ON A BRIGHT, KANSAS summer morning, Emily Dumler was at a friend’s house with her children when her stomach began to hurt. She went to the restroom and was shocked to see blood in her stool. Her doctor advised her to go straight to the emergency room, which became the first stop in a 43-day hospital stay, and resulted in the removal of her spleen and an eventual diagnosis of non-Hodgkin’s lymphoma, a cancer that starts in white blood cells.

For the next two years, Dumler received a barrage of treatments. She went through rounds of chemotherapy and her bone marrow was seeded with steam cells in the hope they would generate a healthy new crop of blood. The cancer continued to progress, however, and in May 2015 Dumler was given fewer than six months to live.

Her physician told her about an experimental new treatment and helped her secure the last opening in a trial taking place at the MD Anderson Cancer Center in Houston. Clinicians there extracted T cells from Dumbler’s blood, genetically modified them to make them targeted cancer killers, then reinfused them into her body. The process is called chimeric antigen receptor T cell therapy, or CAR T cell therapy, and Dumler was the third lymphoma patient in the world to receive it.

Dumler returned home nine days after the treatment, and within weeks she felt better. Scans in August 2015 showed that her cancer had gone completely, and tests since then have found no sign of its return. “Everything I do today, I didn’t think I would ever be able to do again,” Dumler says. “This treatment gave me back my life.”

For the past 45 years, researchers have been looking for ways to amplify the body’s immune response to cancer. Among these, CAR T has been a recent standout, achieving initial remission rates of as high as 90% in some cancers. In 2017, the U.S. Food and Drug Administration approved the first of two new CAR T therapies to treat blood cancers, and the American Society of Clinical Oncology named CAR T the 2018 Advance of the Year.

But while CAR T therapies have proved very good at their current FDA-approved use—curing blood cancers—their track record has been mixed for the much more common varieties of cancer that begin in epithelial cells and form solid tumors. Almost 90% of cancer deaths in the United States—including those from breast cancer, prostate cancer and colon cancer—fall into the second category. The fight against these solid tumors has always put immune cells at a disadvantage. Tumors often grow in parts of the body that are difficult for immune cells to reach, and create microenvironments that can impede any immune cells that do mount an attack.

CAR T’s proponents believe that each of these obstacles could be overcome by clever engineering. This enthusiasm has not only sparked hundreds of investigations, but also renewed interest in older methods of extracting and manipulating T cells to fight cancer. Steven Rosenberg, chief of surgery at the National Cancer Institute and a pioneer of cell-based cancer therapy, continues to pursue a related method in which T cells are also multiplied outside the body and reinjected, which has shown great promise against solid tumors. “We’re working around the clock to try to solve the challenges,” Rosenberg says. “The main stimulus for this research is that we know it can work.”

THE CHIMERA WAS A mythical beast that contained parts from three different animals. Chimeric antigen receptor therapy, similarly, is a curious hybrid not found in nature. At its base is a human T cell, which normally scours the blood stream for threats to the body. When it finds them, it marshals a wider immune response, sometimes attacking directly, and it recognizes these threats by means of a claw-like extension called an antigen receptor. In CAR T cell therapy, clinicians take these T cells out of the body and reprogram their antigen receptors to look for specific proteins found on cancer cells—and instructs them to attack every time.

Since the approval of the first two CAR T therapies in 2017, studies of Tisagenlecleucel (Kymriah) in treating acute lymphoblastic leukemia have shown initial remission rates of 80% to 90%. Axicabtagene ciloleucel (Yescarta) kept 39% of treated patients in remission from refractory non-Hodgkin’s lymphoma for more than two years.

Now many research centers are working to refine CAR T, focusing on the indications that have been proven to work—treating cancers of the circulatory and lymphatic systems. A chief goal for the next generation of treatments is to mute side effects, which can be dangerous and sometimes fatal.

One thread of current research is looking at why some CAR T therapies are less harmful than others. Although both Yescarta and Kymriah target the same antigen—CD19— they use different methods to kick T cells into action. Yescarta delivers a fast and intense T cell response, while Kymriah takes a slower, steadier approach.

Jeremy Abramson, director of the Jon and JoAnn Hagler Center for Lymphoma at Massachusetts General Hospital, is the principal investigator of a trial testing liso-cel, another CAR T therapy for treating non-Hodgkin’s lymphoma. Liso-cel uses a mechanism of action similar to that of Kymriah, and “it looks like those two products do have a more favorable safety profile,” he says. Liso-cel differs from Kymriah, however, in the mix of CAR T cells that it uses. T cells come in two major subtypes, CD4+ and CD8+, and liso-cel administers these to patients in a 1:1 ratio. Studies in mice suggested that the fixed ratio improves effectiveness, though it remains to be seen whether this will prove true in humans, Abramson says.

