patients with hepatocellular carcinoma: A meta-analysis. J Gastroenterol

Hepatol, 2017. 32(7): p. 1286-1294.

46. Song, M.J., Hepatic artery infusion chemotherapy for advanced

hepatocellular carcinoma. World J Gastroenterol, 2015. 21(13): p. 3843-9.

47. Lai, Z., et al., Lenvatinib, toripalimab plus hepatic arterial infusion

chemotherapy in patients with high-risk advanced hepatocellular carcinoma:

A biomolecular exploratory, phase II trial. Eur J Cancer, 2022. 174: p. 68-77.

48. Zhang, T.Q., et al., Camrelizumab (a PD-1 inhibitor) plus apatinib (an

VEGFR-2 inhibitor) and hepatic artery infusion chemotherapy for

hepatocellular carcinoma in Barcelona Clinic Liver Cancer stage C

(TRIPLET): a phase II study. Signal Transduct Target Ther, 2023. 8(1): p. 413.

49. Kudo, M., et al., Sorafenib plus low-dose cisplatin and fluorouracil hepatic

arterial infusion chemotherapy versus sorafenib alone in patients with

advanced hepatocellular carcinoma (SILIUS): a randomised, open label,

phase 3 trial. Lancet Gastroenterol Hepatol, 2018. 3(6): p. 424-432.

50. He, M., et al., Sorafenib Plus Hepatic Arterial Infusion of Oxaliplatin,

Fluorouracil, and Leucovorin vs Sorafenib Alone for Hepatocellular

Carcinoma With Portal Vein Invasion: A Randomized Clinical Trial. JAMA

Oncol, 2019. 5(7): p. 953-960.

51. Kondo, M., et al., Randomized, phase II trial of sequential hepatic arterial

infusion chemotherapy and sorafenib versus sorafenib alone as initial therapy

for advanced hepatocellular carcinoma: SCOOP-2 trial. BMC Cancer, 2019.

19(1): p. 954.

52. Long, Y., et al., Sorafenib plus hepatic arterial infusion chemotherapy versus

sorafenib alone for advanced hepatocellular carcinoma: A systematic review

and meta-analysis. J Gastroenterol Hepatol, 2023. 38(4): p. 486-495.

ACCEPTED Downloaded from http://journals.lww.com/international-journal-of-surgery by BhDMf5ePHKav1zEoum1tQfN4 a+kJLhEZgbsIHo4XMi0hCywCX1AWnYQp/IlQrHD3i3D0OdRyi7TvSFl4Cf3VC4/OAVpDDa8K2+Ya6H515kE= on 08/05/2024

53. Llovet, J.M., R. Montal, and A. Villanueva, Randomized trials and endpoints

in advanced HCC: Role of PFS as a surrogate of survival. J Hepatol, 2019.

70(6): p. 1262-1277.

54. Yang, J.D., et al., A global view of hepatocellular carcinoma: trends, risk,

prevention and management. Nat Rev Gastroenterol Hepatol, 2019. 16(10): p.

589-604.

55. Cheng, A.L., et al., Updated efficacy and safety data from IMbrave150:

Atezolizumab plus bevacizumab vs. sorafenib for unresectable hepatocellular

carcinoma. J Hepatol, 2022. 76(4): p. 862-873.

ACCEPTED Downloaded from http://journals.lww.com/international-journal-of-surgery by BhDMf5ePHKav1zEoum1tQfN4 a+kJLhEZgbsIHo4XMi0hCywCX1AWnYQp/IlQrHD3i3D0OdRyi7TvSFl4Cf3VC4/OAVpDDa8K2+Ya6H515kE= on 08/05/2024

Figure 1 The flowchart of study identification and selection

ACCEPTED Downloaded from http://journals.lww.com/international-journal-of-surgery by BhDMf5ePHKav1zEoum1tQfN4 a+kJLhEZgbsIHo4XMi0hCywCX1AWnYQp/IlQrHD3i3D0OdRyi7TvSFl4Cf3VC4/OAVpDDa8K2+Ya6H515kE= on 08/05/2024

Figure 2 Comparisons of patients’ survival and radiological response between

the HAIC group and the Sorafenib group. (A)Forest plots of overall

survival (OS) and progression free survival (PFS);(B) Forest plots of objective

response rate (ORR) and disease control rate (DCR) ACCEPTED Downloaded from http://journals.lww.com/international-journal-of-surgery by BhDMf5ePHKav1zEoum1tQfN4 a+kJLhEZgbsIHo4XMi0hCywCX1AWnYQp/IlQrHD3i3D0OdRyi7TvSFl4Cf3VC4/OAVpDDa8K2+Ya6H515kE= on 08/05/2024

