All right, good afternoon, folks. We'll get started. The sessions seem to be running pretty close together, and they had a lot of great questions in the last one. My name is Dr. Ryan Mabes. I'm from the Naval Medical Center in San Diego. I'll be the moderator for this session. Before we begin, just a few housekeeping remarks. If possible, move to center of row. Understand it's good to spread out and have some space, of course. Please notice the location of the exit doors. If possible, please silence your cell phone. If you wish to leave the session for any reason, please feel free to exit quietly. Once it's concluded, please exit quickly. The next set, there may be another session after this one. Probably not. And information about continuing education credit is available on the Congress app. All speakers will disclose any conflicts, so why don't we go ahead and get started. Our first speaker is Denise Lopez-Domowitz, who will be speaking on red blood cell ATP depletion and altered ATP export by red blood cells in sepsis. Thank you. Thank you. Hi, guys. Good afternoon. Thank you all so much for having me. I'm currently a third-year fellow in pediatric critical care at Duke University Medical Center, and I'm honored to be able to speak to you all today. The title of my talk is Decreased Export of Vasoregulatory ATP from Red Blood Cells in Murine Sepsis, and I'm excited to share this research with you. I've had the pleasure of working in the lab of Tim McMahon in the Division of Adult Pulmonary Critical Care at Duke, and the experiments related to this project were funded with both NIH and VA support. In addition to that, Alex Roda from the Division of Adult Critical Care, of Pediatric Critical Care Medicine, also graciously granted me a stipend to help some of this research. As we know, sepsis represents a severe systemic inflammatory response to an infectious agent, which can lead to multi-organ failure and death. It leads to 720,000 annual hospitalizations in U.S. intensive care units alone, and is a leading cause of death in children worldwide, with a 10% to 15% mortality reported in some studies. Sepsis-induced red cell dysfunction, or SIRD, represents a unique class of organ failure related to the inability of red blood cells to deliver oxygen adequately to meet tissue metabolic demands, and is only recently being recognized for its significant contribution to sepsis pathology. The focus of my project is to more fully understand the effect of sepsis on the erythrocyte, and the ways in which red blood cell dysfunction can contribute to sepsis pathophysiology. In order to understand the role of the erythrocyte in sepsis, we must first understand the important role that red blood cells play in health. Normal red blood cells have been shown to help modulate local vascular tone and regional blood flow to match oxygen demands through a phenomenon known as hypoxic vasodilation. Red blood cells have an intrinsic store of nitric oxide species, and can release S-nitrosothiol, or SNO, which is shown here in green, in response to local tissue hypoxia. These nitric oxide derivatives function as a potent vasodilator when taken up by vascular endothelial cells, and leads to local vasodilation and promotes oxygen delivery to the hypoxic tissues. This is one way in which healthy red blood cells help to modulate regional blood flow in response to hypoxia. Similar to nitric oxide, ATP is also a potent vasodilator, and is exported by red blood cells in response to hypoxia and acidosis. As such, it can help to match oxygen delivery to tissue demands. ATP export is primarily mediated through a Ponexin-1 hemichannel, which is pictured here. Ponexin-1 expression has been described in both human and murine erythrocytes, and we have previously shown that ATP export in a Ponexin knockout mouse is attenuated. This is seen to correlate with an attenuated response in hypoxic vasodilation as well. These findings support the important contributions of erythrocyte ATP export to meet the metabolic oxygen demands of tissues. In sepsis, several key functions of red blood cells become altered and contribute to pathophysiology, which is seen clinically. Red blood cells show altered oxygen affinity, decreased membrane deformability, and increased aggregation and adhesion. However, the underlying mechanisms that contribute to these alterations are not yet fully understood, and we believe that altered ATP may significantly contribute to some of these defects. Therefore, we hypothesize that in sepsis, dysfunctional red blood cells show altered ATP export, and this contributes to inadequate oxygen delivery for tissue demands. It has already been abundantly shown that ATP export is increased in response to hypoxia. We worked to develop a murine sepsis model for our lab and analyze how ATP export responds to hypoxia in the setting of sepsis. We first asked if there was a change in ATP export in sepsis compared to control, and then if that change in ATP export changes over time. We chose cecal ligation and puncture, or CLP, for our sepsis model, as it reliably produces consistent degrees of sepsis and has been extensively studied. It also mimics sepsis, which is seen in pediatrics, with characteristics similar to necrotizing enterocolitis and other intra-abdominal pathologies. After cecal ligation and puncture surgery, blood was initially harvested on day three and then exposed to hypoxic conditions in a bulb tonometer. The ATP in the supernatant and total cellular ATP are measured and compared across conditions. A hemoglobin assay was also completed, which allows us to correct for the degree of red blood cell lysis in our samples. Our red blood cell preparations only have a small degree of lysis, 1.1 to 1.3 percent, and there was no change in lysis seen based on the exposure to varying oxygen tensions. From these experiments, our results show that there is decreased ATP export in the murine sepsis model after hypoxic exposure. Across the x-axis, here you can see the condition, control