Editor's note: This story was updated in June 2020 to reflect the drug's FDA approvals after publication.
Over its lifespan, the drug has starred in a drama with enough plot twists to rival a soap opera.
When gemtuzumab ozogamicin (Mylotarg) arrived on the scene in 2000, it was the first so-called “magic bullet” drug on the market, a targeted therapy that steers a cell-killing toxin to leukemia cells.
Then, 10 years later, it racked up another, more ignominious first. After a more extensive FDA-required clinical trial showed no clear evidence of the drug’s benefit to patient survival, it was the first medication to be withdrawn from the market after having been authorized through the FDA’s accelerated approval program, which is designed to speed urgently needed therapies to patients.
Now, new evidence may be bringing it back to life. Researchers are zooming in on a specific group of patients who could benefit from it in the latest example of the promise of precision medicine in cancer care. And its story offers important lessons about how we study new therapies, said one of the scientists whose research has helped lay the foundation for its reemergence.
“Our search for one drug that cures everything is really misplaced,” said pediatric leukemia specialist Dr. Soheil Meshinchi of Fred Hutchinson Cancer Research Center, where the drug was first developed more than two decades ago. “The drugs that failed in the past, [it] may not have been the drug’s failure but our failure to match them to the appropriate target.”
Gemtuzumab ozogamicin, or GO, links a toxin to a protein called an antibody that, because it can precisely bind to a specific target, delivers a cancer-killing punch to tumor cells while avoiding damage to most healthy cells. GO's antibody component binds to a molecule called CD33 found on the surface of certain leukemia cells. The cancer cell then engulfs the drug, at which point the toxin is released and the cell dies.
When the drug emerged on the scene, research groups around the world launched studies to test it in different dosages or regimens and in new populations of patients. One of these studies was a large, randomized multicenter trial testing the effects of adding GO to a standard chemotherapy regimen in pediatric and young adult patients with newly diagnosed acute myeloid leukemia, or AML.
The results, published in 2014 after the drug had been pulled from the market, showed a small benefit to patients’ disease-free survival but an increase in post-remission death due to treatment toxicity. In the years since, the research team, which includes Meshinchi, has been carefully digging into their findings.
“The initial data wasn’t really that impressive,” Meshinchi said. “So we started looking for populations that may benefit from this.”
The new study that Meshinchi and collaborators published on Friday in the Journal of Clinical Oncology is the latest of the group’s retrospective analyses, and it joins the growing body of research from trials in the U.S. and Europe finding evidence of the drug’s benefit to certain subgroups of AML patients.
Led by Meshinchi’s collaborator, Dr. Jatinder Lamba of the University of Florida College of Pharmacy, the researchers found that about half the patients on the trial had a single-letter change in a very particular spot in the DNA code. This alone isn’t noteworthy: Throughout our genomes, each of us has numerous instances where single letters of our DNA are different than those of another person, and most of them have little to no effect on the functioning of our bodies.
But this one was different, the researchers discovered: It results in CD33 proteins that are missing the part that binds to GO.
This surprising change opened up a huge gap in patient outcomes on the trial, Lamba and her colleagues found. Among participants without the key binding site in some or all their CD33 proteins, the drug made absolutely no difference in the risk of relapse. This contrasted dramatically with the effect of the drug in patients with all their GO binding sites intact: Among these patients, those who got only chemo had a 49 percent chance of relapsing, but those who also got GO had only a 26 percent risk of relapse.
The researchers observed similar results in disease-free survival, a measure of the time between when a patient successfully completes the initial course of treatment and when cancer symptoms reemerge.
“This is really amazing,” said Lamba, who specializes in understanding the interplay between a patient’s genomic variation and the effects of AML drugs. “I think the results are pretty convincing, and we should use them to make much better calls [about treatment] for the benefit of the patients.”
She added that their results could also have implications for other CD33-targeted drugs for AML that are currently in development.
The difference in outcomes with GO in the group of patients that was biologically able to respond to the drug was far more dramatic than what scientists who study this disease are used to, Meshinchi said.
“In AML, our clinical trials are designed hoping to detect a 6 to 8 percent difference in outcome [between patient groups],” said Meshinchi, who was the study’s senior author. “If we get an 8 percent improvement in survival for AML that’s a home run. That tells you how difficult a disease this is to cure.”
