A little over a year ago, DDNews published an article on the challenges facing new oncology drugs as they moved through the clinical trial process (‘Good enough’ is no longer good enough; March 2013). The article suggested that the bar against which new therapeutics were being measured was constantly moving and that it was insufficient to match standard of care (SoC), but rather the expectation was significant improvement beyond SoC.
With this as the backdrop, DDNews wanted to look further into this challenge by examining the clinical trial process itself and, more specifically, asking whether the gold standard endpoint of oncology clinical trials—overall survival (OS)—was a bar set too high, particularly for drugs being tested in advanced, late-stage cancer populations.
Where the conversations were expected to deal with the financial and practical burdens of achieving an OS benefit, they quickly evolved into moral and ethical conversations about conflicts between what was best for the clinical trial and what was best for the patient as an individual.
Is OS improvement over-reaching?
“In the days of radiation and chemo, both of those treatments directly affected the tumor,” explains Rob Petit, chief scientific officer of Advaxis. “Radiation beam hits the tumor and the tumor shrinks. Chemotherapy gets into the tumor and it kills tumor cells. It was pretty straightforward that we were directly treating a tumor.”
And because patients who saw tumor shrinkage typically lived longer, he suggests, this was seen as a surrogate endpoint for OS. Thus, while maintaining OS as the ultimate goal of any treatment, companies turned their focus to this and other surrogate endpoints.
“The three endpoints that people typically look at in clinical studies are response or tumor shrinkage, progression-free survival (PFS) and OS,” says Oncolytics CEO Brad Thompson. “PFS was introduced to allow people to run clinical studies quicker.”
“If you’re looking at just tumor burden reduction, you can actually do that very quickly,” he adds. “You can demonstrate literally in weeks whether you’re getting an additional tumor burden effect with a new therapy. I think there’s been a large undervaluation of that as an endpoint.”
He is quick to caution, as history has proved, that response rate (RR) “is a different endpoint than OS and people shouldn’t try to equate the two.”
As Chris Smyth, managing director of Novella Clinical’s Europe/Asia operations, explains, the FDA created its accelerated approval program based on the assumption that these surrogates lead to OS improvements.
“Thus, if you can demonstrate improvement in one of those surrogates for OS, then you will get accelerated approval,” he says, but quickly adds, “but you also have an obligation then to go on with your study and demonstrate OS. If you fail to do that, you then lose your approval.
“That’s what we saw with Avastin obviously three years ago in breast cancer. There was a similar thing with Avastin in glioblastoma just earlier this year, where they were seeing great PFS but that didn’t translate to OS.”
Recognizing the frequent disconnect between improvements in OS and its surrogates, oncology groups around the world continue to maintain OS improvement as the gold standard, no matter what the costs.
Cornerstone Pharmaceuticals’ vice president of regulatory and clinical affairs, King Lee, highlights the two main challenges of an OS endpoint.
“Using OS, you have to track the patient for longer periods of time, thereby making your trial longer,” he explains. “And also sample size has to be higher because it is not something you can measure easily, like blood pressure. With 100 mm Hg vs 110 mm, you can have some kind of statistics that show a difference between the two numbers.”
Thompson calculates the numbers of patients involved to highlight the challenges.
“If you could double OS, you could show that in a couple of hundred patients. If you want to do a 10-percent improvement, you’re talking thousands of patients to do it to the statistical level that everybody would prefer to see. How do you run a study like that?”
“Even big pharmas have trouble enrolling 5,000 patients in a clinical study,” he adds. “In glioblastoma, for example, there may only be 10,000 new cases a year in the United States, so you’re talking about running half of those patients in the entire country in a clinical study. It’s just not practical.”
“One of our tricks is to use patients at later stage of disease. They don’t live as long and so improvements in OS can be demonstrated in a relatively short period of time,” says Lee.
Smyth, however, suggests that this presents its own problems.
