When the first targeted oncology therapy — Novartis Pharmaceuticals Corp.’s Gleevec (imatinib) — came to market in the U.S. more than 20 years ago, industry experts believed it was ushering in a slew of similar agents focused on a particular mutation, or biomarker. The field, however, did not quite live up to lofty expectations. And while more research has grown the industry exponentially in recent years, it still faces some challenges. But no one can deny that biomarkers have had a profound impact on cancer care.
Agents that target drivers of cancer “have fundamentally changed the way that we treat a wide range of cancers,” maintained Josina Reddy, M.D., Ph.D., global head and vice president of product development for lung, agnostic, skin and rare cancers at Genentech, Inc., a member of the Roche Group, during a recent webinar sponsored by her company. “And in this era of cancer care, identifying those biomarkers in a patient’s tumor tissue or using blood is an essential step in developing a personalized treatment plan, and for a growing share of patients, that treatment plan might include those targeted therapies.”
But in order to be treated with targeted, patient-specific therapies, people must be tested for all of the biomarkers relevant to their cancer before starting treatment, she noted. “Remarkably, though, a large share of patients don’t get the comprehensive biomarker testing that they need. If we look at lung cancer, for example, in a recent study, only about three-quarters of patients with non-small cell lung cancer were tested for biomarkers at all, and fewer than half of them received comprehensive testing for all of the potentially relevant biomarkers using what’s known as next-generation sequencing. And there are now more than 20 FDA-approved targeted therapies for patients with non-small cell lung cancer. Some of these are suitable only for a small number of patients — let’s say 2%, 1% or even less. But if you or one of your loved ones is one of those 2%, these targeted treatment options could really make a difference. And taken together, these seemingly small percentages of patients add up to a substantial percentage of cancer patients who could benefit” from biomarker testing.
According to Reddy, multiple factors may impede biomarker testing, including “the fast pace of scientific advances and challenges with keeping up,” as well as “skepticism about whether the cost and effort of comprehensive testing is worth it, especially when some of the relevant biomarkers are pretty rare. But research now suggests that the cost of comprehensive testing with next-gen sequencing is offset by avoiding wasted time and missed opportunities for therapy.” Logistical issues may involve whether enough tissue has been collected or potential access to liquid biopsies. In addition, “insurance coverage can be uneven, lab capacity and turnaround time can vary, and some of this can be a specialty challenge in community-based oncology practice, where many patients get their care. So, in short, it’s complicated. It’s difficult, but it’s necessary to get patients and their families up to speed quickly after a diagnosis so that they can insist on the best possible care.”
The inability to undergo such testing means that patients do not have access to potentially life-prolonging medications. “Unfortunately, like many other aspects of our health care system, this access barrier falls more heavily in some communities than in others,” she stated. “In an analysis that was done by Flatiron several years ago, patients with lung cancer who were Black, older or on Medicaid had lower odds of getting next-gen sequencing compared to patients who were white, younger or commercially insured.” In the previously mentioned study, about half of white patients with lung cancer received next-generation sequencing compared with about 40% of Black patients.
“Personalized cancer care should be a source of better lives for all and not a driver of even deeper inequities, and we need to fix these gaps now,” she asserted.
Moderator Matthew Herper, senior writer of medicine and editorial director of events at STAT, explained that the term “biomarker” can mean a variety of things, “everything from DNA sequencing of tissue samples to your blood pressure.” Within oncology, biomarkers can “distinguish tumors as opposed to just where the tumor is,” and this approach has “really changed the way cancer is thought of, is treated and the way that drugs [have been] developed for the past 20 years.”