Even the existing CAR T therapies would likely be more effective if they were used in earlier stages of lymphoma. Currently they are approved only to treat patients who have received two other forms of cancer therapy without success. “These patients are usually fairly beat up, and they often have a lot of lymphoma in their bodies,” Abramson says. Starting treatment earlier might also enable researchers to collect healthier T cells that haven’t been weakened by chemotherapy. Three clinical trials exploring this approach are underway.

On the whole, the future appears bright for using CAR T cell therapy in the treatment of blood and lymphatic cancers. “We’re offering a potentially curable therapy to previously incurable patients,” Abramson says. But many have their eyes on a different prize—a form of the therapy that would work against the much larger class of solid tumors.


T CELLS CAN EASILY find their way to cancers of the blood or lymph nodes because these are where T cells naturally circulate. But solid tumors can occur anywhere in the body, including in regions such as the brain, where immune cells rarely go—and where an overactive immune system can cause multiple sclerosis or other kinds of collateral damage.

Some researchers believe that the best way to sidestep that problem is to deliver CAR T cells directly to the tumor. Christine Brown, associate director of the T Cell Therapeutics Research Laboratory at City of Hope in Duarte, California, has pioneered a way to get the cells near the tumors of glioblastoma, the most common form of brain cancer. She injects the cells into the tumor and into two cavities called the lateral ventricles, where cerebrospinal fluid is made. Together, these delivery methods have shown promising results in preclinical trials.

In 2015 Brown and her team engineered CAR T cells to target the antigen IL13Rα2, a substance produced by glioblastoma, and used this method to infuse several doses of CAR T cells into a man with aggressive cancer. The patient’s brain tumors completely disappeared. Tumors that contained less of the same antigen later sprung up, however, and he died 20 months later.

Delivering CAR T cells close to a solid tumor is only part of the solution. The microenvironment around tumors lacks oxygen and nutrients, which smothers and starves immune cells. It also contains chemical signals that fool the immune system into thinking it isn’t needed. “The solid tumor puts up lots of biologic and physical fences,” says Marcela Maus, director of cellular immunotherapy at Massachusetts General Hospital Cancer Center, whose group is also investigating CAR T therapy for glioblastoma.

Some labs are experimenting with also using checkpoint inhibitors, established immunotherapy drugs that can shut down the false signals that a tumor broadcasts. Still others have engineered countersignals into the CAR T cells themselves. These can both scramble the tumor’s protective signal and call for additional immune support.

Perhaps the most daunting task with solid tumors, however, is pointing T cells toward the right antigen—the molecule on the cancer cells that they must latch on to. For a CAR T treatment, an ideal antigen is one that’s expressed by most of the cancer cellsbut by very few healthy cells. In cancers of the blood, researchers have found such an antigen—CD19—but finding an equivalent target in solid tumors has been more difficult. One batch of CAR T cells was designed to target HER2, an antigen characteristic of some aggressive breast cancers. The treatment, however, seemed to attack cells in the lung that naturally expressed HER2, with fatal results.

Maus’s team is developing a CAR T therapy that targets two or more antigens. This allows them to narrow in on more specific cells in the same way that an internet browser can get more specific when it uses more than one search term. And like a browser search, Maus says, her CAR T cell can use basic logic. With an “AND” function, it requires both antigens on the same cell to fire. It can also be programmed with an “OR” function, which means that the CAR T cell works if either of the receptors is engaged—an ability that allows the researchers to use one treatment to target, for instance, two different types of cells in the same tumor.

Another researcher—Robbie Majzner, a pediatric oncologist at Stanford University School of Medicine—led a group that screened 388 tumor samples from various forms of pediatric cancer. The researchers found that one antigen—B7-H3—was present on 84% of cases, and present at high levels in 70% of cases. Yet one prototype after another of B7-H3 CAR T failed in vitro. Finally, after more than 15 formulations, they produced a version that worked. It took hold in mice with the bone cancers osteosarcoma and Ewing sarcoma. “In the first few days, the tumors continued to grow,” Majzner says. “But then the T cells began chipping away at them until they were gone.”

More than 270 CAR T cell trials had been registered at the U.S. National Library of Medicine by the end of 2018, with about one-third of that number investigating CAR T cells for the treatment of solid tumors. Maus is realistic about the obstacles. “It is going to take some time,” she says, “but we’re working hard, and there is good reason to hope.”


