The young surgeon was mystified. A fist-sized tumour had been removed from the stomach of his patient 12 years earlier, but his doctors had not been able to cut out many smaller growths in his liver. The cancer should have killed him, yet there he lay on the table for a routine gallbladder operation.

The surgeon, Dr Steven Rosenberg, examined the man's abdominal cavity and liver, feeling for hard, dense tumours - but he could find no trace of cancer.

It was 1968. Dr Rosenberg had a hunch he had come across an extraordinary case in which the patient's immune system had vanquished cancer. Hoping there was an elixir in the man's blood, he got permission to transfuse some into a patient dying of stomach cancer. The effort failed. But it was the start of a lifelong quest.

"Something began to burn in me", he would write later, "something that has never gone out".

Half a century later, Dr Rosenberg, now chief of surgery at the National Cancer Institute in the United States, is part of a small fraternity of researchers who have doggedly pursued a dream - turbocharging the body's immune system so more cancer patients can experience recoveries like that long-ago patient's.

Dr Rosenberg, Dr Carl June of the University of Pennsylvania and Dr Michel Sadelain of Memorial Sloan Kettering Cancer Centre have been at the forefront of this research for decades - labouring in separate labs in an intense, sometimes cooperative, sometimes competitive pursuit to bring to fruition a daring therapy that few colleagues believed would work. Now, versions of the therapy for a limited number of blood cancers are nearing approval by federal regulators, and could reach the market as early as next year.

The technique, known as cell therapy, gives each patient an individualised and souped-up version of his or her own immune system, one that "works better than nature made it", as Dr June puts it.

The patient's T-cells, the soldiers of the immune system, are extracted from the blood, then genetically engineered to recognise and destroy cancer. The redesigned cells are multiplied in the laboratory, and millions or billions of them are put back into the bloodstream, set loose like a vast army of tumour assassins.

This radical, science-fiction-like therapy differs sharply from the more established type of immunotherapy, developed by other researchers. Those off-the-shelf drugs, known as checkpoint inhibitors, release a molecular brake on the immune system, freeing it to fight the cancer much as it fights infections by bacteria or viruses.

Cell therapy, in contrast, is brewed especially for each patient, one of the many challenges faced in broadening its use. So far, patients treated with this therapy number in the hundreds, not thousands. And for now it works for certain types of blood cancers only, not common malignancies such as breast and lung cancer. Researchers are also trying to determine how to control potentially lethal side effects. Recently, a clinical trial was halted briefly after three patients died of brain swelling.

Still, cell therapy has produced complete remissions in some patients who were out of treatment options - stirring excitement among doctors and patients and setting off a race between companies to bring the treatments to market.

Getting to this point has taken decades of painstaking work, with many false starts and setbacks.

"It was conceivable we were pursuing a ghost," Dr Rosenberg recalled.

When he arrived at the National Cancer Institute in 1974, his first attempt at immunotherapy was to give patients T-cells harvested from pigs. That failed.

He then began giving patients interleukin-2, or IL-2, a protein made by the body that spurs T-cells to proliferate. In some cases, he treated patients with their own white blood cells that had been incubated in IL-2. The treatments sometimes set off such a violent immune-system reaction that patients had to be placed in intensive care.

From 1980 to 1984, he treated 66 patients without success. Then, in late 1984, he encountered patient No. 67, navy officer Linda Taylor, who had melanoma and whose personnel file carried the stamp "death imminent". She is still alive; her case and others catapulted Dr Rosenberg and IL-2 onto the cover of Newsweek and various newspapers. Some colleagues at the National Cancer Institute began referring to him as Stevie Wonder, thinking he had developed a swelled head.

But IL-2's vaunted prowess fizzled out - it helped only a small percentage of patients with melanoma or kidney cancer.

Dr June started working with T-cells at the Naval Medical Research Institute in the mid-1980s. He and colleague Bruce Levine found a way to multiply T-cells in huge numbers outside the body, a method still used today. And in the mid-1990s, working with gene therapy firm Cell Genesys, Dr June began trying to genetically modify patients' T-cells to kill HIV, the virus that causes Aids.

But when his wife, Cindy, the mother of their three children, developed ovarian cancer in 1996, his research turned personal.

He had tried everything to save her, including the primitive immune therapies under development. But she died in 2001.

"A lot of other scientists would have been disillusioned by the failure, in his case the personal tragedy," said Internet billionaire Sean Parker, who is funding some of Dr June's work.

Instead, Dr June, who had moved to the University of Pennsylvania, stopped treating patients, and devoted himself to creating cell therapies for cancer.

"Things that were back-burner on cell therapy became front-burner," he said.

In the 1980s, scientists began experimenting with gene therapy, putting new genes into cells of the body to treat disease. Dr Sadelain, while still a graduate student studying immunology at the University of Alberta, told colleagues he thought the technique could be used to supercharge T-cells to fight cancer.

"At the time it sounded very pipe dream," said Dr Douglas Green, who was one of Dr Sadelain's doctoral thesis advisers and is now chairman of immunology at St Jude Children's Research Hospital.

But Dr Sadelain, he continued, "believed in the approach and pursued it relentlessly". After earning his PhD, Dr Sadelain headed for the Whitehead Institute for Biomedical Research in Cambridge, Massachusetts, to learn how to do gene therapy, using disabled viruses that could not cause disease to deliver genes into cells. By 1992, he had demonstrated that he could genetically engineer mouse T-cells.

He then moved to Sloan Kettering. In 2003, he and his colleagues - including his partner and now wife, Isabelle Riviere - showed that genetically engineered T-cells could eradicate certain cancers in mice.

