John L. Marshall, MD; Claudine Isaacs, MD; David Spetzler, PhD, MBA


June 22, 2015

Editorial Collaboration

Medscape &

This feature requires the newest version of Flash. You can download it here.

John L. Marshall, MD: Hello. I am John Marshall, professor of medicine and director of clinical research at the Lombardi Comprehensive Cancer Center at Georgetown University in Washington, DC. I want to welcome you to this edition of Medscape Oncology Insights, coming to you from the 2015 Annual Meeting of the American Society of Clinical Oncology (ASCO), where the weather has just been terrible. It's freezing cold, there is a blustery wind—a nor'easter is coming through at the end of May.

What we are going to talk about today is genetic profiling. Molecular testing is everywhere. All of these companies are popping up. For almost every cancer that we take care of, there is some sort of additional genetic testing that is being done now. Is this broad molecular profiling essential to moving forward? Is it even advisable? Or is it, in fact, kind of unnecessary? We are getting a lot of these questions, and to help answer them, I have invited two good friends and world-class experts to come and discuss this with us. First is my good friend and long-time colleague, Dr Claudine Isaacs, professor of medicine and oncology, co-director of the Breast Cancer Program, and co-director of the Fisher Center for Familial Cancer Research at Georgetown University. Claudine, welcome. Thanks for joining us.

Claudine Isaacs, MD: Thank you.

Dr Marshall: And joining us is David Spetzler, who I recently met as we partnered through a bunch of projects. He is adjunct faculty in molecular cellular biology at Arizona State University in Tempe, and chief scientific officer at Caris Life Sciences in Phoenix. David, thanks for leaving the sunshine for this lovely Chicago weather. Thanks for joining us.

David Spetzler, PhD, MBA: Thank you.

Molecular Profiling: A User's Manual

Dr Marshall: What we are going to do is really drill down on two topics. First, we are going to pick apart molecular profiling. David, I am going to primarily pick on you for that. What does it mean? What are the different tests out there? I also want us to focus on what to do when we get these test results back. What do they mean? How do we use them? How do they affect patient care? How do I deal with these funny genetic mutations? We are going to cover all of that.

So, Dave, walk us through what most people know or have heard about next-generation sequencing. We know about some immunohistochemistry (IHC) tests for certain tumors. What do you think is the state of the art today? Walk us through some of the big tests.

Dr Spetzler: I think that the state of the art today is a combination of a multitude of technologies. When we evaluate the underlying molecular components of a cell, the components range from oligonucleotides and the genomic status of that cell to regulatory pathways of RNA, and finally to proteins and protein complexes.

Dr Marshall: Where do you think all of the answers are? Everybody is saying it's the DNA.

Dr Spetzler: I don't. I think that there are some clear answers in DNA, but for the most part, there are so many feedback mechanisms within the cell that we have to look much more broadly than that. If you look only at the DNA, then you are going to miss epigenetic alterations. You are going to miss regulatory components from RNA-induced silencing complexes. You are going to miss protein expression. Depending on how the proteins are assembling, you can have widely varying activity, and that is going to be key to understanding what is happening, especially in light of some of the new and exciting immunotherapies that are coming out.

Dr Marshall: I know that for colon cancer, we do some RAS gene testing as well as IHCs. In breast cancer, you have some standard tests for both genes and proteins, right? What is the standard profile today in breast cancer?

Dr Isaacs: Breast cancer is a little unusual because, for a while, we have been doing subtyping using standard immunohistochemistry and genetic analysis. We are using ER, PR, and HER2 as our standards. The question is, should we be doing more than that on metastatic sites?

Dr Marshall: But even for HER2, you use two different modalities.

Dr Isaacs: We tend to start off with immunohistochemistry and then we reflex to fluorescence in situ hybridization (FISH) if there is a result that gives us an indeterminate finding.

Dr Marshall: We pick on each other about this. We are still debating about which is the right way to test for HER2, aren't we?