versus cecal ligation and puncture, or CLP, and along the y-axis, you have nanomolar, extracellular, or exported ATP concentrations. This change in ATP cannot be explained by the minimal degree of lysis that we had measured in our samples. In the sepsis model, there is a decrease in ATP export when compared to controls on sepsis day three, with ATP export being about half of that which is seen in controls. This is also interesting in that normal erythrocytes have abundantly been shown to have increased ATP export in response to a hypoxic challenge. We then asked whether this decreased ATP export might be due to decreased intracellular ATP content, and we also saw decreased intracellular ATP on sepsis day three. Again, here you can see control compared to CLP along the x-axis and nanomolar concentrations of ATP on the y, but now we're looking at intracellular concentrations of ATP. There was a significantly decreased intracellular ATP in the sepsis cohort compared to the control mice. We hypothesize that this may be due to increased ATP expenditure in the setting of sepsis. This finding made us question what was happening earlier in the sepsis time course. So we next began to take a look at sepsis day one in a similar experiment. Here you can see, again, back to extracellular or exported ATP concentrations in response to hypoxia from control compared to day one sepsis and day three sepsis. We found that ATP export is decreased in the sepsis model on day one just as it was on day three. While the graph may show an increase in ATP export on day one compared to day three, I'm sorry, on day three compared to day one, this was not statistically significant. Interestingly, the decreased ATP export is seen on day one of sepsis despite there being a markedly increased intracellular ATP. Compared to both control and day three of sepsis, red blood cell intracellular ATP is increased on day one as you can see here. This may be a response to stress hormone release in the setting of acute infection, which prompts increased energy production. However, by day three, this increased intracellular ATP appears to be exhausted. It remains to be seen if this is due to exhaustion of ATP stores versus decreased ATP production. However, the fact that ATP export is not increased at either sepsis time point that we analyzed suggests that it is either being used by the erythrocyte or can no longer be adequately produced by sepsis day three. So, in conclusion, sepsis-induced red cell dysfunction is a critical form of organ dysfunction that requires further study. And cecal ligation and puncture is a useful model for inducing forms of sepsis that show signs consistent with CERD. In sepsis, ATP export from red blood cells is decreased in response to hypoxia from days one through three. This may have significant clinical implications and amplify oxygen deficit in tissues, given what we know of ATP's role in the panexin knockout model and how ATP contributes to hypoxic vasodilation. However, this decreased ATP export is in spite of an increased total intracellular ATP on day one of sepsis. This rise in intracellular ATP on day one is no longer seen on day three and may reflect increased energy expenditures or cellular exhaustion throughout the course of sepsis. Moving forward with this information, we hope to replicate these cecal ligation experiments in our panexin knockout mouse. It'll be interesting to assess their baseline, how their baseline attenuated ATP export affects their development of sepsis if they potentially have a more severe course or a less severe course. Additionally, we plan to express, to look at the in vivo effects of this decreased ATP export in the sepsis mice through experiments that look at blood flow, specifically mean arterial blood pressures and hind limb blood flow. And finally, we hope to understand if this decreased intracellular ATP that is seen in our sepsis model is due to increased energy expenditures or alterations in energy production from the cells themselves. So with that, I'd like to thank my primary research mentor and our division of critical care at Duke, who have been instrumental in getting me here. And so thank you all very much for your support. And I open it up to questions. If there are any questions for Dr. Lopez, please go ahead and come up to the microphone. Very nice presentation. So I have two questions. How is the expression of this channel in the mice, the CLP mice? Is it increased? Is it reduced? So we have not specifically looked at the expression of that channel in the CLP mice. But we plan to do the CLP ligation experiment in mice who are a knockout of that specific channel. Yeah, but in that case, it's going to be knocked down. Correct. I'm asking you, in baseline situation, what is expression? So it's affected by sepsis? I don't know the answer to that question. That's not something that we've specifically looked at, but that is a great question and worth investigating. And another question is, you showed the data that in day one, the production of ATP is increased, and then it goes down. So what happened is probably metabolism. Did you check the adenosine level? That is an excellent next step sort of in the direction that we're going to look at specifically how that energy expenditure is coming into play and why there is an increase on that first day. But I suspect that it's potentially due to, as mentioned, the stress response and increased energy production. Thank you. Thank you. Hi, Burkett Heiles-Lasse from Stanford. I, too, have two questions. But the first one was just wondering what your Kaplan-Meier curves for your CLP model looked like, because day one, day three, you might have a survival bias, depending on severity of CLP model. That's true. So the CLP model that we used does a ligation of the entirety of the cecum. So we are consistent within the days. And our Kaplan-Meier curves have overall been relatively consistent. We initially looked at just the day three, and our mortality rate overall was about 30%, which is consistent with what's been previously published. 