ACCEPTED Downloaded from http://journals.lww.com/international-journal-of-surgery by BhDMf5ePHKav1zEoum1tQfN4 a+kJLhEZgbsIHo4XMi0hCywCX1AWnYQp/IlQrHD3i3D0OdRyi7TvSFl4Cf3VC4/OAVpDDa8K2+Ya6H515kE= on 08/05/2024

Figure 3 Associations of the treatment effects between intermediate clinical

endpoints and HCC patients’ survival after receiving HAIC or Sorafenib

(A) Correlation analysis between PFS and OS; (B) Correlation analysis

between ORR/DCR and OS/PFS; (C) Correlation analysis between ORR/DCR

and 1y-OS/1y-PFS. ACCEPTED Downloaded from http://journals.lww.com/international-journal-of-surgery by BhDMf5ePHKav1zEoum1tQfN4 a+kJLhEZgbsIHo4XMi0hCywCX1AWnYQp/IlQrHD3i3D0OdRyi7TvSFl4Cf3VC4/OAVpDDa8K2+Ya6H515kE= on 08/05/2024

ACCEPTED Downloaded from http://journals.lww.com/international-journal-of-surgery by BhDMf5ePHKav1zEoum1tQfN4 a+kJLhEZgbsIHo4XMi0hCywCX1AWnYQp/IlQrHD3i3D0OdRyi7TvSFl4Cf3VC4/OAVpDDa8K2+Ya6H515kE= on 08/05/2024

Figure 4 (A) Multivariable HRs of all potential prognostic factors. Prognostic

factor (number of studies), Left column: HR and 95% CI from pooled analysis

(pooled data if ≥2 studies were included in analysis).

ACCEPTED Downloaded from http://journals.lww.com/international-journal-of-surgery by BhDMf5ePHKav1zEoum1tQfN4 a+kJLhEZgbsIHo4XMi0hCywCX1AWnYQp/IlQrHD3i3D0OdRyi7TvSFl4Cf3VC4/OAVpDDa8K2+Ya6H515kE= on 08/05/2024

Table 1 Baseline Characteristics of included studies

Study Grou

p

Pati

ents

Gend

er

(M/F

)

Age

HBV

(Y/N

)

Child

Pugh

(A/B/

C)

BCL

C

(A/B/

C)

AFP

(ng/ml)

Tumo

ur

Size

( )

Hirami

ne et al

(2011)2

7

HAI

C 45 32/1

3

69.6(

47-

84)

11/3

4 45/0/0 n/a 8.8*103(

0-55.9)

≥5,

n=22

<5,

n=23

Soraf

enib 20 17/3

69.6(

44-

83)

5/15 n/a n/a 7.3*103(

0-97.3)

≥5,

n=8

<5,

n=12

Jeong

et al

(2012)2

6

HAI

C 21 21/0 51(33

-75) 18/3 10/11/

0

0/0/2

1

≥400,

n=12

<400,

n=9

n/a

Soraf

enib 20 11/9 60(49

-75) 13/7 14/6/0 0/0/2

0

≥400,

n=13

<400,

n=7

n/a

Shioza

wa et

al

(2014)2

4

HAI

C 77 64/1

3

67.3±

6.8

14/6

3

49/26/

2

0/42/

35

≥101,

n=47

≤100,

n=30

n/a

Soraf

enib 47 43/4 69.4±

8.2 8/39 39/8/0 0/24/

23

≥101,

n=23

≤100,

n=24

n/a

Nemot

o et al

(2014)2

5

HAI

C 8 6/2 74.9±

3.4 n/a 4/4/0 n/a 279±41

8

5.0±2.

8

Soraf

enib 12 6/6 80.2±

5.4 n/a 10/2/0 n/a 2027±5

219

4.2±2.

