SPECIAL REPORT: An aside on side effects
Hair loss. Nausea and vomiting. Weakness. Tiredness. Frailty.
Put the words "cancer," "chemotherapy" or "radiation" in front of anyone in the public, and these are the phrases you are likely to hear in response.
Sure, all drugs carry the risk of side effects, but oncology therapeutics hold a special place in people's psyches. At the same time, when presented with the diagnosis of cancer, few patients are going to say, "thanks, but no thanks."
Side effects are the new normal when it comes to having your cancer treated—part of the risk-benefit equation—and it is accepted.
Cost of doing business?
Even in clinical circles, there has become a certain expectation when treating patients with oncology therapeutics. When presented with a new therapeutic on the market, oncologists can look at a safety profile that includes 75 percent of patients experiencing some degree of neutropenia (for example), and respond that this is within normal parameters for other drugs in the same category and/or for the same condition.
The clinicians aren't necessarily blasé about the side effects. They simply don't feel like there are any alternatives out there.
Unfortunately, this belief itself can have a cost when a new drug enters even the research phase.
One of the major side effects associated with immuotherapeutics targeting the EGFR tyrosine kinase (e.g., gefitinib and erlotinib) is a significant, if not necessarily debilitating, rash. And despite a lack of supporting evidence—at least initially—many oncologists believed the rash was a sign that the drug was working, and sold the benefit to patients.
Over the last handful of years, further studies on these two drugs have shown that patients with non-small-cell lung cancer (NSCLC) who experience the rash do seem to perform better in treatment, so there may have been some merit to this belief.
But then came the introduction of another member of this family, nimotuzumab, an anti-EGFR for which rash was not a significant side effect. Anecdotally, while the absence of rash might have been hailed as a step forward, it met with resistance. Without the rash, some clinicians found it hard to believe that the drug was actually providing benefit, despite clinical evidence in conditions such as glioma.
The expectation of side effects has become so deeply ingrained in this category that the absence of a key side effect was grounds for disbelief by some.
The development of new drugs has typically focused on efficacy first (benefits), safety second (risks). But as was discussed last month (see "Good enough is no longer good enough," ddn March 2013), efforts to raise the efficacy bar have become more difficult in recent years.
So, if the industry can't necessarily raise the benefits side of the equation, can it lower the risks side? Do side effects have to be the new normal? Or is there a way to either prevent or better ameliorate some side effects?
Easy dose it
Some companies have refused to give up completely on the efficacy side of the equation, believing there is still room to improve therapeutic efficacy without necessarily increasing side effects. For Lawson Macartney, CEO of La Jolla, Calif.-based Ambrx, it's about finding the right balance between the two sides, and his company is betting on improving therapeutic potency.
Often when cancer patients experience debilitating or life-threatening side effects, their oncologists will either reduce their dose or stop therapy—"a treatment vacation"—until they can recover. Ambrx is taking a similar approach in the development of its therapeutics, but rather than wait for something to go wrong before lowering a drug dose, the company wants to use minimal dosing ahead of time by making its drugs that much more specific and potent.
"The old cancer medicines were notoriously non-specific," Macartney explains. "They've been basically toxins and have had horrible side effects. Modern agents are much more targeted, which means that you can give less of it. Unfortunately, in the old days, not until you were able to show efficacy and safety were you then able to move molecules like Herceptin or Tykerb upstream in the treatment paradigm, to say, adjunctive therapy or in chemo-naïve patients."
Macartney sees antibody-drug conjugates (ADCs) as the next step in this evolution, smart bombs that use tumor-directed antibodies to deliver a chemotherapeutic payload, the two held together with a chemical linker.
ADCs are not new. What is new is the specificity with which they are constructed.
"In traditional molecules that are currently available, the linker attaches at several different sites on the antibody, which on the plus side means that you potentially get lots of therapy molecules," Macartney says. "On the down side, however, you may get none or you may get so many that it actually interferes with the antibody binding."