30% on day three? Mm-hmm. OK. And then the other question was, I'm curious what SNO levels are doing at day one and day three. Is this a compensatory mechanism that we're seeing? Have you looked at other vasodilatory agents? That is an excellent question. And ultimately, I would love to have an answer to that. I do have all of those samples currently stored to be able to analyze the SNO levels. And unfortunately, our machine that analyzes that has broken. So I apologize. Technical difficulties. All right. Thank you very much. Thank you all. All right. Our next speaker is Dr. TJ Darden, who will be speaking on circulating leukocyte transcriptomes in chronic critical illness versus rapid recovery sepsis patients. Thank you. Hello, everyone. My name is DeJoya Darden. I am currently a research fellow at the University of Florida in the Sepsis and Critical Illness Research Center. I'd like to thank the SECM for the opportunity to present our research. I'd like to thank Dr. Maves for moderating this session as well. I will be discussing the circulating leukocyte transcriptomes in chronic critical illness versus rapid recovery sepsis patients. I have no disclosures or conflicts of interest. Sepsis is currently the leading cause of in-hospital mortality in the United States. Fortunately, over the last 10 to 20 years, inpatient mortality to sepsis has declined. However, this has not translated to completely improved or accelerated patient recoveries. Advancements in detection and in treatment of surgical sepsis have led to two common clinical trajectories in those individuals that do survive sepsis. There's those that have rapid recovery, and there are those that develop chronic critical illness. Chronic critical illness is characterized by prolonged hospital course with intensive care utilization of 14 days or greater and ongoing organ dysfunction. These patients have been found to have persistent immunosuppression and prolonged impairment of host protective immunity. Rapidly recovering patients are defined as those patients that do not meet the criteria of chronic critical illness or early death, and they spend less than 14 days in the ICU with evidence of organ recovery. Although the majority of rapidly recovering patients have good long-term outcomes, more than 40% of chronic critical illness patients actually have poor one-year outcomes. Therefore, we aim to determine whether blood leukocyte transcriptomic profiles at post-sepsis day 14 could identify the underlying immunosuppressive and inflammatory mechanisms in chronic critical illness patients versus those who rapidly recover in survivors of surgical sepsis. In order to do this, we prospectively followed 120 trauma and surgical ICU patients who were either admitted with or subsequently treated for sepsis, severe sepsis, or septic shock according to sepsis 2 definition. Now, these patients were enrolled before the implementation of sepsis 3, before everyone murders me for using sepsis 2. Patients were categorized as either chronic critical illness or rapidly recovering patients, and all sepsis patients were managed using evidence-based protocols derived from the surviving sepsis campaign that emphasizes timely administration of broad-spectrum antibiotics, source control, and intravenous fluid administration. In addition to evaluating septic patients, we also recruited 41 healthy age and sex match controls. Human whole blood was collected from the patients at discharge or day 14 for total leukocyte RNA isolation. Samples consisted of 79 chronic critical illness patients, 41 rapidly recovering patients, and 41 control patients. The next few slides I will highlight the demographics of both the rapidly recovering patients and chronic critical illness patients. You'll note the chronic critical illness demographics in the third column and rapidly recovering patients in the fourth column. Between the groups, there were no significant differences in sex, age, race, or BMI. We also did not note any difference in the number of patient comorbidities. However, patients with chronic critical illness did demonstrate greater organ dysfunction at 24 hours after sepsis protocol onset, as indicated by their higher APACHE-2 scores. Patients who developed chronic critical illness were more likely to have increasing sepsis severity, exhibiting a higher incidence of septic shock. As you can see, 41.8% of them had septic shock versus the 19.5% of the rapidly recovering patients. Unsurprisingly, chronic critical illness patients had a higher max SOFA and max multiple organ failure incidence. However, there were no differences between groups with respect to their primary admission diagnosis. There was also no significant difference detected between groups with respect to in-hospital mortality. Although the only patients that did die in the hospital were chronic critical illness patients. Notably, there were also differences in discharge dispositions, with most of the rapidly recovery patients having good disposition, as we expect. They were discharged home with home health care and rehab. And those with chronic critical illness were more likely to have poor disposition, being discharged to LTACs, SNFs, hospice, or other hospitals. Now to the meat. The analysis of the genome-wide expression patterns revealed differential expression of 323 unique genes in the chronic critical illness versus the rapidly recovering patients at day 14. As you can see here in this gene expression map, heat map, the chronic critical illness expression pattern at day 14 looks very, very different from the control patients. However, for rapidly recovering patients, it looks more similar. It's not exactly the same. However, it does look more similar to the control patients. We used ingenuity pathway analysis as well, which is based on the transcriptome expression to predict significant up or down-regulated hematologic systems and functional pathways. We found