Getting Chemotherapy Directly to the Liver: The Historical Evolution of Hepatic Artery Chemotherapy

Experimental Trials 1950 -1960

“The regions best adapted to this type of therapy will be determined by the anatomy of the arterial blood supply. A tumor, located in an area whose blood supply is derived mainly from one artery, would be most suitable”

Klopp, 1950 (1)

In the 1940’s, the newly discovered chemotherapy agents were administered intravenously(2). By the middle of the century physicians and scientists were searching for ways to achieve regional therapy that would maximize cytotoxic concentrations and minimize systemic side effects. Two physicians independently developed and published their pioneering techniques for arterial chemotherapy in 1950 and laid the groundwork for regional chemotherapy: Calvin T. Klopp, a surgeon from George Washington University, trained at Memorial Hospital in New York City (now named Memorial Sloan Kettering Cancer Center, MSKCC), and Howard R. Bierman, a physiologist and oncologist from City of Hope Medical Center in Los Angeles. Both physicians described novel techniques for regional chemotherapy with nitrogen mustard, each on a case series of patients with various tumors(1, 3).

Klopp designed his trail after an accidental administration of nitrogen mustard to the brachial artery of a patient with Hodgkin’s lymphoma(1). As he described it: “An erythema occurred which persisted several days and was followed by vesiculation and ulceration of the hand and forearm. Eventually the intense local reaction subsided and no irreversible changes were observed”. This mishap demonstrated that arterial infusion could be a safe route of administration for chemotherapy.

Klopp and Bierman were driven by the rationale that if the effects of chemotherapy could be localized to the region of the tumor via the arterial route, effective treatment would be possible with minimal damage to normal tissues in other areas. Bierman developed a technique for percutaneous catherization of the aorta and celiac access through the brachial artery, publishing an abstract in April 1950 that was accredited with being the first report on the administration of chemotherapy through an artery(3). In the same month, Klopp presented at the American Surgical Association conference his experience using a polyethylene tube to cannulate vessels feeding neoplastic tumors, including the external carotids, iliac and temporal arteries. He suggested a fractionated dosage regimen so the catheter could be left in the artery for a week or more(1). Preliminary results of this work, published some 6 months later, demonstrated that intraarterial delivery allowed for a higher dosage compared to what was given intravenously, due mainly to a lesser depressant effect on the hematopoietic system.

Following these publications, arterial administration of chemotherapy was studied for various cancers, including in the liver. At the beginning of the 1950s, single doses of chemotherapeutic agents were introduced directly into the hepatic artery(4). In 1952, Dr. George T. Pack, Chief Surgeon of the Gastric and Mixed Tumors service at Memorial Hospital NYC, described in the institution’s annual report a joint research program with Dr. David A Karnofsky from the hospital’s Division of Experimental Chemotherapy Service. The two departments examined the effect of nitrogen mustard injected through the hepatic artery on primary and metastatic cancers of the liver. In his report, Dr. Pack stated that 18 patients were recruited to the study, without elaborating on the outcomes. To the best of our knowledge, no dedicated paper focusing on that study was published, although partial results on one patient were eventually reported in a case series published by Dr. Karnofsky in 1953(5): a 46-year-old male with primary carcinoma of the liver who received intrahepatic injections of nitrogen mustard and had a clinical response that lasted 3 months. The method of administration (direct surgical or angiographic catheterization) was not mentioned in either Dr. Pack’s report or Dr. Karnofsky’s published series. In a review published in 1958, some 5 years later, Dr. Karnofsky concluded that nitrogen mustard therapy to the hepatic artery failed to achieve clinical improvement(6). This is interesting because, in a review in 1985(7), credit for the first use of hepatic artery infusion chemotherapy (HAI) was given to Byron and co-workers, including Bierman, who published their results in 1961(8).

The rationale for HAI was strengthened in animal experiments conducted in 1954, thanks to a chance discovery during an investigation of the hepatic circulation in rabbits(9). Charles Breedis, a pathologist at Pennsylvania University School of Medicine, noted that hepatic tumors failed to stain when India ink was injected into the portal vein, whereas hepatic tissue between the tumors became intensely black. He concluded that metastases in the liver are perfused almost exclusively via the hepatic artery, whereas normal hepatocytes derive most of their blood supply from the portal vein.

