Bench to Bedside: Asthma Heterogeneity, Part 1 – Prescott Woodruff
30
August

By Adem Lewis / in , , , , , , , /


Hi, my name is Prescott Woodruff. I’m an Associate Professor of Medicine in the Department of Medicine at the University of California, San Francisco, in the division of Pulmonary and Critical Care Medicine. And I’m going to be talking today about molecular phenotyping of asthma. This is intended to be the first part of a two-part talk on asthma. The second part will be given by Joe Arron and as a companion piece to this talk. And here’s an outline of what I’ll talk about today. I’m going to talk a little bit, first, about background information on asthma, then about molecular phenotypes of asthma, and then finally the rationale for blood biomarkers in asthma. And you’ll see how this leads into Joe’s talk when you get to Joe’s. So, first, by way of background, asthma is a chronic disorder of the airways of the lung. And it’s chronic, but it has episodic symptoms. Those symptoms include airflow obstruction, and then wheezing and shortness of breath. But, underlying that episodic disease is chronic… a chronic problem. And there are several aspects to this chronic problem. One is airway hyper-reactivity, that is, that the airways are more sensitive to stimuli and constrict more easily to stimuli than in patients without asthma. In part, this is due to abnormalities in the airway smooth muscle, which surround the airway. Another major feature of asthma — as we know it for the last 10 or 20 years — is that it involves inflammation of a specific type: a lymphocytic and eosinophilic inflammation of the bronchial mucosa, of the tissue that makes up the airway. And this is an important part of this talk, because this understanding that asthma is an inflammatory disease, and an inflammatory disease with lymphocytes and eosinophils, has been the foundation for therapeutic approaches that are anti-inflammatory in asthma. And has led to a lot of advances in asthma therapy. Finally, one of the other chronic problems in asthma is that there can be airway wall remodeling, that is, changes to the structural cells in the airway, probably as a consequence of the inflammation. But that this airway wall remodeling may perpetuate airway narrowing and symptoms in asthma. I want to say a few more things about the inflammation that underlies asthma, and you probably can’t see all of the detail in the slide, but I’ll point it out for you. The inflammatory problems in asthma begin with the airway epithelium, which, when barraged by a variety of environmental insults, can produce mediators that activate the immune response. And the way it activates the immune response involves both activation of innate cells, here, and adaptive cells, such as T cells. At the end of this process of immune activation, however, the production of T helper type 2 cytokines from T helper type 2 cells is really thought to be one of the centrally important processes that cause those chronic abnormalities I described in asthma. There may be other sources of these T helper types 2 cytokines. Here’s one — IL-13 produced by natural helper cells. And IL-13 could also be produced by basophils and other innate cells. But, from one source or another, these T helper type 2 cytokines, which are IL-4, -13, and -5, in particular, are likely to cause the end-organ problems that we see in asthma, especially IL-13. Now, because asthma is a disease of smooth muscle abnormality, where the smooth muscle is twitchy, and inflammation, most of the pharmacological therapies for targ… for asthma target one or the other of these processes. Most of the old-time medications and those that are still currently used to control symptoms are bronchodilator therapies. They are beta adrenergic agonists or anticholinergic medications that target the receptors that mediate smooth muscle contraction. And their use relaxes the smooth muscle and opens up the airway. But for the last 20 years, we’ve recognized that asthma is an inflammatory disease, and have developed effective anti-inflammatory therapies for asthma. And when you use these on a daily basis, you can prevent the symptoms and reduce your need for bronchodilator therapy. So, those drugs include the inhaled corticosteroids, the leukotriene antagonists, and also an anti-IgE monoclonal antibody. These are all drugs that could be used in pharmacologic therapy for asthma, and currently are used with great success. They could be used as controllers, for example, the anti-inflammatories that I described, or as relievers, when the bronchodilator therapies are given as a short-term preparation, or they could be used in combination, with both anti-inflammatory therapies and bronchodilator therapies given together in a single preparation, and those are commonly used therapies for asthma. Now, the way these are used in asthma is these