that there was down-regulation of 25 functional pathways key to lymphocyte and innate immune cell function in chronic critical illness patients. These were not seen in the rapidly recovering patient. In fact, only four pathways were significantly down-regulated in the rapidly recovering patients at day 14. And all of these were important to lymphocyte movement and migration. Here I've listed just a few of those predicted IPA pathways seen only in chronic critical illness patients. All of these are important to immune function, key aspects of innate and adaptive immunity. Interestingly, we also found three pro-malignancy pathways up-regulated in chronic critical illness, and this was not seen in the rapid recovery patients. We also used IPA to predict upstream regulators based on the significant genome expression. These here are significant predicted upstream regulators for the chronic critical illness patients at day 14. Up-regulated pathways are denoted in orange, while down-regulated pathways are in blue. Upstream IPA analysis revealed completely different transcriptional regulation with chronic critical illness having activation of things like monocyte stimulating, colony stimulating factor, which is a pro-inflammatory protein, and also inhibition of IL-21 and forkhead BOX-01, which are both important for immune suppression and regulation of cell metabolism and differentiation. All of these have been targets for immunotherapy in cancer, so these regulators could be potential targets for immunomodulation and sepsis for chronic critical illness as well. Here, I have the predicted upstream regulators only seen in the rapidly recovering patients. Interestingly, IPA analysis predicted an increase in influence of beta-estradiol and IL-4 on gene transcription. There's also a significant decrease in NF-kappa B, which is a pro-inflammatory protein that was not seen in chronic critical illness at day 14. In conclusion, chronic critical illness patients clearly illustrate a pattern of dysfunctional and simultaneous inflammation and immunosuppression at day 14 after sepsis. In addition, their gene expression patterns reveal specific targets for therapeutic modulation. Since some chronic critical illness patients do recover, further analysis may allow the application of personalized medicine after sepsis. For example, prediction of which patients will require immunomodulation and the type of therapy required. I would like to thank the NIH for funding our research, as well as everyone involved in the Sepsis and Critical Illness Research Center at UF has made this work possible. I'd also like to thank my mentors, Dr. Lyle Moldauer and Dr. Phil Ephron, for their continuous support, as well as our surgery department chair, Dr. Upchurch. Thank you. I'd be happy to take any questions. Thank you. Nicely done. Thank you. Just had a question about the changes in cell populations you might predict in people that are very sick as opposed to ones that might recover. For instance, monocyte depletion or lymphocyte populations tend to be reduced in the more critically ill patients. Could you just comment on that and how that might factor into your results? Yes. So what we did here is we looked at total leukocyte populations, but like you said, the populations of the leukocytes are definitely different between chronic critical illness and rapidly recovering patients. We did not specifically in this one look at those populations. However, in the future, we're actually hoping to look at single cell RNA sequencing so that we can see the difference in these populations and the difference in the expression patterns or gene expression patterns for these populations, as well. That was what I was getting at. Yeah. Any more questions? Great job. Thank you so much. Thank you. All right. Our next speaker is Dr. Tuka Al-Khatib coming to us from East Tennessee University. Good evening, everyone, and thank you, Dr. Mabes, for the introduction. So my name is Touka Al-Khatib, and I am a PhD student in the internal medicine at East Tennessee State University. And today I'm going to be talking to you about S100A9-induced reprogramming of myeloid cells during sepsis. Now according to sepsis, as most of you know, in 2016, the definition has been reinstated as being a life-threatening organ dysfunction caused by dysregulated host response to an infection with or without septic shock. And of course, it's one of the most deadliest syndromes in the critical care units, and it affects mainly the critical ill patients in those units with their immune suppressed system and the many comorbidities that they may have. Currently in the US, there is more than a million sepsis cases annually that are present, and it's the most expensive condition being treated in the US hospitals. Now for the sepsis pathophysiological phenotype shift, it first occurs usually as an early acute hyperinflammatory stage, where it's also known as SIRS, systemic inflammatory rapid response syndrome. And this is where some early deaths around 20% may occur. Now if these patients survive, they might survive and have low-grade inflammation after the early initial stage of sepsis, and around 70% of those patients may further go on to something called a chronic immunosuppressive sepsis stage, where there's a higher chance of secondary infections to affect these patients, as well as around a 70% increase in mortality in these sepsis patients. And this may continue for weeks to months, even after they leave the ICU. And so the cause of, nowadays, the cause of most sepsis deaths is no longer inflammation alone, but instead the lack of the immune homeostasis and the persistent immune suppression and organ dysfunction. Now in my research over here, I'm going to be talking a lot about a type of cells called myeloid-derived suppressor cells, and where these cells come from is in the bone marrow. So we have the hematopoietic stem cells that go and mature into common myeloid progenitors that then develop into immature myeloid cells. And