On the day I arrive at MD Anderson Cancer Center in Houston to meet James Allison and his longtime collaborator Padmanee Sharma, they are nowhere to be found. The previous day, one of their colleagues informs me, Allison was summoned up on stage by Willie Nelson, in front of 60,000 people at a rock festival in Austin, to deliver a harmonica solo. They are still on their way back.

By now, Allison is almost used to adulation. There are even murmurings that his work in cancer immunotherapy might win him the Nobel Prize. Twenty years ago, he was the first to show it’s possible to turbocharge the body’s response to cancer with a drug that releases the immune system so that it destroys tumors on its own.

The drug he identified to do that, called Yervoy, went on sale in 2011 to treat metastatic skin cancer. In lucky patients, it causes otherwise fatal tumors to melt away. By last year, worldwide sales of Yervoy and two newer drugs had reached $6 billion a year, and the medications had been given to more than 100,000 people. This transformative new class of immunotherapy agents, known as checkpoint inhibitors, is acknowledged to be the most important advance against cancer since chemotherapy.

Allison, who is 68, is an unimposing man, with a slight Texas drawl and a stringy mane of white hair. He still finds it hard not to cry when he meets cancer survivors saved by his discovery. But I had gone to talk to him about unfinished business. That is because for every miracle cure, for every Jimmy Carter or 22-year-old melanoma patient pulled back from death, there are many more people who, for reasons that no one understands, can’t be saved. Of all patients dying from all types of cancer in America this year, only one in 12 would be expected to benefit from any immunotherapy drug. Some even argue that direct-to-consumer marketing, including a Super Bowl ad, has created dangerous expectations. Patients cashing in their last chance will, more likely than not, find themselves among the large majority for whom drugs like Allison’s don’t yet work.

Allison has known about the shortcomings longer than anyone. He says they dampen any sense of triumph and shadow him at award banquets. Sometimes, he stays awake at night. “About 22 percent of melanoma patients that get a single round of treatment with Yervoy are alive 10 years later,” he said after receiving a Lasker Award in 2015, and then added solemnly: “We got to get that up, and we got to do it in more kinds of cancer.”

At MD Anderson I was introduced to what Allison calls the “platform.” It is a large-scale effort to determine why the immune system at times acts like the perfect weapon but in other cases fails to kick into action. Sharma, a Guyanese immigrant and practicing cancer doctor, oversees the collection of tumor samples from 100 of Anderson’s 165 cancer trials that involve immunotherapy. The tissue is then scrutinized by her lab and Allison’s for clues to how the battle is proceeding. “What is the immune response doing that leads to tumor rejection? What is the immune response doing that it stops rejecting the tumor and [it] starts growing again?” Sharma asks. “Those are big questions that we still need to understand.”

The answers can’t come too soon for some. The pharmaceutical industry and research institutions are in the midst of a pell-mell sprint into thousands of clinical trials based on new immunotherapy agents. As of October, by one tally, more than 166,736 patients were being sought to fill slots in studies of drugs involving a single protein, called PD-1. The overall number of immunotherapy trials probably tops 3,000, says Jeff Bluestone, an immunologist at the University of California, San Francisco, who also serves as president and CEO of the Parker Institute for Cancer Immunotherapy.

But a growing number of researchers fret that the flood of clinical trials is uncoördinated, redundant, and potentially counterproductive. That is because in many cases, the basic science remains little understood. “This is not sustainable,” Ira Mellman, the keynote speaker at the annual meeting of the Society for Immunotherapy of Cancer, told his colleagues when he took the stage last fall. Mellman, a vice president at the biotech behemoth Genentech, put up a byzantine diagram, consisting of concentric circles crammed with small type. The visually overwhelming slide showed trials under way to test immune-boosting therapeutics. His industry, he said, is “[throwing] plates of pasta against the wall, and hoping that something is going to stick.”

Mellman told me that while Allison hadn’t invented immunotherapy, his drug had been the one that clarified its potential. Now, he says, Allison’s is one of the “few serious efforts” to better understand the mechanisms by which the immune system is killing cancers and the reasons why, too often, it is still not seeing them. “We would have a much better shot at doing what’s best for patients, doing best for science, if we understand mechanisms,” he told me. “You can just wildly try different things and hope that something works, or you can go back and try again and understand the basis of all of this. Until we know that, we’re not going to really understand why some respond and some don’t.”

