Scientist Stories: Nicholas Leeper, New Insights into the Pathobiology of Atherosclerosis

So today, as John pointed out, we'll talk about a process you may or may not have heard about called epherocytosis. And just before we start, there is a disclosure. We have licensed some intellectual property to a company known as 47 Inc., which is an oncology company but does have some relevance to what we'll talk about today. So our group is interested in the genetics of cardiovascular disease. And the reason that we do this is because, you know, the field of cardiovascular medicine has really shown an incredible improvement in the trends towards survival over the last 4 to 5 decades. But many of these trends have begun to plateau and, in fact, are actually starting to worsen in some areas. And I think that a lot of this is because so far we've really only been able to address the risk factors for disease, not the disease process in the blood vessel wall. And so if you think about it, a lot of these improvements are due to smoking cessation programs, control of blood pressure, and perhaps most importantly, control of cholesterol. But I really like a slide like this because it shows that even on top of really maximal cholesterol-lowering therapies, which are game-changing for our field, you can see that the residual risk is quite high. And so we clearly have a lot of work to do, and this helps explain why cardiovascular disease remains the world's leading cause of death. And, you know, we see this all the time where there is this so-called hidden risk, and, you know, people can do everything wrong and lead long, full lives. I mean, Winston Churchill was perhaps one of the least healthy people you can imagine. He smoked and drank, he was overweight and had a stressful life. But he didn't pass away of his stroke until age 93. And so to be born in the 1800s and to lead that kind of life with those risk factors is remarkable. It's telling you something about his biology that's different. And then John has seen this slide before, but this is the other side of the coin. And a person that, you know, really has done everything right. And this is Jim Fixx, who is the kind of the father of the fitness Revolution in the United States, wrote this New York Times bestseller. It's a picture of his legs on the COVID And the story is, of course, that he died of a sudden cardiac death at age 52 while out running. And on autopsy was found to have triple vessel coronary disease. So despite him going on all the TV shows talking about the benefits of exercise, clearly, you know, his relatively healthy lifestyle wasn't able to overcome his genetics. And so we've seen this borne out. If you look at heritability and twin studies, you know, the estimates vary, but the lifetime risk for coronary disease and stroke is maybe as high as 50%. So this suggests that even if you could optimally control for all of our risk factors and get people's blood pressure and sugar levels and cholesterol levels optimally controlled, that we have all of these black bars remaining that we currently can't account for. In our minds, this must be due to genetics or other unmeasured environmental factors, and we have focused on trying to understand that heritable risk, thinking that this can help us develop new therapies. And so, you know, like many others, we have tried to take advantage of this kind of wonderful era we're living in in human genetics. And so with the completion of the Human Genome Project and then the advent of high-density genotyping, arrays, the platform, the GWAS platform has been introduced in the last decade. And, you know, I guess the first study that came out was in 2005, and this was for macular degeneration. And I think this is a powerful example because when you look in an unbiased fashion across the genome, investigators were able to find that the top hit for this disease was in complement factor H. And I'm not in this field, but I understand that people weren't focused on this at the time, so it really allowed that area to kind of open their eyes to a new pathway that might be involved and that may not have been discovered otherwise. In the next year, there were several more GWAS studies reported, and again, another example, if you look at the hits for inflammatory bowel disease, the pathways that were implicated included autophagy genes, you know, again, a pathway that wasn't really being suspected as a causal driver for Crohn's disease and ulcerative colitis. And then you can see the examples kept coming, and, you know, once again, if you look at type 2 diabetes, the top loci were not in insulin sensitivity pathways, but actually were in genes that regulate beta cell senescence. So perhaps this will explain why our insulin sensitizers haven't worked as well as we thought they should, and perhaps could direct us towards other areas of research. And this technology really exploded, and you can see that the NIH tracked all of these different hits that had been found. And pretty quickly, you