Transcription:
Functional Genomics for Precision Medicine in Rheumatoid Arthritis and Scleroderma

Melanie Cole (Host): Welcome to Better Edge, a Northwestern Medicine podcast for physicians. I'm Melanie Cole and today, we're talking about Functional Genomics for Precision Medicine in Rheumatoid Arthritis and Scleroderma. Joining me is Dr. Debbie Winter. She's an Assistant Professor of Medicine and Principal Investigator of the Winter Lab of Functional Genomics in the Division of Rheumatology at Northwestern Medicine.

Dr. Winter, it's a pleasure to have you with us. And I'd like to start, before we get into your studies and research, which is fascinating by the way; can you tell us a little bit about yourself and how you got interested in this particular topic?

Deborah (Debbie) Winter, PhD (Guest): Yeah. Actually, I came here somewhat circuitiously. I started off being trained in Computational Biology and Bioinformatics where I got a general training on different quantitative approaches, statistical and ways of handling big data to understand biological questions. I did my PhD at Duke University in Computational Biology, looking at basic gene regulation and focusing mostly actually on human cell lines.

But then I was more interested in understanding how the ways genes are regulated actually contributes to biological function. And so I did my postdoc at the Vitea Institute in Israel, where we applied these similar techniques to understand how immune cells are specified and differentiate and adapt their function in different conditions. Particularly I focused on macrophages. And which are present in nearly every tissue of the body, where they perform specific functions, but then also have this role in inflammation. And in studying that, I got really interested in how their normal desirable functions get perturbed in disease.

And that's what brought me to Northwestern and the Division of Rheumatology where I could understand the role of macrophages in diseases like rheumatic diseases, rheumatoid arthritis, scleroderma and sort of come at that question from this new perspective of functional genomics.

Host: That's so interesting. And thank you for sharing that with us. So, can you provide an overview of the ongoing research that you're doing in the field of functional genomics? How is the Winter Lab of Functional Genomics, which is very cool; you have it named after you, at Northwestern Medicine, bringing precision medicine to rheumatic diseases?

Dr. Winter: Yeah, So I think what I want us do is start by unpacking that phrase. So, we're going for this precision medicine based approach, which basically is the intersection of personalized medicine and evidence-based medicine. So, we hear a lot about this in the cancer field, the idea of personalized medicine, instead of just treating every patient the same, let's cater our treatment to them. And that's clearly an important aspect when it comes to very heterogeneous diseases like rheumatoid arthritis, for example. But then we also want to combine this with evidence-based medicine. We could hypothesize just looking at individual patient, what might be best for them, but better than that, is to collect a vast amount of data and use algorithms and different approaches and validations to determine how you go from that person's individual profile, how they fit into what we've seen previously, what they're most likely to respond to, how the disease is most likely to progress.

And then the other side is the functional genomics, and this term is meant to contrast with, for example, genetics, which is just based on the DNA sequence that we're born with. That DNA sequence is static. It stays the same through your entire life. So, it can't be used necessarily to map disease progression or disease response or changes over time. Also what we've learned, particularly in these complex auto-immune diseases, is that it doesn't account for much of the variation that we see or much of the susceptibility to these diseases. It only accounts for a small proportion and the other part, comes in from the environment and, and what you're exposed to in your life and how you live your life. That's why twins who they have the exact same genetics, identical twins, don't both end up getting the same diseases. And so in order to account for this; the combination of genetic and environmental factors, we want to go to the next level, which is what your cells are actually doing, how they take their genetics, which is basically the programming and the influence of the environment and act out their function.

And the important thing when we do this is we can't take a whole tissue. We can't take all blood. Or an entire piece of a lung or something like that and get a sense of what, what is actually happening. We need to focus in on specific, a specific population or several specific populations and understand how compared to their normal function, they are now functioning in a different way. And we do this by looking at how the genes are regulated differently or how the epigenomic programming is changed. And this can give us a better sense of the disease in that moment. It can change as the disease progresses and with disease response and vary between people. And this allows us to get a better picture of the disease at that moment.

