Overall, compelling response rates and rates of conversion to resection for patients with initially unresectable colorectal liver metastases treated with hepatic arterial infusion suggest consideration of earlier initiation of the technique in chemotherapy-naïve patients as well as its adoption in patients who have failed first-line systemic chemotherapy before proceeding to second- or third-line regimens.
Nancy E. Kemeny, MD
Memorial Sloan Kettering Cancer Center
The advent of hepatic arterial infusion (HAI) chemotherapy for treating patients with resectable colorectal liver metastases has increased survival outcomes to previously unimaginable levels. In 2017, findings from the largest retrospective study ever conducted to assess the impact of HAI in this population showed that the addition of this therapy extends survival by nearly 2 years.1 However, questions remain about the effectiveness of its use in patients with initially unresectable colorectal liver metastases.
In a review paper published online recently in JAMA Surgery, our team at Memorial Sloan Kettering (MSK) Cancer Center in New York, New York, provided a comprehensive overview of the role of HAI in this difficult- to-treat subset of patients. We discuss the biological rationale, the evolution of combining HAI with systemic chemotherapy, recent evidence for conversion to resection using HAI and systemic chemotherapy compared with systemic chemotherapy alone, and the toxicities and adverse effects (AEs) associated with HAI.2
Overall, compelling response rates and rates of conversion to resection for patients with initially unresectable colorectal liver metastases treated with HAI suggest consideration of earlier initiation of the technique in chemotherapy-naïve patients as well as its adoption in patients who have failed first-line systemic chemotherapy before proceeding to second- or third-line regimens.2
We pioneered HAI at MSK and have been continuously expanding and improving its clinical application for the past 30 years. We have generated most of the evidence to date supporting its use in patients with initially unresectable colorectal liver metastases. The broader oncology community is increasingly acknowledging its potential: Several North American centers have opened new HAI programs, and some new trials in progress may provide additional insights for improving patient outcomes even further.
The Biological Rationale for HAI
Colorectal cancer (CRC) is the third most common cancer in the United States, accounting for an estimated 145,600 new cases and 51,000 deaths in 2019.3 More than half of patients with CRC develop liver metastases.4 Five-year overall survival (OS) exceeds 50% for select MSK patients with limited colorectal liver metastases who undergo complete resection.5-8 Most patients experience recurrence after hepatectomy, but about 20% are cured after resection on long-term follow-up in our series.5
Resection is the primary contributor to longterm survival, but only 15% to 20% of patients with colorectal liver metastases present with resectable metastases.9-11 Major work in the field has focused on converting more patients to resection. Newer systemic therapies have been transformative for downstaging patients from having initially inoperable liver metastases to potentially resectable liver metastases. These therapies include irinotecan12 and oxaliplatin13 as well as the combination regimens FOLFOX (leucovorin, fluorouracil, and oxaliplatin), FOLFIRI (leucovorin, fluorouracil, and irinotecan),14 and FOLFOXIRI (fluorouracil, oxaliplatin, and irinotecan), with or without agents that target EGFR or VEGF.15-17
However, modern chemotherapy rarely cures patients with initially unresectable colorectal metastases or those with extrahepatic disease. For those patients, median survival is 22 months to 24 months,18 median progression-free survival (PFS) is 10 months, and response rates in the second-line setting are only 10% to 30%.19-21
Since the hepatic artery carries the main blood supply to liver metastases, we can deliver agents with high first-pass hepatic extraction directly to tumors via HAI while sparing healthy liver tissue by perfusing the agents, mainly through portal circulation.22
HAI is typically administered via the gastroduodenal artery through a surgically implanted pump, hepatic arterial port, or percutaneous catheter connected to an external pump. Limiting systemic toxic effects with HAI also means that we can treat patients concurrently with systemic chemotherapies at near-full doses.23-25 The most common AE from HAI of floxuridine is hepatic enzyme elevation and, rarely, biliary sclerosis. To decrease hepatic toxicity, we administer dexamethasone in the pump and adjust doses appropriately. Irinotecan and oxaliplatin have also been used in HAI, mostly in Europe and Asia.26,27
Combining HAI With Systemic Chemotherapy
Between January 1996 and December 2000, the multicenter Cancer and Leukemia Group B 9481 trial compared HAI alone with systemic chemotherapy alone. The trial randomized 135 patients with initially unresectable colorectal liver metastases to receive either HAI consisting of floxuridine and leucovorin or systemic bolus fluorouracil and leucovorin.
The HAI group fared significantly better than the systemic chemotherapy group, with a longer median OS of 24.4 months versus 20.0 months (P = .003), a response rate of 47% versus 24%, and time to hepatic progression of 9.8 months versus 7.3 months, respectively. However, patients in the HAI group had a significantly shorter time to extrahepatic progression: 7.7 months versus 14.8 months for the systemic chemotherapy group.28
The results of this trial generated lively debate about the relevance of HAI alone in patients with initially unresectable colorectal liver metastases. However, by the time of its publication in 2006, most institutions had already moved on to use HAI in combination with systemic chemotherapies, and several phase I and phase II trials were underway.