You end up with a really heterogeneous population of molecules that can be difficult to characterize and may interfere with specificity and potency.
Rather than rely on the native amino acids for chemical linkage, Ambrx engineers the antibody with a specific non-natural amino acid located optimally for greatest therapeutic potency.
"The goal is to decrease the dosage necessary for efficacy, increase the relative efficacy and by doing this, improve the therapeutic window between where we start to see efficacy and where we start to see side effects," adds Macartney.
Perhaps the poster-child for ADCs currently is Seattle Genetics' Adcetris (brentuximab vedotin), which has been FDA-approved for Hodgkin lymphoma and systemic ALCL, but is actively being tested in a variety of other cancers.
Several other companies are following closely behind, however, including the Phase III programs of Genentech/Roche's trastuzumab emtansine ADC against HER2+ metastatic breast cancer and Pfizer's inotuzumab ozogamicin ADC against relapsed/refractory acute lymphoblastic leukemia (ALL) and relapsed/refractory non-Hodgkin lymphoma.
The true test of Ambrx's lead oncology product,αHER2-ADC, is still to come as it is in preclinical phases, but the company is in active discussions with several pharmaceutical companies to bring it into the clinical domain.
As Macartney explains, to increase the likelihood of success and proof of concept, the obvious first target is breast cancer, but as HER2 is expressed in a number of other tissues, other cancers are viable targets downstream.
Looking for trouble
The ultimate goal, of course, would be to build a therapeutic product that only hit its intended target and offered few side effects from off-target interactions or its own intrinsic toxicity.
Over the last decade or so, computational scientists and informatics companies like ACD/Labs and Simulations Plus have tried to find ways of predicting potential side effects, at least at the small molecule level, by performing comparative analyses of known and experimental molecules, using techniques such as quantitative structure-activity relationship (QSAR), with varying degrees of success.
One potential reason for the variable success of many of these methods for identifying off-target effects by drugs may be the inadvertent separation of a therapeutic's molecular identity and its pharmacological effects.
"In the modern era of molecular biology, we can look at the structures and sequences of protein targets, and perhaps say, I know this is a serotonin type 1 receptor, I know what it looks like and I can see what drugs bind to it," says Michael Keiser, founder and chief operating officer of SeaChange Pharmaceuticals. "Because of that ability, however, we've kind of forgotten how to organize therapeutics by their pharmacological effect. So you might look at the highly related targets for a particular drug, but you might not think to look for a different sort of target."
Keiser and his colleagues at University of California-San Francisco (UCSF) and the Novartis Institute of BioMedical Research (NIBR) have decided to go back to the older idea of looking at the chemical structures of small molecules rather than a protein target and comparing that small molecule to all of the ligands of each of a panel of pharmacologically important targets.
"On the one side, we have a single drug, and on the other side, we have all of the known ligands for a single target, and we compare those two-dimensional structures in a statistical way," he says. "If we do the comparison right, then the shape of the probability distribution is what's called an extreme value distribution."
Their similarity ensemble approach (SEA) allows SeaChange to link potential drugs to an entire panel of molecular targets and predict whether the drug might interact with one or some of those targets. Once they have the target profile, they then ask out of those target profiles, which targets are associated with particular side effects.
As presented in Nature in last year, the group tested SEA on a panel of 656 prescribed drugs against 73 known targets and identified 1241 potential side effect targets, of which 348 were known in NIBR's proprietary data. A further 151 side effects, previously unidentified, were later confirmed through lab testing at NIBR.
Said UCSF professor Brian Shoichet when the results were announced, "The biggest surprise was just how promiscuous the drugs were, with each drug hitting more than 10 percent of the targets, and how often the side effect targets were unrelated to the previously known targets of the drugs. That would have been hard to predict using standard scientific approaches."
Other groups are following a similar path to preemptively identifying potential side effects in drug candidates, including Véronique Stoven and colleagues at the Mines ParisTech Centre for Computational Biology.