“When you get to the later stages of cancer, the life the patient has is shorter and therefore the ability to demonstrate a difference between two treatments can be more difficult,” he explains. “It’s almost like a big tanker floating in the seas; once the momentum starts, it is very hard to reverse that.”
Lee also looks for situations where patients have limited existing treatment options, suggesting that the standard for OS improvement is likely to be lower.
“If the current therapy only can provide prolonged survival for a few months, then the hurdle we have to climb over to prove that our drug is better than the current therapy would also be lower.”
Making things more complex, because patients need to die before OS can be calculated, they typically receive several treatments over the course of their remaining life span.
This then complicates the identification of any subsequent OS benefit, Lee adds, because you can never be certain which therapeutic the patient received extended survival.
If Thompson had his way, such clinical trials would focus purely on RR as a primary endpoint and OS as a follow-on endpoint. He sees PFS purely as an often-inaccurate surrogate for OS and therefore as not worth pursuing.
“That would allow us to run studies with smaller numbers of patients because honestly, the number of patients needed to show tumor burden changes in a randomized study is much smaller,” he says. “We would have products that wouldn’t be making OS claims initially, but instead you’d be saying this product kills more tumor than the SoC when added to SoC.”
Not just numbers
Whereas the goal remains showing statistical and significant improvement in OS, however, it is important to remember that those statistics are based on something.
“What I think people tend to forget is what’s important to the patient,” says Thompson. “One of the two things is getting rid of the tumor now or demonstrating that there is less tumor.”
“That’s why we have such high satisfaction rating with surgery and radiation,” he suggests. “Surgery is the ultimate in quick response. When you operate on somebody, they go in with a tumor and they come out with the tumor gone.”
Thompson says patients are also interested in living longer, but not at any cost.
“People have heard for the last 50 years stories of patients forgoing therapy that might have extended their lives because the therapy was worse than the lifespan extension,” he says. “Who wants to get second-degree radiation burns on their chest?”
Lee agrees, raising the conversation to a moral level.
“Improvement in survival without any improvement in quality of life (QoL) may not be an ethically meaningful therapy for patients with end-stage disease,” he warns. “They don’t live very long anyway, and then you prolong their survival for a few months but the QoL for the patient is not good.”
“In a clinical trial, you don’t know for sure that your therapy is effective or not, and the patient has a very short period of time for you to test the drug,” he continues. “If the drug is not effective, then you’re basically using up the remaining time of the patient.
“If there are no signs of efficacy, the patient should have the option of opting out of the clinical trial so that he or she still has some time to participate in another clinical trial involving a different investigational therapy.”
The patient as a person
As suggested by the recent unfortunate circumstances of Adrienne Lotton, a patient who found herself in the placebo arm of a trial for a new drug targeting late-stage melanoma (Human compassion vs. scientific rigor; March 2014 DDNews), as companies and oncologists continue to push the envelope in the development of new therapeutics, there is an increasing push back from patients who only have eyes for their own situation and want to be treated as human beings and not data points.
“The fundamental culture conflict or disconnect between the patient and the FDA is that the FDA makes decisions for the public based on results across an entire population, whereas for an individual who has a terminal disease, the risk of the entire population is sort of insignificant compared to mine as a member of that population,” says Petit. The fundamental question on each patient’s lips is: “Will the therapy work for me?”
While acknowledging the existence of some patients who take an altruistic approach to their participation in a clinical trial, the majority, Petit suggests, see the situation as him or her vs. the cancer.
Putting it harshly, he says, “The rest of the population doesn’t matter if you’re that one person in 100 that it doesn’t work for.”
Whereas such personal consideration might have once been seen as emotional noise in an analytical process, opinions are changing, according to Smyth.
“In the past, oncology was about just trying to dose a drug as high as you could get so that it kills the cancer but not the patient,” he says. “I think now there is much more consideration for QoL of the patient.”
With this in mind, he suggests there are four stakeholders in the question of oncology clinical trials.
“One is the patient,” he elaborates. “They’re looking for a cure for themselves or at least relief of symptoms. Second is the clinician. They’re looking for a treatment that they can give to their patients that will be effective.”
