In April 2011, Christopher Barker, a radiation oncologist at Memorial Sloan Kettering Cancer Center in New York, received some unusual news about a participant in a clinical trial. The patient was battling a second recurrence of melanoma that had spread to several areas of her body. After more than a year on the experimental drug, her tumors had only gotten bigger, and after one near her spine started causing back pain, her doctors arranged for local radiation therapy to shrink the tumor and give her some relief.
But the tumor near her spine was not the only one that shrank. “From one set of images to another, the radiologist noticed that there was a dramatic change in the extent of the melanoma,” Barker says. Although only one tumor was exposed to radiation, two others had started shrinking, too.
The striking regression was a very rare effect of radiation therapy, Barker and his colleagues concluded, called an abscopal response. “It’s not common,” says Barker. “But we see it, and it’s pretty remarkable when it happens.”
Although the abscopal response was first recognized back in 1953, and a smattering of case reports similar to Barker’s appeared in the literature throughout the 1960s, ’70s, and ’80s, the mystery behind the abscopal response largely went unsolved until a medical student named Silvia Formenti dusted it off.
While studying radiation therapy in Milan during the 1980s, Formenti couldn’t shake the idea that local radiotherapy must have some effect on the rest of the body. “When you burn yourself, the burn is very localized, yet you can get really systemic effects,” says Formenti, now chair of the department of radiation oncology at Weill Cornell Medical College in New York. “It seemed that applying radiotherapy to one part of the body should be sensed by the rest of the body as well.”
The primary goal of therapy with ionizing radiation—the type used to shrink tumors—is to damage the DNA of fast-growing cancer cells so they self-destruct. But like burns, radiation also causes inflammation, a sign of the immune system preparing for action. For a long time, it was unclear what effect inflammation might have on the success of radiation therapy, though there were some hints buried in the scientific literature. For example, a 1979 study showed that mice lacking immune cells called T cells had poorer responses to radiation therapy than normal mice. But exactly what those T cells had to do with radiation therapy was anyone’s guess.
In 2001, shortly after arriving in New York, Formenti attended a talk by Sandra Demaria, a pathologist also at Weill Cornell. Demaria was studying slivers of breast tumors removed from patients who had received chemotherapy and had found that in some patients, chemotherapy caused immune cells to flood the tumors. This made Formenti wonder if the same thing could happen after radiation therapy.
In addition to fighting off illness-causing pathogens, part of the immune system’s job is to keep tabs on cells that could become cancerous. For example, cytotoxic T cells kill off any cells that display signs of cancer-related mutations. Cancer cells become troublesome when they find ways to hide these signs or release proteins that dull T cells’ senses. “Cancer is really a failure of the immune system to reject [cancer-forming] cells,” Formenti says.
Formenti and Demaria, a fellow Italian native, quickly joined forces to determine whether the immune system was driving the abscopal response. To test their idea, their team injected breast cancer cells into mice at two separate locations, causing individual tumors to grow on either side of the animals’ bodies. Then they irradiated just one of the tumors on each mouse. Radiation alone prevented the primary tumor from growing, but didn’t do much else. Yet when the researchers also injected a protein called GM-CSF into the mice, the size of the second tumor was also controlled.
GM-CSF expands the numbers of dendritic cells, which act as T cells’ commanding officers, providing instructions about where to attack. But the attack couldn’t happen unless one of the tumors was irradiated. “Somehow radiation inflames the tumor and makes it interesting to the immune system,” Formenti says.
Formenti and Demaria knew that if their findings held up in human studies, then it could be possible to harness the abscopal effect to treat cancer that has metastasized throughout the body.
Although radiation therapy is great at shrinking primary tumors, once a cancer has spread, the treatment is typically reserved for tumors that are causing patients pain. “Radiation is considered local therapy,” says Michael Lim, a neurosurgeon at Johns Hopkins University in Baltimore who is studying ways to combine radiotherapy with immunotherapy to treat brain tumors. But, he adds, “if you could use radiation to kindle a systemic response, it becomes a whole different paradigm.”
When Demaria and Formenti first published their results in 2004, the concept of using radiation to activate immunity was a hard sell. At the time, research into how radiation affected the immune system focused on using high doses of whole-body irradiation to knock out the immune systems of animal models. It was counterintuitive to think the same treatment used locally could activate immunity throughout the body.
That perspective, however, would soon change. In 2003 and 2004, James Hodge, an immunologist at the National Cancer Institute and his colleagues published two mouse studies showing that after radiation, tumor cells displayed higher levels of proteins that attract and activate cancer-killing T cells. It was clear radiation doesn’t just kill cancer cells, it can also make those that don’t die more attractive to immune attack, Hodge says.
This idea received another boost in 2007 when a research team from Gustave Roussy Institute of Oncology near Paris reported that damage from radiation caused mouse and human cancer cells to release a protein that activates dendritic cells called HMGB1. They additionally found that women with breast cancer who also carried a mutation preventing their dendritic cells from sensing HMGB1 were more likely to have metastases in the two years following radiotherapy. In addition to making tumors more attractive to the immune system, Hodge says, the damage caused by radiation also releases bits of cancer cells called antigens, which then prime immune cells against the cancer, much like a vaccine.