The group examined the chemical substructures and known side effects of almost 1,000 drugs to identify statistical links between the two, and then used this information to identify potential side effects for compounds that did not have side-effect data. In one example, they noted a link between the anti-obesity compound rimonabant and key phrases such as "borderline personality disorder" and "PTSD."
In 2008, rimonabant was pulled from the European market for side effects that included severe depression.
Both Stoven and Keiser see strong utility for their methods early in the drug discovery process.
"A toxicologist has a hard position, because more often than not, you don't get to evaluate or even see any of the compounds until they're quite a ways along," Keiser says. In other words, a point at which many choices have been made and you're faced with a decision to kill a program: "One of the reasons for this is that the assays needed to make sure there are no negative effects are expensive," Keiser notes.
"Computationally, instead of only being able to evaluate four compounds, you can contribute earlier on when you are still looking at a thousand candidates, and differentiate them not just on efficacy but also on potential off-targets," he adds. "You aren't replacing safety assays, but you are highlighting what products to test."
But even if you could limit the potential side effects of drugs coming down the discovery and development pipeline, those drugs won't likely hit the market for another decade. What can we do for patients now?
With the advent of personalized medicine came the opportunity to identify what patients would most likely benefit from what drugs. But even here, with all of the diagnostic tools at our disposal and an expanding arsenal of targeted therapeutics, the focus has been on the cancer, rather than the patient.
"Everyone is looking at the genomics of tumors, and they're trying to develop therapies based on that," says Ed Rubenstein, president and CEO of InformGenomics. "One fundamental problem with this approach is that cancer by its nature is a fundamental instability of DNA, so what people may be looking at is the consequence, rather than the cause of cancer."
Furthermore, he stresses, cancer is constantly evolving, responding to Darwinian pressures. When you expose a heterogeneous population of malignant cells to chemotherapy, you kill off the easy ones. The more unstable, harder-to-treat ones continue to evolve.
InformGenomics looks at the problem from another perspective.
"A patient's inherent DNA is relatively stable, aside from epigenetic changes, but there are probably genetic bases for every side effect," Rubenstein explains. "And there is lots of evidence to support that the biologic pathways responsible for these side effects are under genetic control."
Thus, like SeaChange Pharmaceuticals and the group at Mines ParisTech, InformGenomics is applying biological networks with its OnPart platform, but in its case, to side-effect pathology and risks, rather than small-molecule characterization.
Rubenstein is quick to point out, however, that this is not the typical biomarker diagnostics effort typically associated with personalized medicine (see "A companion in your corner," ddn October 2012).
"Genes work together in networks," he says. "If you look at most of the predictive biomarkers or genetic markers, they're not network-based, they're standard-frequency statistically based. Genes talk to each other, there's lots of biologic feedback and interaction, so we are using advanced Bayesian networks to discover those genes that interact with one another. The way we're doing that is using SNPs as markers."
By linking patterns of SNPs to various biological networks, the company found it could create individual networks for individual side effects. Thus, Rubenstein explains, the nausea-vomiting networks had neurotransmitter and neurosignaling terms, which said they were on the right track. Similarly, some of the genes that were mapped by the diarrhea network have been implicated in other inflammatory diarrhea pathologies like Crohn's disease and ulcerative colitis.
For InformGenomics, however, it is not just about statistics. It is also very much about the patient and his or her desires, concerns and fears.
Thus, a key component of OnPart is a validated patient questionnaire—the Preference Assessment Inventory—that quantifies the patient's attitudes about potential side effects.
"It empowers patients because not only does OnPart determine their own genomic risks, but it also captures how they view those side effects, how important they are to them in their day-to-day life," Rubenstein explains, giving the example of a concert pianist or computer coder whose biggest fear is peripheral neuropathy. "That information goes back to the oncologist so they can have that real informed-consent discussion about the risks and benefits of chemotherapy options."
