TALK TO CANCER researchers or oncologists and they’re likely to express  respect or even grudging admiration for their nemesis. Cancer is ceaselessly smart, inventive and resilient, they’ll tell you, and though it may be put on the defensive by targeted therapies or other treatments, it invariably finds a way to fight back. That’s why the survival statistics for many kinds of cancers have been stubbornly slow to improve.

What cancer warriors really need is an ally. Maybe a physiological process on the inside that is just as resourceful and powerful as those tumor cells. And now, finally, they may have found what they’ve been looking for: the human immune system.

That’s actually an old notion—that the body’s natural immunity, trained to fight invaders, can be agitated to become more hostile toward tumor cells. But until recently, this approach—immunotherapy—very rarely worked, and it sometimes exacted a terrible toll. The revved-up immune cells may also turn against healthy tissue, which can produce rashes and damage the liver, bowel, lungs and other organs.

Still, a quite small percentage of patients who underwent the harrowing process had a positive long-term response. “So it was impossible to give up on the idea of immunotherapy, but also impossible to offer any individual patient much optimism,” says David Fisher, chief of dermatology at Massachusetts General Hospital. “There has been incredible excitement about immunotherapy for a long time, but also enormous frustration because of the inability to harness the response for more people.”

Another problem was that, even with the immune response, cancer found a way to survive most of the time. The body’s immune cells can’t be constantly destructive, and once they’ve done their job, a cellular mechanism blocks their activity. Tumor cells also made use of the same device, evolving so that they also could step on the brakes at tactical moments, prompting a slowdown in the immune response that might otherwise defeat them.

That braking happens at molecular checkpoints, places on immune cells where activity can be dialed back, and it is at precisely these sites where new-style immunotherapy focuses its efforts. The first drug designed to keep cancer from subverting the immune system’s efforts—an approach known as checkpoint inhibition or checkpoint therapy—was approved for the aggressive skin cancer melanoma in 2012. Since then, checkpoint therapy has begun to revolutionize cancer treatment, proving effective against lung cancers, kidney cancer, bladder cancer, Hodgkin’s lymphoma and mesothelioma, among others. Still, checkpoint therapy leaves many cancers—including major killers such as breast, colon, prostate and pancreatic cancer—relatively unscathed. The big challenge now is to figure out how to extend the benefits of the new treatments to more patients.

Those investigations start with a piece of the puzzle that has been on the table for quite a while—that cancerous cells arise all the time, and that immune cells routinely clear them away. So researchers knew that an anticancer immune response exists; they just needed to determine why it too often fails. “Cancer engages in a years-long struggle with the immune system,” says Gordon Freeman, a cancer biologist at Dana-Farber Cancer Institute in Boston whose basic discoveries are part of the foundation upon which checkpoint therapy has been built. “If cancer grows, it means the immune system is losing.”

It’s now known that there are many layers of immune suppression, and cancers exploit all of these checkpoints to avoid being killed, says Arlene Sharpe, an immunologist at Harvard Medical School who, with Freeman, discovered some critical immune cell checkpoints that cancer cells exploit. As scientists peel back layer after layer, they’re discovering new insights into the intricacy of how cancer’s cellular activity slows down the immune system, and coming up with strategies to keep that from happening.

BEGINNING AS EARLY as a century ago, physicians occasionally encountered patients whose advanced tumors spontaneously retreated or disappeared completely. Sometimes that happened after an infection, suggesting that when immune cells were stimulated to fight the infection, they also started going after the cancer.

Those observations led to the theory of immune surveillance. Dendritic cells, one of several kinds of immune cells, are designed to look for infectious, foreign or damaged cells that don’t belong in the body. When they find such a cell, dendritic cells display antigens (small protein fragments) from the offending cell on their surface—dendritic cells are known as antigen-presenting cells, or APCs—to educate young, “naïve” T cells in the lymph nodes about the dangerous cell’s identity. This activates the T cells as killer T cells that patrol for cells with that identity and destroy any that they find.

But surveillance for errant cells isn’t the immune system’s only job. It also has to teach tolerance, so that T cells learn to avoid attacking healthy cells, to coexist with benign bacteria in the gut and to stop the inflammatory immune response once an infection has been defeated, because continuous inflammation is harmful. That’s where checkpoints come in—they help activated T cells learn to tolerate the cells they had previously been taught to attack.

In the mid-1990s, Sharpe and Freeman elucidated functions of the first checkpoint, a cellular receptor called CTLA-4 that appears on the surface of T cells only after they’ve been activated to fight cancer or another invading cell. They showed that other immune cells have a protein “key,” called B7, on their surface that locks into CTLA-4, and when that happens, it signals the T cell to stand down—it steps on the brakes of the immune response.

The discovery of the CTLA-4 checkpoint inspired James Allison, an immunologist now at MD Anderson Cancer Center in Houston, to come up with a radically new anti-cancer strategy in the late 1990s. His idea was to create an antibody that blocks access to the CTLA-4 receptor and prevents tumor cells from using the checkpoint to stop the attack by immune cells.


