Hepatic Artery Infusion Chemotherapy for Primary and Secondary Malignancies of the Liver: State of the Art and Current High-Level Evidence

Abstract

Background

Hepatic artery infusion chemotherapy (HAI) has been proposed as a valuable adjunct for multimodal therapy of primary and secondary liver malignancies. This review provides an overview of the currently available evidence of HAI, taking into account tumor response and long-term oncologic outcome.

Summary

In colorectal liver metastases (CRLM), HAI in combination with systemic therapy leads to high response rates (85–90%) and conversion to resectablity in primary unresectable disease in up to 50%. HAI in combination with systemic therapy in CRLM in the adjuvant setting shows promising long-term outcomes with up to 50% 10-year survival in a large, non-randomized single-center cohort. For hepatocellular carcinoma patients, response rates as high as 20–40% have been reported for HAI and long-term outcomes compare well to other therapies. Similarly, survival for patients with unresectable intrahepatic cholangiocarcinoma 3 years after treatment with HAI is reported as high as 34%, which compares well to trials of systemic therapy where 3-year survival is usually below 5%. However, evidence is mainly limited by highly selected, heterogenous patient groups, and outdated chemotherapy regimens. The largest body of evidence stems from small, often non-randomized cohorts, predominantly from highly specialized single centers.

Key Message

In well-selected patients with primary and secondary liver malignancies, HAI might improve response rates and, possibly, long-term survival. Results of ongoing randomized trials will show whether a wider adoption of HAI is justified, particularly to increase rates of resectability in advanced malignant diseases confined to the liver.

Introduction

The liver has the unique anatomic configuration of a dual blood supply over the portal vein (PV) and the hepatic artery (HA), an important consideration for any systemic treatment of liver tumors [1]. Primary and secondary tumors derive their blood supply mainly from the arterial-rather than the portal system and are therefore susceptible to intra-arterial chemotherapy [2]. Tumor growth is reduced by inhibition of angiogenesis and induction of tumor cell apoptosis [3–5]. Hepatic artery infusion chemotherapy (HAI) is applied by an implantable pump with a catheter inserted into the gastroduodenal artery (GDA). The chemotherapy dosage in the tumor is increased up to 400-fold compared to systemic treatment, while the normal hepatocytes receive a lower dose due to the different perfusion pattern. Due to the high first-pass effect of drugs in the liver, fewer systemic side effects occur [6]. Implantation of a HAI pump is a standardized procedure that can nowadays be performed minimally invasive either laparoscopically or via robotic approach [7–9].

Colorectal cancer is the fourth leading cancer worldwide in terms of incidence with over 1.9 million new cases in 2020 [10]. Colorectal liver metastases (CRLM) are the main prognostic driver. Approximately 25% of the patients present with synchronous liver metastases, while another 25% will develop metachronous CRLM at a later stage [11]. Only 20–30% of those CRLM are resectable upon first detection; therefore, a multimodal treatment approach is crucial and involves both systemic and local therapies. Current systemic chemotherapy (SCT) regimens combine two to three agents, i.e., 5-Flurouracil (5-FU), leucovorin, and/or oxaliplatin or irinotecan. Furthermore, modern chemotherapy combined with biological agents has increased response rates to 60–70%. Ultimately and including downstaging by HAI, up to 40% of initially unresectable CRLM can be converted to resectable disease, leading to a prognosis that is comparable to upfront resectable CRLM [12, 13].

On the other hand, primary liver cancer is less common with 0.9 million new cases worldwide in 2020 [14]. Hepatocellular carcinoma (HCC) and cholangiocarcinoma (CCC) are the most frequent primary liver tumors and often diagnosed in an intermediate or advanced stage. In this subgroup of patients, current treatment recommendations include transarterial chemoembolization (TACE), selective internal radiotherapy (SIRT), and systemic therapy [15, 16].

Similar to CRLM, primary liver tumors are amenable to intra-arterial therapy due to their supplying arteries, the so-called feeding vessels. Especially the management of CCC, a heterogenous and difficult to diagnose entity, is challenging due to often advanced disease stage at the time of diagnosis [17]. Multimodal treatment for downstaging is of paramount importance to convert unresectable to resectable disease in primary liver cancer [18].

