A potential role for perioperative hepatic arterial infusion chemotherapy?


  • N. Joseph Espat M.D., M.S.

    Corresponding author
    1. Division of Hepatobiliary Surgery, Department of Surgery, University of Illinois at Chicago, Chicago, Illinois
    • Division of Hepatobiliary Surgery, Department of Surgery, MC 958, University of Illinois at Chicago, 840 S. Wood Street, MC/958, Chicago, IL 60612
    Search for more papers by this author
    • Fax: (312) 355-1987

  • See referenced original article on pages 590–7, this issue.


The liver is the most common site of distant failure and cause of disease-related mortality in patients with advanced colon carcinoma. In the study by Sadahiro et al., a randomized controlled study evaluated the effects of prophylactic hepatic arterial infusion chemotherapy for the prevention of liver metastasis in patients with TNM Stage II or Stage III colon carcinoma.

See also pages 590–7.

The liver is the most common site of distant failure and cause of disease-related mortality in patients with advanced colon carcinoma.1, 2 Synchronous hepatic metastases from colon carcinoma portend a poor prognosis and similarly, the subsequent identification of hepatic metastasis denotes disease progression and is strongly associated with a negative survival outcome. Considering these observations, the prevention of liver metastases would be expected to result in an improved survival advantage for patients with colon carcinoma.

The obvious question is, of course, how to prevent hepatic metastases? For patients with lymph node-positive (American Joint Committee on Cancer Stage III) disease, we generally can expect that there is a “high risk” for eventual hepatic metastasis. Efforts at preventing the progression of colon carcinoma to the liver after resection of lymph node-positive tumors have focused largely on systemic chemotherapy with the overall aim of systemic rather than regional disease eradication and/or inhibition of residual tumor cell growth. Several key observations have been made that should encourage the consideration of other potential treatment strategies:

  • 1Radiographic assessment for hepatic disease has limitations dependent on the imaging modality employed; thus not all “negative” assessments are in fact negative.3
  • 2Hepatic micrometastatic disease derives a blood supply from the portal vein,4 but to my knowledge no conclusive evidence of decreased hepatic metastases or survival advantage has been demonstrated by randomized trial meta-analysis for portal vein infusion therapy.5
  • 3Measurable hepatic metastatic disease is predominantly vascularized via the hepatic artery; compared with portal vein infusion,4 hepatic arterial infusion (HAI) chemotherapy delivery results in a significantly improved radiographic response. However, survival outcome data have, to my knowledge, been inconclusive.6
  • 4Hepatic extraction of specific chemotherapy agents is widely variable (fluorodeoxyuridine [FUDR] > 5-fluorouracil [5-FU]) and arterial delivery reportedly achieves higher hepatic concentrations compared with portal vein infusion.7

These admittedly specific and selected observations can be employed to structure a potential patient treatment strategy. Considering that patients with lymph node-positive colon carcinoma may already harbor micrometastatic hepatic disease at the time of colon resection and because the liver is the most common site of failure in patients with colon carcinoma, therapy therefore should be directed toward this potential hepatic disease with the best hepatic-extracted drug infused via the vascular route that achieves the highest hepatic drug concentrations.

In this issue of Cancer, the results of a clinical trial founded on some of these observations are reported in the article by Sadahiro et al.8 In this report, a prestudy statistical power analysis was performed; 316 patients with preoperatively staged clinical Stage II or Stage III (according to the 1997 version of the International Union Against Cancer TNM staging system) colon carcinoma (cecum to rectosigmoid only) were enrolled. Patients scheduled to undergo colon resection were randomized to either HAI or control treatment arms. HAI patients received a 3-week course of continuous HAI of 5-FU (1 week of HAI prior to colon resection and 2 weeks postoperatively). Control patients underwent colon resection only. Patients who were enrolled in the study but who did not receive any HAI were analyzed on an intent-to-treat basis and comprised the HAI-2 group. With a median follow-up of 59 months, HAI treatment was found to result in a significantly improved disease-free survival, overall survival, and liver-metastasis-free survival for patients with Stage III disease.

Clarifying comments regarding the study by Sadahiro et al. are necessary. Clinical staging rather than pathologic staging was employed to determine enrollment eligibility into the study, but pathologic rather than clinical stage of disease was employed to determine which group actually benefited from the treatment protocol. The study employed 5-FU (at a dose of 250 mg/day plus dexamethasone) rather than FUDR because of Japanese regulatory issues. The limited duration of arterial infusion reported in the study (3 weeks) was achieved by transfemoral to hepatic artery catheters (without reported arterial complications), not HAI pumps. Chemotherapy beyond the 3-week study period was administered at the discretion of the treating physician (81% of patients in the HAI arm and 78% of patients in the control arm received adjuvant therapy beyond the trial with fluoropyrimidine-based regimens).

As would be expected, the overall disease-free survival for patients with Stage II disease was not found to be different between treatment arms, but was significantly so for patients with Stage III disease who were treated with HAI (P = 0.0009), with a risk ratio for reduction in death from disease recurrence of 60%. When evaluating overall survival, there again was no significant difference detected between the treatment arms for Stage II patients, whereas Stage III patients in the HAI arm fared significantly better (P = 0.002) than those in the control arm, with a 63% risk reduction reported. Analysis of liver metastasis-free survival revealed quite similar outcomes.

The study findings of improved overall disease-free survival, a risk reduction in disease recurrence, and improved liver metastasis-free survival are encouraging; however, the rationale for the extra week of preoperative HAI therapy was not presented.