MSI-H+ Is Effective Biomarker
One biomarker that works well is microsatellite instability-high (MSI-H+). This particular example, explained Suzanne Topalian, M.D., professor of surgery and oncology at Johns Hopkins University School of Medicine and associate director of the Bloomberg-Kimmel Institute for Cancer Immunotherapy, is “a marker generally of very high mutational burden in a cancer.” Topalian, whose focus is on cancer immunotherapy, pointed out that “very highly mutated cancers seem to be more susceptible to anti-PD-1 [i.e., programmed cell death-1] therapy.” In 2017, the FDA approved the biomarker “for patients with any type of cancer to become eligible to receive anti-PD-1 therapy,” making it the first approved biomarker for all cancer types.
The MSI-H+ and programmed death-ligand 1 (PD-L1) are “very different biomarkers, but both are FDA approved to identify patients who are candidates for anti-PD-1 immunotherapy,” said Topalian. More specifically, “a PD-L1 in contrast to MSI-H+ is a marker that can change over time, and it’s a marker that can have different levels of expression in different sites of metastatic disease in individual patients. So there’s a lot more biological variability, which has made it more difficult to find the right place for this biomarker, to find the right way to use it effectively to inform cancer treatment and care.
“MSI-H+ is not going to change over the evolution of a cancer, and it’s not going to change in different metastatic sites in an individual patient,” she continued. “It’s going to be there. And so if you could test for this 10 years before you think about giving immunotherapy to a patient, you can go back to an old tumor and find this marker, and it can then help you with your strategy for planning out treatment for that patient. But PD-L1 immunohistochemistry is very different than that. PD-L1 has not become an across-the-board biomarker the way MSI-H+ is, but it has been shown to be helpful in certain cancer types in certain disease settings for identifying patients who are more likely to respond to an anti-PD-1.”
Added George Demetri, M.D., director of the Sarcoma Center at Dana-Farber; director of the Ludwig Center at the Dana-Farber/Harvard Cancer Center; and executive director for Clinical and Translational Research at the Ludwig Institute for Cancer Research, “MSI-H+ is hard-wired into the cancer. It is not going to change; it’s locked in, whereas a lot of other so-called biomarkers come and go — the cancer wants to put it on its surface and show it to the internal environment or it doesn’t. And that’s the problem with something like PD-L1.”
Asked if cancer is outsmarting us without the use of biomarkers, Demetri responded, “Yes, absolutely. It’s using every tool of evolution to change its shape, to change the way it shows itself. It’s a pretty sneaky little critter there.”
He explained that the “important thing about MSI-H+ is it’s also mechanistic.… If you’ve got a mutation, it turns on an enzyme, so it’s not stopping. If you can stuff the mouth of that enzyme, it can’t do anything, and it might stop the cancer, and that’s kind of what Gleevec does. With MSI-H+, what it is is a low-fidelity replication system, so the cells are making copies of each other, but each copy starts looking weirder and weirder, and the immune system recognizes things that don’t look human. That’s why our immune system is good at fighting off bacteria and fungi and — when provoked with the right mRNA [i.e., messenger RNA] — COVID. So I think that’s the issue. It can be very mechanistic, and when you see a mechanism-directed biomarker, it’s usually much more reliable, much more powerful and means a lot more for patients because you can really count on it, and it can help us pick the right, very reliable, active drug or therapy for that patient.”
Anna D. Barker, Ph.D., chief strategy officer at the Ellison Institute for Transformative Medicine of USC and distinguished visiting fellow of complex systems at Arizona State University, noted that 21 years ago when Gleevec was approved, she was “pretty familiar” with the BCR-ABL gene associated with the drug. “I think the assumption then — and actually it sort of paved the way for what we talk about today in terms of precision oncology — is that everyone thought there was going to be this rampant, sort of number of fusion genes, that the BCR-ABL gene was going to be replicated a thousand times, and we were going to find thousands of fusion genes.” These are genes involved with cancer that get fused with something else, and “the cancer is using that to turn something that the body usually turns off on all the time,” explained Herper.
However, continued Barker, “that didn’t happen. But it’s happening now in that in all of those years, we’ve been looking for fusion genes again and again and again. Now we’re beginning to see how important they really are because now we’re talking about fusion genes all the time once again.”