Current indications for HAI are (1) downstaging of unresectable tumors, (2) adjuvant treatment after resection, or (3) a palliative therapeutic approach. HAI is usually combined with SCT. In this article, we provide a review of the highest level of evidence on the current application of HAI for primary and secondary liver tumors. The evidence level was appraised by the Centre for Evidence-Based Medicine (http://www.cebm.net).

Indication and Contraindication

For the preoperative work-up, multiple considerations have to be carefully taken into account when assessing patients’ eligibility for HAI.

Arterial Perfusion Anatomy

A prerequirement is the single arterial inflow to the liver without inflow to adjacent organs. Furthermore, a patent GDA with an adequate diameter is required to deliver the agent to the HA. High-quality preoperative imaging studies are mainly derived from CT-angiography while catheter-based angiography is rarely performed. Images need to be available preoperatively for the evaluation of the vessel anatomy, anomalies, and patency. Aberrant, replaced or accessory vessel configurations need to be identified [19].

Liver Function

Damage to the liver parenchyma has either already occurred or contributed to the development of a primary liver tumor, in addition the damage of past and future drug toxicities has to be anticipated. Pre-existing liver disease and portal hypertension may increase morbidity from the pump placement and also reduce tolerability of chemotherapy which reduces the therapy effect. Currently, there is no clear, evidence-based cut-off, below which the liver function is considered insufficient for HAI.

Liver function tests are manifold and generally indirectly reflect the liver function. Basic tests include excretory and synthetic function, namely bilirubin, albumin, and coagulation factors. Ascites may indicate liver dysfunction or advanced disease stages. Patients must also be able to tolerate the often simultaneously administered SCT. HAI in patients with Child-Pugh A liver disease has shown to be feasible and safe and can potentially offer a less toxic treatment option than systemic monotherapy in patients with Child-Pugh B cirrhosis [20, 21].

Surgical Technique

The implantation of a HA catheter is a complex technical procedure which requires an experienced hepatobiliary surgeon (Fig. 1). Minimally invasive and especially a robot assisted approach have been adopted, nevertheless the open approach remains a preferred technique [22]. The following steps need to be followed during the implantation of the pump:

• 1.Exploration of the abdominal cavity to identify extrahepatic disease.
• 2.Division of the falciform ligament and ligamentum teres.
• 3.Definition of the arterial anatomy by identification of the right gastric artery, the common HA, and the GDA.
• 4.Isolation of the GDA and dissection of lymphatic tissue from the hepatoduodenal ligament to gain sufficient space and to identify accessory and replaced arteries to the liver that need to be ligated to avoid extrahepatic application of chemotherapy.
• 5.A transverse skin incision in the left middle abdomen for the subcutaneous pump.
• 6.The catheter is inserted into the abdominal cavity and the GDA is clamped proximally and ligated distally.
• 7.The catheter is inserted through a transverse arteriotomy until the tip of the catheter lies just outside the HA lumen, and the position of the artery is secured with a nonabsorbable suture.
• 8.Cholecystectomy is always performed to prevent chemotherapy-induced cholecystitis.

Fig. 1.

Hepatic arterial catheter placement in the GDA.

Surgical complications comprise HA thrombosis, dissection of the HA due to implantation, catheter dislocation, catheter leakage or misperfusion, and pocket infections. Failure of the pump can occur. Further, morbidity can be caused by the application of the chemotherapy and includes chemical hepatitis, biliary sclerosis, or gastrointestinal symptoms like ulcers, gastritis, or duodenitis.

Percutaneous implantation has been described via a femoral arterial access or less often through other peripheral arteries [23–25]. The catheter is placed in the GDA and fixed with coils. The pump or port was similarly placed in the abdominal wall.