these immature myeloid cells, under normal conditions, can differentiate into macrophages, endodendritic cells, and granulocytes to go and function and just kill any infection and be useful in the infection stage. However, in sepsis, especially in chronic late sepsis, these immature myeloid cells have been shown to increase in significant amounts, and they have been shown to not differentiate into any of these mature cells, as well as they have shown to be more of an immunosuppressive mediators, and so they act more as an immunosuppressive. So one of the causes that might cause chronic sepsis for these cells to increase and not differentiate into any of these is that maybe the environment, for example, the chronic sepsis environment, such as an increase in cytokines, IL-10, might further just cause these cells to stop differentiating. And usually during the sepsis, if there's an emergency or an inflammation-induced myelopoiesis in the bone marrow, this is where it starts to increase in number, and this is used to preserve the innate immune cell function. Now in our lab for the sepsis model, we also use CLIP, cecal ligation and puncture, to treat myelopoiesis-induced polymicrobial sepsis in mice, in wild-type mice. And so the cecum is exposed after the abdomen, after an incision is made in the abdomen. There's a puncture made twice with a 23-gauge needle. The fecal material is extruded in a small amount, and we use a sham as a control for that surgery, where the cecum is taken out of the abdomen, but no ligation and no puncture is occurring to that cecum. And so with this level of injury over the past 10 years, it has been shown that with our model, the sepsis develops into an early hyper-inflammatory stage in the first five days after this CLIP surgery, and a late immunosuppressive stage after day five. And this closely emulates the human general peritonitis. And so this is a diagram also showing that we have an early sepsis after CLIP and a late sepsis just for comparative reasons, because we're trying to test what's going on in this chronic late sepsis stage. Now for S100A9, it has been known to be a known contributor to acute and chronic inflammatory processes, and it's induced in myeloid cells by many signals, including inflammatory mediators. It's released from a type of cells called GR1CD11B. And these GR1CD11B is what we use in our lab to characterize the MDS seeds. Now during early sepsis, they are not immunosuppressive. So during my talk, I'm going to be talking about GR1CD11B if they're in early sepsis. However, in late sepsis, I'm going to be referring to them as MDS seeds. But they are both the same. The main difference is that these are in late sepsis are immunosuppressive, and in early sepsis they're not. Now S100A9 release was found to be released from these cells during early sepsis. However, in late sepsis, the release from the MDS seeds does not happen in late sepsis. And what occurs is that the S100A9 actually moves from the cytosol to the nucleus. And this is associated with an expansion of the MDS seeds, as well as chronic immunosuppression and mortality. So S100A9 that was mainly known to be a pro-inflammatory mediator can also promote anti-inflammation and immunosuppression in late sepsis. So since S100A9 moves to the nucleus and increases MDS seeds, along with comparing early sepsis and late sepsis with LPS stimulation, the S100A9 protein is tremendously secreted from the early sepsis after LPS stimulation. However, in late sepsis, the secretion decreases significantly. Over here. And so next we know that S100A9 plays a role by going to the nucleus, increasing MDS seeds. So we were thinking that since the parent cells, they have exosomes, and these exosomes might contain DNA, RNA, protein that might affect the S100A9 protein function. So exosomes were shed from the late sepsis MDS seeds. And they were cultured with naive GR1 CD11B cells. And LPS was either not added or added to all. So either exosomes from sham, early sepsis, or late sepsis. And looking at the first two, with LPS and no exosomes, S100A9 is secreted. In sham and early sepsis, exosomes is still the same result. But when you add exosomes from late sepsis, the S100A9 protein significantly decreased in secretion. So this might, we might hypothesize or think here that these exosomes contain some mediators that affect the S100A9 function. So we were looking to see, did it have to do with the immunosuppressive phenotype of MDS seeds and increasing the immunosuppression? So exosomes taken from sham or early sepsis MDS seeds cultured with GR1 GD11B cells along with T cells. We were looking to see what's happening to the T cell proliferation and interfering gamma to look at the immunosuppression function. And so exosomes from the late sepsis MDS seeds actually switched the naive GR1 CD11B cells into the immunosuppressive phenotypes by decreasing the T cell proliferation as well as decreasing interferon gamma secretion. And looking at, using PCR, looking at the hot air level in the exosomes as well as in the RNA in late sepsis MDS seeds, you can see that hot air M1 expression is significantly increased. So hot air M1 is significantly increased in the exosomes and hot air M1 is actually a long coding RNA. And so this long coding RNA might have to do with the protein-protein modifications. And so we were thinking if you knock down hot air M1, what will happen to the immunosuppressive function of these MDS seeds? And so knocking down the hot air M1 in the late sepsis MDS seeds increased T cell proliferation and increased interferon gamma. And so now it did the opposite. Now it attenuated the immunosuppressive function of late sepsis MDS seeds without hot air. Now since hot air M1 is a long coding RNA and it may affect protein-protein modification, so we were looking to see does it affect S100A9 protein transportation from the cytosol to the nucleus? And so here is an RNA protein co-immunoprecipitation of S100A9 along with hot air and they did co-immunoprecipitate