“So this,” he said, “is as big a home run as you could possibly have in AML.”
Meshinchi remembers saying “wow” when he first learned of this result, about a year ago. And he promptly headed down four stories in Fred Hutch’s clinical research building, to the office of his mentor, Dr. Irwin (Irv) Bernstein.
Meshinchi asked Bernstein: What do we know about CD33?
Bernstein is one of the best people on the planet to answer that question. A specialist in blood cancers and the biology of blood-forming cells of the bone marrow, Bernstein was one of several investigators that discovered the protein now known as CD33 independently around the same time in the early 1980s. He and his colleagues helped prove which cells CD33 is found on and which it is not. (He is also one of the scientists responsible for the “CD” naming system — CD stands for "cluster of differentiation" — now used for the blood-cell proteins that have become targets for many current treatments and diagnostics used in leukemia and lymphoma.)
GO germinated in Bernstein’s lab, in the fertile soil of his research on CD33. Bernstein had shown that CD33 was present on leukemic precursor cells but not on normal blood stem cells, suggesting that if CD33-positive cells could be selectively eliminated, the bone marrow would be able to resume its normal blood-making functions. When his team developed an antibody specific to CD33 in the 1980s, he said, the implications for cancer treatment were obvious immediately. “We envisioned the translations right away for antibodies, to couple them with drugs,” he said.
After an early attempt to use Bernstein’s antibody to deliver targeted radiation to cancer cells, the researchers pivoted to the idea of using the antibody to deliver a toxin instead.
Bernstein’s pursuit of this idea had led him to calicheamicin, a bacterial poison that was discovered from a sample of Texas soil that a pharmaceutical industry researcher had gathered on vacation. It is so toxic that it could never be given systemically, like traditional cancer chemotherapies are. But industry research showed that it could kill cancer cells in test tubes in the lab; if it could be targeted just to cancer cells, the researchers hypothesized, it might prove to be an effective therapy in AML patients.
As any scientist could tell you, research is often filled with scientific pitfalls, ideas that don’t go anywhere, experiments that don’t work as planned. But for Bernstein, the only real roadblock in all his years of initial research on this drug was the challenge of convincing a pharmaceutical company that it would be worth their while to pursue a CD33-targeted drug, given the small market offered by the relatively rare AML. And then, once that was done and preclinical experiments were ongoing, doing the convincing over and over again, as the original partner company was bought by another that was bought by another in an avalanche of industry mergers.
The frustrations of those years are still fresh for Fred Hutch Deputy Director and Executive Vice President Dr. Fred Appelbaum, who had launched the first trial of Bernstein’s antibody in patients and had been collaborating on the drug’s development ever since.
“Each time it went from one company to another, Irv and I would fly to the East Coast and convince the company not to kill this project, that this was a great proof of principle,” Appelbaum remembered.
Finally, in the late 1990s, the drug was tested in humans for the first time in a Phase 1 trial at Fred Hutch.
Through the accelerated approval program, the FDA granted the drug company approval in May 2000 to market GO for adults with AML that had come back after previous treatments ― a group of patients desperately lacking treatment options. The drug benefited from all four mechanisms the FDA has to get much-needed therapies to market faster. At the time it was the first new drug brought to market for AML in 15 years.
The approval was based on the pooled results of 142 participants in three Phase 2 trials that found evidence of the drug’s potential for benefit to these patients —but no definitive proof that it helped them survive longer.
Typically, a drug can only receive FDA approval after a randomized, controlled trial offers solid evidence of its safety and efficacy against disease. But through accelerated approval, drugs that fill an unmet need can be brought to market under a lower standard of proof — with the requirement that the pharmaceutical company then conducts a larger Phase 3 follow-up trial to rigorously test the new drug’s benefit to patients.
The following four years were “painful,” Appelbaum remembers, as the drug company (Wyeth) and the independent network of investigators slogged through pilot studies, back-and-forth discussions about trial design, and the complex series of regulatory approvals to get GO’s required Phase 3 trial off the ground nationwide. The first patient was finally treated on the study in August 2004.
In 2009, things took an unexpected turn: According to a preliminary data analysis, GO did not improve survival when it was combined with chemo, and it was associated with an increased risk of death from treatment toxicity.