Colorectal Liver Metastasis

HAI without Systemic Therapy in Unresectable CRLM (Level of Evidence 1a)

The sole use of HAI without systemic therapy was abandoned due to the risk of extrahepatic disease progression without systemic therapy and limited benefit [26–29]. A meta-analysis compared mostly fluoropyrimidine-based HAI only with SCT for unresectable CRLM [30]. Except for one trial, the included studies were performed between 1987 and 2006. The pooled results of nine RCTs suggested a higher percentage of patients with tumor response undergoing HAI (43% vs. 20%; relative risk of downstaging: 2.10 [95% CI: 1.59–2.79; p < 0.001]). Although the authors found that HAI had a better overall survival (OS; hazard ratio [HR] 0.83 (95% CI: 0.70–0.99), p = 0.04), this finding was derived from a significant advantage against a subgroup of the control arm, in which more than 1 third of the patients had no therapy at all in the SCT group. Moreover, the analysis has to be interpreted with caution due to the significant heterogeneity among the included studies. Most of the included RCTs used fluoropyrimidine-based SCT, which is inferior to current combination therapies. Therefore, the effect of the evaluated SCT may be underestimated with regard to OS and tumor response.

The influence of the different HAI agents must also be taken into consideration. Some RCTs used floxuridine as therapeutic agent, with the advantage of higher liver extraction concentrations than 5-FU. But the high extraction rate can also be seen as a disadvantage in diseases not confined to the liver, as systemic blood concentration of floxuridine possibly is too low to treat extrahepatic metastases [31]. Lorenz and Müller [32] found a 3.2-fold higher incidence rate of extrahepatic metastases in patients with HAI-floxuridine (41%) compared with HAI-5-FU and folinic acid (13%). Another disadvantage of floxuridine is the hepatotoxicity, which can lead to primary biliary sclerosis and cholecystitis. 5.4% of patients treated with HAI-fluorodeoxyuridine (FUDR) died as a result of biliary sclerosis. In 10% of the HAI group and 5.6% of the control group, treatment was terminated due to severe toxicity.

Regarding the control SCT group, the current gold standard of 5-FU in combination with oxaliplatin and/or irinotecan and an EGFR- or VEGF-inhibitor was not part of most studies. Even though, OS rates were predominantly not different between the treatment groups, only hepatic tumor progression rates were lower in the HAI group [32].

In the SCT group severe diarrhea was the leading side effect also with dose-limiting toxicity and subsequent liver damage (e.g., biliary sclerosis). The latter can be reduced using dexamethasone. The dose-limiting toxicities in the SCT arm were diarrhea and myelosuppression.

HAI in Combination with SCT for the Treatment of Unresectable CRLM (Level of Evidence 2a)

The importance of SCT to enable resectability was emphasized by Lam and colleagues in a RCT, demonstrating response rates in patients with SCT of 64%, whereof 23% underwent curative liver resection [33]. A combined SCT and liver-directed therapy to improve response rate and progression-free survival (PFS) is an appealing treatment approach to further increase the chance of downstaging. The first study describing HAI with SCT was published in 1989 by Safi et al. [34] as phase I study comparing HAI-FUDR alone against HAI-FUDR and SCT but without a significant difference in response rates. The response rates in the CT-scan were 50% in both groups. No improvement of short-term survival was seen in either group during the 36-month follow-up; however, in patients with response to either HAI +/− chemotherapy the survival time increased significantly. RCTs in this area are rare and evidence mainly comes from a large cohort from the Memorial Sloan Kettering Cancer Center (MSKCC).

A phase II study combining HAI with SCT found the combination to be safe and effective. D’Angelica et al. [35] combined HAI-FUDR with best SCT based on prior patients’ chemotherapy history in patients with unresectable CRLM. In 76%, a tumor response was accomplished and most importantly in 47%, a conversion to resectable disease with prolonged survival was possible. Also, the combination HAI-FUDR with bevacizumab was investigated but showed an increased biliary toxicity, so the authors advised against this combination therapy. Considering surgical morbidity, 35% of the patients suffered from minor complications and only 1 patient had a major complication.

The long-term results from this MSKCC cohort including more patients described an even higher conversion to resection rate (52%) of prior unresectable CRLM. The 5-year survival rate in this single-arm study was excellent with 36% [36].