Checkpoint discovery

Cancer is personal for Allison. At 10, he held the hand of his mother, Constance, in tiny Alice, Texas, and wondered at the burn marks up and down her neck. He had not expected her to die. Only later did he learn that the marks were from radiation, and that cancer had killed her. By the time he was 15, cancer had consumed two of his uncles.

When Allison first began to chart a scientific career, he says, he recoiled from cancer. Back then, it seemed, there were few real clues. And immunology, the field he had picked, had a particular reputation for serving up fool’s gold when it came to the disease. “I couldn’t get any purchase on it,” he recalls. “I wasn’t going to go crashing into something until I knew how it worked.”

At that time, in the 1970s, T cells—those tiny assassins that allow the body to fight off infections—had only recently been discovered. Allison was fascinated to learn there were molecular-level sentinels that patrolled the human body looking for trouble—that “if they see something wrong, they just deal with it.” He thought: “What could be cooler than that?”

The existence of such immune cells did raise an obvious question: if T cells were designed to protect the body by killing infected and diseased cells, how was it that cancer managed to elude them? By then, there were hints that sometimes tumors did in fact succumb. In the 19th century surgeons had inoculated cancer patients with heat-killed bacteria, with inconsistent results. In 1980, a Time magazine cover spotlighted a scientific frenzy around a molecule called interferon, which sends the immune system into overdrive. But the treatment was indiscriminate, as likely to harm a person as heal. “It was crazy, because people were doing things and they didn’t understand how they worked,” Allison remembers. “People just said: ‘Oh, well. It causes T cells to grow. So we put tons of it into people.”’

Allison instead began studying the molecular receptors present on the T cell’s surface. One of his most important findings was to locate a receptor called CD28 that acts like a gas pedal. When it gets engaged, it is one of two key signals—in addition to a receptor that actually locks onto a tumor cell and functions somewhat like an ignition switch—that a T cell needs to initiate an attack.

Even when those switches were flipped to the “on” position, however, such attacks were often short-lived and sometimes failed to start up at all. By 1992, Allison thought there might be a third switch. The most likely candidate: CTLA-4, a mysterious receptor sometimes spotted on T cells. But both Allison and Bluestone, the UCSF immunologist, found that this molecule behaved unexpectedly. When proteins bound to it, it didn’t turn a T cell on—it turned it off. These molecular brakes were called checkpoints.

Scientists subsequently demonstrated why evolution might have favored checkpoints. When they created mice lacking CTLA-4, their T cells ended up attacking their own bodies after an infection. Without an off switch, the mice “died within a few weeks, of massive autoimmune disease,” Bluestone recalls.

Bluestone initially saw the chance to develop new types of immune--suppressing drugs—say, for organ transplant patients. But Allison saw a different possibility. Releasing these brakes might strengthen the immune system’s response against cancer. One of Allison’s graduate students had already developed an antibody able to stick to a T cell’s CTLA-4 receptors, essentially jamming the switch. Allison instructed a postdoc to inject the antibody into mice riddled with tumors. The results, he recalls, “were spectacular.”

“The tumors were cured,” Allison says. “I mean, it was 100 percent and zero percent—no statistics necessary.”

Miracle drugs

The drug, the first of the checkpoint inhibitors, would become known as ipilimumab or Yervoy, and it is now sold by Bristol-Myers Squibb, a pharmaceutical company headquartered in Manhattan. Human studies began around 2000 on 14 patients stricken with metastatic melanoma, who were steeling themselves for their finals days in hospice. But after the trial began, three saw their tumors shrink. Allison, who moved to New York City’s Memorial Sloan Kettering in 2004 to be closer to the trials, soon met one of the patients his drug had saved. Sharon Belvin was in her 20s, and had just finished college and gotten married, when metastatic melanomas appeared in her lungs, liver, and brain. She was terminal by the time her physician enrolled her in the first phase II clinical trial. The day Allison met her, she’d been in remission for a year.