Host: Wow. So Dr. Winter, as you're telling us your most recent studies and findings in rheumatic diseases, using these functional genomics approaches, explain the various types of data that may be obtained from patients. Give us an overview, what you're doing and how you're doing it.

Dr. Winter: So, the two main types of data that we consider to be functional genomics would be gene expression data, or transcriptomics and chromotome state, or epigenomic data. And these data types can come either as what we call bulk data. So based on many cells that fit a category like macrophages. So, RNA Seek would be an example of gene expression of macrophages, or it can come in a single cell, a type of data where we have information on the individual cells that make up a population and both data have advantages and disadvantages related to their cost and the types of questions that you might want to address. An example of our use of bulk RNA Seek is our study in rheumatoid arthritis. In our studies in rheumatoid arthritis, we rely on a novel approach called minimally invasive ultrasound guided synovial biopsy where basically a little needle is injected into the joint of a patient with rheumatoid arthritis and we extract tiny pieces of tissues. We actually do multiple passes to get a sampling of tissue from the joint and then this sample can be processed for RNA Seek. And in a study that we published a few years ago, we showed that you could use whole tissue, which would be it all the different cell types that are found in the joint versus just subsetting out the macrophages in that tissue, and we found that there was a lot of information that would be missed when using a whole tissue versus individual macrophages, for the reasons that I described earlier. When we're using macrophages, we can have a better idea of how their function is changed.

Whereas when we have a mix of multiple cells, it becomes a lot harder to understand what's gone different. We might be able to list descriptively what genes are changed, but to understand which cells have changed their expression or how that has changed their function is a lot harder. So, in this study where we ended up focusing on macrophages, we showed that by doing gene expression profiling of the macrophages in different patients, we were able to identify modules of co-regulating genes and then basically classify each patient by their expression of these modules. Then you could use these modules to correlate with disease and clinical characteristics like disease severity or disease activity as measured by DAS 28 or as measured by the number of tender joints that the patient exhibited. And what we found is that the research and modules that correlated with disease severity in this sense, and other modules could correlate, for example, what treatment the patient was on. And this was a pilot study that we've been working the last couple of years to expand in a larger study, where we will specifically get patients before they're assigned to a new biological treatment and after, and develop an algorithm, by which we can predict which patients will respond to which treatment. And we're going to build this algorithm piecemeal from people on different treatments. But the final goal is that a patient could come in, we would run this algorithm on them and it would tell us which treatment they're most likely to respond to. Another example of a study we're doing in scleroderma or systemic sclerosis is exciting because we're actually using cells that can be extracted from the blood of patients.

So instead of undergoing any complicated procedure, we can just look at their blood and this is particularly important in scleroderma because so many tissues can be affected in this disease, the lung, the kidneys, the skin. It's hard to get one answer about the status of a patient or their disease heterogeneity. So in this study, we've collected from patients and controls and these patients have early sclerderma. And we've isolated particularly from their blood, the circulating monocytes, which play a role in inflammation and fibrosis. And what we found is we can actually use these profiles to create disease subtypes, or transcriptional subtypes of these patients and these subtypes will exhibit sort of different disease characteristics. And we're still expanding this study to more patients. And we want to learn more about the implications over time as their disease progresses and what these subtypes mean for the patient's future.

And then once we have completed this study, the idea is that we will have then identified disease subtypes that should be, that should be treated differently and this will inform clinical decision making in the future.

Host: So that segues so nicely, Dr. Winter into this question of how it translates to patient care, the potential applications to precision medicine. How can rheumatologists evaluate the quality of functional genomics data and determine its relevance to clinical decision-making?