Refining techniques 1960-1970

“Several questions must go unanswered at the present time: What is the most effective technique of hepatic artery infusion? What, for example, is the drug of choice? Is the presence or absence of a response related to any intrinsic features of the tumour?”

Editorial, British Medical Journal, 1968(10)

Research in regional chemotherapy to the liver broadened in the 1960s with experiments involving various cytotoxic drugs delivered via continuous intra-arterial infusion. Intraarterial differs from intravenous administration by the relative pressure against which the infusion system must operate in that the former that necessitates mechanical devices that serve as pumps to propel the agents. Early infusion devices posed technical difficulties and ominous risks that made them possible only in a hospital setting, often with little or no response. Dr. Bayard Clarkson from Memorial Hospital NYC reported in 1961 on experience with hepatic artery infusion of Fluorouracil (FU), Floxuridine (FUDR), 5-fluoro-deoxycytidine (FCDR) and Methotrexate(11, 12). Of 17 cases with metastatic carcinoma to the liver or primary hepatocellular carcinoma, only nine patients showed any measurable response, and these were described as “quite transitory”. Arterial chemotherapy at the time was associated with complications and mortality. A fatal air embolism was reported in one patient after an infusion bottle “had inadvertently been allowed to empty completely during infusion”(13). Health care personal were instructed that “infusion bottles should always be connected in tandem and the last one always kept full in reserve, so as to avoid the possibility of air embolism”(11). A case series from the University of Wisconsin(14) reported the following: “When our program was initiated in 1961, a modified Baron food pump was used. This proved unsatisfactory, required considerable supervision, and six patients developed air emboli with transient neurological deficits”. The devices also tended to be bulky and noisy(15).

An important improvement in HAI was achieved thanks to the development of an external spring-activated pump designed for continuous infusion (Figure 1) that allowed the patient to be treated ambulatory. In 1963 Watkins and co-workers(16) developed a portable miniature chronometric pump for outpatient use that had to be refilled with drugs every 5-7 days.

Figure 1.

External pump for ambulatory chemotherapy infusion. A patient with advanced metastatic adenocarcinoma of the liver receiving continuous ambulatory arterial cancer chemotherapy through a treatment catheter placed in the hepatic artery using a low volume portable pump apparatus. Reprinted from Sullivan⁴, with permission.

In 1961, one of the very first techniques for surgically-positioned hepatic artery cannulation was described by Theodore R. Miller et al(17), affiliated with Memorial Hospital NYC. His method entailed cannulating the gastroepiploic artery with a no. 6 ureteral catheter that is passed into the gastroduodenal artery and then retrogradely into the hepatic artery (Figure 2). The catheter is brought out to the abdominal wall through a gastro-epiploic stump. This allowed for two continuous administration methods: angiographically by percutaneous catheterization (based on Bierman’s method that was later modified by Byron et al.(8)) or transabdominally via laparotomy. The surgical method has the advantages of enabling full abdominal exploration to rule out nonhepatic metastases, direct evaluation of the extent of liver involvement, and avoidance of extra hepatic infusion by ligating near-by arteries. Despite these advantages, percutaneous access was still used, especially in patients with extensive liver metastases who were deemed unfit for laparotomy, and in cases of marked hepatomegaly, where exposure difficulties may make it impossible to directly catheterize the hepatic artery. There were, however, complications associated with percutaneous catheterization: while a catheter may be originally well-positioned in the hepatic artery, it could migrate so that drug is infused into the stomach, duodenum, or spleen instead of directly into the liver; and the indwelling intra-arterial catheters were associated with sepsis, bleeding and thrombosis(18). Operative placement was shown to be associated with lower rates of thrombosis, bleeding, catheter failure, and displacement compared to percutaneous insertion(7).

Figure 2.

Hepatic artery catherization achieved by inserting a catheter into the right gastroepiploic artery and threading it retrogradely into the hepatic artery. (A) An infuser device attached to a stump of the gastroepiploic artery supplies constant intrahepatic artery chemotherapy (illustration by Frank Henry Netter, reprinted with permission from Elsevier.).