1

Kawao

ka et al

(2015)2

3

HAI

C 136 123/

13

67(30

-85)

33/1

03

136/0/

0

0/1/1

35

415.3

(2.6~19

38000)

4.5

(1~1.8

)

Soraf

enib 41 29/1

2

69(30

-81) 36/5 41/0/0 0/3/3

8

208.0

(3~8563

2)

4

(1~19)

Fukuba

yashi

et al

HAI

C 128 113/

15

67

(65.5

±9.3)

33/9

5 79/49 n/a

18471.9

±

77176.9

>5,

n=66

≤5,

n=62

ACCEPTED Downloaded from http://journals.lww.com/international-journal-of-surgery by BhDMf5ePHKav1zEoum1tQfN4 a+kJLhEZgbsIHo4XMi0hCywCX1AWnYQp/IlQrHD3i3D0OdRyi7TvSFl4Cf3VC4/OAVpDDa8K2+Ya6H515kE= on 08/05/2024

(2015)2

9

Soraf

enib 72 51/2

1

69

(68.9

±9.8)

33/3

9 61/11 0/0/7

2

13462.3

±

46460.9

>5,

n=28

≤5,

n=44

Kondo

et al

(2015)3

4

HAI

C 44 32/1

2

71

(54-

84)

1/43 31/13/

0

0/16/

28

588.5

(3–

207890)

n/a

Soraf

enib 83 74/9

70

(37-

88)

15/6

8 78/5/0 0/58/

25

199.7

(1.6–

529490)

n/a

Song et

al

(2015)2

0

HAI

C 50 38/1

2

54.3

± 9.9 44/6 45/5/0 0/0/5

0

<200,

n=15

≥200,

n=35

<10,

n=22

≥10,

n=28

Soraf

enib 60 44/1

6

55.8

± 9.0

41/1

9

47/13/

0

0/0/6

0

<200,

n=20

≥200,

n=38

<10,

n=31

≥10,

n=29

Hatook

a et al

(2016)3

0

HAI

C 65 59/6

67

(46-

84)

15/5

0 65/0/0 0/42/

23

415.3

(5-

191500)

4 (1.5-

14)

Soraf

enib 58 43/1

5

70

(50-

88)

13/4

5 58/0/0 0/52/

6

449 (5-

2446)

3.2

(0.5-

19)

Nakan

o et al

(2017)2

2

HAI

C 44 33/1

1

63.4±

10.0 n/a 44/0/0 n/a

<1000,

n=22

≥1000,

n=22

7.42±

3.33

Soraf

enib 20 17/3 65.4±

8.1 n/a 20/0/0 n/a

<1000,

n=5

≥1000,

n=15

7.42±

5.42

Yang

et al

(2017)2

1

HAI

C 54 50/4 54.4±

11.0

44/1

0

25/29/

0

0/0/5

4

<400,

n=23

≥400,

n=31

12.5±

4.6

Soraf

enib 53 39/1

4

58.0±

9.2

43/1

0

34/19/

0

0/0/5

3

<400,

n=23

≥400,

n=30

9.2±5.

1

Morigu

chi et

al

(2017)1

9

HAI

C 32 29/3

65

(40~8

1)

12/2

0 32/0/0 n/a

466.1

(5.1~34

0140)

7.47

(0~17.

9)

Soraf

enib 14 12/2

68

(53~8

2)

4/10 14/0/0 n/a

416.9

(4.3~21

1634)

6.58

(3.27~

10.8)

Terashi

ma et

al

HAI

C 139 111/

28

69

(≥69,

n=74)

36/1

03

139/0/

0 n/a

≥400,

n=53

<400,

n=86

≥5,

n=51

<5,

n=88

ACCEPTED Downloaded from http://journals.lww.com/international-journal-of-surgery by BhDMf5ePHKav1zEoum1tQfN4 a+kJLhEZgbsIHo4XMi0hCywCX1AWnYQp/IlQrHD3i3D0OdRyi7TvSFl4Cf3VC4/OAVpDDa8K2+Ya6H515kE= on 08/05/2024

(2017)3

2

Soraf

enib 51 45/6

69

(≥69,

n=27)

16/3

5 51/0/0 n/a

≥400,

n=16

<400,

n=35

≥5,

n=8

<5,

n=43

Choi et

al

(2018)1

8

HAI

C 29 25/4 60.3±

9.5 21/8 27/2/0 n/a

260.0

(3.6~84

604.6)

<10,

n=14

≥10,

n=15

Soraf

enib 29 27/2 60.2±

7.3

18/1

1 25/4/0 n/a

130.8

(2.0~22

5971)

<10,

n=12

≥10,

n=17

Lyu et

al

(2018)2

8

HAI

C 180 160/

20

51

(25-

77)

156/

24

119/6

1/0

0/4/1

76 467.3 11.6±

3.8

Soraf

enib 232 216/

16

51

(16-

82)