How is this done? To fight cancer, T-cells must be able to recognise cancerous cells. Each T-cell in the body has unique receptors - sort of like claws that jut out from the surface. T-cells patrol the body looking for protein fragments that indicate a cell might be infected by a bacterium or virus. If one of the claws latches on to such a fragment, the T-cell destroys the cell displaying it.

But cancer cells are mutated versions of the body's own cells, not outsiders. T-cells do not always recognise them as something to kill.

So scientists like Dr Sadelain decided to put a new claw on the T-cells, one that could recognise cancer by latching on to a telltale protein on cancer cells.

The new claws came from another part of the immune system known as antibodies. Drug companies already knew how to make antibodies with claws that would bind to specific proteins in the body.

But the claw was not enough. Once a claw binds to a target protein, it needs a molecule to signal the T-cell to go into killing mode. Yet another signal helps sustain the killing. The DNA instructions for all three components are inserted into the patient's T-cells.

Since this concoction is part antibody and part T-cell, it is called a chimera - like the monster of Greek mythology that is part lion, part goat and part serpent. The claw is called a receptor and the protein it binds to on the cancer cell, the target, is called an antigen. The whole construct is called a chimeric antigen receptor (CAR), and the use of it to treat cancer is called CAR T-cell therapy (CAR-T).

Dr Sadelain was not alone in this work. Dr Zelig Eshhar, an Israeli scientist, is credited with developing one of the first crude CARs around 1989. Dr Rosenberg, always on the lookout for new types of immunotherapy, invited Dr Eshhar to be a visiting scientist in his laboratory at the National Cancer Institute.

Another early developer was Dr Dario Campana of St Jude Children's Research Hospital.

As the first decade of this century neared its end, the three pioneers were ready for the big moment - testing their treatments in patients. They scrambled to be the first to announce peer-reviewed results.

Dr Rosenberg and his colleagues published first, in the journal Blood in 2010. They described a single patient with lymphoma whose tumours shrank after treatment. (The patient later received more therapy, and has been free of cancer since.) But the approach really attracted attention the following year when Dr June reported that two of three patients with chronic lymphocytic leukaemia went into complete remission.

One of them, Mr Doug Olson, a chemist from Tinicum Township, Pennsylvania, left the hospital and immediately bought a sailboat. He also took to running half-marathons.

"It was like this weight that had been sitting there was gone," said Mr Olson, who is free of cancer nearly six years later.

Mr Bill Ludwig, a retired captain of the New Jersey Department of Corrections, had already paid for his own funeral when he started treatment in August 2010. Once his genetically engineered T-cells were unleashed in his system, his lungs started to fail, his legs ballooned, his blood pressure dropped and he began hallucinating.

When he emerged from the ordeal, doctors searched for cancer. Detecting none, they ordered another test, certain of error. But there was no mistake. Over 2kg of tumour had been destroyed.

Mr Ludwig, now 71, and his wife bought a recreational vehicle. "We're trying to make up for lost time," he said. He has celebrated the high-school graduations of five grandchildren and welcomed his first great-grandchild.

As for Dr June, Mr Ludwig said: "It's hard to describe someone who basically saved your life. He lost the one he loved, and turned around and saved me years later."

Yet, for all the excitement, there are reasons for caution. The CAR therapy works now only for patients with some B-cell lymphomas and leukaemias, which account for only about 80,000 of the 1.7 million cases of cancer diagnosed a year in the US. It has not been used successfully to treat malignancies of the lungs, breast, prostate, colon or other organs.

"The solid tumours that kill over 90 per cent of people do not respond to anything we have now," Dr Rosenberg said.

Because it is personalised, cell therapy is likely to be frightfully expensive - probably hundreds of thousands of dollars per patient, although the firms bringing these treatments to market have not yet said how much they would charge.

Producing the re-engineered cells is lengthy and complex. Some patients have died during the two to four weeks needed to genetically modify and multiply their cells.

And the therapy itself can be arduous. First, patients get chemotherapy to wipe out many of their existing T-cells, to make room for the engineered ones. Once those enter the body, they can set off a ferocious immune response as well as temporary neurological problems such as memory loss, seizures and hallucinations.

The big thrust now is to expand the use of cell therapy to additional types of cancer.

The key is to find protein targets that the engineered T-cells can latch on to, to kill cancer cells. Ideally, such a protein should be on all the tumour cells, so the entire cancer would be eradicated. But it should not be on healthy cells, or they would also be destroyed, causing side effects.

"T-cells are very powerful," said Dr Campana, now at the National University of Singapore. "In the same way that they can eliminate cancer, they can also kill you."

It might turn out that the best target for each patient will be unique to that person. Scientists are experimenting with DNA sequencing and other techniques to find the best mutated protein in each person's tumour at which to aim the claw.

"Think of how dauntingly personalised this is," Dr Rosenberg said. "We are using their own cells to treat a unique mutation in their own tumour."

He said this approach might allow cell therapy to be used for most patients.

Many other improvements are on the runway.

Dr Sadelain and Dr June are working on "armoured CARs" that not only bind to the target but also produce immune-stimulating chemicals. French firm Cellectis has treated two babies with an off-the-shelf CAR treatment that does not require each patient's cells to be processed. Bellicum Pharmaceuticals is working on genetic switches that dim or shut off the CAR if the treatment is endangering the patient.

"We're in the Model T version of the CAR now," said Dr Levine, now the director of the cell production facility at the University of Pennsylvania. "What's coming along are Google CARs and Tesla CARs."

NEW YORK TIMES