The application of irinotecan, oxaliplatin, and 5-FU in combination with intravenous cetuximab achieved response rates of 41% with an OS of 25.5 months. About 30% of patients accomplished resectability [37–40]. Kemeny et al. [41] reported even better numbers, 92% response rate and a resectability rate of 47% comparing HAI-5-FU and SCT with oxaliplatin and irinotecan (Fig. 2) [42]. An overview of studies is shown in Table 1.

Fig. 2.

Summary of the application of hepatic arterial infusion chemotherapy with clinical outcomes.

Table 1.

Studies assessing the outcome of HAI

Author	Year	Country	Patients, n	Treatment (HAI)	OS	Response rate, %	Complete resection, %	Study design
Focan et al. [39] 1999	1999	Belgium	56	FUDR + SCT 5-FU with different infusion rates	n.a	38–48	22–27	RCT
Carnaghi et al. [40] 2007	2007	Italy	39	5-FU + SCT oxaliplatin and folinic acid	21 months	41	21	Single-arm phase II
Boige et al. [42] 2008	2008	France	44	Oxaliplatin + SCT 5-FU and leucovorin	16 months	87	18	Prospective
Kemeny et al. [41] 2009	2009	US	49	Floxuridine and dexamethasone + SCT oxaliplatin and irinotecan	39.8 months	92	47	Phase I
D’Angelica et al. [35] 2015	2015	US	49	FUDR + SCT depending on history + bevacizumab	38 months	76	47	Single-arm phase II
Levi et al. [37] 2016	2016	France	64	Irinotecan, oxaliplatin, 5-FU + SCT cetuximab	25.5 months	41	30	Single-arm phase II
Pak et al. [36] 2018	2018	US	64	FUDR + SCT depending on history + bevacizumab	38 months	73	52	Single-arm phase II

HAI, hepatic arterial infusion chemotherapy; SCT, systemic chemotherapy; US, United States; FUDR, fluorodeoxyuridine, 5-FU, 5-FU, n.a. not applicable.

Adjuvant Therapy of CRLM (Level of Evidence 1b)

A recent meta-analysis included eight RCTs and 652 patients [43]. The pooled HR for OS was 0.91 (95% CI: 0.72–1.14) and therefore showed no difference in patients with or without HAI. Half of the studies investigated SCT as control group, whereas the other half compared to no treatment. When adding non-randomized studies, there was a benefit of HAI in terms of OS (HR: 0.77; 95% CI: 0.64–0.93). Important for the interpretation of the results, only two RCTs were published after 2010 and only two studies included more than 100 patients.

In further subgroup analyses including non-randomized studies, a benefit was predominantly shown using floxuridine (HR: 0.76, 95% CI: 0.62–0.94) and the combination of HAI with adjuvant SCT (HR: 0.75, 95% CI: 0.59–0.90). A major limitation of the included studies is once again the heterogeneity in disease burden, response-, and resectability assessment. SCT regimens are mainly a combination of 5-FU, oxaliplatin and irinotecan, but various new biologicals may be added. With modern agents applied through HAI, response rates of CRLM increase. The opportunity to initiate HAI in patients who progress under SCT even after multiple lines of treatment is a promising expansion of life-prolonging therapeutic options in a challenging to treat advanced disease.

Again, the largest study evaluating outcomes in this setting comes from MSKCC. In patients who underwent complete CRLM resection with HAI, 10-year OS was 51% compared to 31% in patients receiving SCT only [44]. The 10-year follow-up was updated from their institutional database and improved to an 57% OS [45]. Of note, few centers have published more than an initial series of their programme and consensus and benchmarks that should be achieved in terms of patient outcome are lacking.