Clinicians at Lahey Hospital & Medical Center in Burlington, MA contributed greatly to the development and popularization of HAI during this time. In a 1964 article in the New England Journal of Medicine, Dr. Robert D. Sullivan and colleagues(19) described a technique for placing the catheter that included ligating all non-hepatic branches followed by verification of hepatic perfusion with fluorescent dye injection. This was one of the first reports of a technique to reduce non-hepatic perfusion, and gained wide acceptance for its reduction of the chemotoxic side effects to adjacent non-hepatic organs. The technique did not include a concurrent cholecystectomy. On the face of it, Sullivan et al.’s results(19) appeared superior to those of systemic therapy, but in the absence of adequate controls fair comparison could not be made.

The revolutionary implanted pump 1970 – 1990

“A burgeoning of interest in regional approaches to therapy of the liver”

Opfell, 1985(7)

Enthusiasm surrounding HAI waned in the 1970s due to the huge effort and costs involved in the set-up and maintenance of the systems, and the potential morbidity, not to mention the reduced quality of life for patients attached to an extracorporeal pump for weeks to months(7).

Technological advances came to the rescue in 1970 in the form of an entirely implanted pump that made it possible to administer arterial infusions to outpatients safely and reliably with relatively few restrictions on their lifestyle(20, 21). This was achieved thanks to Perry Lynnfield Blackshear, Jr., a mechanical engineer from the University of Minnesota. His work, a spinoff from missile technology, resulted in the development of an implantable subcutaneous pump with a self-contained “inexhaustible” power source. It contained a fluorocarbon liquid in equilibrium with its vapor phase, resulting in a vapor pressure that provided the power source. It was first used as a delivery vehicle for heparin(22). The co-inventor, surgeon Henry Buchwald, experimented with the infusion device as a mean of infusing FUDR into the hepatic artery (Figure 3)(23), publishing his results in 1980. The pumps were more reliable, almost free of malfunction or failure, and had fairly accurate flow rates. This allowed the patients freedom to swim and shower and perform other activities, without the necessity for frequent loading. Further technological advances in HAI included silicone catheters, which are less prone to promote thrombosis(15).

Figure 3.

Implantable infusion pump developed by Blackshear et al. in the 1970s (20-22). (A) Patient with a subcutaneous implanted drug pump. (B) Ex-vivo and in-vivo demonstration of an Infusaid drug infusion pump refill by injection, in 1983. The full 50 ml pump delivered fluid at a constant rate of 2.5-3 ml/day. Reprinted from Balch et al⁴⁹ with permission.

With the promising technical achievements of an implantable pump, more attention was paid to patient selection, type of cancer and type of chemotherapy. The main focus of HAI was on antimetabolites. Research had shown that FUDR and FU have short serum half-lives(24) and are effective against colorectal cancer, which often metastasizes to the liver. Ensminger et al.(25) demonstrated in-vivo the extent to which these agents are catabolized by liver enzymes: more than 94% of the FUDR was removed during the first pass in the hepatic arterial circulation. Indeed, he described it as “nearly a one pass phenomenon”(25). These and other data showed that when there is sufficient hepatic detoxification, the systemic toxicity seen at higher drug dosages decreases, further strengthening the theoretical advantage of hepatic arterial infusion as a promising therapeutic approach.

A more aggressive surgical approach was put forth in 1985 to avoid regional misperfusion of the drugs. Up until then, two of the most concerning side effects of HAI were gastritis and upper gastrointestinal hemorrhage(18). To avoid gastroduodenal toxicity, particular care was taken to ligate all branches of the gastroduodenal artery. Daly et al.(26) reported ligating the right gastric artery and small branches to the duodenal bulb. Hohn et al.(27) offered an even more aggressive surgical approach to avoid toxicity: ligating all vessels supplying the superior border of the stomach, and a cholecystectomy to prevent chemotherapy-induced cholecystitis. With such measures, surgeons succeeded in achieving a negligible incidence of gastroduodenal and pancreatic toxicity.

Research on HAI flourished in the 1980s. In 1985, reported responses varied from 25 to 83% in colorectal liver metastasis, and from 12 to 75% in primary liver cancer(7). An important finding in those years was the adverse event of sclerosing cholangitis in patients treated with FUDR infusion, reported in 1985 by Kemeny et al.(28). A timeline summarizing the reviewed achievements in HAI from 1950 to 1990 is presented in Figure 4.

Figure 4.

Timeline with important achievements in hepatic artery chemotherapy and publication trends between the years 1950-1990. Annual publication entries in Medline (PubMed) for the terms: hepatic artery infusion chemotherapy.