can see that this became something where there literally are thousands of genes and diseases that have been independently verified. And this has, you know, really been a rich resource for future research because our belief is that unlike proteomics and other things that can be influenced by the environment, it's pretty hard for genetics to lie. So it may be hard for us to figure out how these pathways are causing disease, but they almost certainly are associated with the disease at least. And I think that, you know, in the cardiovascular arena, you know, we are no exception, and so this is the data that you can generate from a GWAS study, and this is called a Manhattan plot, and they will lay out the different chromosomes on the X-axis, each one's in a different color, and what you'd like to see is a high rise above the background, so they'll take people with the disease and those without, so here this would be heart attack or coronary heart disease, and what you'd like to see is a SNP or a variant that's overrepresented in people with the phenotype compared to without, and the higher the skyrise, the stronger the association. And what you can see is that there is, you know, an undeniable hotspot of risk at the chromosome 9p21 locus. And this locus is really interesting to us for several reasons. The first is that these are very common, so it's not like this is a rare mutation that is, you know, may or may not have an important public health impact, but actually these variants have a minor allele frequency as high as 50%, meaning that about 1 in 5 people in this room will carry 2 copies of the risk allele. So, this is what 23andMe is testing when they check your blood for your heart attack risk. You know, the public health impact is pretty significant. As I said, about a fifth of us carry two copies, and I think what's most interesting is that this locus is completely risk factor independent. So we know beyond any shadow of a doubt that this isn't just another pathway that raises your blood pressure or raises your cholesterol. And this to us suggests that if we can figure this out, that we could have a new pathway to approach and perhaps something that's additive on top of our existing therapies and not just another LDL-lowering drug, for example. And so I won't go through all of the molecular genetics, but I will say that this locus has been confusing because it's in a so-called gene desert and it wasn't clear what the causal gene was. We've now performed eQTL analyses and we believe that the causal gene is actually a cancer gene known as CDKN2B, and this may come back later as we look at the intersection between cancer and cardiovascular disease. But in any event, our laboratory has chosen to focus on this locus and we generated a knockout mouse that lacks this CDKN2B gene that we think is a candidate to explain how this heart disease is promoted. And as expected by the human genetics, the knockout mice had, you, you know, much larger atherosclerotic plaques. And so this wasn't really a surprise to us based on what the GWAS studies had shown. But what was quite interesting was the phenotype of the plaques. And so for those of you who work with ApoE mice, you've seen these types of pictures of the lesions, which, you know, will vary, but typically are fairly dense and hypercellular and oftentimes will have a fibrous cap, which would suggest kind of more of a stable plaque phenotype. But our knockout mice had this really amazing phenotype where they universally had these very large acellular necrotic cores. And so this, to us, looked like there was a problem with the accumulation of apoptotic and necrotic debris. And we wondered if this was the process that was dysfunctional in our knockout mice. And in particular, as we were reading on PubMed about how this could be happening, we came across a term called efferocytosis. And this is from the Greek meaning to carry the dead to the grave. And efferocytosis is actually pretty complicated. You know, when I went to medical school, I thought when a cell died, it just got eaten kind of automatically. But it turns out this is a very highly regulated process. And so as soon as a cell begins to undergo programmed cell death, it will secrete molecules called FynI ligands. And these are basically chemoattractants that will tell the phagocyte to come to the site of injury. Very quickly, they will upregulate so-called eat-me ligands, and these are basically cell surface molecules that will interact with receptors on the macrophage to identify those for clearance. And then very quickly, when this process works, the corpse is engulfed and degraded. And if this goes well, it's now known that the macrophage will secrete a number of anti-inflammatory cytokines. And this is, we think, part of a mechanism where it tells other macrophages that the injury or insult has been contained and that no additional recruitment of monocytes is necessary. And so we wondered if our locus was regulating this somehow and that this was