The notion of drug resistance existed at the time, but it had never been studied at a molecular level, and this drug provided a unique opportunity. Because Gleevec worked by binding to Bcr/Abl and shutting it off, Sawyers figured that resistance was ascribable to one of two things: Either the drug was no longer blocking the kinase, or the kinase was no longer the sole trigger for the disease.
Just as Lydon had had to invent a way to screen Zimmermann’s experimental compounds for anti-kinase activity, Sawyers now had to invent a way to study resistance at the molecular level. Once again, methods were being invented for a need that had never existed. Although the concepts he was testing were straightforward, conducting the actual experiments was not. Looking at what genes or proteins were activated or inactivated at different moments in the progression of a disease was totally new. There were no pathways in place for investigating the cellular changes that caused cancer to become resistant to a drug. Sawyers was the first explorer to map this uncharted territory.
At this point, there were plenty of patient samples to work with. Using the same kind of gel analyses with which Owen Witte parsed Bcr/Abl nearly twenty years earlier, Sawyers could measure the level of Bcr/Abl in the samples from patients who’d stopped responding to Gleevec. The pattern was “absolutely crystal clear,” said Sawyers. The levels were high at the start of treatment, plummeted when the patients were doing better, and then crept up again as patients relapsed. Those high levels were evidence that the drug was failing to inhibit the target.
It was one thing to confirm that the kinase was no longer being blocked in patients who’d become resistant. But why? What was going on that would prevent the drug from doing what it did in so many other patients? Why were patients in more advanced stages of the disease more likely to stop responding to the drug? Why did some chronic-stage patients not respond at all? One possibility was that the drug was being metabolized more quickly in patients who didn’t respond compared with those who did. If the chemical was broken down and eliminated from the body too quickly, it might not get a chance to enter the bloodstream and plug the binding site in the malignant cells.
Sawyers had begun tackling resistance while the phase II studies were still underway, and a few months after the drug was approved in 2001, he was ready to publish his first findings. In all of the samples he had studied, Bcr/Abl had been reactivated after its initial shutdown. The kinase had stopped its incessant phosphorylation and then started right back up. He had discovered that in some of the samples, a single amino acid on the enzyme had been changed at an area where a critical bond with the drug was supposed to take place. In others, cells were producing excessive amounts of the already mutant bcr/abl gene, leading to excessive amounts of the Bcr/Abl enzyme, more than the kinase inhibitor was able to block. Why those changes occurred was still a mystery, but the brambles obscuring the problem were being cleared.
The timing of his report was an unexpected counterbalance, coming so closely on the heels of one of the most celebrated drug approvals in the history of cancer care, an event that had been lauded as the beginning of a new era for targeted drugs, a hallmark of personalized medicine. Sawyers rightly feared the media’s spin on his publication, and begged the Wall Street Journal reporter who interviewed him to not turn the phenomenon of resistance into melodrama. “It’s amazingly simple and teaches us how to potentially get around it,” was the message Sawyers wanted to convey about the mechanisms he’d uncovered. Instead, the headline in the June 22, 2001, issue of the paper read, “Gleevec Shows a Weakness in Fighting Advanced Cancer.” Sawyers still sighs about it years later, now from his office at Memorial Sloan-Kettering Cancer Center, where his research focus has shifted to prostate cancer and other solid tumors.
These mechanisms explained acquired resistance, the cessation of a previously encouraging response. But they didn’t explain the patients who did not respond to Gleevec at all. Those patients didn’t have excessive amounts of bcr/abl, and they didn’t have the altered amino acid. Plus, these “upfront resistance” patients often had changes in their blood cell counts after taking the drug; they just never had any decrease in the number of cells containing the Philadelphia chromosome. And it was far more common to see upfront resistance among patients with more advanced disease; 48 percent of blast crisis patients did not respond to the drug, not even with a noteworthy change in blood counts. As Sawyers and others were seeking the hidden trick that some cases of CML seemed to be playing on the drug, others were trying to find routes around it. In particular, patients who had not had a cytogenetic response after being on the drug for a year were being offered higher doses, 800 milligrams as opposed to the usual 400 milligrams. That approach was sometimes sufficient to overcome resistance, but not always.
