Some immune systems defeat cancer. Could that become a drug?

Is there a way to use the body's way of fighting cancers to make a new drug?

Perhaps, according to preliminary research studies.

The idea is to exploit what is known about the growth of cancers. While many grow and spread and are deadly without treatment, some go away on their own or simply do not progress. They remain in the body, harmless and causing no symptoms. It's contrary to conventional wisdom.

But Dr. Edward Patz, who spent much of his career researching cancer at Duke University, has long been intrigued by cancers that are harmless and has thought they might hold important clues for drug development.

The result, after years of research, is an experimental drug, tested so far only in small numbers of lung cancer patients. The results are encouraging, but most promising experimental drugs fail after larger, more rigorous studies.

That hasn't stopped Patz from recently starting a company, Grid Therapeutics, hoping that the experimental drug will turn out to be a new type of cancer treatment.

Researchers not associated with the effort say they are intrigued by the concept.

"It's still very very early," said Dr. Roy Herbst, a lung cancer expert who is chief of medical oncology at Yale University. "But I like it."

So does Dr. David Barbie, chief of thoracic oncology at the Dana-Farber Cancer Institute.

But he cautioned that there is a long way to go from the concept to proof that the drug is causing tumors to stop growing or go away. At this point, Barbie said, efficacy "is still an association."

Patz's quest began decades ago. As a cancer researcher at Duke, he wanted to figure out why cancers that look similar on scans would, on inspection by pathologists and on genetic analysis, behave so differently. Some were indolent, and some were aggressive.

He decided to look for clues in samples from lung cancer patients.

"We had been collecting samples for 25 years from patients who were newly diagnosed," Patz said.

He had tumor tissue and blood samples, and records showing which patients had quickly succumbed to their cancers and which had not.

Were there distinctive genes in the tumors of those whose cancers did not recur, Patz asked? Or were particular blood proteins shed from these tumors?

His team came up empty-handed.

Then they looked in the patients' stored serum -- the clear yellow part of blood without the red blood cells -- for a sign that something had attacked cancer cells.

He found the clue in an antibody, GT103. After studying the serums of several hundred patients, Patz proposed that the presence of GT103 was a sign that a lung cancer patient might "do much, much better if they have early-stage disease."

It looked as if the antibody could prevent cancer cells from using a molecular shield -- complement factor H -- against the immune system. The antibody blocks that shield. When the GT103 antibody is available, an arm of the immune system kills the cancer cells.

The next step was to produce large quantities of GT103 to use as a possible treatment. Barton Haynes of the Duke Human Vaccine Institute supplied the antibody.

Further research showed that the antibody worked in animals. At that point, Patz left Duke to start his company. It began the first stage of drug testing in lung cancer patients, which looks only for safety.

Dr. Hirva Mamdani of Wayne State University in Detroit, Dr. George Simon of OhioHealth and their colleagues at Duke and the Moffitt Cancer Center in Florida conducted the study. It included 31 patients whose standard lung cancer treatments had failed.

The tumors didn't shrink, but they stopped growing, at least temporarily, in one-third of the patients, Mamdani said.

Simon was encouraged.

"I gave it to a patient whose tumor had grown through multiple lines of therapy," he said. "We didn't have a whole lot to offer him." But with the experimental drug, he added, the cancer stopped growing for nine or 10 months.

The next step was to combine the antibody with a pembrolizumab, an immunotherapy drug made by Merck and sold as Keytruda.

The researchers hoped the two drugs would complement each other. The antibody Patz had isolated activates one arm of the immune system. And the Merck drug unleashes another.

"Our antibody specifically kills tumor cells, but it may not kill all of them," Patz said. By adding the Merck drug, he said, "we are trying to train the immune system to take care of everything else."

Did it work?

It's difficult to know. If a significant percentage of patients had been cured, it would be obvious. But that did not happen. Tumors seemed to stabilize in most, but some then grew again.

The problem, Patz said, is that the tumors might be full of immune cells that cause inflammation, with few cancer cells. He saw that with one man's tumor that seemed to be growing. But actually, it was inflammation causing its size.

"What we find and measure on imaging are not all cancer cells," Patz said.

But there was an astonishing outcome in one of Mamdani's patients.

His tumor, she said, completely disappeared. For two years now, scans have not found any evidence of disease.

"He is no longer on any treatment," she said.

What is needed now is a much larger study with a control group that did not get the antibody, researchers said.

"We have to study hundreds of patients," Simon said.