the reason that the apoptotic cells were not being cleared and were accumulating. And it's really interesting because it turns out that our body actually turns over hundreds of billions of cells every day. This is part of our normal homeostasis. And pathologists have long known that if you biopsy throughout the, a healthy body, it's almost impossible to find any examples of apoptotic tissue. And they think that this is because our capacity for phagocytosis is so high that these cells are recognized and cleared within a matter of seconds. And so pathologists have found that the only place where you can actually find examples of tunnel-positive apoptotic cells are in very high turnover organs like the bone marrow, the tonsil or the thymus. And if you see here, you can find a tunnel-positive cell in the tonsil, but it's almost always surrounded by a blue macrophage. And so they think that this means that this cell was actually— you just got kind of lucky. It was already in the process of being cleared. And if you had taken this biopsy 5 minutes later, it would have been gone. And so these investigators looked across the body, and it turns out the one exception to this is in the atherosclerotic plaque.. And so this is the place where you find free apoptotic tissue in so-called the healthy human being. And they didn't really understand why these cells aren't getting cleared, but they did note the difference. And in fact, they quantified it. And it turns out that efferocytosis is impaired almost 20-fold in the plaque compared to elsewhere in the body. So this was a really interesting observation. So we wondered if our locus was associated with any of these so-called eat-me molecules that regulate cell clearance. And so we performed genome-wide co-expression modeling and looked at the module to which CDKN2B was assigned. That's our candidate gene. And it turned out that it was associated with exactly one of these so-called Eat Me ligands, a gene called calreticulin. And again, we had never heard of this, so we had to go to the literature. But it turns out that calreticulin is kind of the grandfather of phagocytosis. And there's a beautiful paper in Cell which showed that calreticulin is absolutely required for dying cells to be recognized for clearance. This is the Eat Me molecule. It's counterbalanced by a molecule called CD47 that I'll come back to in a moment. But these investigators found that calreticulin is necessary for engulfment, and if it's missing, these cells are almost invisible to the macrophage, and it's like they have a cloaking device. And so we looked, and it turns out that our CDKN2B-deficient cells have low levels of calreticulin. Same thing for humans born with an IP21 risk allele. We've shown that they have reduced levels of calreticulin in their plaque. And the end result is that these cells become inedible. So here is a macrophage labeled green. These are apoptotic bodies labeled orange. And when you co-culture them, within the first hour, about 20% of these cells will be ingested. These are double-positive cells, which we think represent a macrophage that's eaten an apoptotic cell. But if you look at cells that are CDKN2B deficient, they basically resist clearance. And so their likelihood of being phagocytosed is reduced by 50% in vitro. Now it's difficult to quantify this process in vivo. I actually would say it's impossible to quantify it in vivo, but we did look in the plaque by electron microscopy and we saw that normally macrophages in the plaque— excuse me— will eat 10 or 20 apoptotic bodies at a time. But in our knockout mice, we found no examples of this. And in fact, all we could find were examples of cells undergoing secondary necrosis. And we think this is part of the problem because if these apoptotic cells are not recognized right away, they undergo secondary necrosis and they have breakdown of the cell membrane, which leads to the release of these inflammatory organelles into the growing plaque. And so our mechanism that we think happens normally is that when a cell in the plaque dies via this mechanism, which I'm skipping for the sake of time today, but it's been published, these will upregulate calreticulin on the cell surface. And this is very quickly recognized by the LRP receptor on the macrophage. And when this happens, the macrophage is able to engulf this cell It actually also will activate a number of homeostatic pathways that are protective for the plaque, including cholesterol efflux. And then this macrophage will secrete anti-inflammatory cytokines as a way to show that everything's OK and that no more inflammation is necessary. But we think that in people born with the risk allele, people who have that 9p21 allele, the 1/5 of us in the room, our cells— I have 2 copies, by the way. Our cells have low levels of calreticulin, and so when these cells die, they can't get recognized. And this macrophage actually becomes frustrated. We don't quite understand why this happens, but he will release extra TNF-alpha and is