‘It’s Almost Like Back to the Future’
She elaborated that “more and more we’re starting to test for them, and more and more we are starting to find them,” such as ROS1 in non-small cell lung cancer. “You can find them with next-genome sequencing, but it turns out it looks like it’s going to be better with transcriptomes. So it’s almost like back to the future.…Where we were 21 years ago is where we would have liked to have been 21 years ago, and that’s where we are now. I think we’re really just starting to see the power of these combinations. I mean, MSI-H+ is a very complex biomarker.…This complexity of biomarkers is the future. And I think that just as you have MSI-H+, and you have biomarkers like PD-L1, we see the same thing in breast cancer or colon cancer, we’re seeing these complex biomarkers come together to tell a bigger story.”
As technology has improved, asserted Demetri, researchers have been able to identify more fusions. Previously, analyses focused on DNA, which “sticks around forever; it’s easy to understand. But DNA turns into RNA, and then the RNA turns into proteins. The RNA is more evanescent; it comes and goes, and it’s more fragile. But that’s really where it’s easier to find these fusions. So the adage of a carpenter is only as good as their tools really applies here because we didn’t have great technologies to find these fusions when all we were looking at was DNA and old technologies. But the new way of looking at RNA makes it easier, and now we can actually find it in other ways where the technology has gotten better.”
When Herper asked how tumor mutational burden (TMB) as a biomarker compares to MSI status, Topalian explained that “as we gained more experience and information from testing, immune checkpoint blockers like anti-PD-1s in different cancer types, in pulling all the information together, people started to realize that it was the cancers that tended to have more mutations in them that were more likely to respond. A great example is lung cancer in smokers. Those cancers are caused by a carcinogen, and they tend to have high mutational burdens. And they’re also more responsive to anti-PD-1s than a cancer like pancreatic cancer, which has a very modest or low tumor mutational burden.”
However, applying that TMB association with anti-PD-1s to a biomarker “required establishing a threshold, like a cut-off point above which you would say, ‘This tumor has so many mutations, it’s going to respond to anti-PD-1s,’ and that would qualify that person to receive an anti-PD-1 therapy,” she continued. “That’s where the problem came in. It’s very difficult to define where that threshold is. Eventually the FDA did approve one anti-PD-1 drug, pembrolizumab, for any type of cancer that has a TMB that is higher than 10 mutations per megabase in the tumor. So that’s about 300 mutations per tumor genome. But there’s now information that maybe one cut point like that is not going to be the best for every cancer type. And so this has now started an ongoing conversation in the literature about whether a certain type of cancer needs a higher cut point or a lower cut point in order to use this marker in the best way. It’s still an unresolved issue.”
Herper noted that many biomarkers are rare, so are they helpful in common cancers? Barker said she believed they are, but “I think there are some big bumps to get there. But if you look at something like HER2,” a common biomarker in breast cancer, “it turns out that one way or another, it’s helping nearly every single patient. I think there are going to be biomarkers like that. This is a receptor, it gets amplified, you’re going to be in one class of patients, but we’re starting to use these biomarkers very effectively to subtype these diseases.” Subtyping tumors can lead to more effective treatment of them, she contended, and this approach is used in various cancers.
“Breast cancer set the standard for one of the first biomarkers,” stated Demetri. “Did it have a hormone receptor in it, and should you use hormone-blocking therapies to treat it? That was a biomarker that saved the lives of millions of women around the world with hormone therapies.” He said he expects that “we’re going to find more and more reliable biomarkers — they’re probably going to have to be multiple — that will point to combinations almost certainly because most of the common cancers are pretty complex things. They’re not simple like GIST [i.e., gastrointestinal stomal tumor] with one or two driver mutations. They’re not simple like HER2-driven breast cancer, which has that dominant HER2 focus driving the cancer. They’ve got all sorts of little things driving them, some of which don’t stick around. They can change,…they can shift, they can use this epigenetic shifting to turn from blue to green. They’re chameleons. And I’m not worried about that because I do think our technology will catch up to it, and we’ll have three or four combinations just like we do for very severe infections. Look at multidrug-resistant tuberculosis. You still can come up with multiple drugs to overcome that.”
