 We have started discussion about how to make contents for high-throughput assays and one of the technology platform which we are discussing is protein microarrays. Protein microarrays without need to purify the protein of interest that is what you heard from the last lecture from Dr. Josh Leibar about his technology development aspect of nucleic acid programmable protein arrays or NAPA. In the same line, let's continue our lecture with our distinguished invited faculty Dr. Josh Leibar to talk about some applications of these technologies. Biomarkers are very valuable for variety of applications which we want to decipher. For example, you want to monitor a drug response, you want to detect a disease at early stage, you want to follow the disease progression or you want to follow how long a patient can survive, you want to see that a disease might recur right. So, there are variety of ways different type of biomolecules starting from the proteins another biomolecules could be used as biomarker to indicate variety of physiological states and therefore biomarkers could be diagnostic biomarkers, prognostic biomarker, recurrent biomarkers and you can name you know many biomarkers for different type of applications but exactly what these biomarkers are and how they can be used especially in the clinical settings in the clinical scenario. Dr. Josh Leibar is going to talk to you today in much more detail about biomarker based assays and of course protein microarrays based applications will follow that how you can use protein microarrays for the biomarker discovery programs. So, in this slide Dr. Josh is going to talk to you about what are biomarkers and what are your considerations when you are thinking about a biomarker discovery program or you want to discover some biomarker, what should be your criteria to determine the sensitivity and the specificity of the biomarkers. He is going to talk to you about some of these basics in today's lecture. Today what I thought I do is take a moment and talk a little bit more about biomarkers both in the context of what they're useful for and also how do you evaluate the quality of a biomarker. So, here's the definition that I put down. So, it's a measurement of some kind with the intention of providing information about a clinical state or a biological state of the organism. So, if you go to a doctor's office and they measure your blood pressure that's a biomarker. They're using the blood pressure to get a sense of your health and they're going to use that to predict the likelihood of getting hypertension or cardiovascular disease. Right? If you go to the doctor and they measure your cholesterol that is a biomarker. It doesn't exactly measure the health of your heart, but it's a predictor of the health of your heart. Right? So, anytime that you measure something, the goal is to predict the outcome and that's important to remember because you're not exactly measuring the outcome. You're measuring a predictor of the outcome and that predictor may not be perfect. Some predictors make mistakes. Right? Okay. So, the type of biomarkers that I work on are proteomics biomarkers. I'm not going to ask you what proteomics is because I know all of you know that. But let me mention that there are two general approaches to proteomics. One is this abundance based approach and we talked about that a couple days ago. That is essentially measuring the amount of particular proteins in the blood or in tissue or in any other setting. And typically on the abundance based approach you measure the difference in the abundance of specific molecules in the healthy state and the disease state. And you look for differences that are biomarkers, that they're predictive of the outcome. Right? Some proteins will be different, but it has no predictive value. It's just random variation. And the job of a scientist is to figure out when those differences are predictive and when those differences are just random variation or sometimes they're a little bit in between. Maybe they're mildly predictive, but not predictive enough to be useful. The other type of marker, the other type of proteomics that we'll talk about is functional proteomics. And that's the kind that I do. That's producing proteins and then studying their function. And once again, we use that to look for biomarkers. Okay, so this is a common issue that comes up a lot in biology. As some of you may know, I'm also the editor of the journal of proteome research. I'm one of the editors there. So I get papers all the time from scientists who want to publish biomarkers. And this is one of the most common mistakes I see all the time in the marker field. So imagine that you're measuring some value of some molecule. Let's say it's a protein that you discovered when you did your mass spectrometry or an antibody marker that you discovered on a protein array. And you see that in the case of the disease, it's much higher than in the case of the normal. And you say, wow, look at that difference. It's the mean value over here is much better than the mean value over here. And therefore, I've got a biomarker. Okay, so does anybody see any problem with that? What? The overlap, that is the problem, right? The overlap, this right here. This particular measurement, while probably significant in terms of the biology, is not a good biomarker. So those two values are statistically significant. There's no question. They will have a very good p value. So you'll be, you know, you'll be tempted to say it's a good biomarker. But let's consider the use of the biomarker. If the value is down here, you could very safely say that that's normal. You could say, if you measure that in a person, no problem, that person's healthy. If you measure the value up