more likely to assume a foam cell phenotype. But what we really worry about is that this cell will undergo secondary necrosis and will release all of these inflammatory tissue into the growing necrotic core. And so we published those findings a couple of years ago now, and since that time we've been thinking very hard about how do we translate these results from bench to bedside, because that's what John taught me and this is what we all, you know, kind of come to work for trying to do. And so it's interesting because now if you read the textbooks, epherocytosis has made it onto the main page here, and you can see that people are starting to appreciate that this process and these cells not getting cleared may be a major driver of growth of the necrotic core and the inflammation here that drives plaque rupture. And so we've been wondering how we can reactivate this process, if we can almost leverage the immune cells to begin to clear out this tissue and shrink the core as a novel approach. And, you know, as we thought about it, we realized that you know, one way would be to increase the expression of these Eat Me molecules. So if you could find a way to turn up the calreticulin on these dying cells, that should work. But, you know, the cell is already dying for a reason. It's already diseased, and we didn't really think that we had an easy way to adjust the behavior of those cells. And instead we said, well, can we find a way to block the countervailing Don't Eat Me signal? So instead of stepping on the gas, could we take away the brake and this should have the same physiological effect. So before we could do this, we had to see if these "don't eat me" molecules are present at all. Like I said, calreticulin is the "eat me" molecule, CD47 is the "don't eat me" molecule, and we just didn't know if this was present in the diseased blood vessel. So we looked at coronary samples taken from rapid autopsy at Stanford and in healthy coronary arteries, so this is a person that died from a, from a motor vehicle accident. You can see there's very little red staining here in this healthy artery. But in a person who died of an acute coronary syndrome, we saw this really dramatic upregulation of CD47 in the plaque and really this kind of intense localization to the necrotic core. And so this was, you know, qualitative, and we wanted to see if this would be borne out quantitatively. So we contacted our collaborators at the Karolinska Institute who have a very beautiful dataset called the BIKE Biobank, which is a carotid endarterectomy biobank. And we saw— I'm sorry that this is defective here— but in people with atherosclerosis compared to without, in both the discovery and the validation sets, we saw that there was a higher level of CD47 in people with vascular disease compared to healthy controls. As we acquired more coronary samples over time, we saw there was a stepwise increase in the CD47 expression as the plaques progressed. And you can see that it really seemed to correlate with the phenotype of plaque vulnerability. We saw the same thing in our mice. As you fed them high-fat diet, their aortic expression of CD47 increased over time. And we used an HRP-tagged antibody to show that there really is, again, kind of local augmentation of CD47 in the plaque. I think also pretty interestingly is that we found that it wasn't simply a marker of disease versus no disease, but also was a marker of symptomatology. So amongst those carotid patients who had their plaques taken out, about half were taken out for asymptomatic disease. You know, a doctor had heard a bruit and decided to operate. And the other half were taken out for symptomatic TIA or stroke. And we found in 3 separate cohorts that patients with symptomatic cerebrovascular disease, like stroke, had higher levels than those who had asymptomatic disease. So to us, this suggests that CD47 is up as disease progresses and also may be a marker of plaque vulnerability. So this is a problem because what we know from the literature is that our body is very careful about how it recognizes and clears out dying cells. Because I told you, if we turn over 100 billion cells each day and they have to be cleared within a matter of seconds, that means that our macrophages and phagocytic cells are very primed to clear these cells. And we have to be careful to avoid the off-target clearance of healthy cells. And so it's known that CD47 is expressed on almost all cells, and this is one of the key ways that we don't have off-target clearance of healthy tissue. And authors have shown that one of the first things that happens during apoptosis is that CD47 becomes downregulated. And this is part of how the cell identifies itself as being diseased or dying and identifies it for clearance. So here's some of the original data showing in lymphocytes at least that as these cells are exposed to increasing time of an apoptotic stimulus, you can see that the CD47 levels drops very dramatically. And they think that this is part of how the cells