“The regulatory agencies are looking for a demonstration of both safety and efficacy, and that efficacy measure is often OS,” he continues.
“The fourth group I lump together as the payors. What they’re looking for is value for money and effectiveness,” Smyth concludes.
Thompson takes it back to the discussion of what the patient actually wants in the process.
“If all you do is short-term reduce the tumor burden in a patient and you do it in a safe manner without a lot of side effects, that should be a registration endpoint,” he says. “We should be able to design clinical studies that accommodate patient-centric decisions being made that might degrade the quality of your clinical study and just be able to statistically account for that in an upfront fashion.”
“We need to manage the potential clash that happens when a very conservative regulatory agency meets an evolving field,” adds Robert Pierce, chief medical officer of OncoSec Medical Inc., who offers an example of a numbers versus patients quandary.
“I was engaged in designing a clinical trial, and I was fighting with my own biostats people because they were maintaining we needed to randomize and we couldn’t have crossover to have the chance that it could be a registrational trial,” he recounts. “All of the treating clinicians and myself were strongly of the opinion that it was unethical to put those patients on chemotherapy because we know it not only doesn’t work, but it has a significant morbidity and mortality. We just got stuck in a not-very-thoughtful loop.”
Echoing those sentiments, Thompson talks about the growing activism by oncologists.
“You see it in every clinical study that you run,” he claims. “There is always a percentage of patients that are enrolled on a clinical study that first, may not even have the cancer you’re interested in; and second, are kept on or taken off a study for reasons that are patient-centric, not trial-centric.”
Such actions may be good for the patients, he says, but they degrade the company’s ability to use the clinical study.
“People complain because from a clinical trial perspective, it’s difficult and expensive and can really muck up a study, and that really inhibits the ability of a product to get registered and get used by patients generally,” he continues. “On the other hand, these doctors are just caring about their patients. You should be able to introduce a mechanism in a clinical study that accounts for that.”
But oncologists aren’t the only ones starting to play a more proactive role in patient treatment. Patients too are starting to stand up and gang up via social media.
“You see it with the blogosphere,” Pierce says. “Patients are talking, and there is an emerging consciousness within these patient populations of what’s working and what’s not.”
The PD-1 program at Merck, on which Pierce worked before moving to OncoSec, capitalized on this, “Not in a conscious way, but in that all of the melanoma patients were blogging and those studies enrolled faster than anyone has ever seen anything happen because it caught fire out there in the consciousness of the patients,” he recalls.
It can’t just be about the numbers, for Lee.
“Let’s say, based on a clinical trial, a drug has been shown to be effective in 20 percent of the patients,” he gives an example. “We don’t individualize for that patient, we just give the same drug to all patients with the same disease because there is a 20 percent chance that it may work.”
“That may be good from the marketing point of view because you sell a lot of drug, including drugs to the 80 percent of patients in whom the drug may not work, but it is not very ethical,” he argues.
This brings the conversation around to biomarkers.
Biomarker boom and bust
“The biomarker aspect is very important,” enthuses Petit. “First of all, if you can identify in advance a subset of patients who express a marker that’s been associated with successful treatment, or even if you can identify the patients who would not benefit from that treatment, you can avoid treating people who would have no likelihood of benefiting.”
From a clinical trial perspective, he suggests, you can select your patients up front and sweeten the pot, so you can have remarkable clinical trial results early on with a smaller number of patients.
“I am a huge fan of biomarkers analysis, and now we have the technology to do it in a timely and broad-based fashion, which we didn’t have just a few years ago,” adds Thompson.
“As an example, if you’d asked me this question five years ago, we’d have been able to look at just the Ras marker itself. That would have cost us $3,000 to $4,000 for that one marker alone, and it might have taken us a couple of months to get the data back,” he recalls.
Now, he says, the tissue samples sizes are very small; 50 ng is enough to do a sample analysis that takes a fraction of the time.
