Genentech’s Souped-Up Herceptin: The Odyssey Toward a More Powerful Breast Cancer Drug
Scientists have been dreaming for decades about drugs that go after cancer cells like heat-seeking missiles. The idea is to destroy the enemy and spare other cells from the collateral damage that so many cancer drugs cause. Now after more than 10 years of sustained effort by one of largest teams ever assembled at Genentech, and some gut-wrenching ups and downs, that vision is on the verge of turning into reality.
This story began when Genentech, the South San Francisco biotech pioneer now owned by Roche, made history back in September 1998. That’s when it got the green light from the FDA to sell a genetically engineered antibody designed to seek out a mutated protein found on malignant cells in about one-fourth of women with breast cancer. The drug, trastuzumab (marketed as Herceptin), has been proven over the years to work even better than Genentech first imagined it would in women with early stages of cancer. The therapy is now one of the biggest selling biotech products in history, generating more than $5 billion in annual worldwide sales.
Yet from the moment of the FDA’s approval, a small group of scientists in-house were thinking even bigger. They suspected that their engineered antibody, even though it could hit a specific target on cells, would fail to land a real knockout punch against tumors. The scientists had shown that drugs of trastuzumab’s generation could interrupt signals cancer cells need to grow, and they later realized that the drugs could also sometimes provoke an immune system reaction against the tumor. But they weren’t a cure.
So scientists at Genentech, including Mark Sliwkowski, wondered if a targeted antibody would work even better if they could attach potent little toxins to it, making it into a “smart bomb” or an “empowered antibody.” Nobody has turned this concept into a commercial success after 30 years of trying, but Genentech and its partner in Waltham, MA, ImmunoGen (NASDAQ: IMGN), are betting they have finally been able to pull off this daunting scientific trick.
Evidence to support the claim arrived last December at the San Antonio Breast Cancer Symposium. A next-generation form of Herceptin, jointly developed by Genentech and ImmunoGen, called T-DM1, partially or completely shrank tumors in about one-third of 110 patients who were extremely sick, whose cancer was still worsening after they had endured an average of seven prior rounds of therapy. The most common severe adverse event was a depletion of blood platelet cells, which was found in 5.5 percent of patients, researchers said. The balance of safety and effectiveness was so convincing that Genentech plans to seek FDA approval before the end of this year to start selling T-DM1 in the U.S.
Important as it may be for breast cancer patients, the finding has re-energized the once moribund field of antibody-drug combinations, often called “antibody drug conjugates.” Genentech is so bullish on the concept that it now has 50 such drug candidates moving through its internal R&D pipeline, following the lead of T-DM1. This idea of “bombing” the tumor, once written off as a pipe dream by most people inside Genentech and academic research, has morphed into one of the company’s top strategies for fighting cancer, according to Sliwkowski, a longtime champion of the concept.
“We take this far more seriously than we did previously,” Sliwkowski says. “We are investing a lot of money and resources into this. That should answer your question.”
One of Genentech’s collaborators on antibody-drug conjugates, Seattle Genetics, says T-DM1 and another drug it is developing for Hodgkin’s disease called SGN-35, are at the forefront of a new wave of cancer therapy. “These are both exciting programs that have the potential to help thousands of cancer patients,” says Seattle Genetics CEO Clay Siegall. “I believe they represent only the beginning of what is possible with empowered antibody-based therapies.”
The FDA approval of Herceptin was pivotal, but T-DM1 began its journey inside Genentech as far back as late 1997, Sliwkowski says. Genentech in those days wasn’t really a cancer drug company. It was better known for its work in cardiovascular disease and with genetically engineered growth hormones. It had just gotten its first taste of success in cancer with rituximab (Rituxan), a hit drug for lymphoma.
Scientists at Genentech got fired up about another first-generation antibody for cancer, the original Herceptin, when the pivotal Phase III results arrived in late 1997 that proved the drug was likely good enough to win FDA approval. The thought of creating a more potent second-generation drug was floated internally, but didn’t get much traction right away, Sliwkowski says. That’s because the company was so preoccupied with getting its original Herceptin application filed with the FDA, passing muster at an expert advisory panel meeting, and lining up proper diagnostics to make sure the drug got to the right patients.
“We were a pretty lean and mean team, and didn’t have a lot of extra time to think about second generation drugs and all that kind of stuff,” Sliwkowski says.
Sliwkowski, who joined the company in 1991 and did a lot of work on the original Herceptin, started thinking more seriously about a more potent form of the drug the day after FDA approval in September 1998. “Not that the date means anything to me,” he joked.