in late sepsis MDS seeds. But when we removed hot air from the late sepsis MDS seeds, S100A9 that usually moves from cytosol to the nucleus if hot air is present, here it actually happened the opposite without hot air. It just stayed in the cytosol. Now doing the opposite in early sepsis cells, we thought of transfecting these cells with hot air M1 and so the hot air M1 level will be increased significantly. S100A9 protein was there. They co-immunoprecipitated with hot air. And also in early sepsis where there's no immunosuppressive of MDS seeds supposed to happen when we added hot air M1, S100A9 transferred from the cytosol to the nucleus. And here's the hot air M1 level and these early sepsis transfected with hot air. So after expressing the hot air M1 in early sepsis and seeing that they might do something to this S100A9, move it from cytosol to nucleus, we tried looking at the immunosuppressive phenotype. And so a hot air M1 increased significantly in early sepsis, increased the arginase and the IL-10 levels significantly. And so these two are mediators of the immunosuppressive environment phenotype. And as well, they decreased TNF alpha significantly, which is a pro-inflammatory cytokine. So the main conclusions from these studies are that differential expression of long-coding RNA hot air M1 in late sepsis may contribute to the sepsis-associated immunosuppression. And hot air M1 may regulate the S100A9 transport from cytosol to nucleus during late sepsis. And so it inhibits the S100A9 secretion. Nuclear S100A9 may be responsible for switching the GR1CD11B myeloid precursors into immunosuppressive MDSCs. And so molecular targeting of hot air M1 to inhibit MDSC expansion in sepsis immunosuppression might improve the sepsis survival. Finally, I want to thank my supervisor, my PhD supervisor, Dr. Ilghazar, as well as the research assistants and postdocs working in our lab, Isato and AJ. And this research was, of course, supported by NIH grants. And I thank you all for listening. And I'll take any questions if anyone has any. Thank you. Hello. Hi. Nice talk. I might have just missed this, but from which body site did you harvest these MDSCs? Yes. Thank you for asking that question. So for the GR1CD11B or the MDSCs, these were taken from the bone marrow of the mice. So after we do the sepsis experiment in these clipped mice, we either wait five days or we wait more than six days. And we euthanize the mice, and we take the bone marrows and extract the cells using magnetic isolation. Hi. Burkett, again. I was just wondering how, two questions. One is, how did you guys do your exosome isolation? Was it differential centrifugation, or was it a kit? And then the second question was, what are the transcriptional regulators of hot air M1, and how does it vary in early and late sepsis? So for the first question, how do we get the exosomes? So this was using a kit from QIAGEN. It's very simple. They do filtration. You do centrifugation. And then it removes mainly everything that isn't exosomes or vesicles. And so in the end, you do get the exosomes. So we're certain that if it's a high centrifugation process, you could have other things like nuclear DNA, ATP, other modulators that might impair immune function. So that could be a potential confounder. For this one that we've used, the exosomes were... So before we even do this purification, we even filter before that, just to get rid of any other materials. And then this kit itself, it's very helpful because it removes everything that we don't need. It's any of the other proteins or the nuclear nucleic acids or anything that's not needed. And can you repeat your second question, please? What are regulators of hot air M1, upstream regulators? What makes it shift in levels? So that's a very good question. That's actually going to be in our future work of when hot air is present, how does it remove S100A9 and move it from the cytosol to the nucleus. So what we're thinking is that, of course, the immunosuppressive environment of sepsis along with IL-10, especially IL-10 and arginase 1, when these are present, so hot air M1 might maybe act as a transcription cofactor. And so it might change some of the promoters of these immunosuppressive cytokines. But I'm not aware of anything in the literature right now that might increase hot air M1. But in humans, for example, in mature myeloid cells, they are high. But in precursors, they're very low. However, in our situation here in sepsis, in late sepsis, they're very high in precursors of myeloid cells. So that means that they have to do somewhat, in our case, with the sepsis pathophysiology. Great job. Thank you very much. Thank you. All right. Our next presentation is going to be by Elliot Krauser, who's going to be talking about effects of monocyte distribution with and white blood cell count on a sepsis prediction algorithm. Sir, thank you. Thank you. Thanks for staying, everybody. I'm going to… A bit of a departure from the lab talks that were excellent earlier in this session. More translational here. I've been working for more than five years with Beckman Coulter to develop the biomarker of sepsis. And now we're trying to merge that with a machine learning algorithm. And I'm going to present this work today. And it's an interesting combination now of two companies and academic centers. And BARDA has now funded the project. So sepsis prediction algorithms can assist clinicians in a more timely diagnosis of sepsis. Machine learning algorithms have been, as you probably heard at this meeting, a number of those have been put forth and seem to be helpful for early detection of sepsis. We also published recently that elevated changes in the monocytes in the peripheral blood can be leveraged to detect sepsis. This is actually possible in the CBC. You get a CBC with differential, it can be added to the differential as a parameter. So readily available at the initial encounter with the patient. So we thought, well, why don't we combine them both together