Dr Isaacs: We are. Interestingly, in breast cancer, where we have been doing ER and PR testing for decades now, it took us until about 2 years ago for ASCO and the College of American Pathologists to come out with guidelines for how ER and PR should be performed and how we should interpret the results,[1] a test that we have been doing for decades.

Can You Trust Genetic Test Results?

Dr Marshall: As this new stuff is emerging, we often think of it as black and white, that the test is positive or negative. Talk to me a little bit about quality—tissue, tumor variability, all of those things. How reliable are these tests? can get widely differing results from what should essentially be the same underlying sample.

Dr Spetzler: It varies incredibly by laboratory. I think one of the key missing elements of molecular profiling is standardization of quality metrics across the entire industry. Everyone does it a little differently and everybody thinks about it a little differently, and depending on who you are working with, you can get widely differing results from what should essentially be the same underlying sample.

If we take, for example, the ER/PR question, testing of the incoming antibody lots is essential to a high fidelity result. Most people are relying on the manufacturer's quality control system, which often relies on just a recombinant protein test. I personally think that that is a very poor quality-control metric. Looking at a native system, looking at tissue lysate, and ensuring that that antibody has a high degree of specificity for its target is essential to interpreting the results properly. We actually perform a Western blot and then do confirmatory mass spectrometry on the antibodies that we use in clinical assays to be sure that we are hitting the right targets.

Dr Marshall: We have a couple of issues about tumor size. If you only have a very small amount of tumor, you are limited. Talk to me a little about the way tumors are stored. Most are formalin-fixed. How much of this stuff needs to be in fresh tissue? We see some assays that need to be fresh and some that can be formalin-fixed.

Dr Spetzler: It depends on the maturity of the technology. For the most part, we have been able to get the information out of formalin-fixed samples. It takes a little bit of extra work in preparation, but we can get good-quality results. There is definitely a risk for overfixation, where you end up with crosslinking of the nucleotides. That can inhibit your ability to amplify the DNA and get a good molecular result. But for the most part, I think we are doing a relatively good job of that.

Dr Marshall: What is the correct tumor descent in breast cancer? On a new diagnosis it is the primary, right?

Dr Isaacs: If you are doing molecular sequencing in breast cancer, you are typically doing that in the metastatic setting. We generally try to do a biopsy closest to the time that we are going to use those results. In breast cancer, there have been studies looking at changes to the standard evaluations that we do—looking from the primary to metastatic site—that show that ER, PR, and HER2 status can change.[2,3] As we use even broader testing to examine many more genes and proteins, it makes sense to look at the most proximate biopsy to make determinations.

Dr Marshall: And how often would you do that in breast cancer? Just once with metastatic disease, or are you finding that it might be logical for patients who have had several years of survival with metastatic disease to do that more than once?

Dr Isaacs: If we are going to be using next-generation sequencing techniques or various broader looks at the tumor, we really want to get the information closest to the time when we will be either changing therapy on the basis of that molecular testing or using the testing as one more piece in the jigsaw puzzle for making a decision about what to do next. My own recommendation is that if we are going to do testing, we should do it close to the time that we are going to be making a decision.

Dr Marshall: We in colon cancer have been a little slow because we have had this belief that, say for RAS testing, the primary is the way it is always going to be, but my experience over the past year or so is that the more testing we are doing,[4] the more we are seeing changes. One of the pushbacks we always get about this type of testing is tumor variability. Studies have looked at isolated balls of tumor and seen different profiles from different segments.[5] How are we going to deal with that variability in making therapeutic decisions?

Blood-Based Biopsy to Aid Therapeutic Decisions

Dr Spetzler: That is a very tough question. One of the things that we can do is look at the underlying structure of the tissue and try to profile the multitude of clones that are present. But relying on tissue for that is going to be a flawed approach. We eventually have to get to a blood-based biopsy-type technology that is capable of looking at all of the clones that are present in the cancer.

Dr Marshall: Is this the circulating cancer cell? There is a lot of work like that in breast cancer. Is that what you are thinking?

Dr Isaacs: Yes. People are looking typically at circulating cancer cells within the setting of clinical trials to try to evaluate their utility.

