An opposite effect of chemotherapy?

Blogging on Peer-Reviewed ResearchLet’s face it. These days, research papers in the peer-reviewed biomedical scientific literature are becoming more and more complex and difficult to understand. For many journals, it seems, if you don’t have at least seven meaty, dense, multipanel figures (preferably some of which with flashy color confocal microscopy), you don’t have a prayer of getting published. Along with the requisite multiple figures, it seems, the prose has become more impenetrable even over the 23 years or so since I graduated from college and embarked upon my current career. Scientific papers in the old days (say, as recently as 50 years ago) weren’t like that, and 100 years ago they really weren’t like that. It wasn’t because color printing was too difficult and expensive, either. It just seems to me that, back then, scientists were better at telling a story, using the data as the basis. There were times when I used to go into the really old stacks in the library and just browse old medical and scientific journals, amazed at how much easier they were to read than the present day.

That’s why it’s a breath of fresh air to come across an article that, while perhaps not written in the compelling prose of the days of yore, at least reports simple and compelling experiments that don’t require poring over each sentence and figure in order to figure out what the authors are trying to say and sometimes still not being sure. Even better, it’s an article1 that’s right up my alley in that it’s about metastasis and it’s in the latest issue of Cancer Research. It’s interesting because it both confirms and suggests a seemingly paradoxical effect of a form of chemotherapy.

The article, by Robert M. Hoffman and coming from the Departments of Surgery and Orthopedic Surgery at the University of California, San Diego and the School of Medicine, Kanazawa University, Japan, studies the question of whether chemotherapy can actually have “opposite” effects, namely effects that enhance tumor growth rather than inhibit it or kill tumor cells. Before we can understand what this study is about, it’s first necessary to understand that the effects being studied are not effects on the tumor cells. The chemotherapeutic agent being studied, cyclophosphamide, a nitrogen mustard alkylating agent, does indeed kill the tumor cells in the tumor models being study quite handily. It’s also a commonly used cancer chemotherapy for a fairly wide variety of malignancies, usually as part of combination chemotherapy for breast cancer and a few other solid malignancies, as well as some leukemias and lymphomas.

Prior observations have shown that pretreatment with some chemotherapeutic agents or radiation therapy can actually facilitate the formation of metastases. In the case of cyclophosphamide, in mouse models of cancer, this effect is observed when mice given this drug and then tumor cells are injected 24 hours later. Because of the pharmacokinetics of the drug, there is no chemotherapeutic agent left by the time the tumor cells are injected, meaning that the effect is entirely on the host. Speculation has centered on either vascular damage as a cause, allowing tumor cells to leak out into the tissues from the blood vessels. Another proposed mechanism is that cyclophosphamide either kills macrophages or prevents them from killing tumor cells trapped in the small blood vessels. Given that a critical step in metastasis is the ability of tumor cells, after lodging in the microvessels, to extravasate into the surrounding tissue and proliferate, the vasculature is the most likely target.

To study this phenomenon, Hoffman’s group took HT1080 human fibrosarcoma cell line. When these cells are injected into mice, the vast majority of them do not form metastases. Given their low baseline rate of metastasis formation, they make a good model system to test manipulations that increase tumor cell aggressiveness. To do these tests required a very simple experimental design. Nude mice (a strain of mice with compromised immune systems that will permit human cells to form tumor xenografts) were either pretreated with cyclophosphamide or vehicle, and then 24 hours later HT1080 cells were injected into the epigastric cranialis veins of the mice. These cells had been labeled with fluorescent markers that allowed the nuclei and cytoplasm to be visualized as different colors, which allowed in vivo imaging of the tiny tumor deposits.