At the ASCO Gastrointestinal Cancers Symposium in January, InformGenomics presented the findings of a study of OnPart in 384 patients who received the modified FOLFOX regimen +/- bevacizumab for a variety of cancers to see how accurately the platform could predict a set of side effects. Their predictive networks had accuracies beyond 90 percent.
In announcing the findings, Lee Schwartzberg, lead investigator and Medical Director of The West Clinic, said, "This allows us to customize our chemotherapy regimens and side effect control interventions for best patient care. These side effects can impair function, create inefficiencies in medical practice and are costly to patients and payers."
InformGenomics is also developing a similar system for patients receiving stem cell transplants. According to Rubenstein, about 40 percent of patients receiving high- dose chemotherapy and stem cell transplant will develop serious oral mucositis, meaning lesions of the mouth, serious infection, prolonged hospitalization and an increased risk of mortality, but it is impossible to predict which patients using clinical factors.
"Prophylactic drugs are available to prevent the oral mucositis, but they are very expensive, and if only four out of 10 patients are going to benefit, it is hard to justify the use of that agent," he says. "If we can predict who is going to get serious oral mucositis, then it allows an already approved drug to be targeted appropriately to the patients who are going to benefit. We see that as another way to take cost out of the system."
So where are we with those prophylactic treatments?
Making things better?
As founding chair of the NCCN Febrile Neutrogena Panel, member of the NCCN Antiemetic Guidelines Panel and member of the MASCC board of directors, Rubenstein is well positioned to understand the armamentarium of side effect treatments.
In the earliest days of oncology, drug toxicity essentially forced oncologists to develop supportive care as patients were almost as likely to die from their treatments as from their cancer, he explains.
"That led to the advent of prophylactic antibiotic therapy for neutropenic fever and the development of pheresis machines and blood component therapy."
"The second wave of cancer supportive therapy was the development of effective antiemetics, which was usually a combination of things that had lots of side effects, but then the 5-HT3 receptor antagonists were developed," he adds. "Around the same time, you got better advances in antibiotics and then the development of blood cell growth factors."
Those advances continue. As mentioned earlier, oral mucositis can be an incredibly debilitating side effect of many chemo and radiation therapies, afflicting almost 500,000 cancer patients each year in the United States alone. At present, there are few therapeutic options available to patients other than pain management, Access Pharmaceuticals' MuGard being a major player.
Last summer, at the MASCC conference, Access presented Phase IV studies of MuGard in patients undergoing chemoradiation therapy for head-and- neck cancer, showing significant reduction in mouth and throat soreness and delay in mucositis onset, as well as reductions in weight loss and use of opioid medications.
Less advanced in development is the defensin mimetic brilacidin from PolyMedix, which is just entering Phase I trials. Part of a new class of antibiotics, brilacidin has shown antibacterial, antibiofilm and anti-inflammatory activity in various animal models, leading to an NCI grant last September for the Phase I study the company hopes to initiate this year.
Slightly ahead of brilacidin is Soligenix's oral mucositis candidate SGX942, a fully synthetic pentand radiation therapy. The company describes the product as an innate defense regulator that stimulates the innate immune response to damage. In preparation for its Phase II study, which the company expects to initiate in the second half of 2013, Soligenix announced the formation of a mucositis medical advisory board in February.
As Rubenstein suggests, great strides were made in the treatment of chemotherapy-induced nausea and vomiting (CINV) with the advent of the 5-HT3 receptor antagonists, the most recent entrant being Eisai's Aloxi (palonosetron), but even here, companies continue to seek improvements.
For example, last September, A.P. Pharma resubmitted its NDA to the FDA for its lead product, APF530. Essentially the 5-HT3 receptor antagonist granisetron, APF530 has been formulated for a more durable response on a single injection. The hope is to extend granisetron's current indication for acute-onset CINV (first treatment day) to include delayed-onset CINV (up to five days), putting it more on par with Aloxi, which is indicated for both acute and delayed onset.