In some ways, Barker says, oncologists have always sensed that radiation works hand-in-hand with the immune system. For example, when his patients ask him where their tumors go after they’ve been irradiated, he tells them that immune cells mop up the dead cell debris. “The immune system acts like the garbage man,” he says.
Now, immunologists had evidence that the garbage men do more than clean up debris: they are also part of the demolition team, and if they could coordinate at different worksites, they could generate abscopal responses. With radiation alone, this only happened very rarely. “Radiation does some of this trick,” Formenti says. “But you really need to help radiation a bit.”
Formenti and Demaria had already shown in mice that such assistance could come in the form of immunotherapy with GM-CSF, and in 2003 they set out to test their theory in patients. They treated 26 metastatic cancer patients who were undergoing radiation treatment with GM-CSF. The researchers then used CT scans to track the sizes of non-irradiated tumors over time. Last June, they reported that the treatment generated abscopal responses in 20% of the patients. Patients with abscopal responses tended to survive longer, though none of the patients were completely cured.
As the Weill Cornell team was conducting their GM-CSF study, a new generation of immunotherapeutic drugs arrived on the scene. Some, like imiquimod, activate dendritic cells in a more targeted way than GM-CSF does. Another group, the checkpoint inhibitors, release the brakes on the immune system and T cells in particular, freeing the T cells to attack tumors.
In 2005, Formenti and her team found that a particular checkpoint inhibitor worked better with radiotherapy than alone and later reported that the same combination produces abscopal responses in a mouse model of breast cancer.
In 2012, Formenti had an unexpected chance to test this treatment in the clinic when one of her patients who had read about her research requested that she try the combination on him. The patient had run out of options, so Formenti’s team obtained an exception to use the immunotherapy ipilimumab, which she had used in her 2005 study and had only been approved for melanoma, and proceeded to irradiate tumors in the patient’s liver. After five months, all but one of his tumors had disappeared. “We were ecstatic,” Formenti says. “He’s still alive and well.”
The availability of checkpoint inhibitors seems to have opened the floodgates. Since the US Food and Drug Administration approved ipilimumab in 2011, there have been at least seven reports of suspected or confirmed abscopal responses in patients on checkpoint inhibitors, including the one Barker witnessed. Contrast that with the previous three decades, where less than one per year was reported, according to one review. Almost all of the recent cases involving checkpoint inhibitors have been in patients with melanoma, since that’s where the drugs have mainly been tested. But, abscopal responses with or without immunotherapy have been reported in patients with cancers of the liver, kidney, blood, and lung.
There are now dozens of clinical trials combining radiation with a range of immunotherapies, including cancer vaccines and oncolytic viruses. “There’s quite a nice critical mass of people working on this,” Formenti says. She and Demaria are now finishing up a clinical trial in lung cancer patients using a protocol similar to the one that worked so well in their original patient.
“I think we know that people who respond to checkpoint inhibitors already have more immune-activating tumors,” Demaria says. The question now, she says, is whether radiation can expand the 20% of people who respond to the combination therapy.
One solution might be to match combinations to particular patients or tumor types. Demaria’s team is collecting blood and tissue samples from patients in a Weill Cornell lung cancer trial to look for differences in the immune responses of those who do and don’t generate abscopal responses. Such changes in the number or status of a cell type associated with particular outcomes are known as biomarkers.
So far, there is little data about how the two types of responses differ. Barker and his team did publish measurements of a broad range of immune markers from their patient who experienced an abscopal response. “We didn’t really have a lot of clues in terms of what we should look at,” he says. They observed a bump in activated T cells and antibodies specific to tumor proteins following radiation, followed by steady declines of both as the tumors regressed. But, he says, there was no “smoking gun” that could explain why this particular patient responded the way she did.
Understanding how the immune system responds to immunotherapy and radiation will be key to optimizing the combination of the two. “One needs to do these combinations to try and improve the outcome on both sides of the equation,” says William McBride, a radiation oncologist at the University of California, Los Angeles. There’s still controversy, for example, over whether the immune system responds better to high doses of radiation over short periods or low doses over longer periods. “We think we know the best sequence of therapy based on the pre-clinical studies, but that hasn’t been confirmed in clinical studies yet,” Barker says. “If we had a biomarker that would tell us in what way you should give the radiation, that would be enormously valuable.”
Demaria says her research suggests that more tumor damage is not always better and that high radiation doses may be counterproductive, activating feedback responses that suppress immunity. She’s currently comparing immune signatures of different radiation regimens in mice. So far she says regimens that make the cancer look and act like virally-infected cells tend to elicit the best immune responses, but there is a long way to go in translating that work into humans.
“Things are moving faster than they have for a long time, but at this point there are still a lot of unanswered questions,” she says.
Fortunately, she and Formenti have plenty of motivation to work on those questions. Demaria says she still remembers examining a bit of tumor that was left behind after that first lung cancer patient received treatment. It was full of T cells which had presumably destroyed the cancer. “It’s the picture you never forget,” she says. “It is probably the biggest satisfaction to see somebody’s fate turned around by what you can do.”