and see if that works even better. And this is just preliminary data from the machine learning side. A company called Desena, this company here, put together a really, one of the most innovative things about their algorithm actually is the ability to take data from the EMR, analyze it in a cloud-based kind of de-identified fashion, and then reconnect that information if you meet certain criterion such that they could actually make a phone call to a nurse who is taking care of a patient saying, your patient seems to be at risk for sepsis. And then they actually conducted a study, a smaller study in an ICU setting, and a prospective study. And they did just this. They called the nurses, and then time to antibiotics or intervention was assessed, and the outcome was assessed. And actually, dramatically positive results. There was a 21% reduction in length of stay, 58% reduction in mortality, and almost three hours reduction in the administration of antibiotics, mostly because this four-hour look-ahead, you could actually predict who was going to get sepsis later based on the criterion that we normally use. So it was a predictive model, not a detection and diagnosis model. And then the monocyte story, which I was going to merge with the other. Monocytes are interesting and important, as we've talked about in this session. The immune cells can be pro-inflammatory in the early phases of sepsis, and they can actually mediate an anti... or they can be participant in the immune suppression of late sepsis. They don't work as well in late sepsis. Interestingly, though, and I was at a meeting about 10 years ago in Germany, where they were... I met a member of the Beckman Coulter team, who was a pathologist, and they said, well, monocytes get really big, or actually neutrophils, they said at the time, get really big in the setting of infection. And I said, well, what if I hit my toe with a hammer? Wouldn't they also get big? And they said, no. It's only infection. And I said, well, if that's true, then this would be a pretty interesting parameter for use of detection of infection or sepsis. We actually looked at neutrophils first, but it turned out monocytes were more exact and better indicators of sepsis. And it wasn't the absolute size of monocytes or the average size of monocytes. It was actually a change in the distribution width or the range in size of the monocytes. So on the left-hand side here, you'll see a non-septic patient. Beckman Coulter analyzer can differentiate the different cell populations by light scatter properties. And then on the y-axis, you see the size of the cells. And as you can see here, the monocytes, they all conform to a range right in here on average. But when a person is septic, there are fewer of them. And then the range in size really increases off the scale of this particular analysis. Same was true of neutrophils, though, interestingly. What we also showed, I won't get into all the details of that study, what we showed that when the MDW wasn't giving you the same information as the white count, it was complementary to the white count. If you added the performance of an abnormal MDW in the setting of an abnormal white count, was more predictive of sepsis. So in this particular study, we're going to put them all together, we used a machine learning approach called gradient boost and trained the data sets using an algorithm based on vital signs and also adding in either the MDW or white count or both. And this was conducted in four hospitals in California and all had access to, they had access, full access to the electronic medical record as well as the MDW value in some of them. So the vital signs that they used for the training of the algorithm included diastolic and systolic blood pressure, heart rate, temperature, and respiratory rate. And the white count and MDW were collected all on a Beckman Coulter, all the hospitals used the same system. The standard use for the diagnosis of sepsis was based on a publication in JAMA by Ree recently, basically two service criterion based on the sepsis 2 standard. We use sepsis 2 because if you go to sepsis 3 and by the time you're in organ failures, your risk of death is about 23%, 23% or so. And that actually represents a subpopulation of sepsis deaths. If you were to include sepsis 2, all sepsis 2, more than half the deaths occur in people who present initially without organ failure. So in order to be inclusive, we don't want to miss people that are going to die of sepsis, so we want to include those people too. And so not only would you have to have a service criterion, but there had to be some evidence that microbial testing, somebody suspected infection. When we did just a vital signs analysis or vital signs with white blood count, there was an inclusion of almost 10,000 patients, 10% of whom met the sepsis criterion. There was only limited data from MDW, and so we'll show that data kind of separately. So looking at these patients who presented to the emergency department with vital signs only as the machine learning algorithm components, you can see that the negative predictive value was really quite good. So if a person did not conform to these vital sign criterion, they almost assuredly did not have sepsis. But the positive predictive value was poor, and that's a problem on the clinical side where you're getting a lot of false positive results. In the non-emergency department, the performance was a little bit better, but when you added the white count, you could see the positive predictive value improved, no significant change in negative predictive value, and likewise an improvement in the positive predictive value in the non-ED or admitted patients. And the area under the curve was very good for all of these analyses. Now again, it's very limited data relatively when you're adding in the monocyte distribution width, but if you look at the performance width when you add the monocyte, either the monocyte distribution by itself is better than the white count in terms of