Ongoing studies like the French multicenter combined phase II and III PACHA-01 trial (NCT02494973) include patients with high-risk disease, defined as at least four surgically or interventionally treated CRLM [46]. Patients in the interventional arm receive HAI with oxaliplatin and 5-FU, those in the control arm FOLFOX. The trial is designed to assess the efficacy after curative-intent surgery on RFS at 18 months. If 39% (21 of 54) of the patients show no sign of hepatic recurrence, the trial will be continued to a phase III to demonstrate superiority of the HAI arm in terms of 3-year RFS. Results are expected in 2028. Furthermore, the multicenter phase III PUMP trial is currently conducted in the Netherlands (Netherlands Trial Register 7493) and compares HAI combined with systemic 5-FU to no therapy after resection of CRLM [47]. Patients with low-risk disease, corresponding to a clinical risk score of 0–2 are enrolled and the primary endpoint is PFS. An overview of ongoing studies is shown in Table 2.

Table 2.

Ongoing randomized controlled trials assessing the effect of HAI on CRLM registered on clinicaltrials.gov

NCT	Acronym	Size, n	Intervention (HAI)	Control	Efficacy endpoint	Phase	Country
NCT05103020		100	Oxaliplatin + iv FOLFIRI + bevacizumab or cetuximab	FOLFIRI+ bevacizumab or cetuximab	Curative-intent resection rate, OS, PFS, ORR	2	Korea
NCT04898504	EXCALIBUR1+2	45	5-FU	Arm 2: transplant	OS, QoL	2	Norway
				Arm 3: 2nd line iv chemotherapy
NCT02494973	PACHA-01	220	Oxaliplatin, iv 5-FU + LV5	mFOLFOX6	RFS	2/3	France
NCT02885753	OSCAR	348	Oxaliplatin, iv 5-FU + LV5 + panitumumab or Bevacizumab	FOLFOX + panitumumab or bevacizimab iv mFOLFIRINOX + panitumumab or bevacizimab	PFS	3	France
			Oxaliplatin, iv FOLFIRI + panitumumab or bevacizumab
NCT02102789		142	FUDR + iv mFOLFOX6	mFOLFOX6	R0 rate, ORR, R0/R1 resection, RFS, OS, PFS	3	China
NCT01312857		75	FUDR + iv FOLFIRI + panitumumab	FUDR + FOLFIRI	RFS, OS	2	US
NCT00200200		73	Bevacizumab 5-FU + iv oxaliplatin + LV5	5-FU + irinotecan + LV5	PFR, OS	2	US
NCT03500874		288	FUDR + iv FOLFOX	FOLFOX	RFS, OS	3	China
NCT03678428		160	Oxaliplatin FUDR + iv irinotecan	FOLFOXIRI	ORR, R0 rate, tumor regression, PFS, RFS, OS	3	China

Primary Liver Tumors

Hepatocellular Carcinoma (Level of Evidence 1b)

Management of HCC is guided by the 2022 updated BCLC algorithm [48]. Surgical resection is only offered to patients with small HCC, without macrovascular invasion and preserved liver function. Transplantation is mainly reserved for patients with HCC within the Milan criteria or adapted versions in most countries [49].

For intermediate or advanced-stage HCC, there are several options to deliver local and systemic therapies. Transarterial therapies play a role in the management of HCC outside the transplant criteria or as bridging procedure. Besides catheter-based embolization such as TACE or locally delivered radiation, HAI is part of multimodal therapy options in this algorithm. In practice, HAI is mainly combined with systemic therapy since HAI alone did not result in a survival benefit [50].