“She hugged me,” Allison recalls. “Her husband hugged me, and her mother and father were there, and they all hugged me. It was just sobbing, and everybody was really happy. I walked to my office and I had a lot to think about. I cried all the way there.”

Allison says by that time he was aware of his drug’s limitations. It didn’t help everyone, and it didn’t work in most cancers. And if he needed a reminder of the stakes, it came in 2005, when Allison’s brother succumbed to prostate cancer after eight years. The same year, doctors found early-stage cancer in Allison’s own prostate. He had surgery rather than chance drug treatment.

As soon as cancer researchers learned that Yervoy worked on some previously incurable patients but not on others, many asked the obvious question: was it possible the body had more than one checkpoint? Another molecule, called PD-1, was quickly identified and successfully targeted with checkpoint inhibitors. Allison’s Yervoy was approved in 2011 by the U.S. Food and Drug Administration for patients with melanoma. Three years later the FDA approved Merck’s PD-1 inhibitor pembrolizumab (Keytruda) and a similar drug, also from Bristol-Myers Squibb, called nivolumab (Opdivo). One or both have since been approved to treat some types of lung cancer, kidney cancer, and Hodgkin’s lymphoma, creating the most important new class of cancer drugs in a century.

Gatling gun

The day I arrived at MD Anderson to tour the platform, an Argentinean immunologist, Luis Vence, greeted me in a fluorescent-lit hallway. Our first stop was a lab where he swung open the door of a refrigerator-size machine to reveal 28 black canisters arrayed around a central hub, like the bundled cylinders of a Gatling gun. When cancerous samples come in, they are treated with fluorescent antibodies designed to stick to CTLA-4, PD-1, and other molecules on the surface of immune cells. The machine can then, in a few seconds, use a laser to scan all 10,000 or so cells from a biopsy, count them, and separate them by type. Vence compared it to sifting through multicolored ping-pong balls.

In a nearby lab, one of his colleagues, Jorge Blando, directed me to a microscope through which I could see a panorama of a cellular battle under way. The slide contained a slice of bone marrow riddled with tumors. These were recognizable by their larger, fuller-shaped cells. Among them were the tiny immune cells, stained brown, that had infiltrated and begun to attack. Others seemed to hover around the periphery. How many eventually make it in—and how long they survive to keep fighting—determines whether the tumor is defeated.

“What you are looking at in cancer is natural selection at a high speed,” Vence says. “When you treat it with chemotherapy, maybe you destroy 99 percent of the tumor. But the 1 percent that is left is resistant to chemotherapy. That’s the one that grows back and basically kills you.” This explains why even the latest targeted drugs—those designed to hit very specific molecules on, say, a breast cancer cell—typically extend patients’ life by only a few months.

Yet Vence and others believe that the immune system is inherently capable of spotting and countering any move a cancer makes. How else to explain how some advanced melanoma patients, who had tumors in their lungs or brain, are disease-free years after a course of Yervoy infusions? “The beauty of immunotherapy,” Vence says, “is that the immune system can evolve at the same time, along with the tumor. It can keep up much easier.”

It was Sharma who had the idea for the platform. When she began it, few volunteers were yet receiving Yervoy, then a relatively new and unproven drug. So Sharma persuaded patients slated to have less serious tumors removed by surgery to take small doses. A biopsy sample was collected before the drug was administered. Then, comparing the initial cancer and the excised tumor, the lab could use state-of-the-art technology to track the immune response and begin to examine why it didn’t always work. Sharma’s first finding came fairly quickly. In tissue from bladder cancers treated with CTLA-4 antibodies, readings from the Gatling gun showed that T cells possessing a molecule called ICOS were “off the charts.” Sharma’s reaction was elation mixed with confusion. T cells with ICOS on them had previously been found only in the tiny sacs in the lymph nodes known as follicles, and they were believed to suppress immune responses, not enhance them. Allison decided to engineer mice whose tumors triggered ICOS. In their tumors, CTLA-4 was four times as effective. ICOS, it turns out, was part of cascade that made T cells attack tumor cells more effectively.

“I can’t believe we missed this,” Allison remembers telling Sharma. “This is amazing.” He’d been scooped by his collaborator and felt blown away. They’d been spending more and time together, talking on the phone and working on science. Now he blurted out: “I love you!” Sharma recalls plowing forward with the conversation as if nothing had been said. But he had said it. The pair were married in a small ceremony in 2014.