By sequencing DNA from the cells of patients resistant to Gleevec, the underlying mechanism gradually came to light. Both upfront and acquired resistance, Sawyers could see, were caused by the same phenomenon that led to CML in the first place: genetic mutations. “As we looked deeper, and as others looked deeper, it became clear that there were many different mutations that could cause resistance,” said Sawyers. By 2002, Druker and other researchers who’d been investigating the problem at OHSU had found several dozen mutations occurring in addition to the original bcr/abl abnormality. Sometimes, patients with upfront resistance had mutations that were not present in patients who responded well to Gleevec. Patients who relapsed appeared to have acquired mutations during the course of their care. Genetic alterations that had not existed at the outset were present a few months into treatment. And, Sawyers showed, when certain mutations were induced in samples from CML patients, the cell-killing effects of the drug stopped. The connection between additional mutations and lack of response was just as direct as the connection between Bcr/Abl and response.
But knowing that mutations were present still didn’t fully explain the problem. Why would a newly accrued mutation prevent kinase inhibition? What was going on in the cell? What did it look like?
Those questions raised one that had until then gone largely unvoiced. Was Gleevec really working as everyone thought it was? The drug had been designed to target the ATP binding site on the Bcr/Abl tyrosine kinase, the exact point where the phosphate adhered to the enzyme. But there was really no way to confirm that this binding was actually happening inside the cell because no one could see it. From the chemists at Ciba-Geigy to the clinical trial investigators, everyone had made educated guesses about what the kinase looked like. No one knew for sure.
Just as Sawyers was puzzling over these questions, a man named John Kuriyan, at the University of California–Berkeley, came up with the answer. The solution had come from x-ray crystallography, a technique that was not advanced enough to be useful when Jürg Zimmermann was sketching out the design of the experimental molecule using pencil and paper. Kuriyan had shot x-rays at crystals and watched how the resulting beams of light bent as they hit the kinase, each angle enabling him to build a picture of the molecule. The technique brought the Bcr/Abl kinase out of the dark room of the cell and into the light of a computer monitor.
That image was immediately indispensable. Kuriyan could show exactly where the molecule bound ATP and exactly how Gleevec interrupted that process. Sawyers regularly flew from Los Angeles to Berkeley, where he would don 3-D glasses and stare at Kuriyan’s computer while he manipulated the image of the kinase, rotating it to reveal its every angle. For Sawyers, it was like traveling to the dark side of the moon and realizing that it held the secrets of the universe.
The images took him right back to where Zimmermann had started when he crafted the anti-kinase molecule: the shape of the enzyme. Now, armed with this vibrant image, the kinase looking like an enlarged kidney bean with swirling ribbons delineating its various atoms and chemical bonds, Sawyers could see why additional genetic mutations stopped Gleevec from working. The mutations changed the shape of the kinase. When that happened, the fit that Gleevec had to the ATP binding site was no lo
nger snug. Bcr/Abl could escape its grip and resume its haywire activity.
That explained the resistance. Cancer cells accrue mutations over time, and that evolution gradually changes the structure of the cells. As a result, a component that starts out just a bit different from its corresponding part in a normal cell eventually becomes totally foreign. Sawyers, Druker, and others had also figured out that some CML patients have other mutations present from the start that prevent them from ever responding to Gleevec.
Sawyers was thrilled to understand resistance, but he knew that the next important question now had to be tackled. Could the science behind resistance lead to better drugs? The success of the first tyrosine kinase had depended entirely on the fact that it blocked Bcr/Abl and only Bcr/Abl. The shape of the kinase and the medicine were thoroughly specific, which is why the treatment worked. How could a drug target a kinase that was pulled out of shape by new mutations? What were the chances of creating a chemical with enough wiggle room to block an even-more-mutated kinase?