The efficacy and safety of a combination of sorafenib and HAI have been assessed in multiple RCTs [51–54]. Ikeda et al. [51] demonstrated three-fold higher response rates (7.3% vs. 21.7%, p = 0.09) and a better median survival when patients were treated with sorafenib and HAI Cisplatin compared to sorafenib alone (10.6 vs. 8.7 months, HR: 0.68; 95% CI: 0.44–1.05; p = 0.073). Similarly, in a randomized multicentre setting He et al. [52] demonstrated a much higher response rate (40.8% vs. 2.5%, p < 0.001) and an almost twice as long OS (13.4 vs. 7.1 months, HR: 0.35; 95% CI: 0.26–0.48; p < 0.001) when HAI with FOLFOX was added to sorafenib versus sorafenib alone for patients with HCC and PV invasion. Conversely, the largest randomized multicentre phase III trial, including 206 patients, found no significant difference in OS when combining sorafenib with HAI administering low-dose cisplatin and FU (11.5 vs. 11.8 months, p = 0.955) [53]. Another small randomized trial including patients with major PV thrombosis showed higher response rates in combination therapy of sorafenib and oxaliplatin/5-FU HAI versus sorafenib alone (41% vs. 3%, p < 0.001) [54]. Also, the median OS was 16.3 months with sorafenib plus HAI and 6.5 months with sorafenib alone (HR: 0.28; 95% CI: 0.15, 0.53; p < 0.001). In these mentioned trials, overall adverse event rates were similar in the HAI and the control arm. Catheter-related complications were low, with a reported incidence of 6–12%. Importantly, only one reoperation was required for an adverse event across the four RCTs. While RCTs demonstrated the feasibility of HAI, the added survival benefit of combining sorafenib was inconsistent. Ultimately, a confirmatory large multicentre RCT is needed to assess the efficacy of HAI in HCC. Table 3 lists ongoing randomized studies registered on clinicaltrials.gov.

Table 3.

Ongoing randomized controlled trials assessing the effect of HAI on HCC identified on clinicaltrials.gov

NCT	Size, n	Intervention (HAI)	Control	Efficacy endpoint	Phase	Country
NCT03949231	200	PD-1/PDL1 inhibitor	PD-1/PDL1 inhibitor	OS, PFS	3	China
NCT04777942	320	Neoadjuvant TACE-HAI + surgery	Surgery	PFS, OS	–	China
NCT04424043	280	Neoadjuvant TACE-HAI + surgery	Surgery	PFS, OS	–	China
NCT03780049	304	FOLFOX + H101 (recombinant human type-5 adenovirus)	FOLFOX	OS, PFS	3	China
NCT03009461	64	FOLFOX + Sorafenib	Sorafenib	OS	2	China
NCT05231382	426	Raltitrexed, oxaliplatin (SALOX)	FOLFOX	PFS, OS, Objective response	3	China
NCT03591705	240	TACE + FOLFOX	FOLFOX	PFS, OS, Downstaging rate		China
NCT04181931	320	nTACE-HAI (FOLFOX) + surgery	Surgery	PFS, OS		China
NCT04178642	138	Idarubicin-lipiodol	No adjuvant therapy	RFS, OS	2	France
NCT04595864	40	Neoadjuvant mFOLFOX6	Surgery	OS, RFS, PFS	3	China
NCT04947826	100	FOLFOX + HLX10 (PD-1 antibody) + HLX04 (VEGF antibody)	FOLFOX	ORR, PFS, OS, TTR	2	China
NCT05313282	140	mFOLFOX7 + apatinib + camrelizumab	Apatinib + camrelizumab	PFS, OS, TTR, DOR, ORR, DCR	3	China
NCT03469479	252	FOLFOX	Surgery	OS, DFS	3	China
NCT05263219	230	FOLFOX + DEB-TACE	FOLFOX	OS, ORR, DCR, PFS, conversion rate to resection	3	China
NCT03722498	100	FOLFOX	Sorafenib	PFS, OS, ORR	2	China
NCT04667351	400	Oxaliplatin, leucovorin, and 1,200 mg/m2 5-FU	Oxaliplatin, leucovorin, and 2,400 mg/m2 5-FU	PFS, OS, ORR, DCR	3	China
NCT05007587	60	mFOLFOX + lenvatinib	Oxaliplatin/raltitrexed + lenvatinib	ORR, DCR, OS, PFS	1	China
NCT05198609	214	FOLFOX + camrelizumab + oral apatinib	Camrelizumab + oral apatinib	OS, PFS, TTR, DOR, ORR	3	China
NCT05171166	156	FOLFOX + donafenib + TACE	Donafenib + TACE or DEB-TACE	PFS, OS, ORR, DCR	2/3	China
NCT04687163	400	FOLFOX 130 mg/m2 + sorafenib or lenvatinib	FOLFOX 85 mg/m2 + sorafenib or lenvatinib	OS, PFS, ORR, TTR	3	China

Cholangiocellular Carcinoma (Level of Evidence 2b)

Patients with CCC often present with advanced disease, and survival is limited. For intrahepatic CCC (iCCC) in early stages, surgery can improve 5-year survival rates by up to 40% [55]. The role of transplantation remains controversial in unresectable disease as selection criteria are currently not well defined. Moreover, while some advanced have been made in systemic treatment for patients with advanced iCCC and other biliary tract cancers, median survival with the most active regimen is still only just above 12 months [51]. Local treatment modalities such as radio- and chemoembolization or HAI are appealing treatment options when the tumor is confined to the liver.