therapies are used in an empirical fashion. That is, asthma is treated empirically, according to clinical severity and responses to treatment, and not according to the underlying biology. This figure gives you a sense for it. In this figure, there are steps to care, where each step involves the addition of a new medication that is added when a patient has continued symptoms or is perceived to have not responded to that prior step in therapy. And usually what happens is you start with bronchodilators and then add inhaled corticosteroids or other anti-inflammatory medications. You may then add longer-acting bronchodilators. But ultimately, you may get to a step which requires even more significant, and perhaps expensive, biological therapies for your asthma. And this is an area that is of course of interest to pharmaceutical companies, but all these areas, all these steps in asthma care, to me are quite interesting. Because these steps are taken on the basis of symptoms and not on the basis of the underlying biology. And it’s possible with that, whether you’re going up and adding therapies or deciding when to subtract therapies, that understanding that biology could be really helpful in making those decisions. So, why does it matter that you might understand the biology underlying a patient’s asthma in order to treat them? Well, here’s just a general way of thinking about personalizing medication use to the underlying biology. When we put on our hat as scientists, we study, typically, molecular mechanisms. We study molecular mechanisms because if they truly reflect the cellular pathological and physiologic features of a disease that make up the clinical phenotype of a disease, then that will be extremely powerful. Because targeted therapies developed against that molecular mechanism will effectively treat people with that disease. And that is extremely powerful in our modern approach to medicine. On the other hand, physicians often recognize… recognize that patients are heterogeneous, that people who we label as having asthma may represent groups of people with distinct clinical phenotypes. They may differ in subtle ways. What we don’t often understand is the reason. But if that reason is because there are differences in the underlying cellular pathological and physiologic features, and that reflects distinct molecular mechanisms, then the drugs we will have developed to target the first molecular mechanism will fail to work in a significant percentage of patients. And that seems to be happening in asthma and could be happening in other complex diseases. This section, then, is about the molecular phenotypes of asthma, that is, how could you then start to subset patients based on their underlying biology? Let me get back to this figure, then, on the Th2 cytokines in asthma, and remind you that we were saying that T helper type 2 cells and T helper cytokines are important in the effect of inflammation on the end organ, in this case, the airways. And let me remind you that interleukin-13 is one of those really important T helper type 2 cytokines. One of the reasons it’s important is that if you overexpress interleukin-13 in the mouse you can generate a lot of the pathological changes that you see, and that constitute airway remodeling in humans with asthma. You can see mucus cell metaplasia, so there’s more mucus in goblet cells, and therefore more cough and sputum production. You can see subepithelial fibrosis. And that could lead to narrowing of airways on a more chronic basis. In addition, interleukin-13, when it interacts with its receptor-bearing cells, can cause the recruitment of eosinophils, ultimately, through the production eosinophil… eosinophilic chemotactic proteins, such as eotaxins. And this cytokine and others can affect IgE class switching, so that immunoglobulin E is produced, which exacerbates asthma by arming and activating mast cells and basophils. So, interleukin-13 is thought to be one of the really important T helper type 2 cytokines that underlies inflammation in asthma. Based on that understanding, my mentor at the time, John Fahy, and I designed a clinical study to understand how interleukin-13 or other Th2 cytokines affect epithelial cells, which are resident lung cells in the airway. And what products are produced by those epithelial cells? Because we had the model in our head… heads that these products could influence the mesen… mesenchymal cells to induce remodeling or every hyper-responsiveness, and the phenotype of asthma. Now, actually, this model is really oversimplified. And it’s even a little out of date. But we designed a study to investigate it and that led to the discovery of certain biomarkers of IL-13 in the airway. And here’s the study we designed to investigate it. This was a human study. We were interested in human asthma. And we were trying to identify epithelial genes that