A review of the development of HAI must include mention of other locoregional therapies. In 1960, regional perfusion of the liver, created by temporarily excluding the liver from the general circulation, was tested in patients with liver cancer using nitrogen mustard, without success(29). The basis for transarterial hepatic embolization (TAE) was established in the 1960’s after it became evident that hepatic artery occlusion, caused by prolonged catheter placement in patients receiving HAI via percutaneous catheterization, is well tolerated by patients(30). But only in the 1970’s did angiographers utilize embolization agents to treat liver tumors. In the 1980’s this treatment modality further evolved to include transcatheter chemoembolization (TACE)(31).

Current perspective

In the past three decades, the introduction of prospective randomized controlled trials yielded more grounded evidence of an improvement in response rate and survival over systemic therapy in selected patients. Since the first randomized trial on HAI in MSKCC was published in 1987(32), more than 100 publications from the center have demonstrated the efficacy of regional chemotherapy to the liver(33-35).

HAI has shown survival benefit after metastasectomy in colorectal cancer, and as conversion therapy for initially unresectable disease. When compared with chemotherapy alone, the use of adjuvant HAI resulted in a 50% reduction in liver related mortality at 5 years (5.8% versus 14.3%)(36). In a retrospective single center propensity-score matching analysis, patients who received perioperative HAI had a median survival of approximately 2 years longer than patients without HAI (67 versus 44 months)(37). A recent review from our Center(34) concluded that HAI should be initiated earlier in patients with initially unresectable colorectal liver metastasis (CRLM) who are chemotherapy naïve, and that HAI can augment rates of resectability in patients who have failed first-line systemic chemotherapy. Data from large-scale randomized clinical trials are lacking, yet multi-center collaborations have been formed to mitigate this gap. The PUMP trial, a randomized controlled trial investigating the efficacy of HAC for low risk CRLM, is currently underway(38).

In advanced HCC, a meta-analysis found HAI to be superior to systemic therapy (sorfenib) in overall survival of patients with portal vein thrombosis (39). The largest trial demonstrated a difference in median survival of 1.6 months (median overall survival 7.1 with HAI versus 5.5 with systemic therapy)(40). There is also evidence that HAI is a promising modality for the treatment of intrahepatic cholangiocarcinoma, improving local disease control(41). A meta-analysis of regional therapy for unresectable intrahepatic cholangiocarcinoma reported the longest median survival with HAI (22.8 months), followed by radioembolization (13.9 months) and TACE (12.4 months)(42).

Technical achievements in the last two decades include surgical placements by laparoscopy(43) and robotic surgery(44), and a wakening interest in percutaneous radiology-guided catheterization(45). A recent review of the technical aspects of hepatic artery infusion pump placement describes the MSKCC technique(46).

HAI’s potential is being increasingly acknowledged by the oncology community, with new programs opening in North America, Asia and Europe. The challenges with HAI have to do with the need to adapt a coordinated interdisciplinary approach to ensure the program is successful and risks are minimized. HAI therapy requires close collaboration between surgical oncology, medical oncology, interventional radiology, and oncology nursing. The experience accumulated by leading centers can help new programs, by shortening the learning curve and mitigating early adaptive phase complications. An additional aspect hampering wider adaptation has been inconsistent supply of infusion pumps. Albeit, as demand for HAI increases, manufacturers will adjust accordingly.

The benefits of HAI should be weighed against the possible risks. One of the major disadvantages of liver-directed therapy with FUDR is the development of biliary sclerosis(47). In a recent trial, this complication developed in 6.7% of 208 patients treated for colorectal liver metastases (36). While these patients may need biliary stent placement, experience at MSKCC suggests that there is no difference in survival among those with biliary complications who received salvage therapy (stenting or dilation) compared to those without complications (47). These results are possible with mitigating strategies that limit the risk of biliary sclerosis, such as co-administration with dexamethasone (48), and optimization of its management. An additional caveat of HAI is that it might complicate and even contradict other liver direct therapies in the armamentarium, such as yttrium-90 radioembolization and transarterial chemoembolization.

CONCLUSION

The realization that regional chemotherapy to the liver carries the significant advantage of maximizing local concentration of cytotoxic agents and minimizing unwanted systemic effect led to collaborations between surgeons, medical oncologists and engineers aimed at developing improved intrahepatic chemotherapy administration techniques.