High-level evidence for HAI in iCCA is scarce. In a single-arm phase 2 trial of HAI-FUDR combined with systemic gemcitabine and oxaliplatin, 38 patients with initially unresectable tumors demonstrated a radiologic response rate of 58% and disease control in 86%. The effect was independent of lymph node status. Four patients (11%) even were converted to a resectable stage [56].

Long-term data from prospective studies have also shown excellent tumor control in patients with unresectable iCCC treated with HAI with 48% of patients showing partial response as per RECIST criteria and 3-year survival of almost 40% without surgical resection [57–59]. Kasai and colleagues reported a case series of 20 patients with advanced iCCC with a 60% response rate when patients underwent combined therapy with 5-FU and pegylated interferon with HAI. However, 25% of the patients in this cohort already had lung metastasis [60]. Ghingirelli reported a disease control rate of 91% in a series of 12 patients treated with a combination of gemcitabine and oxaliplatin [61].

The latest published bi-institutional study including 268 patients from the Netherlands and the MSKCC reported a longer survival in a HAI cohort as compared to a cohort treated with systemic gemcitabine and cisplatin (OS at 3 years 34.3% in the HAI group vs. 3.5% in the SCT only group) [62]. The adjusted HR for HAI was 0.27 (95% CI: 0.19–0.39). Of note, all patients receiving HAI were treated at MSKCC whereas the SCT only group comes from two Dutch centers. This design impairs comparability of the groups and increases the risk of unmeasured confounding.

Treatment of histologically proven multifocal iCCC with HAI-5-FU in 141 patients was compared to resection (R) in 178 patients in a retrospective multicentre cohort [63]. Eligible patients treated over a 18-year period had a higher tumor burden in the HAI group with regard to bilobar disease (HAI: 88.0% vs. R: 34.3%), median tumor size (HAI: 8.4 cm vs. R: 7.0 cm), and proportion of patients with 4 or more lesions (HAI: 67% vs. R: 24%). The 30-day mortality in the resection group was eight times higher in the resection group (HAI: 0.8% vs. R: 6.2%), while the median OS was comparable (HAI: 20 months vs. R: 19 months). Based on these results, the indication for resection of multifocal iCCC needs reconsideration when no survival benefit can be achieved as compared to HAI; however, the risk of surgical resection was significantly higher.

Unresectable iCCC have been treated with intra-arterial floxuridine and a favorable 3-year OS of 39.5% compared to SCT (0%) was achieved in a meta-analysis of 4 studies and 144 patients, including the abovementioned phase II trial by Cercek et al. [56, 64]. Local therapy in this setting is especially appealing because patients usually die from biliary obstruction and liver failure rather than a systemic disease progression. To further assess the benefit of HAI in addition to systemic therapy with gemcitabine and oxaliplatin in patients with unresectable iCCC, a multicentre phase II RCT is currently recruiting patients (NCT04891289). CCC in other locations has rarely been treated with HAI. A phase II trial conducted in China assessed tumor response in 37 patients with advanced perihilar CCC. The administered HAI consisted of oxaliplatin and FUDR, maintenance therapy was oral capecitabine [23]. The overall achieved response rate was 68% and the disease control rate was 89%. Table 4 lists ongoing randomized studies registered on clinicaltrials.gov.

Table 4.