here, you can say very clearly, that person has the disease. But as she pointed out, look at how much there is overlap here. If you measure anything in this range, maybe it's disease, maybe it's not disease. The separation is not clean enough to make this a good biomarker. And so that's why we're going to talk a little bit today about how do we define good biomarkers. Okay, so there are lots of different ways to classify biomarkers. So you could definitely talk about the uses. What other types of ways to classify biomarkers? What? Yeah, what you're measuring? What the material is you're measuring? So what type of molecule? Absolutely. What other, so that's the type of molecule? What any other? What about the source of the molecule? Is it blood? Is it cerebral spinal fluid? Is it urine? So where you get the molecule is also you can you can be looking at DNA protein or or or lipids. So that's the type of molecule, but you could also be looking at blood markers or urine markers. And then the last thing I would say in terms of classifying markers is the level of validation. So that's what that's what's shown here. So you can classify them by what you're going to use them for. Are they prognosis, prognosis, diagnostic, that sort of thing. You can talk about where you're going to get, you know, where you're going to get the material, what the type of material is, and then how well validated it is. There's also a level of validation for the biomarker itself. And we'll we'll come back to that. So here are some of the clinical uses that we have for biomarkers. Okay, so let me let me remind you what they are. So one of the one of the applications of biomarkers is what's called disease stratification. And what that means is there are times when you have a particular disease, and you you know that the population of individuals who have this disease are not all alike, right? Some of them have one subtype of disease, and others have another subtype. Does any know any no examples of cancers that have different subtypes? Breast cancer is the classic, right? So breast cancer has five or six known different subtypes. They're they're classified based on the molecular classification, looking at different surface markers like the estrogen receptor, progesterone receptor, the HER2NU receptor, but also looking at other genes that are or expressed or not expressed in the tumor. And based on that gene pattern, the different subtypes of breast cancer will have different prognosis. It will those different subtypes will respond to different therapies. And so the markers, the gene expression that you look at, those are biomarkers that help you stratify patients. And that turns out to be very important because by stratifying patients into different groups, now you know better how to treat them and what to tell them in terms of expectations for their disease. Okay, what's a companion biomarker? Anybody hear that term before? So anybody familiar with the drug Herceptin? Herceptin is a drug that we use to treat people with breast women with breast cancer, who have HER2NU positive disease. So why is it important that they have HER2NU positive disease? That's right. So Herceptin specifically targets HER2. So if there's no HER2 on the cells, Herceptin will probably not be very effective. So when the drug Herceptin first came on the market, it was a revolution. It was the first drug in the modern era that was based on specifically targeting a biochemical pathway that we knew was related to cancer. All the drugs before that were basically toxic chemicals that kills dividing cells. Cytoxin, Adriumycin, all those drugs, they kill cells because they're dividing. They were not selective for cancer pathways. Herceptin was the first molecule that was developed to target a particular pathway. And yet when they gave it to women with breast cancer, the response rate was probably 15%. So that means that when they gave it to 100 women, 15 of them would have some response, 85 of them would have nothing. And of course this was a drug that was going to cost in the US between $50,000 and $100,000. So they're taking a very expensive drug and 85 out of 100 women would get no benefit at all. And probably the bigger issue is those women would be waiting and waiting to get a good therapy while they were on a therapy that was doing them no good. So what did they do? Well, someone reasoned maybe we should only give this drug to women who have her two new positive cells. Because if they don't have her two new positive cells, they probably won't respond. So if you now do a diagnostic test and look only at women who have her two new positive cells, what do you think the response rate is there? Well, over 60%. So you went from almost nobody responding to more than half of women responding. So all of a sudden the drug companies and the insurance companies said, look, before you give Herceptin, you need to first test for her two new positivity. And only women who have her two new positive are eligible to get the drug. And that way they could have a much better chance that this drug would be effective. These days with Herceptin and other agents that specifically target the her two pathway, women who have that disease who at one time had one of the worst outcomes in cancer now have one of the better outcomes in cancer. Because those drugs are so selective for their specific subtype of disease. So a companion marker is a blood test that you give together with the plan of giving a drug to determine if that patient will respond to that drug. So it's specifically designed in these days, at least in the US, if you're going to develop a new therapeutic for any type of cancer that targets a pathway, the FDA usually requires that you have a companion