With the help of the Boston venture capital firm Third Rock Ventures, they also started a company called Jounce Therapeutics that is developing a drug to increase ICOS levels. Human tests got under way last year, and although it’s too early to know how the drug is working, the idea has already been remunerative. Jounce went public in January, raising $117 million. Now Sharma drives a Tesla with a vanity plate that reads “ICOS.” On Allison’s Porsche, the plate says “CTLA4.”

A tidal wave

During the same meeting at which -Mellman castigated the industry for spaghetti throwing, I saw Allison huddled over an iPad with another scientist, discussing some of the most recent findings he, Blando, and Sharma have made using their platform. They have been studying prostate cancer, in which no checkpoint drug yet seems to work. “What we found was that prostate cancer is almost a desert immunologically,” Allison says. “It’s a very cold tumor. There’s not much in there.” But Blando’s microscope has revealed that two drugs together might make the difference. Yervoy, he found, is necessary to drive T cells into the tumor, while the addition of a PD-1 drug makes sure they start killing. On the basis of these results and further research, Sharma and Allison convinced Bristol-Myers to combine the drugs in a clinical trial for advanced prostate cancer.

Many immunotherapy trials don’t have as much new preclinical research behind them. One reason is that drug companies are still exploiting the original checkpoint discoveries. Bristol-Myers’s Opdivo has been approved for eight different cancer “indications” in two years, which must be a record. “The pace of the clinical applications of the science is much faster than understanding mechanisms in the lab,” says Gregory B. Lesinski, a scientist at the Winship Cancer Institute of Emory University.

But racing ahead of the science can also incur huge penalties. Last summer, a test of Opdivo as a first-choice treatment for advanced lung cancer led to one of the greatest fiascoes in the company’s history. Bristol had organized a trial that, in seeking the largest market, had essentially taken all comers. Its competitor Merck chose to test its drug in only lung cancer patients whose biomarkers indicated they were most likely to respond. When Merck reported its results in June, they were so good that independent monitors said patients in a control group using chemotherapy could switch to the new drug immediately. Then, in August, Bristol acknowledged that its own test had failed to show a benefit. Shares of the company dropped by 20 percent, and Bristol’s research and development chief eventually stepped down.

The revival of immunotherapy now includes cancer-fighting viruses, genetically reprogrammed T cells, and vaccines designed to make tumors more visible to the immune system. Understanding the best way to put it all together is one of the crucial jobs ahead. At times, the explosion of new activity has tended to diminish the importance of Allison’s drug. Although it is still a billion-dollar-a-year blockbuster, Yervoy is now less often prescribed, in part because of side effects. One analyst called it the “iPod of immunotherapy”—a product overshadowed by the revolutionary change in thinking it caused. “Its importance would be hard to overstate in terms of what it has done to crystallize all the other activities,” says Mellman. “In my opinion, the idea that the immune system could target cancer didn’t start with Jim. But the field did.”

Once a year Allison packs a sold-out venue at the American Society for Cancer Research. There his own band, the Checkpoints, plays to doctors and scientists, nearly all of whom are converts to immunotherapy. Yet Allison still recalls the reviewer who, two decades ago, told a journal to reject his breakthrough paper because “we all know that immunotherapy’s crap. It’s never worked.”

Now that immunotherapy looks like the future, how far can it go? As I stood with Allison and Sharma in the MD Anderson parking lot, saying our good-byes, they seemed hopeful. Allison grabbed a piece of paper and sketched a graph. Start with everyone who has cancer, he said. Then, going out to the right, trace the survivors: how many are left after two months, six months, a year. It’s a line that, for most advanced cancers, drops relentlessly to the dust. But immunotherapy is lifting the curve. In melanoma, there are more and more long-term survivors. Allison calls it “raising the tail.”

“Ultimately, the goal is to try to get the survival rate as high as we can in as many different kinds of cancers as we can,” he says. Allison has finally gotten purchase on the monster that darkened his childhood. And he is not going to let go. 

Adam Piore is the author of The Body Builders: Inside the Science of the Engineered Human.