The approval of oxaliplatin and irinotecan in the late 1990s led to the next generation of HAI trials testing combination regimens.2 For example, a phase I trial at MSK tested HAI of floxuridine combined with 2 separate oxaliplatin-based therapies, irinotecan or fluorouracil and leucovorin, in 36 patients who progressed on first-line chemotherapy. Response rates approached 90% in both groups; median OS was 36 months for the irinotecan group and 22 months for the fluorouracil and leucovorin group.29
A separate phase I trial at MSK was among the first prospective studies to examine the effectiveness of HAI in converting initially unresectable colorectal liver metastases to resectable or abatable disease. Forty-nine patients received HAI of floxuridine with oxaliplatin and irinotecan. The response rate was 92% in this heavily pretreated patient population, Kemeny and colleagues reported. Median OS was 51 months for patients who were chemotherapy naïve and 35 months for those who had been treated previously.30
Systemic options are limited to investigational agents, such as regorafenib (Stivarga)31 and TAS-102 (trifluridine/tipiracil; Lonsurf),32 for patients with initially unresectable colorectal liver metastases who progress on standard chemotherapies, but these agents produce response rates of only 1% to 2%. However, HAI is a viable salvage approach for controlling disease in these patients.
At MSK, we examined results for 110 patients with initially unresectable colorectal liver metastases who were refractory to fluorouracil with leucovorin, irinotecan, and oxaliplatin with or without bevacizumab (Avastin) or anti-EGFR therapy, and who were then treated with HAI of floxuridine and best systemic therapy. The response rate was 35%—a rate that is not possible with systemic chemotherapy alone in refractory patients.33
Discoveries of some of the molecular drivers of disease have led to investigations of combining systemic chemotherapy with HAI, with or without targeted therapies, such as anti-EGFR agents for RAS wild-type mutations and anti-VEGF agents. However, efforts targeting the molecular characteristics of initially unresectable colorectal liver metastases are still in early stages.
Effectiveness of HAI to Convert Initially Unresectable Colorectal Liver Metastases to Resection
The primary goal of multimodal treatment for initially unresectable colorectal liver metastases is to optimize response to make resection possible. Research has shown that patients whose disease can be converted to resection fare as well as those whose disease was resectable up front.10,34
Kemeny and colleagues examined the effectiveness of HAI in converting initially unresectable colorectal liver metastases to resectable or abatable disease in 49 patients. Conversion to resection was achieved in 47% of patients overall and 57% of those who were chemotherapy naïve.30
Despite a substantial burden of disease among these patients, and a strict definition of irresectability, the response rate was 76%. Conversion to resection was the only factor independently associated with prolonged OS and PFS. Three-year OS rates were an astonishing 80% for patients who underwent hepatectomy compared with only 26% for those whose disease remained inoperable.35
MSK investigators recently updated this trial’s results with longitudinal data from an expanded cohort of 64 patients. At a median follow-up of 86 months, 52% of patients had achieved conversion to resection, a rate more than 2 times greater than the historical average. Median PFS was 13 months, and median OS was 38 months, with a 5-year OS of 36%.36
Outcomes were significantly better for chemotherapy-naïve patients compared with those who had been treated previously: Response rates were 86% versus 61% , PFS was 19.7 months versus 10.0 months, and median OS was 76.6 months versus 29.7 months, respectively. Only conversion to resection was significantly associated with OS on multivariate analysis. The approach also demonstrated curative potential: Nine patients (14%) were disease free at about 5 years of follow-up.36
Managing Toxicities and Adverse Effects
It is essential to balance our enthusiasm for liver-directed therapy with an awareness of the toxic effects of this treatment. Allen and colleagues found that 544 patients (22%) treated at MSK from 1986 to 2001 experienced complications. The incidence of complications decreased with surgical experience. Most issues were salvageable, with 80% of pumps functioning for at least 2 years.37
A review of drug-related toxic effects from HAI in 4580 patients treated at MSK found gastrointestinal symptoms in 25%, hepatic toxic effects in 22%, and myelosuppression in 9%. HAI of floxuridine was mainly associated with hepatic enzyme elevations, which lead to biliary sclerosis in some people.38
Another MSK study found that 4.6% of patients undergoing HAI of floxuridine required a stent, and yet there was no difference in survival between those who received salvage treatment with a stent and dilation compared with patients without biliary complications.39 A further study among 50 randomized patients found a higher dose tolerance of floxuridine at 5 months (P = .05) and an increased response rate (P = .03) with concurrent administration of dexamethasone.40
Advancing the Treatment of Colorectal Liver Metastases
Determining up front whether we can resect patients’ metastases requires technical expertise gained in a high-volume institution in a setting of multidisciplinary oncologic care. Further, administering HAI is complex and requires a dedicated team of surgeons, oncologists, radiologists, and nurses to ensure that patients achieve the best outcomes possible.
At MSK, we have pioneered the use of HAI in initially unresectable colorectal liver metastases and generated a substantial portion of the evidence in the field to date. We are currently conducting a randomized phase II study (NCT01312857) of HAI of floxuridine and dexamethasone and intravenous irinotecan, 5-fluorouracil, and leucovorin, with or without the anti-EGFR agent panitumumab (Vectibix), in patients with wild-type KRAS and resected liver metastases.
We are also excited to see results from a phase III trial (NCT02102789) by Chinese investigators who are comparing HAI of floxuridine plus modified FOLFOX6 and dexamethasone versus modified FOLFOX6 alone, with a primary outcome of margin-negative resection. This study may finally answer the question about the effectiveness of HAI plus systemic chemotherapy compared with systemic chemotherapy alone.
We look forward to future advances in the field, including refining patient selection, learning more about the genetic determinants of response, incorporating targeted therapies into rational trial design, and developing multi-institutional registries comparing outcomes for current and alternative liver-directed treatment approaches.