are specifically differentially expressed in asthma and were responsive to steroid treatment. We wanted to do that because we wanted to find epithelial genes that could possibly explain the relationship between inflammation and the symptoms of asthma. And so we wanted to find those that were actually sensitive to steroids and, when blunted by steroids, correlated with improvements in symptoms. So, to do this, we studied three groups of patients. We studied patients with asthma, we studied smokers as a disease control, and we studied healthy controls, in a bronchoscopy-based study, where we actually did fiber-optic bronchoscopy and obtained epithelial cells from the airways in these patients. We obtained some other samples too. We obtained bronchial biopsies, for example. And then, after bronchoscopy, we treated patients in a blinded fashion with either fluticasone, an inhaled corticosteroid, or placebo, and repeated bronchoscopy. We also repeated lung function testing, both at the beginning of the study and after eight weeks of the inhaled fluticasone, so that we could assess who was responding to the therapy. To cut a long story short, here are the results of that initial study. We used gene expression microarrays and applied those to RNA from the epithelial brushings, and asked whether we could identify genes that were differentially expressed in asthma, either induced or repressed, and whether there were genes that were steroid-responsive, and then sought the overlap, the intersection of those two gene sets. We used a very stringent criteria because, frankly, we didn’t want to have a list of dozens of genes to follow up on. We really wanted to have a few. And we found three that really rose to the top of this analysis. Those three were CLCA1, periostin, and serpinB2. It turns out that the first one was already known to have some role in the production of mucus in asthma, through mouse models, and is still the target of study by investigators. Periostin, as we’ll talk about, is a matricellular protein produced by epithelial cells and also likely plays a role in the pathogenesis of asthma. And serpinB2 is a serine protease inhibitor. Its function in the airway, and even its cellular function, is still somewhat debated. Our original goal, of course, was to study these in a cell biological manner, and to study their function and role in mouse models of asthma and then in human asthma. But after meeting Joe and being involved with a project that had a very specific goal, we started to see these three genes as having another purpose in our human study. So, these three genes, as I’ve said, periostin, CLCA1, and serpinB2, were induced in human airway epithelial cells in vivo in asthma. It turns out that in vitro, in our cell culture studies, we found that they were also specifically induced by interleukin-13, and very highly induced. And so we realized that they were a good biomarker of the effects of interleukin-13 in the human airway, in a diseased state. And we found this to be very valuable, because it’s actually very difficult to measure interleukin-13 or any of the other Th2 cytokines directly. And so having a biomarker of their effect was something that was new and quite practical. Here’s a little bit of the data. When you look at it, again, for these three genes from these epithelial brushings, you can see in each instance that periostin, here, CLCA1 and serpinB2 are each highly induced in asthma. But they’re not induced in all asthmatics. There are some asthmatics who are… have low-level expression of these genes. And you can see here, in these three panels, that all three genes are highly intercorrelated. And this observation suggests that there is a specific subset of people — these people, here — who are high for all three genes, and may be different biologically from these people, who are low for all three genes. Another way to look at this is to do a hierarchical cluster analysis of the three genes. And here the genes and their expression levels are represented in the rows, with red being high-level expression and blue being low-level expression. And the columns represent the individual subjects in our initial bronchoscopy study, those that were either asthmatic or healthy. It’s a little hard to see, but these letters are A for asthma and H for healthy. And what this shows, at least in a descriptive way, is that there’s a group of patients who are very high for all three genes, and that all of those people have asthma. And that’s about half of our asthmatics. But in this big block of people who have low-level expression of these three genes, we have intermixed asthmatics and healthy controls. So that about half of our asthmatics were low for those three genes, and indistinguishable from healthy controls. And this was the observation that led to the idea that there really might be two distinct groups, based