Avatars guide treatment, predict effects of cancer chemotherapy


Some people sort out problems with a spouse. Others solve their challenges on a long walk with a dog. Soon, if all goes according to plan, women with triple-negative breast cancer may be able to share their tumors, as well as some of the burden of treatment, with a couple dozen immunocompromised mice.

Known as “avatars,” or “patient-derived xenografts,” these mice are being evaluated as substitute patients in select research laboratories worldwide. They give doctors a way to determine in advance how multiple anti-cancer medications, even combinations of drugs, are likely to affect a particular patient’s tumor.

“Mice, ironically, are becoming an important tool in the shift toward more personalized medicine,” said Geoffrey Greene, PhD, the Virginia and D. K. Ludwig Professor and Chair of the Ben May Department for Cancer Research at the University of Chicago. “Patient-derived xenografts [PDX] enable us to employ mice as surrogates. Before a patient begins chemotherapy, an avatar can help her physician predict how her cancer will respond to a wide range of available drugs and thus make better choices about how to treat that specific tumor.”

Greene’s laboratory will use these murine stand-ins to study treatment of triple-negative breast cancer, a common and difficult to treat form of the disease. It is called “triple-negative” because these tumors lack three crucial therapeutic targets, the receptors silenced by drugs such as tamoxifen and Herceptin, which are mainstays of breast cancer treatment.

A generous $1 million donation from the estate of Ruth Bruch—a former trustee of the University of Chicago Cancer Research Foundation who died from breast cancer in October 2014—makes this program possible.

How it Works

Geoffrey Greene, PhD

Geoffrey Greene, PhD

The process of creating a patient-derived xenograft begins when a tumor or core biopsy is surgically removed from a patient. Part of the tumor is used for standard pathological and genetic profiling. These guide the initial treatment. The rest of it is sliced into pieces. These are surgically implanted under the skin of a few mice, “usually one to three, at first, depending on how much tissue we have to work with,” Greene explained.

These mice have weakened immune systems, so they won’t reject this foreign tissue. “We can even transplant some of the surrounding human tissue along with a cancer,” Greene said, “to see how the tumors interact with that part of their environment.”

About two months later, if all grows according to plan, the transplanted tumors, now much larger, are removed and again divided. Each piece is then implanted in another mouse. After two or three such generations, there are enough mouse models, perhaps 10 to 20 or more, to begin testing.

“First, we confirm that the mouse-grown tumors still resemble the original human version,” Greene said. “If that’s the case, we select available drugs, based on tumor biology and genetics, which are most likely to be safe and effective. Then we test several of those drugs, or combinations of them, in small groups of the avatar mice.” If a drug works well in treating the mice, it is likely to have a beneficial effect on the patient. The success of this approach depends on obtaining tumor tissue early enough to allow for the time it takes to generate PDX mice from a patient’s tumor.

The mice and their offspring can be maintained and monitored for months or years. Greene has nurtured one set of tumor-bearing mice for more than five years. This enables the researchers to follow how cancers change over time and, with luck, it can help physicians anticipate the emergence of drug-resistant tumor cells. Avatars may also help uncover biomarkers that can predict resistance and point to drugs may overcome it.

Studies have found that tumors transplanted into these mice evolve “pretty much in sync with the patient’s original tumor,” Greene said. “The accumulated mutations we find in the transplanted tissue tend to reflect what happens in the original.”

Choosing targets

Chemotherapy 3Triple-negative breast cancer is a good but challenging target for this approach. There is no reliable treatment for this cancer and disease tends to recur, so better drugs are needed. Because these tumors are aggressive, they are more likely to grow in a mouse host. “We can get 50 to 80 percent of those to take,” Greene said. But they can be so aggressive that it becomes “a race against time.”

If the mouse-grown tumors can get ahead of the patient’s disease, “we can begin to investigate treatment options weeks or perhaps months before there’s a need to make clinical decisions,” Greene said. The goal is “to start testing second- and third-line therapies while the patient is still stable on the initial treatment.”

The PDX landscape

Clinical use of avatars as patient proxies is not brand new. There are several university-based programs already in place.

Greene’s colleague, Kevin White, PhD, professor of Human Genetics and director of the Institute for Genomics and Systems Biology at the University of Chicago and Argonne National Laboratory, has been working with several institutions that use mice embedded with human tumors to study the genetics of pancreatic cancer and to use patient-derived xenografts to unravel how patients respond to treatments. Others are looking at bladder cancer and certain forms of leukemia.

There are even commercial vendors, such as Maine’s Jackson Laboratory and New Jersey’s Champions Oncology, that supply mice for avatar-based studies.

There have been some real successes, but the evidence confirming that these mice can genuinely improve patient care and help patients live longer is still being collected. Laboratories using avatars are also learning more about the limitations. Some cancer types do not engraft easily. The transplant environment may differ from the tumor’s original environment, even if non-cancerous human cells are transplanted along with the tumor. And the recent giant steps in immune-based therapies cannot be studied in immunocompromised animals. Still, several studies, including one involving triple-negative breast cancer, found a high correlation between how mouse avatars and human patients responded to various drugs.


Ya-Fang Chang, PhD, a senior Post-doctoral Scholar in the Greene laboratory, works with the lab’s PDX mice.

One high hurdle remains: Avatars are costly. News articles about avatars that looked at the expense of creating and testing avatars for a single patient mentioned charges from commercial suppliers as high as $25,000.

Immunocompromised mice are costly to purchase and can be difficult to maintain. To make careful decisions about therapy, most physician-researchers prefer at least 20 to 25 mice per patient. Testing multiple drugs on those mice can add to the expense. Insurance companies rarely pay for this.

“Research funding from the Bruch estate, at least while the program is relatively small, should support our efforts to bring this approach into the clinical setting,” Greene said.

There may be unresolved issues, but “I think that before too long every major player in cancer research will need to take this model seriously,” Greene said. “We think avatars can become an extremely useful tool. They could tell us in advance if specific, alternative drugs are likely to be better than the current standard of care. That would be nice to know upfront.”

%d bloggers like this: