.jpg)
Listen to this episode
Episode Transcript:
Translational research is an effort to build on scientific advances to create new therapies and diagnostics that advance patient care. In this special edition of Global Medical Office Dialogues, we talk with some of the brightest minds at the forefront of translating science into medicine. In this episode we’ll hear from noted physician-scholar, Dr. Peter Kotanko, research director for RRI, the Renal Research Institute in New York City. Dr. Kotanko is one of the world’s leading authorities on kidney-disease related research, and the head of RRI’s celebrated annual International Conference on Dialysis, Advances in Kidney Disease. I talk with Dr. Kotanko about the quest to reimagine the future of kidney care by both seeing and focusing on new and evolving science that can benefit patients.
Dr. Maddux
We’re beginning our series today on translating science into medicine and I'm here with Dr. Peter Kotanko. Peter, give me some of your thoughts about how translational science gets into the practice of medicine and how we actually see that connection and how does that work in your mind?
Dr. Kotanko
So, thank you for Frank for having me. Yeah, I think there is several areas in the translational science realm, that are of great interest to us as a company, and to medicine in general and I, I would think that the way how we translate actually insights from those areas into practical medicine, this depends a lot on what we are talking about, what areas are we talking about. So, for example, one obvious area I think where translation into clinical application is very very relevant, it's this whole field of omics. So I'm not just talking about genomics, although I will say epigenomics or transcriptomics, proteomics, metabolomics, and more recently also this area of phenomics. In this area I think it's really a combination of gathering data, say from our electronic health record and combining them with respective biochemical measurements. And in this combination of clinic data and data from these various omics fields can be accomplished by developed mathematical models such as for example machine learning or other areas of artificial intelligence so these areas will allow us to bring together these various areas with clinical application. So I think this is a really, really important process for us to further work on so that these various fields of omics don't exist in isolation, and in the clinical data actually provides a means to connect them, and the technical tools we use to do this will come primarily from areas, such as statistics and NDI I think. Other areas like for example, in mathematics there is very interesting developments going on in advanced mathematical modeling of physiological systems, but also same predictive analytics and again, I think it's really this combination of advanced methods with clinical data that provides this added value in the field.
There is other areas which we have not really touched on that much, but I would foresee will become really important in the years to come and this is this whole field of how climate change actually will impact the way how we deliver care, how it will impact on a fundamental operational level what we are, how we react to this new environmental realities.
Dr. Maddux
So, let's step back just a minute. We talked frequently about the art of medicine and the science of medicine but we frequently don't talk as much about the mathematics of medicine and the physics of medicine. Can you speak just a little bit to the use of mathematics and modeling and the gating steps that have gotten us to the point where we can now actually consider mathematics a core component of how we actually look at developing translational understanding of new scientific facts and translating them into usable components in our clinical care?
Dr. Kotanko
Yeah, I mean, mathematics and medicine aren't actually that different, as they are sometimes I think perceived. Student med schools, when they start learning about, you know, physiology, or start learning pharmacology they get in an implicit way exposed, very soon to the application of mathematics in medicine. Think about for example pharmacokinetics, right? It's mathematical models that at the end of the day, describe how drugs distribute in the body how they are excreted, metabolized and so on. But this area if we stick to pharmacology for a second as a simple example. I mean, this could be married with other really interesting areas such as pharmacogenetics. And this brings this to an entirely different level. And that's something I think that would be of course, great interest to us as a company, because I could foresee a future where we bring together mathematical models that describe the distribution of substances in our patients that they bring it together with pharmacogenetics where we understand the impact of certain genetic modifications on say drug metabolism drug elimination. And then really bring this together also with clinical data that report organ function like patients have impaired kidney function. I mean this is something on top of pharmacogenetics that impacts strike metabolism and this already is brought together by mathematics at the end of the day, because we need to have to take a quantitative approach to really get as an output, something that's actionable. Actionable such as for example a drug level or dosing intervals for tracks. So I think that this is actually a good example where we see how various areas of pharmacokinetic, pharmacogenetics and impact organ function are brought together by mathematics. Other examples are in the end we have done this in the past and we are currently doing this at the larger scale where we develop mathematical models of physiological processes, such as for example, bone metabolism or such as erythropoiesis so the development of red blood cells in the bone marrow, which are very generic models and then, personalize these models. We create to say a mathematical twin of a patient, by using the patient's data that inform the parameters of this mathematical model. So, this has turned out to be pretty powerful in several applications. And it allows us to run for example wheelchair clinical trials. So these are just a few examples but I mean, I could go on and on and think about other examples, but just to give you a quick update.
Dr. Maddux
So let's spend a minute, talking about computational biology for a minute and then we'll go back to looking at how that impacts through the avatars and the twins, that you were just mentioning, and this concept of a virtual clinical trial. I want to try to bring it together so that we get a better picture of how we can actually make it useful for those in practice today. So first let's start with the kind of skill sets that people have, you know computational biology and high performance computing solutions are things that have not always been readily available in our field. And now I think both of them are available and some of the work that your group and, and others have done substantially recognizes how we've been able to use those skill sets to create models that have great impact on today's clinical care. Could you describe that a little bit and describe sort of the skill set that people have that they need to have to do this work?
Dr. Kotanko
Yeah, actually, I think key to success is really that people with very diverse backgrounds actually come together, they learn how to speak a common language, and when they say common language I'm not saying that, say the physician has to understand everything what the mathematician is saying or the computer scientist or that the mathematician needs to study medicine. No but just to be able up to a point to really communicate and understand the problem. So this is I think an indispensable prerequisite. The quality to listen to the other. To the willingness to leave safe territory, you know, that so that in other words the mathematician who may reside sometimes in Ivory Tower steps out, and says okay, I'm now really looking into the complexities of a biological system. So, in general, what are the skills that are required for successful modeling of physiological systems, it's applied mathematicians. So these are mathematicians who really not just know the theory of mathematics. But then, how to develop models of reality. And this again is art and science combined. Then what's, of course, necessary is extensive programming skills, data scientists that deal with large amounts of data that at the end of the day are necessary to inform the models to personalize things. And then of course there is a need for, for people with Medicaid and biology background. And it's really... there needs to be this specific chemistry in addition to the formal skills and. And I mean, I hear that at the Renal Research Institute, we're very blessed in having people out of very diverse skill sets and not with a long standing experience on a different team together across traditional boundaries of fields.
Dr. Maddux
So in reality that computational biology expertise really comes through multiple people on the team. Some of them mathematicians, some of them clinicians, some of them, actual biologists and physiologists. Tell us just briefly, give the example of how we have utilized the in silico twins, that you all have built these avatars, to actually influence clinical care through this concept of a virtual clinical trial. I think our audience would love to hear just your version of what the virtual clinical trial looks like.
Dr. Kotanko
Yeah, so we have applied virtual clinical trials in several settings but let me start maybe with the most developed one. So, this deals with anemia management in our patients. So what the starting point was really that we, and when I say we I mean a set, a group that brings very diverse skills to the table. That we started to ask ourselves, how can we in the language of mathematics, describe the process of erythropoiesis of the development of red blood cells in the bone marrow and the fate of red blood cells in the circulation. And it starts just by reading a lot of literature. What is known about the physiology of that? And then, mathematicians step by step with many many feedback loops in the team, of course, start to build, and to develop block diagrams of models say yeah there is this developmental pathway, and there are these influences just in very a 30,000 feet view of the system. Once the team agrees on the plausibility of this, say such a block diagram, then the mathematicians really start to formulate say transitions from one cell type to another, say from, I don't know from stem cells, erythropoietic stem cells to be a few cells. They start to develop mathematical equations that can describe these processes, and for example our anemia model consists of dozens of equations that describe that process. So this, then we have a sort of a mathematical framework. Now, this is very generic. If I may give you an example it's like, you know, describing oh what's the area of a square? Yes, it's the square of the length of the sides. That's a generic description but then you really would want to know, how can you personalize this? And so in other words, translate this generic description into the description of erythropoiesis in a specific patient. And in order to do this, you take clinical data from the patient such as hemoglobin levels say ES8 doses and a few others. And start to... with a process that's called parameter estimation, that you start to estimate key parameters in that model. Every model has one or more parameters that really define the dynamics of the model. And actually it's again it's work of the mathematicians, in combination with biologists physicians to look into something that's called sensitivity analysis to identify those key parameters that define the dynamics of a model, and once, then through the process of parameter identification, these key parameters are identified for a given patient. And just this whole process of parameter identification gives tremendous insights, I can tell you. For example in the NEMA model, one of those key biological indicators is the lifespan of red blood cells, which as we know and helps us up to around 120 days but then in dialysis patients its every 70 days, and in this process of parameter identification, and this is the process of creating their mathematical twin of a patient, we learned for example, we understand in Mr. Smith, or as just the red blood cell lifespan is 65 days, or 80 days name it, And this, so it goes way beyond that. This is one application of the models we have just started to look into. What can be learned about the patient as such does it provide us, even with new diagnostic windows into the patient's physiology? Okay. Then we have the personalized model that mathematical twin, we also call it the avatar. And then we start to do this, not just for one patient but for thousands of patients, which gives us a whole, a large number of mathematical twins of real existing patients. And then we start to quote unquote “experiment” with those mathematical twins. So for example, we test different regiments of applications of an ESA an Ethnopolysis agent. And then we observe if certain interventions cause a more beneficial outcome than others. And by doing so, we learn what interventions, actually, are better than others. And once we come up with good interventions, good algorithms to treat anemia, then we discuss this of course with the respective people in the company, medical office, and others. And eventually, these algorithms are rolled out and really are used in the care of patients. Now this is the second aspect. So the first one, which we started researching is the diagnostic aspect. The second one is to utilize this research for clinical trials to roll out algorithms and there is even a third aspect that's very interesting because once you have created a true mathematical twin of application, then you can start really personalizing anemia therapy, meaning you can use this mathematical twin, and see what kind of say, ESA dosing would be optimal to bring the patient into the hemoglobin target. And we are currently conducting a prospective randomized trial on that. And I can tell you the first results are just mind boggling. The test shows to me again how powerful such an approach can be.
Dr. Maddux
Yeah, and I think the evolution of this is one that progressively becomes more aligned with our precise personalized care that we're looking to engender. I want to go back and just touch on one point, the original avatars were mathematical solutions to these many differential equations that were built as part of the mathematical model of physiology. And originally the high performance computing solutions required, very long times to actually develop one virtual twin of an individual. My understanding is that either the techniques and the methods or the machinery has gotten much much better over time, and we now have the ability to produce many thousands of avatars in a reasonable period of time. Any thoughts on where that's headed in the next iteration of technology?
Dr. Kotanko
Yeah, that's a very interesting question. I think it's really two things that have accelerated this so greatly. First, it's that we have access to much more by orders of magnitude more computational power. So this is a very important development. The second development is that our mathematicians have fully learned actually to optimize the ways how they identify parameters. One of our mathematicians here has developed an entirely new method for optimization procedures, something that's important in that context. And this method allows us to identify parameters in patients, so that they can be avatorized much faster. So I think it's a combination of on one hand, having much more computational capability, but on the other hand, to become just much smarter when it comes to two crucial things such as parameter identification. Where's this heading to? I think there will be developments in both fields. Of course over the years I mean computational power will increase, and with cloud computing we have even access to larger computational capabilities, but I think also that our mathematicians are just really on to improving algorithms, improving ways how to estimate parameters and actually we collaborate with Department of Mathematics to further improve these, these aspects here.
Dr. Maddux
So, Peter I want to go back to a comment you made about omics, in general, and to recognize that, I think in many areas of whether it's genomics or metabolomics or other things there are epi-omic phenomena, epigenetic phenomena, otherwise phenomena that are driving changes that occur based on our environment and our age and all these other factors. Do you imagine that at some point, we might be able to understand and manipulate the epi-genetic framework that we live in to achieve a certain result?
Dr. Kotanko
Yeah. I'm not that convinced, I mean, that this will be possible in the in the near future. But I know that of course there are efforts underway in that respect and attempts actually in animal experiments have been successful. But, to what extent this is possible in humans, I don't know, and I think that this might be a long way to go. One reason is actually that we have just learned to understand what environmental impact, including by the way also psychological stress has on the epigenetic profile. So, yeah, so I don't think we'll see this in the near future.
Dr. Maddux
Yeah, what made me think about it when you were speaking earlier, is when we think about the lifespan of the red cell. Well, that's the other end of erythropoiesis obviously it's the senescence of cells and the question that came to mind is is the senescence of red blood cell lifespan actually something that is variable in an individual human or is it in fact strictly related to their condition of uremia or their condition of renal health otherwise, and it just strikes me I'm not sure we know whether that's actually variable or not.
Dr. Kotanko
Yeah. Actually, I mean, again such a wonderful question. Thank you, Frank for that. So, we and others are conducting research actually into the determinants of red blood cell lifespan. And we over the years we've published on that and what the current understanding in uremia is that it's a combination of several things that define red blood cell lifespan. It says we have shown clearly that certain uremic toxins exert exvivo, so in bench experiments the stress on red blood cells that has to do with the generation of reactive oxygen radicals that eventually result in an earlier eryptoses so that's the red blood cell death. We went back to patients and saw, indeed, that this also happens there, so that a combination that patients with short red blood lifes will be measured with whereas methods have that this is actually associated with uremic toxicity. So this was one interesting part, the other and I find this extremely fascinating is that we found and published this I guess two years ago or one year ago, that in patients who have a low oxygen saturation of hemoglobin molecule, that in those patients where there is no indication that red blood cell life has been shortened. And when we repeat those or mimic these conditions in the lab by putting red blood cells and the conditions of different oxygen saturation and incubate them with uremic toxins, guess what, when we incubate red blood cell lifespans and the condition of low oxygen saturation, plus uremic toxins, the rate of eryptosis goes up. So, one could think of course, that there are ways to increase red blood cell lifespan. It's not fate. It has to do with, with a specific biochemical medier. And, I mean, once we are able to understand this better, maybe we can come up with strategies actually to extend the red blood cell lifespan in patients. This would have tremendous consequences. So we did some mathematic modeling of that. What would it actually mean for patients to expand the red blood cell lifespan from say 70 days or 60 days to 80 days? The impact on the use of ESA’s would be traumatic. So, this is something we are very interested in. We collaborated with a university in Brazil on that. And I think something we should pursue, can we modify the red blood cell lifespan. Yeah.
Dr. Maddux
We're good. Well, Peter. This has been a great kickoff to our discussion on translating science into medicine. There are innumerable numbers of questions to be answered and activities to try to move forward, how we actually improve care to patients. Thank you for talking with me about this initial aspect of that. And we'll talk soon about some other topics.
Dr. Kotanko
Thank you.