on the expression of IL-13 or other Th2 cytokines in the lung. We have some data from this initial data set that corroborates that. This is PCR data from endobronchial biopsies taken from the same subjects. And if you get a big enough piece of tissue, you can do PCR for the Th2 cytokines. And what we found is in that red group, which were high by inference, they really did have elevated levels of interleukin-13 and interleukin-5 at the transcript level in their biopsies. And so we started to call these T helper type 2, or Th2, High asthmatics. Now, the clinical features of these patients were interesting. In certain ways, both Th2 High and Low patients with asthma had similar features. They both had abnormal lung function — that’s the far box, which is a lung function measure. They both had reversibility in their airflow obstruction to albuterol or a beta agonist — that’s the middle box. And both groups had more skin prick test reactivity to allergens than healthy controls. And so, in that sense, they both had clinical features that we usually associate with asthma. But in looking at some of their physiology and their inflammatory features, we could find ways in which they started to differ. Th2 High asthma, the far-right group in each figure, had exaggerated airway hyper-responsiveness. And this is a plot of responsiveness to a drug called methacholine. And the lower this plot, that is, the less methacholine that is required to induce a response, the more sensitive you are. So, you can see, even though both groups are different from healthy controls, Th2 High asthma has greater responsiveness. Both groups had elevated IgE levels, but it was higher… higher in Th2 High asthma. Both groups had blood eosinophils, but, again, it was even higher in Th2 High asthma. And finally, when we got to the lung and looked for eosinophils in the lung, we found that, really, only T helper type 2 High asthma, or Th2 High asthma, had eosinophils in the lung. And we realized that this was a molecular counterpart to a cellular categorization of asthma that had been proposed in the past. And that eosinophils were part of this phenotype. I can summarize the features, then. Both Th2 High and Low asthma have decrements in lung function. They both have bronchodilator responsiveness and allergen skin prick test reactivity. But Th2 High asthma has greater airway hyper-responsiveness, or sensitivity to stimuli, exaggerated IgE levels, and blood and especially bronchoalveolar lavage, or lung eosinophilia. Now, I talked a bit about remodeling as being an important feature of asthma. And the study of remodeling is something that I’ve been doing in the lab for some time. It’s nice to look at pictures of what the airway looks like. Here’s a trichrome stain of an airway biopsy from one of our subjects. This is the epithelial layer. It’s a pseudostratified layer with goblet cells that come out clear, ciliated cells at the apical surface, and basal cells, here. Then there is a basement membrane with sub-basement membrane fibrosis. And then the mesenchyme, which is loose connective tissue and cells. And in asthma you get increased fibrosis. And we can measure that in our biopsies. In asthma, you also get an increase in the number and mucus content of goblet cells, particularly in the number of goblet cells. And we can measure that directly in our bronchial biopsies. You can also measure these other features of remodeling, here, but let me just show you those first two. And what we found is that Th2 High asthma was the group that had patients with fibrosis. Not all of them had fibrosis, but it was really fairly exclusive to the Th2 High group. We also found that Th2 High asthma had modestly increased mucus… mucin stores. So, here is mucin stores measured from those biopsies in the High asthma on the far right, Low in the middle, and healthy controls on the left. And it’s modestly increased. But the way they really differed was in the type of mucin glycoproteins that were produced by those cells. And there are at least three different types of secreted mucins. There’s MUC2, which is elevated in Th2 High asthma, specifically. MUC5AC, which is somewhat elevated in Th2 High asthma. And MUC5B, which is specifically repressed in Th2 High asthma, so that the ratio of 5AC to 5B is very abnormal, and 5B is driven down. It turns out we’re just learning now about the actual biological meaning of each of these mucin isoforms… well, not isoforms… specific secreted mucin genes. And they each play different roles in bacterial defense or antiparasitic defense in the lung, so that the pathological features of Th2 High asthma are really quite distinct in being a low MUC5B state and a high MUC5AC state. That’s all very interesting from a cell biology perspective, but does it mean anything medically? And I remind you that in this study we incorporated these bronchoscopies