diagnostic. You have to have a test that will specifically tell that patient that they're likely to respond. Okay, what's a prognostic marker? It tells you the likely outcome of how aggressive the disease is and whether or not it's going to, you know, what you can expect from the future. So that's basically right. So generally speaking the prognostic markers are not related specifically to therapy, although sometimes they are. But the idea of a prognostic marker is, you know, when you see a patient who has a disease, oftentimes one of the main question that they're going to ask you is, what can I expect? What's my outcome? How am I going to do? And a prognostic marker is intended to tell you that. Okay, what's a disease progression marker? It's a little bit related to this, but it's not quite the same. So once you have the disease, doctors will order that test every time they see you and they will look at that test and say, how are you doing? How is that disease coming along? Are you responding to the therapy? So in the case of cancer, which is what I know best, if we have a patient who has a liver cancer, for example, we might, every time we see that patient order a CEA test, carcinoembryonic antigen. This is a protein that's produced by liver cancer cells or metastatic cells in the liver. And when the tumor is growing, the CEA level goes up. When the tumor is shrinking, the CEA level comes down. And so the doctor will monitor that level, maybe give drug, monitor it again, and use that as a biomarker to tell the doctor how the patient is doing over time to track the disease. So we call that disease progression marker and they're very useful. I mean, in the clinic, we use them for diabetes. Hemoglobin A1C, if you have ever heard of that, that's a disease progression marker. It monitors how the disease is doing, how well is the insulin therapy doing. CEA, in the case of cancer, lots of diseases have these markers that let you know how the patient's doing. Okay, early detection marker, that's kind of obvious from the name. So an early detection marker, the goal is to find the disease very early, usually because you believe that by catching the disease early, you have a better chance of having a good outcome. So the argument is if I can catch the disease at its early stages, then I can treat it early and have and reduce the likelihood of, you know, disease spread or bad disease outcome. Now the tricky thing about early detection biomarkers, if you think about it, is who is the population you're going to use the early detection markers on? Who are you going to give it to? So you could give, so that would be one way to do it. People have a family history, that would be a great idea. Implicit in what you're saying is that you're using early detection markers on healthy people or apparently healthy people, right? So all these other markers that I've talked about before, right, disease stratification, companion markers, prognostic markers, disease progression markers, those are all people who are already in your clinic and they're already sick. And your job is to use these markers to tell you something about the state of their illness. But an early detection marker, that's something that you give to a healthy person to see if they're sick. So it's a very different kind of marker. It also has a lot of implications in terms of cost and usage because if you're going to give a marker to healthy people, right, you need to know that it's a pretty good marker. Otherwise you can cause all kinds of trouble and we'll come back to that in a little bit. Okay and then risk marker. What's a risk marker? Someone back there, I didn't get it. So you could predict the risk of the disease. So the difference between both five and six are markers that you would do on healthy people. The difference being that this marker doesn't really find the disease, it just tells you the likelihood that you might have it. Whereas this marker is really intended to find the disease. You have it right now. If you do a pap smear on a woman, you're looking for the presence of neoplasia right now and asking does she have, you know, the possibility of cervical cancer. If you do a risk marker like a BRCA1 test, you can't say that the person has breast cancer. All you can say is that she has an increased likelihood of getting it. So a genome, getting your genome sequenced, that's a risk marker. That's not a disease detection marker. So certainly, you know, I'm a big believer in this early detection biomarker, mostly because I think it will have important outcomes. And I think I've shown you this already. Just to remind you, what we're looking at here is these are the four most common diseases, cancers in in the U.S. at least. They're all epithelial based cancers. They are the big, big killers of people in our country with cancer. These are survival plots. So the percent of people surviving based on stage of disease. Stage of disease basically tells you how early you caught it. And it starts with the latest stage disease and moves to the earliest stage disease. And what you can see is if you catch the disease early, you have very good survival rates, except maybe lung cancer, which never has a good survival rate. But nonetheless, the survival of early stage disease is always better than survival with late stage disease. So this is an argument for why we want to catch the disease early. Okay, so I mentioned that you could look for biomarkers based on the source of material. And these are just some of the types of materials that you could use to get biomarkers from. Certainly blood is a popular one. Tissue is a popular one. Well, I should say blood is popular. Sputum is good. Urine is good. Tissue, really only in the case of cancer. You