“Two weeks later you have your profile,” which he says is acceptable. “And you can get 300 to 400 gene markers done for about the same price as the one marker a few years ago.”
Pierce is more cautious about the practicality of biomarker diagnostics.
“There is a lot of promise, but there is also a lot of pain and suffering and potentially increased costs of drug development,” he warns.
“We’re currently designing our pivotal registrational trial and have been kicking around the pros and cons of building in a companion diagnostic because we have a hypothesis about which patients will most benefit,” he says. “But the extraordinary cost in time, energy and pure money to co-develop what amounts to almost another drug, the assay can cost almost as much as developing a drug.”
For Pierce, the big issue is that no laboratory test is perfect, that there is always an intrinsic false-negative and false-positive rate to consider.
“The dilemma that treating clinicians are confronted with is what do you do when you have a 5-percent false-negative rate and those patients who would be called negative and not given the drug still have a finite and reasonable chance at having incredible responses of such long duration that you could almost begin to use the word ‘cure’,” he explains.
A lot of the FDA-approved companion diagnostics, he suggests, would only lock out the treating physician from access to those drugs if their patient doesn’t score right for the biomarker, even though they may still respond.
“It is a very convoluted problem involving reimbursement, approvals and physician and patient freedom,” he decries.
Thompson understands the challenges inherent in companion diagnostics.
“Literally, getting a new diagnostic that doesn’t exist developed can be as time-consuming, if maybe not as expensive, as running clinical trials and getting the product approved,” he says. “It is a very different skill set too. You’re looking at a company that has therapeutic expertise being told you have to develop a diagnostic.”
“I actually posed a question to a very high-ranking person in a big pharma company: Well, why don’t you acquire a biomarker development company and whenever you put something into clinical program, you start working on the biomarker?” Petit recalls. “And the answer was: Because there are so many types of biomarkers that require completely different platforms and technologies, there is no one company that has expertise in everything. And you never know exactly how it’s going to work out.”
And, he says, the polymorphic nature of cancer further adds to the complexity.
“We like to think of cancer as this disease or that disease—breast cancer or prostate cancer—but in reality it is a population of diseases,” he says.
He highlights the example of breast cancer, where biomarkers include ER+, PR+, HER2-overexpressing, etc., but he warns there is a whole other level of subdivision that goes down almost to the gene level.
“There is a lot of interest right now in the future of cancer treatment being based on approaching large amounts of data that pertain to a particular patient’s own genetic makeup and the particular aspects of that patient’s tumor,” he says, before piling on the problem with tumor heterogeneity.
“Tumors by definition mutate a lot,” he adds. “So, not only are they all sort of different, but they’re changing constantly. It’s kind of like trying to hit a moving target from a running horse.”
With this variability in mind, tumor profiling has become increasingly important in finding the optimal treatment regimens and combinations that will feature your experimental therapy in its best light.
“Until recently, we simply did not have the tools to do a comprehensive scan for the pathways affecting this patient,” says Padma Sundar, Affymetrix OncoScan product manager. “Now, at least for mutations, we are able to do next-generation sequencing and identify all of the pathways that are being impacted and target them with combination therapies. And, with tools like OncoScan, we can also look at problems with copy number [and gene overexpression].”
The advantage to pharma companies is that if they are developing a therapeutic targeting one pathway, tumor profiling could help predict how best to combine the experimental treatment with other known or experimental compounds targeting the other pathways.
For Cornerstone, however, biomarker analysis is less about patient stratification and more about efficacy, circling the discussion back to surrogate markers. Rather than target surface markers on cancer cells, Cornerstone has focused its efforts on defective glucose metabolism pathways unique to cancer cells.
One could measure the biomarker before giving CPI-613 and after, say, one cycle of treatment, he explains. Then, theoretically, “If the biomarker changes in the direction that we anticipate it would, then we conclude that most likely the patient will respond.”