186/

46

159/7

3/0

0/3/2

29 478.6 11.7±

3.9

Moriya

et al

(2018)3

3

HAI

C 21 16/5

69

(44-

88)

n/a 21/0/0 0/21/

0 n/a n/a

Soraf

enib 45 38/7

73

(43-

86)

n/a 45/0/0 0/45/

0 n/a n/a

Kodam

a et al

(2018)1

5

HAI

C 150 135/

15 68 38/1

12

150/0/

0 n/a 464.2 5

Soraf

enib 134 102/

32 69 22/1

12

102/0/

0 n/a 448 4.2

Kang

et al

(2018)1

4

HAI

C 95 84/1

1

55.3±

7.6

67/2

8

59/36/

0

0/19/

72

12118.3

±

34100.1

8.1±3.

7

Soraf

enib 44 37/7 56.6±

9.0

34/1

0

30/14/

0

0/17/

25

28144.5

±

78708.9

7.4±3.

5

Saeki

et al

(2019)3

7

HAI

C 55 42/1

3

66.7

±

11.4

n/a 36/19/

0

0/55/

0 n/a

7.1

(4.0-

>10.0)

Soraf

enib 78 57/2

1

72.2

± 8.5 n/a 60/18/

0

0/78/

0 n/a

4.0

(2.3–

6.2)

Ueshi

ma et

al

(2020)3

1

HAI

C 429 345/

84 67.4 91/3

38

255/1

74/0 n/a

>400,

n=218

≤400,

n=211

>5,

n=108

≤5,

n=321

Soraf

enib

134

6

1081

/265 69.9 252/

1094

1173/

173/0 n/a

>400,

n=676

≤400,

n=670

>5,

n=275

≤5,

n=107

1

ACCEPTED Downloaded from http://journals.lww.com/international-journal-of-surgery by BhDMf5ePHKav1zEoum1tQfN4 a+kJLhEZgbsIHo4XMi0hCywCX1AWnYQp/IlQrHD3i3D0OdRyi7TvSFl4Cf3VC4/OAVpDDa8K2+Ya6H515kE= on 08/05/2024

AHK

et al

(2021)1

7

HAI

C 20 18/2 n/a n/a 20/0/0 0/0/2

0

1942.53

±

5715.44

n/a

Soraf

enib 29 26/3 n/a n/a 29/0/0 0/0/2

9

26102.5

8±

80097.4

5

n/a

Ahn et

al

(2021)1

6

HAI

C 38 30/8 53.0±

11.6 n/a 27/11/

0 n/a 71341±

14823 n/a

Soraf

enib 35 30/5 58.3±

9.5 n/a 24/11/

0 n/a 69745±

21274 n/a

Han et

al

(2021)3

6

HAI

C 151 136/

15

56.9

± 9.7

128/

23 n/a n/a

909.3

(37–

11,579)

9.0

(6–12)

Soraf

enib 37 31/6

56.3

±

10.1

32/5 n/a n/a

193.2

(12-

3909)

8.8

(6.6-

12.6)

Zaizen

et al

(2021)3

5

HAI

C 88 61/2

7

73.8

± 9.6 6/82 56/32/

0

0/71/

17

4399 ±

24,315

3.72 ±

3.1

Soraf

enib 243 193/

50

72.4

± 9.5

37/2

06

196/4

7/0

0/178

/65

7275 ±

49,351

4.2 ±

2.3

Lyu

Ning et

al

(2022)9

HAI

C 130 115/

15

54

(45~6

1)

120/

10

88/42/

0

0/5/1

25

337.8

(28.5~1

2902)

11.5 ±

4.5

Soraf

enib 132 123/

9

53

(45-

62)

114/

18

93/39/

0

0/9/1

23

304.2

(15.3~3

086.5)

11.0 ±

3.4

Iwamo

to et al

(2022)3

8

HAI

C 418 327/

91

68.6

±

10.9

77/3

41

418/0/

0

0/52/

366

27,729.0

±

154,561.

1

8.29 ±

4.60

Soraf

enib 844 677/

167

70.1

±

9.61

154/

690

844/0/

0

0/305

/532

17,735.3

±

110,114.