Dr. Winter: Yeah. So there's two parts to this. If someone wants to evaluate a paper, they see that describes the study versus someone, someone performing the studies themselves. So, so, I guess let's start with the question of let's say I want to study functional genomics in my patient population, and I'm going to profile them, do gene expression profiling, for example, and collect this data. How can I, how do I know that I have good data? And it's actually a very complicated question and we spend an inordinate amount of time assessing the quality of the data we get to make sure that it's going to be suitable for the type of studies that we do.

We assess many parameters from the quality of the cells that are obtained to the quality of the DNA that comes out, to the quality of the sequencing, to the quality of the of the data itself. And whether we believe that each sample represents a true reflection of the gene expression. And so it's a multi-stage process. And it's a big concern, both for generating the algorithms and testing the algorithms because dips in quality can lead to technical variability and we don't want our algorithms driven by technical variability. We don't want to think that a certain gene is expressed lower in someone when actually it just wasn't picked up properly by the technology.

So a lot of spent on assessing these. And so if you're on the other side of things, if you're presented with a study claiming that they have developed an algorithm that can be used in practice in a real life situation, you want to first understand the quality of the data that was being used to generate the algorithm. The next question is the algorithm itself. Has it been properly trained and validated and the golden standard for these kinds of algorithms or other machine learning type applications is, has it been trained on a vast representative data set that reflects the type of examples that we going to see and optimized appropriately and assessed for how well it works and then has it then been tested on a completely independent data set and shown to be performed while in terms of both specificity and sensitivity. So, sensitivity is its ability to pick up the positive cases and specificities is ability for not to pick up the negative cases for whatever your question is.

So there are multiple levels of quality control that go into both developing such algorithms and such frameworks, as well as for assessing the use in your application. You want to make sure that the situation in which you're applying such an algorithm matches the situation in which it was tested for.

Host: Wow. As we wrap up Dr. Winter, what's next when it comes to this area of study? Tell us about any studies or clinical trials that you're involved with at Northwestern Medicine and what you want the main takeaway from this fascinating episode today for other providers.

Dr. Winter: Yeah, I think a lot of what I talked about today involves using the clinical resources we currently have. So, trying to decide how to of have the strategies that are already available, how to treat this set of patients versus that set of patients or trying to predict, given the treatments we already have, which one, would be best applied to a particular patient. And this is a huge problem. Because we have to know it's the best. It's a better use of our resources, it's a better use of our money if we can go straight to the right solution. And it's a better patient experience, we're more likely to appropriately treat the patient and appropriately address their symptoms and improve their quality of life and stall disease progression.

But I think what was really on the horizon, what the next step is developing new treatments and new therapies to supplement what we already have. And some of the technologies that might be valuable for that direction are gene editing, approaches like CRISPR. And it really, I should be saying genome editing because rather than necessarily knocking on or turning off an individual gene, using the same functional genome generating the goal would be to edit the way that genes are regulated in the way they're programming, so that we can really fine tune our treatment. So, instead of saying a certain gene say, TNF is a problem, let's turn down TNF which will turn down inflammation, but we'll also compromise the ability of the patient to respond to challenges like infection. What we can do is say, well, we actually really only need to turn TNF or downstream factors in the TNF pathway. We really just need to address the downstream factors. And therefore, if we can tweak a regulatory region here or a regulatory region there, we can actually affect the cellular function without affecting the systemic ability of the body to respond. And then this type of targeted treatment, I think is really the next step for treating these complex auto immune diseases by specifically regulating cell function, as opposed to having a more systemic effect, which can have many off target and negative consequences.

Host: What an exciting time in your field, Dr. Winter, you're doing some really incredible research. Thank you so much. And I hope you'll join us again and update us as things change and progress. To refer your patient or for more information, please visit our website at breakthroughsforphysicians.nm.org/rheum. That's R-H-E-U-M to get connected with one of our providers. That concludes this episode of Better Edge, a Northwestern Medicine podcast for physicians. I'm Melanie Cole. Thanks so much for listening.