into a randomized trial. And we treated patients for eight weeks with inhaled corticosteroids or placebo in a blinded fashion. And when we unblinded ourselves to both the treatment group and then also whether they were Th2 High or Low, we found that Th2 High asthma responded to inhaled steroids, whereas Th2 Low asthma did not. And so here’s the data, in black, for the placebo arm, over the eight weeks of treatment. The Th2 Low arm that was treated with inhaled steroids over eight weeks of treatment. But then lung function in the Th2 High… High arm that was treated with inhaled steroids for eight weeks. And that’s the group that showed the expected improvements in lung function. So, this… this set of biomarkers dissected those who were responsive to inhaled steroids from those who were not. So, let me just summarize, then, what this set of studies taught us about molecular phenotypes of asthma. The first is that there is clearly a Th2 High subset of asthma. This set of patients has a distinct pathology, an inflammatory set of characteristics. There’s a lot of eosinophilia, fibrosis, increased mucin, and, especially, differences in mucin gene expression. And other differences I didn’t really have time to show you today. We think this group is driven by T helper type 2 cytokines acting locally within the lung. And you might predict that this group would respond to inhaled corticosteroids. And you might predict it would respond to drugs that block T helper type 2 cytokines, which are an emerging class of therapies that Joe is gonna talk about later. We also learned that there are some patients who are relatively low for T helper type 2 inflammation. They have different cellular pathological and inflammatory characteristics with little eosinophilia, no fibrosis, and normal mucin gene expression. And we don’t actually know the biological underpinnings for their episod… episodes of airflow obstruction and their symptoms. We can treat them with inhaled bronchodilators, and that may control their symptoms, but there’s no analogous controller therapy for them at this point, because we don’t actually understand their disease. And this remains an area of unmet need. Now, why is this useful to have this way to, now, phenotype people at a molecular basis? One is we could research disease mechanisms, and I’ll show a little bit of how this helps research disease mechanisms. And another is that it could help us understand the biological basis for treatment unresponsive patients. That is why, in that first figure I showed, some patients don’t respond at each step, and just have more and more medications used for their disease. It could also help us, as Joe will tell us, figure out what the best medication might be for that patient’s disease. In order to help us with further research in this area, we’ve designed a continuous measure of Th2 inflammation, rather than the crude Th2 High and Low categorization. And this figure shows some of that data from our studies to date in our asthma laboratory. What we did for the continuous measure was to take a scaled mean average of the three genes that serve as biomarkers, CLCA1, periostin, and serpinB2, and use that as one continuous measure, which is indicated here on the x-axis. This would be high-level expression of the three genes, and this would be low-level expression. And what we show here are, in healthy controls, that most healthy controls have fairly low-level expression of the three genes. We show that in patients with asthma who are not on inhaled steroids, that is, the dotted line, that there’s actually a broad population, and perhaps even two populations, of patients with asthma, ranging from very low expression of these three genes to high expression. And when we study our patients who are on inhaled steroids, and on a standardized dose, that have mild to moderate asthma, we find that most of them have driven down the level of expression of these three genes. And therefore likely have lower Th2 inflammation. There may be a small group with continued expression of these three genes. And whether or not this group with continued T helper type 2 inflammation makes up a big percentage of patients with more severe asthma is an important research question. So, let me show you just one research application. We’ve been studying microRNAs in airway epithelial cells and the role they might play in epithelial cell differentiation in the airway. We’ve also been studying microRNAs as potential biomarkers of disease in the lung, because microRNAs can exist in the blood in a form that’s resistant to RNases. And we found several that are elevated in the airway in asthma, in epithelial brushings. Here’s one, miR-663, that’s clearly elevated in patients with asthma. If we plot the epithelial metric for any given patient, the Th2 metric, on the x-axis, and miR-6… -663 on the y-axis, you