36
_______
THE FIRST FIVE YEARS
Following its first approval, a strident pace of expansion brought Gleevec to an ever-widening pool of patients. The first approval, in May 2001, was for patients with CML who’d not responded to interferon or were ineligible for a bone marrow transplant. Novartis had opened numerous clinical trials after the first phase I study had revealed the potential power of the drug. As those subsequent studies proved successful, the drug tumbled forward with more and more FDA and international approvals for the treatment of other CML populations. As other malignancies fell prey to the new drug, approvals expanded beyond leukemia.
In November 2001, Gleevec was approved in the European Union and Japan for adult CML patients who’d not responded to interferon therapy or who were in the accelerated or blast crisis stages of the disease. In February 2002, Gleevec was approved for the treatment of patients with GIST who were not eligible for surgery or in whom the cancer had metastasized following a previous other treatment.
In May 2002, the results of the phase III IRIS study—in which newly diagnosed patients were randomized to receive either Gleevec or interferon plus ara-C as their first treatment—were broadcast for the first time. In responses, survival time, and side effects, Gleevec was consistently superior to interferon. Halfway through the trial, a group of statisticians conducting an interim review of the data insisted that all the patients who’d been randomized to interferon be switched to Gleevec; the advantage was already so unequivocally clear that it would have been unethical to continue giving patients the older medication. The subsequent publication in the New England Journal of Medicine in 2003 showed that after eighteen months, 75 percent of newly diagnosed patients had reached a complete cytogenetic response. The conditional approval of Gleevec for CML would soon become a standard approval. The drug would be widely available for newly diagnosed patients, doing away with the need to first try interferon or undergo a bone marrow transplant.
In December 2002, the drug was approved in Europe for children with CML who were not candidates for a bone marrow transplant. In May 2003, the FDA approved it for children in the early stages of the disease who either had a recurrence after a transplant or for whom interferon had stopped working. In 2006, Gleevec was approved for the treatment of acute lymphoblastic leukemia in cases where the leukemia is positive for the Philadelphia chromosome; a rare skin tumor called dermatofibrosarcoma protuberans; a group of precancerous blood conditions called myelodysplastic syndrome; aggressive systemic mastocytosis, a rare condition that strikes the connective tissue; and two linked blood conditions known as hypereosinophilic syndrome/chronic eosinophilic leukemia. In 2008, the GIST indication was expanded to include postsurgical treatment.
Eventually, Gleevec would be approved in 110 countries. The FDA approvals would span six different diseases. In addition to the growing roster of approvals (all under the name Glivec outside the United States), the continued health of CML patients taking the drug also escalated sales. CML became a peculiar phenomenon: It was the only cancer with an increasing number of survivors. As the population of people with CML continued to grow—more people diagnosed each year added to the number of patients who were no longer dying from the disease—so did sales of Gleevec. In 2006, the sale of Gleevec increased by 17 percent. By 2007, cancers other than CML accounted for 30 percent of the sales. In 2007, annual sales of Gleevec totaled more than $2.5 billion.
Novartis forged ahead into other rare-disease territory. In 2007, a Forbes article titled “Big Bucks,” documenting the strategy, reported that the company had drugs in the pipeline for the treatment of at least six conditions diagnosed in fewer than 200,000 people in the United States per year, the criteria for a disease to be considered orphan. In 2009, Ilaris, the company’s drug for a rare autoimmune disease driven by a single genetic mutation, was approved, with annual sales totaling about $26 million. Gleevec had proved the principle of targeting the underlying genetic cause of a disease, and now it had proved another: Rare diseases are a profitable market. (Novartis’s effort to get Ilaris approved for the treatment of gout, which would have turned the drug into a true blockbuster—defined as annual sales of $1 billion or more—was rejected by the FDA in 2011.)