At the time, most cancer scientists were focused on developing targeted therapies that could deactivate specific mutations in cancer cells to stop their growth. “But we argued that we didn’t need to know which mutation in the cancer cell to target—because immune cells already know how to recognize cancer,” Allison says. All the immune cells needed was a way to keep cancer cells from blocking their attack. And finally, a strong candidate emerged to take that job. An antibody called ipilimumab, or “ipi,” was developed, tested and approved (as Yervoy) for advanced melanoma in 2012. It’s designed to block the CTLA-4 checkpoint, keeping the B7 key expressed by tumor cells from turning off the cancer-fighting immune response.

Yervoy was groundbreaking. Even targeted therapies that can melt away advanced melanoma tumors are only a temporary fix; usually, the cancer returns and kills patients within a year of when they began treatment. But about one in five patients who take Yervoy has a “durable” response of three years or more, and some participants from early trials of the drug have survived 10 years and counting. Yervoy changed the goal of treatment from prolonging the time before the cancer returned (progression-free survival, or PFS) to prolonging life itself (overall survival, or OS).

MEANWHILE, RESEARCHERS were also looking at a second immune system checkpoint—the PD-1/PD-L1 pathway, which Sharpe and Freeman had discovered in 2000—that cancer also perverts for its own purposes. Whereas CTLA-4 braking normally happens early in the immune response, in the lymph nodes, the PD-1 brake is applied later, after activated T cells have traveled into the tumor tissue. PD-1 is a T cell receptor that binds to one of two protein keys (PD-L1 and PD-L2) that can appear on many types of cells, probably to quell inflammation. When those keys lock into the T cell’s PD-1 receptor, the T cell is signaled to tolerate the cell. But some tumor cells, under attack from T cells, begin expressing those key proteins themselves and thus manage to evade the immune response.

Here, too, however, antibodies that block PD-1 on the T cell can release the brake and unleash the repressed immune response. In 2014, the two anti-PD-1 inhibitors Keytruda (pembrolizumab, or pembro) and Opdivo (nivolumab, or nivo) were approved for advanced melanoma.

In clinical trials, about twice as many patients respond to PD-1 inhibitors as respond to Yervoy (40% of all patients who get those drugs, compared with 20% of those taking Yervoy). But with either therapy, the odds of long-term survival is about one in five. And while both types of drugs can cause mild immune reactions—rashes and diarrhea—Yervoy may also result in more serious problems, including liver damage. Yet those side effects ease when patients stop the therapy, and tumors may continue shrinking, suggesting that the immune system remains able to sustain its attack on tumor cells.

Because CTLA-4 and PD-1 function differently, researchers speculated that they could work together, multiplying the benefits. A randomized, double-blind clinical trial published in The New England Journal of Medicine in June 2015 compared, among other things, a combination of Yervoy (anti-CTLA-4) and Opdivo (anti-PD-1) to the effectiveness of Opdivo alone in patients with advanced melanoma. For those who got both drugs, progression-free survival was 27% longer—PFS of 11 months, compared with seven months for patients who got Opdivo alone. (It’s too early to know whether the combination will improve long-term survival.) But taking the two drugs together almost quadrupled the risk of side effects—from 16% to 55%—and about a quarter of the patients on each therapy died of their disease. So without a clear benefit in survival, says Keith Flaherty, director of the Henri and Belinda Termeer Center for Targeted Therapies at Massachusetts General Hospital, it’s hard to know whether the excess toxicity of the combination is worth it.

But those trial results reinforced something that researchers have learned from other laboratory and clinical studies—that the likelihood that a tumor will respond to anti-PD-1 therapy tends to rise or fall in line with how much PD-L1 a tumor expresses. Apparently, if a tumor isn’t exploiting the PD-1 brake to evade immune attack, depriving it of the use of that checkpoint doesn’t do much to heighten the impact of immunotherapy.

Ideally, such a finding could help doctors predict which patients may do well on a particular kind of immunotherapy, in much the same way that genetic analysis of a tumor forecasts a response to targeted therapies. Just check for the prevalence of PD-L1 receptors in a tumor and you could guess whether a PD-1 blocking drug would help that patient. But because the response of some patients bucked this logic, measuring a tumor’s PD-L1 expression isn’t a foolproof indicator on its own.

A BETTER TOOL MIGHT measure the number and variety of genetic mutations in a tumor cell. “Cancer is a loss of control of growth,” says Gordon Freeman of Dana-Farber, “and that requires mutations in the oncogenes that drive cancer’s growth.” But because cancer doesn’t know which mutations will be most helpful in spurring runaway growth, it generates lots of them, randomly. “Some mutations help it grow and some don’t have much effect on growth, but can be targets for immune attack,” Freeman says.

In the work on targeted therapies, scientists focused mainly on the “driver” mutations that help cancer grow, spread and resist therapy, with the goal of using a targeted therapy to disable the genetic anomalies. “We ignored ‘passenger’ mutations that were just along for the ride,” Flaherty says. “We thought they were irrelevant noise.”