can see that they’re very strongly correlated. And it gives you a sense for just how much 663 reflects local T helper type 2 inflammation. In fact, if you do a cell culture experiment where you expose epithelial cells to IL-13, you can now see that miR-663 is expressed almost not at all in the absence of interleukin-13 but goes up with IL-13 exposure to very high levels. And it turns out that you can measure 663 in extracellular fluids and it is another candidate biomarker for investigation in asthma. It could also play a role in epithelial cell biology. But, by having a way to measure the degree of Th2 inflammation in our subjects, we can determine from a list of hundreds of microRNAs, for example, which ones really seem to reflect Th2 inflammation. Here’s an application of a metric that’s more clinical. You might ask the question, does T helper type 2 inflammation persist in patients who have continued symptoms despite inhaled steroids? And this is a repeat study that we did after the first study. We did this with Genentech and Joe Arron, who you will hear from later. And we just studied healthy controls, mild to moderate asthmatics, again, before and after steroids. And now, somewhat more severe asthmatics. Not very severe, but moderately severe asthmatics, who were already on inhaled corticosteroids for their asthma control. And we analyzed the expression levels of these three genes as a marker of Th2 inflammation. And what you can see is, in mild to moderate asthmatics, when you give them steroids the vast majority normalize their three-gene biomarker of Th2 inflammation. But if you take a more severe population, that there are still people with elevated levels, despite inhaled corticosteroids. And that’s the inspiration for using this as a clinical tool to identify the cause of disease in people who require more medications at those higher steps of asthma therapy. So, let me just briefly talk about the rationale for blood biomarkers in asthma. The rationale is that the assessment of these phenotype-specific responses would be a lot easier in clinical trials, and more practical in the clinic, if we had a blood biomarker. I’ve been talking to date about bronchoscopically obtained gene expression studies. The approach that was taken by Genentech and by us was to develop assays for [secreted] proteins in the signature. And one of those was periostin, one of the three genes in that initial signature. And the reason that was interesting is because our data showed that it was secreted in the basolateral media in culture and can be detected in the blood. And here’s just a little of that data. This again is a photomicrograph of an airway biopsy. Here’s immunohistochemical staining for periostin — it comes out brown — in a healthy control and in asthma. And you can see that there is more of it, particularly in the basement membrane region and the… the subepithelial mesenchyme. And here are the blood vessels. There’s a capillary, there’s another capillary. And so it is secreted, basolaterally, by epithelial cells into a region where it can get in the blood. And here is a cell culture preparation that we frequently use, which is growing cells at an air-liquid interface — these are epithelial cells — so that we can sample the cell lysate, we can sample the apical surface, and we can sample the basal liquid. And, after exposure to IL-13, periostin is indeed highly secreted into the basal media. And so, based on that, this was one of several candidates for a biomarker, a blood biomarker, of Th2 inflammation. So, let me just give you a summary and we’ll get to more of the story, the second half of this, later. The summary is that asthma is a heterogeneous disease. And I think this is a worthwhile summary for a lot of diseases. We are increasingly recognizing that cancer is a heterogeneous disease, and even lung cancer, and even specific types of lung cancer are not one disease but many diseases. I think it’s entirely possible that inflammatory diseases that are common, or even metabolic diseases, can have diverse causes and reflect heterogeneity. In this case, only a subset of patients have high levels of T helper type 2 inflammation. And this is the type of inflammation we had been treating very specifically for asthma for 20 years. Patients with Th2 High asthma are distinct and respond better to inhaled corticosteroids. And blood marker… blood biomarkers of Th2 High asthma could be used to guide anti-inflammatory therapy. So, that’s the summary of the first part of the talk. I should acknowledge that a lot of this work was done with NIH funding, and some of it, also, through a collaboration with Genentech. And that the people involved are listed here. But, in particular, John Fahy, and Joe Arron. And I want to thank you for your time.


Leave a Reply

Your email address will not be published. Required fields are marked *