positive predictive value, and when you add them all together, monocyte distribution width, white count, and vital signs, performance was the best. Again a relatively small sample, but it looked to be improved, statistically was improved. Other standards that we use for early detection, we use MUSE for instance in our health system. MUSE we knew was not working very well. It's pretty good at detecting who doesn't have sepsis if it's normal, but if it's abnormal, it's a flip of the coin as to whether they were septic or not. And SOFA was even worse. SOFA scores were very nonspecific. If you look at the change in positive predictive value, because again I think that's a very important parameter when we're looking for an alert in the hospital. Arm fatigue is a big problem in hospitals, and one of the things they hate about the current sepsis, the MUSE for instance, is that they're getting way too many false positives. So if you were to add the white count, as noted before, the change in the positive predictive value from the baseline compared to just vital signs alone was significant when you added each component. Here's an area under the curve kind of comparison. An interesting, not only is the performance of the combined metric very good, area under the curve of .98, but if we look at the three hour predictive value, in other words three hours in advance of the onset of sepsis, the performance is almost equal with an area under the curve of .93. So in conclusion, the machine learning algorithm trained on vital signs together with the white count and the MDW resulted in very good, excellent performance over the currently available prediction methods. These results are going to require further validation and independent and prospective studies to show that they actually change behavior in the hospital setting. And so this is actually the recently funded project by the government, the BARDA program, where we're going to actually not only use the MDW, but some other hematological parameters including red blood cell size change, neutrophil parameters, and we're also going to expand the vital signs to include imaging and all kinds of other clinical information and then retrain all this. And the BARDA program actually is helping with the implementation of these things through, for instance, the FDA, such that they're also involved with how we conduct these trials to facilitate the rapid adoption of the technique or the technology. And then we're going to do prospective studies to show that it actually works. And actually we are looking for collaborators who are interested in being involved in that phase of the project down the road. So I mentioned the other parties, but by name from the Desana Company, and they did a lot of the analysis for this particular abstract, Sidney Lee, Samson Matarazzo, Jacob Calvert, Jaina Hoffman, and the founder of the company, Ritankar Das. Sarah Kehoe is from Danaher Corporation, which owns Beckman and Radiometer and Cepheid. So she's actually doing all the orchestration of all the different parties, including myself. Dr. Tejedor from Beckman Coulter helped, she and I designed the clinical trials. Michael Pan and David Persing have helped in the study design as well. With that, I'll open it up to some questions. Great talk. Did you look at sepsis prognosis at all? You know, we did a very preliminary analysis if you had septic shock or sepsis or just sepsis or infection. It goes up. It appears to go up. The MDW, for instance, goes higher and higher. But we use a cutoff value for some of these things. So you're just getting a binary. We did not try to train it, though, on these more severe phenotypes. I think that's a good idea. Or you're just like wondering if a patient survived or not. Oh. Hmm. No, we haven't done that except based on the severity of sepsis of presentation. How many of your patients were culture positive? Culture positive. Okay. So good question. Historically, in the first study where we validated the MDW parameter, only about 40% have a positive culture. Actually, it was even less than that. And I think that's historically about the norm. But as far as positive cultures in this larger 9,000 patient cohort, I think it was in the 40% range. I don't know if that data is there. I just don't have it in my head. When we talk cultures, we talk bacterial cultures, viral cultures, and probably other forms of fungal cultures as well. So would there be a difference in the MDW-based... You are tying it in with antibiotic administration, right? It would be great to have a test telling us, okay, fine, this patient has sepsis because they infected, but it's not bacterial. I totally agree. We actually tried to do a influenza study towards the tail end of the influenza season a couple of years ago. We got a little bit of data from that. And influenza clearly makes the MDW elevated, for instance, and all those sepsis parameters are very common in influenza. So I think for the influenza, there might be a different pattern though if we go to the next level. If we're looking for these patterns that the machine learning algorithm can detect, I think we will see different patterns emerge that would help us say, okay, I think this is a flu and not a bacterial, and maybe there's a different pattern for fungal infection as well. I think that if we have enough input, we would be able to separate all these. And that is the plan. But I don't know. We'll have to see. We don't have enough data yet. All right. Are there any more questions? All right. Well, thank you so much for attending this afternoon's session, and thank you very much to our four speakers. Those were fantastic talks. I really enjoyed them.

sepsis

red blood cells

multi-organ failure

ATP export

leukocyte transcriptomes

chronic critical illness

gene expression patterns

sepsis prediction algorithm

machine learning algorithms

Immunology

Inflammation

Multiple Organ Dysfunction

Presentation

Select

Professional

Associate

2020

Advanced

Sepsis Resources