There was enormous skepticism inside the company, and outside, at the time. Many researchers had tried to create this kind of “magic bullet,” virtually from the moment the Nobel Prize-winning team of Georges Kohler and Cesar Milstein created the first targeted antibodies in 1975. All subsequent efforts to make antibody drugs loaded with toxins had failed. Often, it was because nobody knew how to properly link a toxin to the antibody. The toxin would often break off in the bloodstream before it ever got to the tumor, making the drug less potent and possibly more toxic than it should have been than if it got to the target and was properly metabolized. “There was a lot of baggage,” Sliwkowski says.
Still, Genentech was in the mood to think big. It had proven its critics wrong about Herceptin. Scientists were skeptical about whether that treatment, like most antibodies before it, would provoke a human immune system reaction against the antibody itself that would render the product useless. Others wondered whether the target it was aimed at (the HER2 protein) was truly more abundant on cancer cells than healthy ones. Overcoming those doubts made a few people within Genentech think they could solve the problems again with antibody-drug conjugates, Sliwkowski says.
“The pendulum of interest in oncology within Genentech swung in a new direction. We were this oncology company now. We were serious.” Sliwkowski says.
They needed to be. The scientists at Genentech figured there were four major variables that needed to be resolved if an antibody-drug combination were ever to work.
First, they needed to aim the antibody at a truly valid target that is involved in an important cancer cell process, is found abundantly in an accessible place on the surface of cancer cells, and has minimal presence on healthy cells. Few targets like this exist in the human body, but Genentech had shown this to be the case for HER2, the target of Herceptin. “We knew HER2 was special, and as validated a target as you could get. We said if we can’t get it to work on HER2, then it ain’t going to work,” Sliwkowski says.
Then came the antibody itself. The good news was that because Genentech had an FDA-approved antibody for breast cancer cells, the manufacturing wing was churning out vats of it by the kilogram, not the usual micrograms that basic researchers need to scrape by with. That meant the scientists had more than enough raw material to run every experiment they could possibly think up, Sliwkowski says. It may sound trivial, he says, but projects in academic labs or small companies often stall because they only have enough resources to produce very small quantities of a drug, forcing them to do pretty limited experiments. Since they had lots of bulk drug, and it easy to link the antibody to all different types of toxins, the scientists went at it, Sliwkowski says. They had the luxury to think systematically, to try hundreds of combinations of antibodies and toxins in the petri dish, review the data, and tinker with one variable at a time until they got what they wanted.
“I didn’t think people were assessing the full potential of these things. I really wanted to turn the crank. And we did,” Sliwkowski says.
The third key variable was with the linker system that was meant to attach the antibody to the toxin. This is the part where Genentech leaned on the expertise of its partners at ImmunoGen, and later, at Seattle Genetics. Getting a linker that was stable in the bloodstream, and get the toxin exactly where it was supposed to go inside the cell, was the key challenge.
The fourth major variable, Sliwkowski says, was figuring out how to deal with the toxin. These molecules, because they are being directed to tumors in small doses, need to be super-potent, i.e. more than 1,000-fold more potent cell-killers than the usual chemotherapy drugs, which circulate throughout a cancer patient’s body. Nobody knew when Genentech first got going on this project whether the antibody, a relatively huge Y-shaped protein in molecular terms, should be loaded with one, two, five, 10 or 20 different toxins to get the ideal tumor-killing punch. They really didn’t where those toxins should be strategically placed on the antibody backbone, or what kind of difference that might make.
But having confidence in the cell target, and the antibody itself, helped scientists narrow down their options from the start. Even so, most of their preconceptions were wrong, Sliwkowski says. Some people thought that if loading more toxins on the antibody would give it more punch. Actually there was a sweet spot in the middle that was most effective. Others, Sliwkowski included, thought that maybe the Herceptin antibody wasn’t really the best vessel because it was so big and bulky, and that an antibody fragment might be a more efficient vehicle for delivering toxin to tumor. Wrong again.
“We did those experiments with fragment. The efficacy was fantastic. Toxicity was horrible,” Sliwkowski says. “It was jaw dropping. You’d injected the rats, and the guys would call you up the next morning, saying ‘they’re all dead.”
Despite the setbacks, Genentech’s senior managers gave this program some vital support early on. When Sliwkowski came asking for significant resources, one key supporter was Richard Scheller, now the executive vice president of research and early development.
