Update on Hepatic Intra-Arterial Chemotherapy

Introduction

About 140,000 people in the United States were diagnosed with colon or rectal cancer in 1996, and 40% of those patients will ultimately die of metastasis from that disease.[1] The liver is the dominant metastatic site in the majority of these patients, and more than 80% of patients with hepatic metastases succumb to liver failure.[2]

Despite the fact that metastatic colorectal cancer is usually a systemic disease, in some patients, the liver is the only site of metastasis. This is evidenced by the fact that, with careful patient selection, resection of isolated hepatic metastases can result in long-term disease-free survival. For example, resection of up to three colorectal cancer lesions in patients without evidence of extrahepatic colorectal tumor leads to a 5-year survival of approximately 25%.[3,4]

The incorporation of intraoperative ultrasound of the liver into the staging evaluation, as well as the application of prognostic factors, such as carcinoembryonic antigen (CEA), lesion size, and stage of the primary tumor, may permit the identification of a group of patients who would enjoy an even better survival.[5] Unfortunately, although about one-third of patients with hepatic metastases appear to have liver-only tumor, in no more than 10% of these patients is the tumor amenable to curative hepatic resection.

The regional delivery of chemotherapy to patients with liver tumors is not a new concept. Even though colorectal metastases appear to migrate to the liver via the portal vein, macrometastases in the liver derive more than 80% of their blood supply from the hepatic arterial circulation; in contrast, normal hepatocytes are supplied primarily by the portal circulation. Therefore, the administration of chemotherapy into the hepatic artery allows for the selective delivery of drug to the tumor with relative sparing of normal hepatocytes.[6] The pharmacology behind this strategy is well-defined[7]: depending on a drug's clearance and toxicity profile, a marked increase in the area under the curve (AUC) may be achieved. If a dose-response relationship for that drug and that disease exists, this could then translate into proportionally greater efficacy.

The regional behavior of colorectal cancer metastatic to the liver and favorable drug pharmacokinetics led to extensive clinical investigation of hepatic intra-arterial (HIA) chemotherapy in the 1970s and '80s, initially via percutaneously placed catheters. The introduction of a totally implantable pump system made this an ambulatory treatment. Encouraging, if not inflated, response rates ranging from 29%[8] to 88%[9] were reported and engendered numerous randomized protocols comparing systemic vs HIA chemotherapy. Despite these studies, the optimal dosing and role of HIA therapy in the management of colorectal liver metastases remain unclear.

This review will summarize and critique the randomized studies and will address the technologic and surgical advances that have now better defined the technique necessary for the safe, successful administration of HIA chemotherapy. A series of phase II trials will be reviewed, and the design of a Cancer and Leukemia Group B (CALGB) randomized study assessing this modality will be described. Although the focus of this review is on liver metastases from colorectal cancer, a brief description of HIA therapy in other malignancies will also be presented.

HIA Chemotherapy in Untreated Patients

Results of Randomized Trials

Four major prospective trials comparing systemic fluoropyrimidine chemotherapy vs HIA floxuridine (FUDR) chemotherapy delivered via the implanted pump in untreated patients with liver-only colorectal metastases were completed in the early 1980s (Table 1).[10-13] In all, 375 patients were enrolled in these randomized studies. In each study, the response rate to HIA chemotherapy was superior to the systemic counterpart. However, in none of the studies did this translate into a survival improvement for the HIA approach.

Two European trials conducted during the late 1980s randomly assigned patients to either best supportive care or HIA FUDR. The studies of Rougier et al[14] and Allen-Mersh et al[15] both demonstrated a survival advantage of HIA therapy, but most of the control patients did not receive systemic chemotherapy, raising the concern that these patients, at least by American standards, received suboptimal treatment.

The latter study,[15] done in the United Kingdom, was conducted with particularly rigorous quality control: Every pump was implanted by the same surgeon (Allen-Mersh). A thorough quality-of-life analysis, administered by dedicated nurses, demonstrated a prolongation of normal-quality survival, with respect to physical symptoms, anxiety, and depression, in patients treated with HIA chemotherapy, as compared with those given best supportive care.

The data of all of these trials have been subjected to two different meta-analyses. Each meta-analysis confirmed the trends that were apparent in the individual studies; namely, that response rates were greater with HIA therapy. However, although there was a trend toward an improvement in survival, only when the patients enrolled on the no-treatment control arms were included did the difference reach statistical significance.[16,17]

Why would the substantial improvement in response rate not translate into a survival advantage? There are numerous possible explanations for this:

• Colorectal cancer is a systemic disease, and thus, even effective regional chemotherapy to liver metastases does not alter patient survival.

Although colorectal cancer metastatic to the liver is usually a systemic disease, as noted above, selected patients with isolated liver metastases from colorectal cancer can be cured with hepatic resection.[3]

Also, most patients with unresectable liver metastases die of progressive liver dysfunction,[2] and even with effective management of liver metastases, most patients treated with HIA chemotherapy in these studies die from advancing liver involvement.[17] Therefore, the argument that systemic therapy is needed is not necessarily supported by the facts.

• Hepatobiliary toxicity, including biliary sclerosis, led to the cessation of therapy in many patients.

Without question, the dose of FUDR in most of the studies--.3 mg/kg/d for 14 days--was too high. This was not appreciated until many patients suffered permanent biliary damage.[18] Therefore, the HIA therapy was overly toxic, and nearly half of the patients stopped treatment for that reason rather than for treatment failure.

• Surgical complications, including catheter dislodgment and gastrointestinal misperfusion (ie, inadvertent administration of chemotherapy to the stomach or duodenum), made this a very difficult therapy to deliver safely.

Reflecting the experience of the operating surgeon[19] and the infancy of the technique,[20] surgical complications in these studies were prohibitive. For example, of 33 patients randomized to HIA therapy in the Mayo Clinic study,[13] HIA treatment was never begun in 5 patients because of surgical complications, was prematurely discontinued in another 5 because of later problems, and was abrogated in 4 others because of gastrointestinal misperfusion.

• Crossover of patients from systemic to HIA therapy may have masked a survival advantage.

This may have confounded the findings of the two largest studies,[10,11] since there is evidence that fluoropyrimidine-based HIA chemotherapy produces a response rate better than 30% in previously treated patients.[17,21]

The results of these trials led to one of two interpretations: either the therapy was not worth pursuing or needed to be improved markedly to make it worthwhile. Focusing on potential areas of improvement, investigators concentrated on patients without radiographic evidence of extrahepatic tumor and pursued two avenues: (1) standardizing the surgical technique of pump placement and (2) optimizing the delivery of chemotherapy from the standpoint of toxicity and response.