The results were rather fascinating. First, it was observed that there were two types of tumor colonies formed by HT1080 cells in cyclophosphamide-pretreated mice. One type of colony formed within the vessels without extravasation, while the other type formed from cells that extravasated into the surrounding tissue. The extravasated tumors tended to occur in the smallest vessels, while in larger vessels tumor cells grew without extravasation. Within a week after injection, the extravasated tumor deposits stopped growing and started to regress, while tumor deposits in the blood vessels continued to grow. Thus, tumor cells proliferated in larger blood vessels, migrated to smaller blood vessels throughout the body, and then extravasated, a result not observed to anywhere near the same extent in the non-treated mice.

The results appear consistent with some sort of host-based cancer cell-killing process that is inhibited by cyclophosphamide. Consistent with this interpretation is the observation that the extravasated tumor colonies gradually regressed, suggesting a slow recovery of this ability to kill tumor cells after an initial insult.

The question, of course, is: What, if anything, does this mean for patients with cancer undergoing chemotherapy? First, let me emphasize that this is a very artificial model of cancer metastasis. Specifically, it’s an induced metastasis model, as opposed to a spontaneous metastasis model, the latter of which involves letting tumors grow under the skin of mice or in other locations and then measuring “spontaneous” metastases at a certain time point. It should be pointed out that patients with cancer do not shed tumor cells into the bloodstream in large boluses of a million cells in a single shot, which scaled up to human size, would be hundreds of millions of cells. In humans, tumors grow slowly in their tissue of origin in a process that takes years to progress from transformation of a cell to the malignant phenotype. Eventually, some cells in the tumor acquire the ability to invade blood vessels, travel to distant sites, and extravasate. The vast majority die, but some manage to form colonies known as micrometastases. In any case, the difference is that a human being with an established tumor is shedding small numbers of tumor cells into the vasculature all the time on a more or less continuous basis. It’s a very different situation than this model. Moreover, the model is set up so that the injection into a vein whose blood flow went to a skin flap, which allowed in vivo imaging of the tumor cells in blood vessels and the colonies that they formed. Finally, since no tumor cells are present during the chemotherapy injection, there is no toxic effect on the tumor cells, whose killing effect could possibly more than balance out the predisposition to metastasis. All in all, there are many aspects of this model that are so different from the real situation that interpretation has to be undertaken with care.

Still, it’s an interesting result, and does seem to jibe with the classical conception that chemotherapy can temporarily suppress the immune system. Certainly, it’s worth pursuing in order to work out the mechanism by which this effect occurs and use that knowledge to come up with strategies to mitigate it. What we have to remember is that it is the overall balance of effects that matter. If chemotherapy, for instance, can predispose to metastasis formation, it is of less concern if the effect is tiny compared to the tumor killing effect. Think of it this way. We already know chemotherapy has a lot of toxic effects on the immune system, the GI system, and other systems with rapid cell proliferation to the point that the complications from chemotherapy can be life-threatening. However, for most cancers the benefits outweigh the risks, and I suspect that, even if this article presents a true result that can be generalized to humans, that will not change.

Finally, there’s another point about this article. I noticed that some of the authors were affiliated with a company known as AntiCancer, Inc., which had apparently obtained a National Cancer Institute small business grant to study this. Given that I have reviewed such grants before on an NIH study section, I know that there needs to be a a business plan and a plan for developing a product as part of the application. This makes me wonder why the investigators were so vague about the potential mechanism for the cyclophosphamide effect and whether they have more of an idea than the article lets on of exactly what is going on in their tumor model. Indeed, it makes me wonder whether the authors actually have some idea of how to block this effect that they aren’t publishing yet. After all, if they didn’t have some idea of how to alter or ameliorate this effect, I tend to doubt they would have gotten the grants to try to make a product.


1. Yamauchi K, Yang M, Hayashi K, Jiang P, Yamamoto N, Tsuchiya H, Tomita K, Moossa AR, Bouvet M, Hoffman RM (2008). Induction of cancer metastasis by cyclophosphamide pretreatment of host mice: an opposite effect of chemotherapy. Cancer Res. 68(2):516-20.