NCI’S STEVEN ROSENBERG, MEANWHILE, has continued to pursue another promising branch of his early T cell research. In the 1980s, he began to extract T cells directly from tumor sites and used those cells—tumor-infiltrating lymphocytes, or TILs—that had broken through the tumor’s defenses as a “starter batch” for further therapies. By 1988, he had refined this TIL treatment to achieve sustained remissions in about 30% of his patients with metastatic melanoma. But further studies showed that although TILs could be grown in the lab from virtually any kind of tumor, only TILs from melanomas had significant antitumor activity. And even for melanoma, a different kind of immune therapy—checkpoint inhibition—proved to be much more effective than TIL treatment.

Yet Rosenberg continued to investigate why TIL treatment sometimes worked in melanoma. That cancer turned out to have unusually high mutation rates—and mutated DNA not only gives rise to cancer, but also to proteins called neoantigens, which are not found naturally in the body. Rosenberg suspected his TIL cells could have been very good at targeting these neoantigens—and that while melanoma may give rise to more of these targeted T cells, he might be able to find and amplify neoantigen-focused T cells in all cancers.

Advances in genetic screening and other technologies made it possible to test this hypothesis. Since 2010, his team has looked at tumors and T cells from more than 100 patients with a variety of solid cancer types. Of these patients, 70% to 80% mounted T cell responses to the neoantigens specific to their cancer. Rosenberg says he believes that this could represent a “final common pathway” that explains the effectiveness of so-called natural immunotherapies—treatments such as IL-2, checkpoint inhibitors and TILs that rely on increasing the number of the body’s own immune cells, without modifying or changing them in the lab. These T cells train themselves to home in on the neoantigens created by mutated genes. “The very products of the gene mutations that caused the cancer are likely the best targets,” Rosenberg says.

His strategy for harnessing the approach is promising, if labor intensive. His team removes cancer tissue that contains T cells and performs whole-exome and RNA sequencing of the tumor. Researchers then perform the same process on normal tissue to identify which mutations are specific to the cancer. After that, they find the T cells that have homed in on those cancer-specific mutations and put them in a medium to multiply the cells in the lab, and finally reinfuse them into the patient.

In 2014 Rosenberg published the first successful TIL treatment of a patient with a common epithelial cancer—bile duct cancer. In 2016 a patient with metastatic colon cancer showed regression of the disease under TIL therapy. That same year, his team treated a woman with breast cancer who had been unresponsive to all other treatments. The screens found 62 different mutations in the patient’s tumor cells and identified strains of TILs that recognized four of these mutations. The researchers then expanded these four strains in the lab and reinfused them. The patient’s cancer regressed and has been in remission for more than three years. Results were published in Nature Medicine in 2018.

This approach, in theory, could work for any cancer because all tumors have mutations that don’t exist in healthy cells, Rosenberg says. And it may be further improved by modifying TIL cells to help them compete more effectively in the tumor microenvironment—for example, by equipping them with ways to compete with deceptive immune signaling coming from the tumor.

TIL is a highly personalized therapy. In the 100 cancers that Rosenberg studied, virtually every patient’s T cells recognized a different neoantigen. “We need to create a unique treatment for each patient because all of their antigens are different,” Rosenberg says.

ONE CHALLENGE FOR BOTH TIL and CAR T cell therapies is their very high price tags. It can cost more than $1 million to treat a patient with Kymriah or Yescarta, once hospital stays and follow-up care are factored in. Because the treatments are labor-intensive and personalized, they will never be inexpensive, but new research into reducing side effects—which may include fever, low blood pressure and cognitive disruptions—could help cut hospital costs.

Automation in the lab is also helping reduce the time needed to engineer and expand T cells, from four months a decade ago to just a few weeks today. In a recent mouse study, Carl June, a pioneer of CAR T cell therapy and director of the Parker Institute for Cancer Immunotherapy at the University of Pennsylvania, demonstrated that the typical nine to 14 days needed for expanding the cells could be reduced to three days—and the CAR T cells infused after three days appeared to be more potent than those that mature longer.

Ultimately, there could be an off-the-shelf CAR T therapy that doesn’t require modifying the cells from individual patients but instead uses T cells from healthy people, and somehow fine-tunes them for a particular patient, perhaps through genetic manipulation. A current study at MGH is testing such cells. “They need a bit more tweaking to make them mass produced and widely available,” Maus says, “but we are making progress.”

Rosenberg has no doubt that there will be ways to simplify his TIL therapy as well and bring its costs down. But with cancer, he says, the hope of a simple, one-size-fits-all manufactured therapy may never have been an option. “Cancer is not one disease,” he says, “so it’s very unlikely that a single drug is ever going to solve this problem.”