WITH EVERY PASSING year after the initial approval in 2001, the question of whether patients’ responses would last persisted. The trials may have been completed, but the experiment continued. There was no track record, no backlog of evidence, no years of experience. At that point, the goal was to reach five years.
At the same time, a major leap had been made in measuring responses. A diagnostic test known as polymerase chain reaction had become widely available since the early clinical trials. The test enabled a deeply penetrating view into blood cells taken from CML patients. It was used to find how many cells contained the bcr/abl gene and was powerful enough to spot a single abnormal cell among 100,000 normal ones. Results were calculated in terms of log reductions, with each one signifying ten times fewer cells with the bcr/abl gene. Now, in addition to hematologic and cytogenetic, another response was added to the mix: molecular. If the polymerase chain reaction test showed a 3-log reduction over time, then the patient was considered to have a major molecular response. It was as close as a CML patient could come to being cured.
By 2006, the phase III IRIS trial had been ongoing for five years. Druker and the other investigators assembled everything they knew about the patients who’d been taking the drug all that time. Their report, published in the New England Journal of Medicine, prompted yet another collective sigh of relief. Of the 350 patients who’d had a complete cytogenetic response within a year of starting treatment, the disease had stopped progressing in nearly all. The IRIS trial had included polymerase chain reaction testing, and the results were startling. Every single patient who had a major molecular response within eighteen months of starting treatment was still alive at the five-year mark. Among the 382 patients who’d been taking Gleevec since the study opened in 2001, 340 were still alive in 2006. “It is currently recommended that imatinib therapy be continued indefinitely,” the authors concluded.
37
_______
THE SECOND GENERATION
Now that Sawyers knew that drug resistance was caused by changes in the shape of the binding site, he envisioned a solution. A drug that still adhered to the kinase but with a structure that allowed for those changes might still block the enzyme from binding ATP. “If you could find a drug that . . . was less demanding, more promiscuous, then it should work,” Sawyers said. As he traveled around to various cancer meetings, he started testing the waters of this new concept, to see if it grabbed anyone’s interest. Soon after one of those meetings, he got a call from a scientist at Bristol-Myers Squibb. When the scientist heard Sawyers speak about the problem, his mind had turned to a compound his lab created.
Called dasatinib, the compound had been created to inhibit T cells in the immune system. CML wasn’t
a target of investigation at the industry lab, but when the drug was screened against a group of kinases, the scientists there noticed that it blocked Abl. The company sent Sawyers a sample of the compound, and he tested it in a collection of cell samples from patients who were resistant to Gleevec. In nearly every sample, dasatinib blocked the kinase. With Gleevec having already charted the course for tyrosine kinase inhibition, the drug went quickly into clinical trials, where its anti-CML activity was confirmed.
As the investigation of dasatinib proceeded, exact patterns of Gleevec resistance emerged. Patients who started taking Gleevec very soon after being diagnosed with CML tended to continue doing well; the chance of relapse among this population was about 4 percent for the first five or six years. But many patients who started the drug several years after the diagnosis tended to stop responding to it at some point. The extra time had left a window for additional genetic mutations. The slow accrual of abnormalities went hand in hand with resistance.
Bristol-Myers Squibb may have been the first company to bring a second-generation tyrosine kinase inhibitor to clinical trials, but Novartis was not far behind. Ever since Gleevec had been approved, Alex Matter had been wondering if the chemists could do even better. Could the compound be made stronger? If so, would that improve patients’ outcomes even more? Matter, Zimmermann, and the other researchers at Novartis were eager to find out.
Diving into a second-generation tyrosine kinase inhibitor stirred up yet another fight for the ever-combustible Matter. “You’re going to destroy our franchise,” the marketing team warned him. If word got out that Novartis was working on another compound, then patients and the public at large would infer that Gleevec was not the best possible medication. Matter insisted that if Novartis didn’t pursue the next such inhibitor, then the competition would.
The Philadelphia Chromosome Page 27