Dr Marshall: I always wonder which cells living in the blood are actually reflective of the tumor. Give us some of the newer technologies that are being explored—how we are going to measure the blood and what we are going to measure in the blood.

Dr Spetzler: Aside from circulating tumor cells, there is certainly opportunity to exploit circulating microvesicles that are going to be shed into circulation much earlier in the cancer event. That is going to be a critical component, because if you have to wait until there is metastatic disease to begin to get that type of real-time feedback, it is going to be less useful than if we can get the information early. Then we can really start to make meaningful contributions to treating patients.

Dr Marshall: I struggle with how we are going to figure all of this out. We can order these tests and get these answers back. We need to work together to figure out the answers to all of this. There are various groups forming—we are forming one ourselves—to try to work collectively. What is the magnitude of the number of patients and amount of data? How big does the haystack have to be in order for us to find these needles?

Dr Spetzler: It is huge. If you do the math for 55 drugs with 60 biomarkers, even assuming equal prevalence of them, you need 2 million patients in order to be powered to assess an individual biomarker's correlation to an individual therapeutic. The days of randomized trials are quickly going to go away because it is just mathematically infeasible to accomplish it. I think performing observational evaluations and analyses are the only things that we can do. It is going to take a concerted effort to track not only the underlying profile, but also the patient's treatment history and final outcome. By combining all of our data, we can begin to have a chance at this, but it is going to take a village. It is not going to be done by any one person or any one group alone.

Dr Marshall: Claudine, I know you and I have been thinking a lot about this. How are we going to apply this broad molecular testing in the clinic and use it? My experience with most oncologists is that they have been doing this testing as an attempt to find an extra therapy. They go through the standard lines of therapy, the patient is still doing okay but has metastatic disease, and they're looking for something more. Tell me your thoughts about this and also about how we could move this more proximally in breast cancer, colon cancer, and other diseases, where it might help guide us away from empirical therapy and more toward molecularly targeted therapy.

Best Uses for Molecular Profiling

Dr Isaacs: In breast cancer, where we have volumes of data on what to do in certain situations, we need to be a little careful and make sure that at least in the first couple of lines of therapy, we are being guided by well-conducted clinical trials. Once we get to the point where we care whether it is treatment A or treatment B and the umpteen other factors that go into it—patient preference, comorbidity—we can certainly use this technique as another piece of that puzzle to help push us.

We have also been using it when we think about phase 1 trials. We are looking at new drugs because there is a plethora of clinical trials that are using good standard agents, but they are adding some targeted therapies. It allows us to try to figure out whether this trial might make more sense for this particular patient based on the profile that we are seeing.

Dr Marshall: Yes, I think that is a great setting. If you have choices and you are unsure, can these tests help guide us? And then we also have the other piece. We are seeing a lot of this in gastrointestinal cancer. In fact, we are seeing some of this at the meeting. We are finding targets like HER2 in colon cancer.[6] Not a huge number, but 5%-7% of colon cancers is a lot of colon cancer. That is higher than the percentage of ALK-positive lung cancer. Do you see that there may be some value in finding those targets? What level of evidence do you think you would need to say that, because a trial showed a particular benefit rate, we should get that drug extended to those types of patients?

Dr Isaacs: For targets in breast cancer, there have been many studies that have validated certain measures, and those are the measures that actually showed us that particular therapies were effective. I think that we need to be careful if we have a tumor that is repeatedly HER2 negative by standard evaluation and then, all of a sudden, it turns up HER2 positive. We need to do a little due diligence and make sure. We also recognize that tumors have different clones, and that you might indeed have a clone now that you didn't recently biopsy that is HER2 positive.

With Antibodies, You Don't Always Get What You Pay for

Dr Marshall: Okay. That's great. So, David, the talk at this meeting was about immune therapy, and your company may be the only one that has a commercial assay for measuring PD-1 and PD-L1. There is a lot of talk about whether this is the right biomarker or not. We have positive trials and we have negative trials. Talk to me about this target and tell me your opinion about whether or not we should be measuring it. Why are we getting variable results from various trials?