Ongoing randomized controlled trials assessing the effect of HAI on CCC identified on clinicaltrials.gov

NCT	Size, n	Intervention (HAI)	Control	Efficacy endpoint	Phase	Country
NCT04891289	164	FUDR + Gemcitabine/oxaliplatin	Gemcitabine/oxaliplatin	PFS, OS	2	USA
NCT03771846	188	FOLFIRINOX	Gemcitabine/oxaliplatin	OS, PFS, TTP	3	China
NCT01862315	55	FUDR + gemcitabine/oxaliplatin	Gemcitabine/oxaliplatin	PFS, response in MRI	2	USA
NCT04961970	18	FOLFOX	Gemcitabine/cisplatin	OS, PFS, TTP	3	China

OS, overall survival; PFS, progression-free survival; TTP, time to progression.

HAI and Interventional Oncology Therapies

Thermal Ablation

To treat small-volume disease, the sequential combination of thermal ablation with either radiofrequency (RFA) or microwave ablation (MWA) was assessed in patients with CRLM and HCC. For CRLM, the largest series comes from MSKCC including 286 patients with 415 CRLM. Interestingly, there were no biliary complications in HAI-naive patients, versus 11% in HAI patients. Predictors for biliary complications were biliary dilatation, bevacizumab therapy and an ablation margin >10 mm. Therefore, the authors advocate for an ablation margin between 6 and 10 mm with a resulting 76% local tumor control rate and improved biliary complication incidence (4%) [65].

In a randomized study, Oyama et al. [66] compared HAI with RFA ablation with RFA only in patients with early stage HCC. The potential advantage of the combined approach is the destruction of tumor tissue with the prevention of recurrence by HAI. Although only including 70 patients, the intrahepatic recurrence-free survival was better in the combination group compared to ablation alone (HR: 0.468; 95% CI: 0.235–0.896; p = 0.022). The primary endpoint, recurrence-free survival at 3 years, was 54% in the combination group versus 34% with ablation alone; however, this 20% difference was not statistically significant (HR: 0.597, 95% CI: 0.320–1.091; p = 0.094) [66].

For advanced HCC, Liu et al. [67] compared HAI in combination with lenvatinib to HAI, lenvatinib and MWA in 150 consecutive patients. The group including MWA showed longer median OS (>30 months vs. 13.6 months; p = 0.010), longer PFS (12.8 vs. 5.6 months; p < 0.001), and longer intrahepatic PFS (14.6 vs. 6.8 months; p = 0.002). Only 1 patient developed bilioma after thermal ablation [67]. The combination of HAI and thermal ablation mainly lacks high-quality trials; therefore, the next step should be to collect more reliable and solid evidence in subsequent prospective studies.

Radioembolization

In a prospective phase I, study from MSKCC SIRT using yttrium-90 microspheres was evaluated as a salvage therapy for liver-predominant CRLM in 19 patients with progression after HAI and SCT. In this heavily pretreated population, median local PFS and OS after SIRT were 5.2 months and 14.9 months. Similar results could be achieved in a retrospective analysis on 103 CRLM patients from MSKCC with a local PFS of 4 months (95% CI: 3.3–4.8) and OS of 11.3 months (95% CI: 7.9–15.1). In multivariate analysis number of extrahepatic disease sites, carcinoembryonic antigen level, albumin, tumor differentiation level, and sum of the two largest tumor diameters were independently associated with OS [68, 69].

Conclusion

HAI might improve response rates, resectability, and long-term survival in well selected patients with primary and secondary liver malignancies. The combination of SCT with HAI is especially promising for conversion therapy of initially unresectable tumors. Other areas of interest consist of the adjuvant setting in patients with high-risk liver disease or the palliative setting for liver-only disease with the potential to minimize systemic adverse events. However, high-level evidence incorporating up-to date systemic treatment is limited, and future studies should focus on homogenous patient selection and standardized response evaluation.

Conflict of Interest Statement

The authors have no conflicts of interest or financial ties to disclose.

Funding Sources

There was no funding for this study.

Author Contributions

C.K., S.S., and P.C.M. performed literature search, drafted the manuscript, and created tables and figures; P.D. created figure and critical revision of the manuscript; V.H., U.K. G.F.H., A.T.B. B.P.M., and O.K. performed literature search and critical revision of the manuscript.

Funding Statement

There was no funding for this study.