are recognized for clearance. So why is the CD47 going up in the plaque if it should be going down as these cells are dying? Well, first we just checked to see what the— if there was a different direction on vascular cells compared to the lymphocytes that have been published. And it turns out the pattern was the same. So if you take a cultured smooth muscle cell and render it apoptotic, the RNA of CD47 will fall over a matter of hours. And the same thing happens on the cell surface. So here's baseline, very high expression, which goes away in culture with Staurosporine. We saw the same thing a third way by cell surface flow cytometry. So it looks like the pattern should be going down, and we didn't understand why it was going up. So, to get at this, we again turn to our bioinformatics. And in collaboration with Jake Lucis at UCLA and Eric Shad at Mount Sinai, we again had access to RNA-seq data from both mice and humans. And we were able to look at co-expression modules in the atherosclerotic vessel to see which genes tend to associate with CD47 in the diseased wall. And the first thing we noticed was that The list was actually fairly small, and there was a real overabundance of inflammatory factors linked to CD47 levels. When we performed pathway analysis, actually inflammation related to cytokine signaling was the top pathway predicted. And then we could ask the computers to look upstream to see if there was kind of a master regulator that should be driving this whole expression profile, and when we did that, the computers predicted that TNF-alpha TNF-alpha should be the top factor responsible for this upregulation. So we went and looked for this, and it turns out that the CD47 promoter does have regions of open chromatin that have predicted binding sites for transcription factors known to be downstream of TNF-alpha, including RELA. And as we looked in vitro, if you expose smooth muscle cells to all of these pro-inflammatory or pro-atherosclerotic cytokines, they really have no effect on CD47 expression. But as predicted, TNF-alpha really dramatically increases CD47 levels by about a 3-fold increase. So what we think happens, and we've now shown this experimentally, is that normally when a cell begins to die, it will lower its levels of CD47 over the first 4 hours.. And this makes the cells more edible. So here is again the ephorocytosis assay, basically looking at the likelihood that these cells become phagocytosed. And you can see that as again the cells die and their CD47 levels drop, they become more edible, they're more likely to be cleared. So this is what we would like to have happen. However, in the presence of TNF-alpha, there's a much higher level of CD47 at baseline.. And as these cells die, the CD47 levels still do drop, but the decrease is blunted. And you can see that even when the cells are fully apoptotic, their level is higher than a healthy cell at baseline. And so the end result here is that because these cells have more "don't eat me" level expression, they are slightly less likely to be cleared at baseline, and they become significantly less likely to be cleared under pro-atherosclerotic conditions. So if we load the cells with LDL and render them apoptotic, the phagocytosis rate falls by over half. So our working mechanism now is that TNF-alpha, which by the way is of course known to be upregulated in the plaque locally, we believe that this signals through the TNF receptor and NF-kappaB1 to directly activate the CD47 promoter. This leads to a decoration of CD47 on the cell surface. Which will signal through this receptor called SERP-alpha on the macrophage, and this provides that repulsive "don't eat me" signal. So the macrophage can't recognize this cell, which becomes secondarily necrotic. And we think that this is part of why the necrotic core is accumulating over time. So knowing this, we wondered if we could target this process and if we could do it in vivo.. And it turns out, just coincidentally, that research that's been going on over the last 5 to 10 years has shown that this efferocytosis defect is actually a major driver of cancer as well. And so scientists have shown that almost all malignancies will upregulate CD47. And probably this is one of the key ways in which these cancer cells evade the tumor cytotoxic macrophage. We think this is part of the reason that the immune system breaks down and can't recognize these cells and clear them. And so groups from the oncology field have been working for several years trying to interrupt this axis, thinking that if they could take away this "don't eat me" signal, that they could reactivate tumor killing. And it turns out that one of the real leaders in this field is a colleague of ours from Stanford named Irv Weissman, who's the director of our Stem Cell Institute. And through a series of really elegant, high-profile publications, They've shown that antibodies that interfere with the CD47 pathway can have dramatic anti-cancer effects, either alone or in combination with other