commentary

Journal of Surgical Oncology 2003;82:131

COMMENTARY

Hepatic metastases of solid malignancies remaindifficult to cure. For some malignancies, particularlycolorectal cancer, the liver is the only obvious site ofmetastases and therefore, liver-directed therapies are alogical therapeutic option. Indeed, clinical trials havesuggested that intrahepatic arterial chemotherapy admi-nistered after resection of hepatic metastases of coloncancer improves clinical outcome [1]. To obtain pre-liminary evidence of safety and efficacy, new strategieswill require testing in animal models. Therefore, animalmodels that mimic metastatic solid malignancies clini-cally would be a major boost to the development ofnovel strategies. In our opinion, to be useful for testingstrategies before human clinical trials, a model mustsatisfy several requirements: (1) the pattern of diseasespread should mimic the human disease, (2) the geneticabnormalities underlying the progression of the tumorshould be similar, (3) the pharmakokinetics, pharma-codynamics, or other factors important to the distributionof the therapeutic agent must be similar in the model andhuman disease, and (4) the innate response of the tumorbearing host to the tumor must be similar in the modeland human disease. Many models fail to satisfy theserequirements. For example, the most common approachis to inject the tumor subcutaneously. While this maybe appropriate as a model of a skin cancer, it does notmimic the pattern of disease for a colon cancer. Otherapproaches, such as injection into the organ from whichthe tumor is derived, do not always completely recapitu-late the spread of the tumor. For example, a tumor cellline injected into the pancreas may cause only peritonealmetastases, and not livermetastases. Finally, somemodelsuse murine tumors, whereas an ideal model would usehuman tumors. Stapfer and colleagues [2] claim to havedeveloped such a model, which was optimized for estab-lishment of hepatic metastases of a human pancreaticcancer xenograft.In their study, Stapfer and colleagues [2] adapted a

method of delivering tumor cells into the portal vein,consistent with the spread of human pancreatic cancer.Multiple nodules developed over time with typicalhistologic features. They then used this model to testthe ability to deliver a matrix-targeted retroviral vectorintraportally. They observed 1–3% transduction effi-ciency in vivo.Advantages of this model are the numerous nodules

that develop in the liver over time and the production ofabundant stromal collagen that allows better efficacy of

the matrix-targeting vectors. Also, by using nude mice,any effects of an immune response against nonselfantigens is eliminated. Disadvantages are the fact thatperitoneal disease apparently does not occur, unlikehuman pancreatic cancer. A model in which both of theseoccurred would be even more desirable. Also, sincethe blood supply to hepatic tumors is typically derivedfrom the hepatic artery, a technique for delivering intra-hepatic arterial therapy, as opposed to intraportal therapy,might also be more relevant.

The establishment of models of hepatic metastases isclearly important for the development of treatments forconditions that typically involve liver metastases. Giventhe possible hepatic toxicity of these approaches, theestablishment of safety in animal models is of para-mount importance. As mentioned above, the demonstra-tion of clinical efficacy for intrahepatic arterial deliveryof chemotherapy to treat hepatic metastases of coloncancer has increased the excitement in liver directedtherapies. Other investigators [3] and our group [4] havedemonstrated the efficacy of intrahepatic delivery of newagents to treat hepatic tumors. It is hoped that thedevelopment of models that predict efficacy in the humanwill increase the ease of translating these observationsinto clinical trials.

Michael Morse, MD

Bryan Clary, MD

H. Kim Lyerly, MD

Departments of Medicine and SurgeryDuke University Medical CenterDurham, North Carolina

REFERENCES

1. Kemeny N, Huang Y, Cohen AM, et al.: Hepatic arterial infusion ofchemotherapy after resection of hepatic metastases from colorectalcancer. N Engl J Med 1999;341:2039–2048.

2. Stapfer M, Hu J, Wei D, et al.: Establishment of a nude mousemodel of hepatic metastasis for evaluation of targeted retroviralgene delivery. J Surg Oncol 2003;82:121–130.

3. Zager JS, Delman KA, Malhotra S, et al.: Combination vasculardelivery of herpes simplex oncolytic viruses and ampliconmediated cytokine gene transfer is effective therapy for experi-mental liver cancer. Mol Med 2001;7:561–568.

4. Selzner M, Bielawska A, Morse MA, et al.: Induction of apoptoticcell death and prevention of tumor growth by ceramide analogues inmetastatic human colon cancer. Cancer Res 2001;61:1233–1240.

DOI 10.1002/jso.10169

Published online in Wiley InterScience (www.interscience.wiley.com).

� 2003 Wiley-Liss, Inc.