Using a panel of tests vs. one for a single biomarker is “the direction we’re going,” declared Topalian. “Right now, we have unidimensional markers for immunotherapy, and we’re trying to develop something deeper than that. It turns out that some of the markers we have do not depend on each other functionally. For instance, PD-L1 expression does not depend on TMB, so that means that you may have more power by combining those two markers in a single test. There’s a lot going on in that area, but so far no home runs [in terms of] something that has made it to an approval.”
Oncology Is Experiencing ‘Huge Shift’
According to Barker, oncology is undergoing “a huge shift.…We’re collecting enormous amounts of data, from all kinds [of sources], especially from molecular profile mutations. In parallel, we’re collecting real-world data.…The future may be that maybe the genes are going to tell us what we see in the data, and we’re going to be able to go back and construct these panels prospectively for cancer.”
Demetri pointed out that in 1997, “lung cancer was just lung cancer.” But now the condition is recognized as consisting of non-small cell and small cell lung cancers, and biomarkers such as EGFR, ROS1, ALK, RET and KRAS are involved. As such, turnaround time for biomarker tests is really important as well to help determine treatment regimens.
In order to get this testing to underserved populations, “data has to rule,” maintained Demetri. “We should practice evidence-based medicine. And if you need that data, you need that evidence, to make a decision for your patient, just picture what you would do for your family member, your loved one. Everybody can picture that. If I cannot make a wise decision without having those data, how can I make a decision? We as professionals need to justify that, mainly to payers, and the payers will say, ‘Well, that’s a lot of money; why should we pay you?’ And then we have to have a strong enough evidence base, that it really is a life-or-death, significant prolongation of life vs. closer death that really drives that to a high-quality health care system.”
But, continued Barker, “when something obviously works [such as Gleevec], then CMS steps up to pay for it, and people in the community hospitals are going to use it, just like people in the academic medical centers. I think our problem with biomarkers is that there’s been a long regulatory learning curve here, a learning curve on both sides of the equation. Now we’ve even got breakthrough status for certain diagnostics. So this whole field has moved very quickly.…This has to be very broadly distributed to our 80% of patients who are treated in community hospitals and all other kinds of places other than our cancer centers.”
As the Biden administration is talking about reigniting the Cancer Moonshot, there has been talk of needing a “learning system in oncology,” Barker said. In the same way, “We need a learning system for biomarkers. You take any one of the diseases we’ve talked about today — we could actually put that dataset together for PD-L1. We could have a national biobank for some of these kinds of things or a data repository for these tumors. We just haven’t done it. Creating a learning system could change a lot of what we’ve talked about.” And it would allow community oncologists to be able to look up their patient in a dataset based on their biomarker profile. “Those are not intractable problems. We should solve these problems.”
In the quest to collecting those data in real time and finding emerging biomarkers, “academic labs and commercial labs are currently trying to better one another because the academics want to stay alive, [and] the commercial labs want to take over,” said Demetri. “It really is almost an arms race of who can do it better. That’s a positive cycle of competition to try to see how fast from the minute a sample can get out of a hospital into a molecular pathology lab, how fast can a high-quality, annotated result be given back to the doctor?”
He explained that a problem is not only the technical element but also education “because a lot of doctors were trained before these data were available. We have to retrain everybody to know what this really means so we don’t see a massive epidemic of misinterpretation.” Noting the amount of patient-advocacy groups on the call, Demetri asserted that “patients are really their best advocates, and they can be the ones who share their understanding of a biomarker or a disease or a term of art with other patients so they can use it wisely because you can’t count on all the doctors being up to speed as much as this field is changing.”