Dr Spetzler: There are a couple of different components to that answer. First, of course, is the underlying quality of the antibody that is being used in the assay.

40% of the antibodies we can buy off the shelf actually bind to a different target than what is on the label of the bottle.

Dr Marshall: I am an oncologist. I don't know why that matters. I think antibodies are perfect and they all work the same.

Dr Spetzler: It is a little shocking to realize that when we order antibodies and test them, we find that about 40% of the antibodies we can buy off the shelf actually bind to a different target than what is on the label of the bottle.

Dr Marshall: So the antibodies that you would use in a Clinical Laboratory Improvement Amendments lab are that variable?

Dr Spetzler: They are that variable. When we get additional lots of the same underlying clone, about 25% of the time there is a perturbation in the performance of that particular lot of antibody. They are incredibly sensitive to changes in purification, storage, or underlying buffer content.

Dr Marshall: Is that true for all of them or is it true for the ones that are newer and that haven't been ironed out quite as well?

Dr Spetzler: RUO antibodies are research-use-only. There are ASR antibodies, which are analyte-specific reagents, and then there are FDA-approved antibodies. As you climb up that regulatory chain, you get an increase in reliability.

Dr Marshall: How do oncologists know what they are getting?

Dr Spetzler: If they don't ask the question, they probably don't know what they are getting.

Dr Marshall: Many times, oncologists aren't making the decision about who is doing the test. The test can be done locally in their own pathology labs or wherever their healthcare system has contracted. Is that right?

Dr Spetzler: That is right.

We need to learn to question our lab results.

Dr Marshall: So, what we need to do is—just like we do when we get radiology reports that say "new nodule, can't rule out cancer"—learn to question our lab results, particularly as these new molecular tests come forward.

Dr Spetzler: Without question.

Dr Marshall: Now let's drill down on the PD-1, PD-L1 story a little bit.

Dr Spetzler: If you assume that some of the negative results are a result of poor quality control, then that makes the positive results a little more enticing. Of course, the alternative could be just as true. What if there is a perturbation—it is not actually hitting the PD-1 or PD-L1—and that is what is showing the positive result? It comes back to needing an aggregation of all the data elements to be able to fully understand what is happening. As you evaluate the intercalating leukocyte and its binding partner on the surface of a tumor, there can be an interaction there. Is it the number of leukocytes expressing these proteins that is actually what is going to determine whether there is a good response or not?

I think we are at a point where we have a lot of questions. We don't necessarily have a lot of answers. For me, it is all about facilitation of exploration of these data elements. Some of them are going to turn out to be good, some of them are going to turn out to be bad. But unless we start to cohesively understand what is out there, we are not going to be able to figure out which is which.

Tales of the Unexpected Mutation

Dr Marshall: That brings us to the next topic. When you do a broad profile, 600 genes and all of this IHC, what is research and what is practice? Claudine, I have been down to your office several times with a report of some patient who has some funny BRCA mutation or the like. We are getting unexpected mutations, and I, as an oncologist, go back to the chart and look at the family history. Did I miss something? How are we going to deal with all of this new stuff that we are going to find, the needles in the haystack that might mean something but might not?

Dr Isaacs: Finding what is suggestive of a germline genetic mutation, not a somatic mutation in the tumor tissue, is something that we need to be increasingly prepared for. As we start to do more and more of this testing, we are going to find unexpected mutations and tumor types that we weren't considering as part of the spectrum. It is quite clear that we need to be prepared for that. Whenever somebody sends off one of these tests, they need to be prepared for something that might not be what they anticipated and that could be reflective of a hereditary predisposition to cancer. Not only does it affect the individual that you are testing, but it could have profound implications for their family members, too. Before you send the test, it is important to discuss with the patient that there might be an unexpected finding. There are some suggestions that you should take a broader family history and tell the patient if there is a suggestive family history. "Maybe we will find something on your tumor. How will we handle this?"