don't often do biopsies. There are a few other diseases where you might do biopsies. But you know, most patients are not thrilled about having parts of them cut out. So don't do that often. But all of these, in one way or the other, have been used as one form of biomarker for people. Okay, and then we mentioned earlier that you can look at, so that was a source of material. Now we're looking at biomarkers by type. Oh yes. Yes. So interesting that you asked that. If you ever, in fact, some of the most famous historical poisonings, right, have been determined by looking at hair and nail, because arsenic shows up in hair and nail. And it's dated. Where it occurs along the nail or along the hair can tell you how long ago the person had arsenic. And so I forget which European monarch was murdered by arsenic poisoning, but they went back to the corpse and measured it. And you can actually determine arsenic in the hair and nails. There are probably other things you can measure there, too. But that's the most famous one. Yeah. Okay. And then so these are, these are types of things that you can measure in sources from all those places. So obviously, you know, all of us are proteomics people, so we're going to be interested in number one, which is why it's listed number one. But you could look at metabolites. You could look at DNA or RNA or other nucleic acids, long chain RNAs or whatever. You could look at glycoproteins. You can look at cells themselves. Certainly, if you look at the immune response, there are two things you can measure here. So in the case of B cells, you would look at antibodies that are in the bloodstream. And that's what NAPA does. NAPA looks at antibodies in the bloodstream. You could also look at T cells. T cells are much harder to look at in high throughput. T cell, the classic T cell assay is called the LE spot assay. And what you do is you have to present the T cell with a presenting cell and an antigen. And then you have to measure the secretion of gamma interferon or something like that. It's a complicated assay and you can't really do it at the scale of omics. And then of course images are also a type of biomarker, right? So x-rays, CT scans, pet scans, all those various technologies. In a sense, they are also measuring things and you can imagine that some of those imaging studies themselves could be early detection biomarkers. In the case of breast cancer, mammography is an early detection biomarker, right? Nowadays, in the U.S., there is a recommendation to do spiral CT scans on very heavy smokers as an early detection biomarker for lung cancer. Okay, but no matter what test you do, this has to be true. It has to be robust and reproducible. Alright, so now I, the last thing I want to mention, so we talked about the use of biomarkers. We talked about the source of biomarkers. We talked about the type of biomarkers. And remember I said the fourth way to classify markers was their level of validation. So let's talk a little bit about what I mean by validation. Okay, so the first thing you have to do if you want to validate a biomarker is you need to define how you're going to use it. So the key is understanding the quality of the marker and what it needs to do determines, is determined by what you're going to do with it. If you're going to use a biomarker for early detection, then you have to remember I'm going to use this biomarker on healthy people. So these people are walking around living their lives and everything is fine. I'm going to do a marker on them and suddenly I'm going to tell them that they may have a disease and now they've got to go to a doctor, they've got to get a biopsy, they've got to do a test, they've got to scan, whatever. All of a sudden you're going to cause them to do a whole bunch of stuff. So you need to know that that marker is really robust or else you're going to cause a lot of trouble. On the other hand, if you already know the patient has cancer and all you're doing is trying to monitor how it's doing relative to the drug, maybe that marker doesn't have to be quite as specific because you already know that they have cancer. That's not a question. All you need to know is the level of the cancer. So the level of specificity for that is not as high as for early detection markers. So the first thing you have to do is define the clinical usage and everything else will follow from that and it's the number one most common mistake made by people who go for biomarkers is they decide they're going to get a biomarker but they never stop to think about what they're going to use it for. I've seen people come up with markers for things that have no value because no one would ever look for that so you have to you have to think about that. So in conclusion I hope you have learned now some basics of biomarker. Of course you know there are a lot of changes happens in the physiological states of any individual and then technology platforms are very robust but still if you think about the you know measuring the slight perturbation slight changes the technologies can have some noise which may appear to us you know these are some changes which are happening from the biology induced you know from the samples. So how to really determine that what small changes we are measuring they are real and these are real biomarker candidates. Determining them is you know really challenging and that's why I think there's a lot of emphasis globally to discover biomarkers and bring new biomolecules for the clinical assays however still our success has been limited but if you follow the basics which is discussed today in the lecture by Dr. Josh LeBan I'm sure our efforts of following up and making success of a biomarker program will be very valuable. Thank you very much.