Biomarkers should be able to help strategize the next therapeutic step for a patient undergoing therapy, perhaps through liquid biopsies, said Topalian. “Information on when relapse is occurring will help us get a jump start on what the next treatment will be.” While liquid biopsies are not “standard practice for most cancers,…I do think there is a lot of study in this area now. One scenario is a patient who starts out with a positive marker in the blood and then is on a treatment, and you want to see that marker dropping so you can follow the trajectory of the marker over time and get a sense of what the outcome” will be. If that marker disappears completely, does this correlate with long-term survival? “These are all the kinds of studies in process now.…One of the challenges with blood-based markers such as ctDNA [i.e., circulating tumor DNA] is sensitivity. I think we’re in the laboratory still trying to improve the sensitivity so that we could pick up, for instance, a tumor relapse after surgery when the tumor [is] still at a microscopic or very small volume.”
Biomarkers have a role across the entire treatment journey “from prognosis to diagnosis,” as well as “treatment planning and monitoring at every phase of the patient’s journey,” including through survivorship, Barker noted. “These biomarkers have applications across that entire space, and more and more, there are going to be all kinds of approvals for these biomarkers for different applications. The FDA says your biomarker must be ‘fit for purpose,’ which means it has to have context. In other words, you have to know what you’ll use your biomarker for.”
While Demetri stated he was against “broad, expensive screening in people who are otherwise healthy,” one area of concern is with so-called cancer predisposition syndromes, where cancer runs in the family, such as with Lynch syndrome. “It’s really important because we think a lot of those young Lynch syndrome patients can be cured with immunotherapy or at least certainly live a long time.…One cancer biomarker is looking at the genes that run through families and seeing if there is a predisposition and then monitoring those people so we can get a cancer as soon as possible, where we know they have a hundred- to a thousand-fold likelihood of developing cancer than the general population.”
Demetri said he’s all for the entire population getting screened via liquid biopsy, but that he worries the technology is not quite there and would produce a lot of false positives at this point. “We’re at the point where liquid biopsy is good, but it’s not as good as we want it to be.”
Looking to the future of biomarkers, Topalian says that she’s “excited about the potential for multidimensional biomarkers.…There are some technologies that wrap this all into one package. One would be gene expression profiling — can we find a group of genes whose expression either up or down can be packaged and will tell us that a treatment might be effective in that situation? Another technology is done on tissue slices, a multiplex staining of different proteins in the tumor.”
Barker looks to “much better, highly effective biomarkers for the big killers, those cancers we haven’t been able to do much about” such as pancreatic cancer and glioblastoma. “What are the unknown unknowns? What are we missing with these cancers? We can’t seem to detect them early enough to actually stop them.” She said she expects that “we’re going to have the ability…to detect cancer before it’s clinically diagnosable. If you had the ability to do that, to deliver drugs and monitor patients,…I think that’s all doable.…I think biomarkers are the key to doing all that.”
In glioblastoma, added Demetri, we’re “starting to see some exceptionally responding patients to immunotherapy, but only a few, so we need to learn why those few people? Why are they so lucky? What is it about their tumor?”
People at an advocate level have various sources of information available, agreed the speakers.
“It turns out that advocates are the new scientists when it comes to biomarkers,” declared Barker. “Once an advocate understands that this is important to their disease, they’re not going to leave their physician’s office unless they get a test for that biomarker.” There are even patient groups, such as a ROS1 group formed around that particular biomarker, she said. Similarly, added Demetri, the NTRK biomarker, which occurs in many cancers, has a global, non-profit patient group — the NTRKers — focused on it.
Training programs are available through groups such as the American Association for Cancer Research (AACR) and the American Society of Clinical Oncology (ASCO). Literature, including white papers from the National Cancer Institute (NCI), is available. And most major oncology meetings have patient-focused tracks on biomarkers, said Topalian.