Was the rationale for “prophylactic” therapy the concern for the potentially inaccurate clinical staging used as part of the eligibility criteria? Or were the authors concerned about the potential for the presence of micrometastatic disease that would not be evident by routine standard pathologic assessment but would be observed by methods such as immunohistochemistry/step-section (which reportedly may occur in as many of 20% of patients9), thus upstaging the patients' pathologic stage? Neither is likely given that patients would be subjected to unnecessary treatment in the case of pathologic Stage II disease or better and the pathologic assessment described was by routine pathologic lymph node assessment rather than immunohistochemistry/step-section. Furthermore, the delivery of preoperative HAI therapy would not be expected to have a measurable systemic effect. Thus, the rationale for this component of the strategy remains unclear.

In the study by Sadahiro et al.,8 one specific question was whether perioperative HAI therapy provided an advantage in liver metastasis-free survival after resection of colon carcinoma; the data supported that this is the case. One potential explanation might be that HAI therapy promoted liver metastasis-free survival by treating already present, nonradiologically evident metastatic disease. However, to my knowledge, no conclusive evidence exists to support the treatment of hepatic occult disease with HAI therapy. Another plausible mechanism for the observed results would be the potential for HAI therapy to eradicate tumor cells released into the portal circulation at the time of colon resection. This hypothetical mechanism would be predicated on the surgical concept that viable tumor cells may be released as the result of tumor manipulation during a procedure, which although instrumental in defining the “no touch technique” for colorectal carcinoma surgery, is not irrefutably scientifically based.10

A further consideration is that in the study by Sadahiro et al., the results may be associated with the perioperative use of chemotherapy and be independent of the route used (i.e., HAI). Are there different mechanisms/benefits for the perioperative administration of chemotherapy in patients with colon carcinoma? This question was previously posed in a CALGB trial that closed prematurely as a consequence of low patient accrual in May 2000.11

Further data from the study by Sadahiro et al. describe the site of first recurrence in patients in the HAI treatment arm to be the liver (4%), lungs (3%), and local recurrences (5%) compared with patients in the control arm of the study, in which the sites of first recurrence were the liver (15%), lungs (6%), peritoneum (5%), and local recurrences (4%); only the liver as the site of first recurrence was found to be significantly different between treatment arms (P = 0.002). It is interesting to note that patients enrolled in the study and randomized to the HAI arm who did not get receive HAI therapy (the HAI-2 group in the intent-to-treat analysis) had a 16% incidence of failure at the liver, which was comparable to the patients in the control arm, further supporting the belief that the benefit of the HAI therapy regimen was a treatment-specific reduction in the rate of hepatic metastasis. Because the site of failure differed only for the liver, the improvement in the disease-specific survival and overall survival rates observed were associated with the reduction in hepatic failure that was achieved. This finding confirms the previously noted association between colon carcinoma recurrence in the liver and survival,2 but it is not biologically clear how this reduction in hepatic recurrence occurred. Thus, the observed improvement in liver metastasis-free survival associated with this treatment strategy is likely related to the use of the regimen in the specific preoperative and perioperative setting of colon resection for the initial tumor.

The potentially difficult issues for adopting this treatment are:

  • 1The provision of cytotoxic chemotherapy to patients (Stage II disease) who under current indications generally would not be treated. However, the authors of the study recognize this issue and would support the use of this approach for patients with confirmed Stage III disease. It is not clear how the preoperative identification of patients with “true” Stage III disease might be accomplished, so in reality some Stage II patients would receive chemotherapy without an indication.
  • 2Arterial infusion chemotherapy has not been widely utilized in clinical practice for hepatic metastasis. Although cooperative trials groups and regional cancer centers have employed HAI therapy, to my knowledge the literature has been divided regarding the potential benefits and systemic therapy remains the mainstay of treatment.12
  • 3When arterial infusion is employed, FUDR achieves higher hepatic concentrations and is nearly completely extracted on the first pass with limited systemic side effects; therefore, FUDR would be preferable to 5-FU.13
  • 4The options for systemic chemotherapy drugs (e.g., irinotecan, oxaliplatin) for hepatic metastasis from colon carcinoma have evolved since this trial was instituted (1994), but beyond FUDR and 5-FU for HAI, limited progress has been documented.

Sadahiro et al. have reported a significantly positive, well structured, appropriately statistically powered and properly analyzed randomized clinical trial.8 However, the benefit of their strategy will have to be confirmed by others. Further studies are recommended to determine whether the results reported were dependent on the “timing” of HAI therapy and the need for HAI therapy both before and after colon resection: preoperative only HAI plus colon resection versus colon resection alone (which is similar to the strategy used by Sadahiro et al.) in which clinical disease stage would be used for enrollment compared with postoperative only HAI plus colon resection versus colon resection alone, in which pathologic disease stage would be used for enrollment. Finally, the question of the potential role of systemic perioperative chemotherapy in patients with colorectal carcinoma independent of whether the HAI route of administration was used remains unanswered.

In my opinion, it is most important to define the value of the preoperative weekly HAI infusion because this is the most potentially problematic component of the regimen because it is based on clinical staging for enrollment. If the preoperative HAI therapy component were demonstrated to be nonessential in achieving a similar outcome, then the use of an implantable hepatic arterial pump rather than a transfemoral hepatic artery catheter would be potentially clinically and cost advantageous by allowing patients to receive the regimen on an outpatient basis. However, either a prolonged (3-week) hospital stay or HAI pump implantation for HAI drug delivery would be cost-prohibitive for adopting this regimen.

Although there are some potentially problematic study mechanics, it is clear that the results of the study by Sidahiro et al.8 raise the possibility of a future treatment strategy with the potential to significantly impact patient outcome.