From this point, events moved rapidly: The trial was halted, the FDA recommended that GO be withdrawn from market, and, in June 2010, the drug company — by now, Pfizer Inc. — voluntarily complied. After Oct. 15 of that year, patients with AML in the U.S. could no longer receive GO without special approval.
Buzz over GO’s potential reemergence on the market began to build almost as soon as it was pulled, however, as each new piece of evidence trickled out from all the other ongoing trials of the drug. Research groups in Europe had published the results of several trials of different regimens of GO in newly diagnosed patients that showed evidence of the drug’s benefit to patients, especially those with low- or intermediate-risk cancers. And last year, building on early research by investigators like Dr. Roland Walter of the Hutch, Meshinchi, and collaborators demonstrated that GO was only effective in their young trial participants who had CD33 levels above a certain threshold.
GO had barely left the scene, and it was already time for a comeback.
In February 2017, with results from all these newer studies in hand, Pfizer submitted to the FDA an application for GO’s approval, once again. The agency is scheduled to discuss the application in a few weeks.
[Editor’s note: The FDA approved the drug in September 2017 for adults newly diagnosed with CD33-positive AML, and for adults and children over age 2 with treatment-resistant, CD33-positive AML. The agency expanded the approval in June 2020 to include babies and children newly diagnosed with the disease.]
From his office at Fred Hutch, Appelbaum reflected on the twists and turns of GO’s story over the past several decades.
“I think we’ve learned an awful lot in the course of this,” Appelbaum said. “It’s been fascinating and it’s also sometimes been disappointing. You hope for better outcomes and then, it’s fun to see that it probably did help some patients get cured with this. You’ve just got to figure out how to keep working on this and make it better the next time.”
Meshinchi feels that there’s enough evidence now to support the use of GO as standard of care for newly diagnosed AML patients who possess only the full-length version of the CD33 protein that can bind the drug. In the latest study with Lamba, the researchers found that whether the patient has the genetic variant they discovered predicts how well GO will work for that patient better than any other predictive factors identified in previous studies.
“What this will allow us to do is actually predict who will respond to Mylotarg, instead of treating everyone with it and exposing them to the potentially significant toxicity it could have,” Meshinchi said.
Meshinchi, Lamba and their collaborators have developed a method to rapidly determine whether a given AML patient has the variant that allows GO to bind to CD33. Their test is now available through the Seattle medical laboratory testing company Hematologics, Inc., and possibly more labs in the near future, to help guide patient care. The European research groups are now using it to go back and reanalyze their samples, too, to confirm the U.S. team’s latest result, the researchers said.
Meanwhile, Meshinchi and his teammates are designing their next big AML trial, which will incorporate GO for patients with the complete version of the CD33 protein. They’re also delving further into the biology of the genetic variant they discovered and testing antibodies that bind to universal portions of the CD33 molecule, with an eye toward making more widely applicable CD33-targeted drugs in the future. They’re also talking about ways to potentially correct the variant in the half of AML patients whose cancers would not naturally be vulnerable to the drug, though this is further down the road.
Meshinchi hopes that other cancer researchers will learn from the story of GO the importance of targeting drugs in development to the populations most likely to benefit from them — and of striving toward big strides forward for patients, not small ones.
“We are trying to design smarter trials. But we’re not there yet,” Meshinchi said. “Part of the issue is, we need to really change our mindset and our expectations. As long as you’re happy with survival rates of 30, 40, 50 percent, and your goal is to improve outcomes by 5 percent, we’re not going to make a lot of gains.”
Spurred by the most recent development in the GO story, Bernstein is now delving back into his research on CD33 biology and writing the latest chapter of his work on this topic. He is gratified to see something he started studying so long ago get resurrected and to offer unexpectedly “spectacular” promise for a subgroup of patients.
“It’ll be a resurgence of the use of the drug when people find out about” the new finding, he said. “It’s quite exciting. I’m glad I’m still around to see this.”
Talk about this story on Facebook.
Susan Keown is an associate editor at Fred Hutchinson Cancer Research Center. She has written about health and research topics for a variety of research institutions, including the National Institutes of Health and the Centers for Disease Control and Prevention. Reach her at firstname.lastname@example.org or on Twitter @sejkeown.