Scheller’s role was important, because he had authority to do something pretty unusual in development, Sliwkowski says. Most basic scientists at an early stage of research wouldn’t bother to think about things like enlisting the help of toxicologists, the people who help assess whether a drug is being properly absorbed and metabolized in the body or whether it’s going to be too toxic. People with that skill are often busy working on drugs in later stages of development. But many drugs that look good in a petri dish stumble later in animals or people when they prove to be too toxic. In this case, Scheller made sure Sliwkowski and his colleagues got important early help from their peers in toxicology.
So Sliwkowski felt he had the resources he needed to help screen out the duds early on, and improve the odds of success.
Even with all that support, the worst was yet to come. By 2002, Genentech had taken its lead antibody-drug conjugate into its first clinical trial. At the time, it used what is known as a disulfide linker. This was thought to have the nifty ability to remain stable in the blood, but selectively release its toxic cargo once confronted by certain enzymes in cancer cells. It looked great on the whiteboard, and it showed a clear affect on cancer cells in the lab. But when this candidate with the disulfide linker entered animal testing, it flopped. “We really tripped up on safety,” Sliwkowski says. [[Correction: 1:25 pm, 6/16/10: An earlier version said the drug and disulfide linker failed in a clinical trial.]]
Though the researchers had poured three years and a lot of resources into the first conjugate, it was time to start over. Genentech started to run its usual battery of tests, altering one variable at a time to see how it changed the results, in cancer cells and healthy cells in the lab.
Fast forward three more years, to Valentine’s Day, 2005. Sliwkowski says that’s the day senior management gave the green light for the R&D teams to do the last experiments they needed to seek FDA clearance to begin a new round of clinical trials. This time, they used a “thioether” linker from ImmunoGen. The big difference with this new linker, scientists learned, was that it instead of releasing the cell-killing agent, it formed a covalent bond between the antibody and the toxin that couldn’t be broken.
Using that new linker meant that the whole antibody-drug package could get internalized into the cell, trafficked to the place where it encountered certain enzymes, and then be degraded, the same way enzymes in your stomach break down the steak you ate for dinner. Through this degradation process, a huge antibody is ultimately broken down into amino acid building blocks, still clinging to the potent toxin, which can then do its anti-tumor thing.
By April 2006, the scientists had raced through their preparations to get what was now T-DM1 ready to enter its very first clinical trial. One year later, Genentech’s senior brass started raving about how the company was seeing anti-tumor activity from T-DM1 in this original study of 18 patients, even in some cases at the tiniest doses that were just intended to assess safety, not effectiveness. When I was a reporter at Bloomberg, Ian Krop, a researcher at the Dana-Farber Cancer Institute, told me, “We’re about as excited about this one as we can get this early in the game.” More details emerged a couple months later at the biggest annual cancer meeting of the year, the American Society of Clinical Oncology.
The scientists were still a little nervous, and a little scared of what might happen, given how they’d been rudely surprised by the failure of the earlier souped-up version of Herceptin, Sliwkowski says. He says it was a sobering moment, to realize that actual patients were gutsy enough to take an experimental drug like this, with so many potential risks that could have emerged.
But those days are now past, Sliwkowski says. Even though scientists expect most of their creations to fail, they now expect T-DM1 to succeed.
“This thing works, and it works pretty well. We expect as we move it up the ladder into earlier and earlier breast cancer, it’s going to work even better,” Sliwkowski says.
That’s not to say this program is risk-free. Genentech is planning to seek FDA approval this year based on a relatively thin body of evidence, from a clinical trial that didn’t have a control group. It’s possible the FDA could convene an expert advisory committee to debate whether to wait for data from more rigorous clinical trials.
T-DM1 also takes twice as long to manufacture as a regular version of Herceptin, and is more complicated to piece together from various contract manufacturers around the world, says spokeswoman Krysta Pellegrino. If the drug is approved for sale, and lives up to its billing in early-stage breast cancer, then Genentech will need to invest more in commercial-scale manufacturing.
And what about those 50 other antibody-drug conjugates at earlier stages of R&D? Will they benefit from the lessons learned through the T-DM1 experience, and truly usher in a new paradigm for cancer treatment? Genentech can’t say that for sure. Each target is different, and so is the way the cell internalizes the antibody-drug conjugate, Sliwkowski says. All the usual questions about the right target, right antibody, right linker, and right number and positioning for the toxins all still apply.
Still, this is a technology that could become a reusable “platform,” for the discovery of multiple products, not just a one-hit wonder or a “me-too” incremental advance. Before I left the Genentech South San Francisco campus last week, I had to ask Sliwkowski how much confidence he has in the 50 other drug candidates, and whether they will represent a new wave of cancer treatment.
“I certainly hope so,” Sliwkowski says. “But you’ll have to hold onto that pipeline chart and come back and see us in 10 years.”