8

3.59 ±

3.39 ACCEPTED Downloaded from http://journals.lww.com/international-journal-of-surgery by BhDMf5ePHKav1zEoum1tQfN4 a+kJLhEZgbsIHo4XMi0hCywCX1AWnYQp/IlQrHD3i3D0OdRyi7TvSFl4Cf3VC4/OAVpDDa8K2+Ya6H515kE= on 08/05/2024

Table 2 Prognosis and response rate subgroup analysis of HAIC versus Sorafenib

OS PFS

Subgroup Reference HR 95%CI Reference HR 95%CI

Median size ≤

5cm

15; 23; 25; 27; 30;

31; 35 0.98 0.80-

1.21 25; 30 0.9 0.51-

1.57

Median size >

5cm

9; 14; 18; 19; 21;

22; 28; 29; 36; 37;

38; 45

0.6 0.46-

0.78

9; 14; 18; 19;

22; 28; 29; 36;

45

0.52 0.40-

0.67

Chemotherapy

naive

9; 14; 16; 18; 19;

21; 26; 31; 33; 37;

38

0.67 0.56-

0.81

9; 14; 16; 18;

19; 26 0.51 0.40-

0.65

CTP-A 15; 19; 22; 23; 27;

30; 33; 35; 38 0.82 0.64-

1.05 19; 22; 30 0.50 0.28-

0.91

Refractory to

TACE 15; 27; 30; 34 1.32 1.01-

1.73 30; 34 1.1 0.80-

1.52

EHS (-) MVI

(-) 15; 23; 31; 38 1.25 0.86-

1.81 n/a n/a n/a

EHS (+) MVI

(-) 31; 38 1.35 0.65-

2.84 n/a n/a n/a

EHS (+) MVI

(+) 19; 30; 31; 38 0.72 0.44-

1.17 n/a n/a n/a

EHS (-) MVI

(+)

15; 18; 19; 22; 23;

29; 31; 38 0.58 0.45-

0.75 18; 19; 22; 29 0.49 0.36-

0.66

Propensity

score

matching

9; 18; 24; 28; 29;

30; 35; 36 0.65 0.38-

1.12

9; 18; 28; 29;

30; 34; 36 0.69 0.51-

0.94

1y-OS 1y-PFS

Reference HR 95%CI Reference HR 95%CI

Median size ≤

5cm

15; 23; 25; 27; 30;

31; 35 1.03 0.81-

1.32 25; 30 0.90 0.51-

1.57

Median size >

5cm

9; 14; 18; 19; 21;

22; 28; 29; 36; 37;

38; 45

0.61 0.52-

0.72

9; 14; 18; 19;

22; 28; 29; 36;

45

0.54 0.41-

0.70

Chemotherapy

naive

9; 14; 16; 18; 19;

21; 26; 31; 33; 37;

38

0.66 0.58-

0.76

9; 14; 16; 18;

19; 26 0.5 0.39-

0.65

CTP-A 15; 19; 22; 23; 27;

30; 33; 35; 38 0.85 0.61-

1.17 19; 22; 30 0.50 0.28-

0.91

Refractory to

TACE 15; 27; 30; 34 1.36 1.03-

1.79 30; 34 0.87 0.63-

1.20

EHS (-) MVI

(-) 15; 23; 31; 38 0.90 0.63-

1.30 n/a n/a n/a

EHS (+) MVI

(-) 31; 38 0.69 0.58-

1.83 n/a n/a n/a

EHS (+) MVI

(+) 19; 30; 31; 38 0.74 0.51-

1.06 n/a n/a n/a

EHS (-) MVI

(+)

15; 18; 19; 22; 23;

29; 31; 38 0.69 0.51-

0.95 18; 19; 22; 29 0.50 0.34-

0.75

ACCEPTED Downloaded from http://journals.lww.com/international-journal-of-surgery by BhDMf5ePHKav1zEoum1tQfN4 a+kJLhEZgbsIHo4XMi0hCywCX1AWnYQp/IlQrHD3i3D0OdRyi7TvSFl4Cf3VC4/OAVpDDa8K2+Ya6H515kE= on 08/05/2024

CI: confidence interval; CTP: Child-Turcotte-Pugh; DCR: disease control rate; EHS:

extrahepatic spread; HAIC: hepatic arterial infusion chemotherapy; HR: hazard ratio; MVI:

major vascular invasion; n/a: not available; ORR: objective response rate; OS: overall

survival; PFS: progression free survival; TACE: trans-arterial chemoembolization

Propensity

score

matching

9; 18; 24; 28; 29;

30; 35; 36 0.78 0.52-

1.16

9; 18; 28; 29;