If you find something unexpected, there are clear recommendations from the American College of Medical Genetics.[7] They recommend that if you find an incidentally found mutation that predisposes the patient to a number of hereditary syndromes—many of them are hereditary cancer syndromes—you have a duty to disclose that to the patient. You need to be prepared for that. You should probably be prepared to have a genetic counselor around to help you. Many people are creating molecular tumor boards that include a broad group of clinicians and molecular and medical geneticists, or a genetic counselor, to help wade through that process.

Another issue that often comes up: We do this testing as people are approaching the end of their life. What happens if you find a BRCA1 mutation and the person that you did the testing on is deceased? You have a piece of information that could have very profound implications for that person's family members.

Dr Marshall: Yes, this is potentially charged information. I want to make sure that I understand this. When you or molecular companies get tissue, you are actually purposely testing the tumors. You select out the tumor.

Dr Spetzler: We do.

When in Doubt, Call in a Genetic Counselor

Dr Marshall: The tumor might have a BRCA mutation, but the germline may not. When we get this test back, we need to remember that the result is for the tumor, right? What would you advise if you got that [a test result showing a BRCA mutation]? Does the patient go straight to a genetic counselor? Do I get him to spit in a cup and send it off to 23andMe? What do I do at that point?

If you see BRCA1 or BRCA2, what you need to do is confirm it with germline testing.

Dr Isaacs: I would recommend, if it's possible, involving a genetic counselor in that process. Some genes are those that you associate with hereditary cancer syndromes. They are frequently mutated in the tumor and they reflect a somatic change. An example is CDH1, which is associated with hereditary diffuse gastric cancer. Most of the times that you see a CDH1 mutation, it is not due to an inherited predisposition but to a somatic mutation in the tumor. That is much more common. If you see BRCA1 or BRCA2, what you need to do is confirm it with germline testing, and I would recommend that you involve a genetic counselor in the process.

Dr Marshall: Any further comment on that?

Dr Spetzler: One component of that is a greater understanding of the underlying molecular data that are being created. Something that might help resolve this without additional cost to the system is an evaluation of the variant frequency of the mutation that is found. If you find a BRCA mutation but it has a 14% variant frequency, the probability that it is a germline alteration becomes very, very low. The risk to the family is now significantly diminished. That might change the decision-making process about when to involve a genetic counselor. That is a level of information that is not typically shared but should be.

Dr Isaacs: I think the point that you are making is that many of the changes that we find when we send off for what we think is genetic testing are, in fact, variants that are of no significance whatsoever. They are benign polymorphisms, and we need to be able to triage that. At present, it is hard to tell from the report what that is, and that is why I think it is important to involve the genetic counselor, so people don't think that they have a mutation. If they had sent it off for germline testing, it would have been reported as a negative result.

Dr Spetzler: I couldn't agree more.

Dr Marshall: One of the most common questions that I get about this is: "I have an ALK mutation in colon cancer. How can I get crizotinib (Xalkori®) and who is going to pay for it?"

I don't have a quick-and-dirty answer to that, other than that some companies will pay. If you ask, some insurance companies and payers will pay for it. ASCO has really taken this on. Richard Schilsky has a very nice position paper[8] on what we would like to happen, which is a partnership between genetic companies; patients; the industry, who would be sponsoring the drug; and regulatory bodies to study the drug. A patient could get crizotinib in exchange for information on the outcomes. The more we build these networks and the more we measure the outcomes, the more likely we can prove that these drugs have activity in these settings.

For the folks out there who are ordering these tests and feeling frustrated because they're getting back answers that don't really help, or those with a fear about payment, I think there is policy change on the way that is going to improve those.

Claudine, David, I want to thank you for taking time out of your busy ASCO schedule to come and talk about this very important subject of molecular profiling. It is my hope, at least, that incorporating this kind of testing will not only improve outcomes but also bring better value—meaning that it will allow us to do a better job, have less toxicity, and have better outcomes for our patients.

Again, thank you for joining us. And thank you guys for joining us in this edition of Medscape Oncology Insights. This is John Marshall, reporting from ASCO 2015.


Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.