anti-cancer antibodies. And they can lead to significant reductions or even cure of basically every type of cancer that they've looked at so far. And so we wondered if these types of therapies would have the same effect in atherosclerosis.. And so now the idea is that if the problem here is this repulsive interaction, could we take away that and begin to trick the macrophages into eating these cells and perhaps shrinking the atherosclerotic plaque? So first we had to see if our antibodies would work on vascular cells like they do in cancer, and the short answer is that they do under— increasingly pro-atherosclerotic conditions, you can see that there is a doubling or even tripling of the clearance rate of these cells, again, in this in vitro phagocytosis assay. And we showed in vivo that we could pretty definitively interrupt this "don't eat me" axis. And here the pathway is that the CD47 interacts with SIRP-alpha, and it's thought that the way that this tells the macrophage not to engage in engulfment is by phosphorylating this phosphatase SHP-1. And we saw that in the control mice there's actually really dramatic activation of SHP-1 in the macrophages. And we think this explains why the macrophages aren't able to participate in phagocytosis. But if we gave these mice the antibody, we could completely shut this down. So here you can see that these same red macrophages have very low levels of SHP-1 activation.. And we think that this proves that the pathway had been interrupted with our antibody. Ultimately, these mice developed plaques that have far fewer numbers of tunnel-positive cells. And although it's difficult to really quantify the phagocytic index in vivo, like I mentioned before, we could find that there was a lower level of these so-called free apoptotic bodies. At baseline, it's pretty common to see these tunnel-positive cells that are quite far away from the red macrophages. We think that this is part of the problem, that these macrophages aren't recognizing this dying cell. But after giving the antibody, those tunnel-positive cells we did find were much more likely to be in close proximity to the macrophage. So again, we think that this shows that these are being recognized and hopefully are in the process of being cleared. Again, with the electron microscope, we showed that all of these free apoptotic bodies and necrotic tissue could be ameliorated with the antibody. I love this picture because you can see that he's had lots of apoptotic bodies taken up and is about to take in quite a few more. And this is qualitative, but it gets at the idea that this is really quite a potent therapy.. And then, you know, at the end of the day, what you really care about is what happens to the plaque. And the first thing we found was that the necrotic cores shown here in baseline, which we had traced out, are dramatically reduced with the anti-CD47 antibody. And then ultimately these mice had much smaller atherosclerotic plaques. So here in the aortic sinus, you know, you can see that this is a pretty small plaque for people that work with the ApoE model. We actually put in, I think, 6 different models in this paper, all of which showed a similar benefit, including here on Fos throughout the aorta. So, you know, the paper was looking really great, and, you know, this looked like a really effective therapy. But the question, of course, is, you know, what about any toxicity? And in particular, if CD47 is so important for self-tolerance, why aren't we inducing the off-target clearance of healthy tissue? So we looked carefully, and first, I have to mention that CD47 is known to interact with thrombospondin signaling, so we wanted to make sure there was no effect on nitric oxide production. And so we looked by the Grease Reaction and others. There was no difference in blood pressure. And we also excluded a role for any effect on cholesterol levels. And an effect for insulin sensitivity, body weight, or diabetes. But what we did find was that the spleens were quite large in the mice, and it really was all due to an increase in the red pulp. And this translated into, you know, a modest but, you know, perhaps worrisome anemia that was accounted for with an increase in reticulocytosis. And it's interesting because if you go back to the original literature, this was actually first identified as a marker of self on red blood cells. And again, with this idea of self-tolerance, it turns out that our red blood cells have a finite lifespan. And part of the way that they get recognized for clearance is by turning off their CD47 as they age and need to be cleared out by phagocytes in the spleen. And so while this might be an acceptable toxicity for a cancer patient, We weren't quite sure this would be okay for all of us walking around after barbecue chicken pizza last night. So the question was, how can we ameliorate this anemia? And so as we thought about this, we, you know, wondered if we could take advantage of the link to the TNF pathway that we had just identified. And