30; 34; 36 0.70 0.51-

0.96

ORR DCR

Reference RR 95%CI Reference RR 95%CI

Median size ≤

5cm

15; 23; 25; 27; 30;

31 5,47

3.12-

9.58

15; 23; 25; 27;

30; 31 1.28

1.10-

1.50

Median size >

5cm

9; 14; 18; 19; 21;

22; 28; 29; 37; 45 4.32

2.75-

6.78

9; 14; 18; 19;

21; 22; 28; 29;

37; 45 1.50

1.23-

1.84

Chemotherapy

naive

9; 14; 16; 18; 19;

21; 26; 37 5.89

3.56-

9.76

9; 14; 16; 18;

19; 21; 26; 37 1.49

1.22-

1.83

CTP-A

15; 19; 22; 23; 27;

30 6.23

3.38-

11.46

15; 19; 22; 23;

27; 30 1.34

1.09-

1.64

Refractory to

TACE 15; 27; 30; 34 4.05

1.42-

11.52 15; 27; 30; 34 1.06

0.77-

1.45

EHS (-) MVI

(-) 15; 23 8.92

3.70-

21.49 15; 23 1.49

1.24-

1.80

EHS (+) MVI

(-) n/a n/a n/a n/a n/a n/a

EHS (+) MVI

(+) 19; 30 2.85

0.72-

11.36 19; 30 1.25

0.64-

2.41

EHS (-) MVI

(+)

15; 18; 19; 22; 23;

29 5.47

2.28-

13.15

15; 18; 19; 22;

23; 29 1.54

1.09-

2.16

Propensity

score

matching 9; 18; 28; 29; 30 3.80

2.00-

7.21

9; 18; 28; 29;

30 1.29

0.98-

1.68 ACCEPTED Downloaded from http://journals.lww.com/international-journal-of-surgery by BhDMf5ePHKav1zEoum1tQfN4 a+kJLhEZgbsIHo4XMi0hCywCX1AWnYQp/IlQrHD3i3D0OdRyi7TvSFl4Cf3VC4/OAVpDDa8K2+Ya6H515kE= on 08/05/2024

Table 3 Subgroup surrogacy analysis of intermediate clinical endpoints

Correlation of treatment effects (R2, 95%CI)

ORRRR vs

OSHR

DCRRR vs

OSHR

1y-OSHR vs

OSHR

1y-PFSHR

vs OSHR

1y-PFSHR

vs PFSHR

Median size ≤

5cm

0.20(-0.98-

0.90)

0.00002(-

0.96-0.97)

0.90(-0.12-

0.999) n/a n/a

Median size >

5cm

0.03(-0.74-

0.56)

0.008(-

0.71-0.61)

0.70(0.44-

0.96)

0.87(0.66-

0.99)

0.978(0.95-

0.997)

Chemotherapy

naive

0.01(-0.76-

0.65)

0.09 (-0.83-

0.51)

0.44(-

0.001-0.92)

0.41(-0.36-

0.96)

0.989(0.95-

0.999)

CTP-A

0.06(-0.93-

0.81)

0.47(-0.98-

0.49)

0.62(-0.32-

0.99)

0.99(0.991-

1.00)

0.96(0.34-

0.99)

Refractory to

TACE

0.015(-

0.97-0.95)

0.54(-0.89-

0.99)

0.97(0.47-

0.99) 0.94 (n/a) n/a

EHS (-) MVI (+)

0.71(-0.63-

0.99)

0.42(-0.99-

0.83)

0.98(0.75-

0.99) 0.94 (n/a) n/a

Propensity score

matching

0.75(-0.99-

0.06)

0.37(-0.97-

0.59)

0.96(0.85-

0.99)

0.85(0.44-

0.99)

0.97(0.90-

0.99)

CI: confidence interval; CTP: Child-Turcotte-Pugh; DCR: disease control rate; EHS:

extrahepatic spread; HAIC: hepatic arterial infusion chemotherapy; HR: hazard ratio; MVI:

major vascular invasion; n/a: not available; ORR: objective response rate; OS: overall

survival; PFS: progression free survival; RR: risk ratio; TACE: trans-arterial

chemoembolization ACCEPTED Downloaded from http://journals.lww.com/international-journal-of-surgery by BhDMf5ePHKav1zEoum1tQfN4 a+kJLhEZgbsIHo4XMi0hCywCX1AWnYQp/IlQrHD3i3D0OdRyi7TvSFl4Cf3VC4/OAVpDDa8K2+Ya6H515kE= on 08/05/2024