OPKO Health, meanwhile, in partnership with Tesaro, is currently in Phase III trials with its CINV candidate rolapitant, a neurokinin-1 receptor antagonist that it is testing in three studies of patients receiving highly- and moderately-emetogenic chemotherapy with standard of care, which includes a 5-HT3 receptor antagonist and the steroid dexamethasone. The company expects the results of the trials for late 2013.
And of course, several companies continue to push various marketed supportive care products through clinical trials, looking for expanded applications. Such is the case with Amgen, which continues to test Neulasta (pegfilgastrin) against febrile neutropenia. In January, the company announced the results of its Phase III PAVES trial in metastatic colorectal cancer, showing significant reduction in the incidence of grade 3/4 febrile neutropenia versus placebo.
Alder Biopharmaceuticals took a step back, however, and decided there had to be something better than the current whack-a-mole approach to tackling side effects; that is, hitting them as they arise, or are expected to arise. The company took a more holistic approach by looking for commonalities in the root causes of many side effects.
Alder focused on the pathways controlling inflammation, and specifically on the cytokine interleukin-6 (IL-6).
"There's a number of places where inflammation plays a role in a negative sense for cancer patients," explains Randall Schatzman, company president and CEO. "First and foremost, the cancers themselves release pro-inflammatory molecules like IL-6, which cause a lot of inflammation secondary to the tumor. And this inflammation is really responsible for a large percentage of the morbidity and mortality."
The second place inflammation plays a role is in tumorigenesis itself and how certain chemotherapies work on the tumor, such as in the development of resistance. And finally the problems that chemotherapy itself causes—e.g., anemia and oral mucositis—side effects of chemotherapy that Alder believes are driven by pro-inflammatory cytokines.
"What we're trying to do is tease out the role of inflammation within this remit of paraneoplastic issues and chemotherapy side effects, because at the end of the day, oncologists want to get as many cycles of chemotherapy into a patient as possible to increase their chances of getting a complete response," adds Jeffrey Smith, Alder's senior vice president of translational medicine. "If the patient is saying time out after, say, three cycles out of six because of the side effects, then clearly, the oncologist isn't winning. It's a suboptimal treatment of the patient's cancer."
As Smith sees it, it's about more than simply controlling side effects to improve patients' quality of life, and therefore making treatment easier. It's also about getting more cycles into the patient to improve the chances of killing the cancer off. It's about both efficacy and safety.
Alder's lead cancer product, ALD518, currently in several Phase II studies, is a humanized anti-IL-6 monoclonal antibody that keeps the cytokine from interacting with its receptor and triggering a variety of cancer-related conditions such as catexia (wasting), anemia, oral mucositis and acute graft-versus-host disease.
Smith sees opportunities for ALD518 not only in preventing some of the side effects of chemotherapy, but also in helping to modulate the body's response to immunostimulatory regimens, using the recent case of children being treated for ALL as almost a proof of concept.
In that study, patients received modified T-cells designed to attack the leukemia, but in some cases, the patient's immune system was overstimulated and the therapy triggered a cytokine storm that threatened to kill the patient. When one oncologist noted that a patient's IL-6 levels had spiked, he prescribed her Actemra (tocilizumab), an antibody used in rheumatoid arthritis that targets the IL-6 receptor. Within hours, the patient stabilized.
Smith thinks a day is approaching when this type of regimen becomes the paradigm, not the anecdote.
"I could foresee more of that, where agents such as an anti-TNF or anti-IL-6 or other agents that modify the immune system are used to damper down unwanted effects of immunostimulatory regimens," he says.
Schatzman also sees the holistic approach as paying off from a payor perspective.
"It's one thing to have an agent that makes patients feel better, and it's another to have an agent that improves outcomes," he says. "I think the burden is on us to do the latter because when you improve clinical outcomes rather than just patient-reported outcomes, payors are more willing to pay."
And at the end of the day, as with any industry, it's about getting paid.