I'll just say that, you know, there are a number of TNF-blocking antibodies available on the market. And if you read the people that write about the inflammatory hypothesis of atherosclerosis, they're always quick to point out that lupus patients and rheumatoid arthritis patients who get put on these TNF inhibitors have lower levels of heart attack than they would be expected to have. So they've always wondered if blockade of TNF-alpha was somehow beneficial for atherosclerosis. So we looked, and it, you know, again, because these, you know, commercially available antibodies, a number of studies have been performed where people have looked at microarrays before and after treatment with these drugs. And when we got access to those publicly available data, we found that indeed in basically all of these studies that compared to baseline, after treatment with these drugs, the human expression of CD47 is reduced. So this kind of confirms our proposed mechanism of action and shows that we can lower this. We found the same thing in our mice, excuse me, where treatment with Etanercept led to a reduction in CD47. So we wondered if we could use combination therapy and reduce the dose of the anti-CD47 antibody and if this would get around some of the toxicity. And so first, again with the electron microscope, we saw that the combination therapy that really induced dramatic phagocytosis in vivo. And the end result was that indeed there was a modest combinatorial effect. But I'll point out again that this dose of the antibody was reduced by 75%. So it shows just how potent this is. And in this case, the mice had no anemia or any toxicity whatsoever. And so we're hopeful that this can be a strategy strategy as we think about translation into humans. So now our working hypothesis is that this is the problem in baseline. We think that this will be important for people who are— who do carry the 9p21 risk allele, but we think that it will work in all comers as based on the microarray data from the biobanks I showed you earlier. But we think that we can interrupt this.. And if we do that, we can begin to clear out these cells and we hope actually begin to induce plaque regression in the necrotic core. So as we think about moving from bench to bedside, we're really in a unique position because we're able to leverage the work that's been going on simultaneously in the cancer arena. These antibodies have now been humanized and we've taken them through nonhuman primate studies and shown that again there is the predicted anemia, but it can be ameliorated with a dose escalation approach or with a priming dose approach. And actually under the leadership of Irv Weissman and others at Stanford, this has now been advanced into two first-in-man Phase I studies. One is at Oxford in patients with AML. The other is at Stanford in patients with solid tumors. And those data are just beginning to report out. The first findings were published in abstract form at ASCO recently. And in humans with malignancy, so far, there have been really no unexpected side effects that have needed to delay that clinical trial aside from the anemia, which again can be largely avoided with a priming dose approach. We're thinking now actively about how to advance this into a Phase 2 cardiovascular trial. I think this is a really interesting opportunity where we can leverage what we're going to learn from the cancer trials and hopefully extend that as we think about the cancer— as we think about the cardiovascular arena. And the hope is that we can use some type of non-invasive surrogate like FDG PET/CT, which shows very nicely the vascular inflammation scan, or perhaps even with a coronary-based imaging scan based on IVUS or OCT. And so I hope that we can have these types of trials underway in the next year. So just to summarize, I think that what we found is that the hypothesis-free human genetics are really a very powerful tool that can provide incredible insights, and I think that, you know, there are several examples. You know, without the GWAS studies, we would probably have never found this unusual tumor suppressor locus at 9p21. I would argue that each of these genes would be— have been unlikely to have been identified without unbiased agnostic genetic approaches, and that these processes like efferocytosis may not have been identified as being so important. I think it's quite interesting that this process seems to be a really important driver of atherosclerosis, and it's hard not to notice. Commonalities with cancer. And importantly, this is not a fixed effect. So we're hopeful that this can become a new targeted therapy for shrinkage of the necrotic core and perhaps also to induce plaque stabilization. So with that, I do acknowledge the lab, the people that have done the work, particularly Yoko Kojima is our first author for this work. She's just an incredible scientist. And the rest of the team. And of course our funding. And so with that, let me stop and thank you again, John, for the invitation.