At the end of the day, OS remains the gold standard and perhaps should be the final target of any effort to develop new oncology therapeutics, but it equally seems obvious that this decision can’t be made in a vacuum and should consider the needs and goals of all parties, not least those of the patients.
“It is possible, it would be relatively easy to do and I think we should just acknowledge patient-centric endpoints as being important,” concludes Thompson.
The special case of immunotherapy
It is difficult enough for a new chemotherapy to prove its worth on top of standard of care, but the growing pipeline of immunotherapeutic approaches by companies such as Advaxis, OncoSec and Oncolytics are highlighting unique challenges to overcoming the therapeutic bar.
“I like to think sometimes of cancer as like a roadmap,” says Chris Smyth, managing director of Novella Clinical’s Europe/Asia business. “There is more than one way to get from A to B. If you put up a roadblock on one road, the cancer is able to select a different path and continue on its growth. So either earlier in the route, you can block and limit the options it has, or later block it at as many different paths as possible.”
“In a way, immunotherapeutics are some of the most powerful options that we’ve seen recently because the inherent mechanism is it can just wipe out the cancer,” he adds. “So, it’s not reliant on the very specific pathways to block.”
“Immunotherapy doesn’t often directly affect the tumor at all,” explains Advaxis Chief Scientific Officer Rob Petit. “Rather, it is an indirect treatment in that you’re actually affecting the immune system inside the patient and stimulating it to fight the cancer.”
This fact raises certain implications on the ability of immunotherapy to help the patient fight his or her cancer.
First is the question of whether the patient’s immune system is healthy enough to be manipulated to fight the cancer, he explains. In particular, he highlights the patient who is already receiving cytotoxic chemotherapy, which is notorious for knocking down the white cell count needed to fight the cancer.
“Number two: It can take a lot of time for the immune system to get built up to fight the cancer, whereas with chemotherapy or radiation, the treatment works right away and you can watch the tumor shrink very quickly,” he adds. For patients in the later stages of disease, that time may be fleeting.
Thirdly, he suggests, therapy may at first appear counterproductive.
“When immunotherapy is working on a cancer, the total mass of the tumor will increase sometimes quite significantly as all these lymphocytes infiltrate and infuse into the site of the tumor,” he says. “So, if you’re just going by the measurement of that mixture of cells—part tumor cells, part lymphocytes—it looks bigger.”
And because tumors can get bigger before they get smaller, therapy may show tumors that were already there but were too small to be clinically detected. Once they get inflamed and infiltrated by lymphocytes, he says, they show up and can look like a new lesion.
OncoSec Chief Medical Officer Robert Pierce foresees a future in which there will be an inversion of what is standard of care and what people go on as first-line therapies.
Such an inversion, however, would mean introducing immunotherapies much earlier into the patient treatment regimen, in places that are now occupied by well-characterized, often-used chemotherapeutics, surgery and radiation. And as Petit suggests, this may not be easy to accomplish.
“Medicine is very reluctant to get rid of something that works pretty well—that is the best they have—and so the challenge is how do you establish that this immunotherapy really does have the activity that you need and then develop sufficient rationale for doctors to begin to start to think about where it starts to make sense to use the immunotherapy earlier in disease, on top of or in lieu of the existing treatment,” he opines.
“At Advaxis, we’ve really tried to approach this challenge in two ways,” he continues. “One is to try to find settings where we can establish a foothold of activity based on the unique mechanism of action. Sometimes you do end up toward the end stage of disease with patients who have limited treatment options and you try to be careful about your patient selection process so you can make sure you get the healthiest of that group just to establish some activity in the disease.
“We’ve also established a completely different side-effect profile that may make it more compatible when used earlier in disease.”
It’s essentially about hedging your bets and about being open to all opportunities, or at least as many as you can afford. Advaxis tends to try to do both late-stage and early-stage treatment as soon as possible so it can establish where its treatments will fit in within the spectrum of how the patient is managed.
The hope is an early bird that is not only special for the company, but for patients, oncologists and regulatory bodies as well.