But as it happens, those passengers are germane to the goals of immunotherapy. In recent years, researchers realized that some passenger mutations produce odd, novel altered proteins—neoantigens—that are unlike anything else the immune system has seen before.  It recognizes them as foreign and mobilizes its offenses against them by activating T cells. That heightened immune response means that tumors with lots of mutations—and abundant neoantigens—are also more likely to be recognized as “foreign” by the immune system. And checkpoint inhibitor drugs may produce powerful responses in these patients.

“Melanoma is the king of mutated cancers because long-lived skin cells called melanocytes spend their lives under the sun exposed to ultraviolet radiation, accumulating thousands of UV-associated mutations,” says Fisher of MGH. That may help explain why melanoma is vulnerable to checkpoint inhibition. Lung cancer in smokers, meanwhile, has nearly as many, thanks to the carcinogens in cigarette smoke. And because bladder cells are constantly exposed to bodily toxins being flushed out, those cells, too, may produce many mutations. It’s probably not coincidental that lung and bladder cancers also tend to respond well to checkpoint inhibitors.

And there’s yet another factor that can help predict who might respond to checkpoint therapy: whether activated T cells have congregated in and around a tumor. “It’s become clear that checkpoint inhibitors don’t trigger the initial immune response,” says Nir Hacohen, director of tumor immunology at MGH. “Rather, in tumors that respond to these drugs, there is almost always a pre-existing immune response, which has likely been blocked by the tumor.” And the more pre-existing killer T cells there are in a tumor, the more likely the tumor will be destroyed once the T cells are unleashed by checkpoint inhibitors.

“In fact, most tumor tissue samples have immune cells within them, and many cancer cell genomes include mutations in genes related to immune resistance,” Hacohen says. “But because immunology was not a mainstream subject in cancer biology for many years, many of these observations were not appreciated by the broader cancer community.” His group has been deciphering which genes and proteins play a role in tumor immunity, using more than 8,000 tumor samples to try to crack this problem.

“WE NOW THINK IN terms of a spectrum regarding which cancers are close to being recognized and cleared by the immune system,” Flaherty says. “Checkpoint inhibition has uncovered one end of the spectrum, and melanoma sits at the extreme end. For cancers on the other end, with no immune recognition, we need other tools. There won’t be one immunotherapy for everyone. We need a personalized medical approach.”

One tool that could fight tumors—especially the less immunogenic cancers that are at the opposite end of the spectrum from melanoma—is a vaccine containing neoantigens culled from a patient’s tumor. “Neoantigen vaccines do not require the immune system to have already generated a response against a tumor,” says Hacohen, who leads one of these efforts. “It’s more like a flu vaccine that can induce new responses.” The idea is to use neoantigens unique to a patient’s tumor to provoke an immune response that will attack the tumor of that individual by recognizing the neoantigens displayed on the surface of that tumor.

Still another tool is a sophisticated procedure called chimeric antigen receptor (CAR) T cell therapy. In this case, the objective is to break the immune system’s tolerance of cancer cells and force it to recognize the tumor, says Marcela Maus, a tumor immunologist who helped develop CAR T cells to treat blood malignancies at University of Pennsylvania’s Abramson Cancer Center and who will be leading a program for CAR T cells at MGH. “We harvest T cells from a patient’s blood, genetically modify them so they will recognize that patient’s cancer cells, cultivate them in the laboratory for about a week, and re-infuse them into the body,” Maus says. Both CAR T and neoantigen vaccines are in early clinical trials.

Another intriguing possibility is that radiation, chemotherapy and targeted therapies—all of which aim to kill cancer cells directly—might make cancers more immunogenic in the process. Dying cells release their internal contents, which may include hidden neoantigens for the immune system to detect. Numerous trials are testing various combinations of checkpoint inhibitors with each of those other approaches.

THE DAWN OF THE checkpoint era has caused enormous optimism among patients, but there is also some confusion, says Fisher. Some people with specific mutations in their cancers are eligible for genetically targeted therapies that will almost always shrink or dissolve tumors—but only temporarily.

That leads to this conundrum. “The prospect of a durable response—what we can even think of as a cure—is significantly higher with immunotherapy than with targeted therapy,” Fisher says. But the proportion of patients who will benefit from current immunotherapy may be lower than the percentage who will respond to the best targeted therapies. So patients desperate for any improvement may gravitate to targeted therapy. “But what we really want is a cure,” he says. “It’s hard to know what the best option is because we can’t yet accurately predict who will be among the group of patients who are helped by immunotherapy.”

International teams of researchers are now trying to develop a clinically useful “immuno-score,” a ranking of how receptive a cancer might be to immunotherapy, that would be based on several kinds of information—how much PD-L1 there is in a tumor; what driver mutations it has; its immunogenic neoantigens and immune resistance genes; how much T cell infiltration there is; and other measures.

For now, patients have lots of choices to make, balancing probabilities, survival benefits and the risk of side effects—and that would be a lot to weigh even without the emotional turmoil of a cancer diagnosis. Eventually, science will help make those calculations more accurate. It may always be a guessing game, but in the future there may be many more winners who can beat cancer at its own game.