All right. So I changed the title of my talk just a little bit about precision medicine in sepsis. Is this a fairy tale? Because I want to oppose this question to us and hopefully by the end of today we'll have some great talking points. So I do have some disclosures. I am not an MD and therefore I am not suggesting any therapeutic treatment for sepsis. My research is funded by multiple parts of the NIH and I have contributed into pharmaceutical sponsored trials for COVID-19. So our lecture objectives today are by the end of this lecture we're going to talk a little bit about that foundational knowledge and what we've learned about from urine models and why these may not have actually translated into successful clinical therapies. And we're going to understand some of the considerations that we need to think about when we want to apply precision medicine to sepsis and with a big focus on enrichment and population design for better and more promising clinical trials. Now this audience needs no background into sepsis, defined as the life-threatening organ dysfunction caused by a dysregulated host response. We've known about sepsis forever. In fact, it was written about by Homer in the Iliad and Odyssey and the term sepsis actually comes from Greek, sepo meaning I rot. So there are mentions of sepsis all the way back in history. I think most of us are very familiar with the sepsis three criteria that, you know, we've been using to make that clinical definition and in fact at this meeting there's going to be the Phoenix criteria for those interested in pediatric sepsis. So we're very interested in defining this syndrome, this dysregulated immune response. And I think through alliances such as the Surviving Sepsis Guidelines through SCCM, through the Sepsis Alliance, there is a more, there's more awareness about sepsis. These campaigns have been great. And we can talk about demographics, we can talk about statistics. No one is going to ever question that sepsis is a major threat globally and in the U.S. due to its cost and its toll on human life. And in fact, what people may not realize is that the first Surviving Sepsis Guidelines were 20 years ago this year. SCCM's Surviving Sepsis Guidelines were first written in 2004. So we are 20 years still using the same therapeutics of early fluid or vasopressor antibiotics and things like that. And so again, even the CDC recently has put out this sepsis campaign, you know, over a million deaths annually. And so in 2017, the World Health Organization actually put forth this sepsis document. It's actually very well done if anyone hasn't read it. But basically they put forth this resolution by the World Health Organization to improve the prevention, diagnosis, and clinical management of sepsis. So members from SCCM, the European Society of Critical Care, and scientists and clinicians from around the world have come together to really address this important clinical phenomenon that hopefully we can contribute to that conversation today. So if you read that World Health Organization, you'll see that there's this diagram in here. And I think anybody who's familiar with the sepsis literature will probably agree that this is probably one of the most comprehensive figures when you're talking about the paradigm of sepsis. Because it takes into consideration a lot of factors that people might not think about. First of all, it talks about the pre-hospitalization state, at risk, and it actually makes a point to focus on the pathogen itself. As an immunologist, we know that the immune response to a bacteria is going to be different than immune response to a fungus, which is going to be different than immune response to a virus. And then it looks at what is actually occurring during the acute sequela of sepsis. But also it looks at outcomes of sepsis. We know that patients are surviving sepsis, which is great. But patients who do survive sepsis and that index hospitalization are also at risk for recurrent infections. They're at risk for recurrent hospitalizations. And part of that is probably attributed to immune paralysis that we don't fully actually understand. So what has the research shown? So if you go to PubMed and just type in sepsis, when I did this PubMed search back in December, I got over 217,000 articles. I did not read them all, I fully admit. But even if you filter to English only, because I only speak English, and exclude preprints, that's still 192,000. I guarantee if you did that same search today, you're going to find even more. And so of these, we have published in peer review journals over 8,000 clinical trials on sepsis. But yet, as Dr. Rumi alluded to, we still have basically our clinical therapeutics, our fluid management, antibiotics, and things like that. And part of this is when we think about translating from the bench to the bedside, or basic science and how we're advancing the field, we have done an amazing job at preventing mouse sepsis. And I use the term preventing because when you read these papers, oftentimes these mice are actually given something and then sepsis is induced. I am not going to walk into the ICU in Kentucky and say, hey, I think I'm going to develop sepsis in 24 hours, please go ahead and pre-treat me. And this is great from a mechanistic point of view. I personally use mouse models in my own lab. We have gained leaps and mountains of information from these models, but we probably need to start thinking about the models that we're using. And if you compare them, obviously we are not mice. But if you actually look at all of the different components of human physiology versus marine physiology, there are some things that are drastically different. First of all, neutrophils are the predominant white blood cell in our circulation. We use the CBC with differentials, one of the most common labs drawn every day. We're looking at that neutrophil count. Mice, it's actually lymphocytes. And so when you're looking at marine data, the predominant lymphocyte in blood circulation of mice are actually lymphocytes. Endotoxins, you see a lot of data about using LPS and endotoxin. Mice actually have an inherently high resistance to endotoxin. In addition to fever, when we have an outward sign of infection, we actually develop hyperthermia or pyrexia, fever. Mice actually develop hypothermia. It's a completely different thing. Their microbiota are different. And I bring these up here because whether or not you may think about these, all of these are actually very important in terms of the immune response. And so I bring this up because when you're looking at sepsis literature and when we're thinking about sepsis as a field, wanting to translate what's been discovered in the lab into the clinic, we really need to be thoughtful and think about the model. There have been lots of models about endotoxemia, where mice have been injected, they have been instilled, they have inhaled LPS, which is supposed to mimic a gram-negative bacteria. Humans do not get exposed to sole LPS. We get exposed to pathogens. And if you actually look in the pediatric space, many pediatric cases of sepsis are new to gram-positive pathogens, yet we still rely on gram-negative models. Cecal ligation and puncture. No surgeon is going to punch a hole in a patient's cecum and then sew them up and see what happens in four to six hours. I actually use a cecal slurry model. It's a useful, great model in our lab. But again, this is not exactly relevant or maybe doesn't mimic the exact phenotype of what happens. And so, for those of us who are on the bench side, or if we're interpreting this primary basic literature, we really need to be thoughtful and think about the models that we want to use to put forth and to translate. And so, what I think about when I think about precision medicine, some of the things that I think that we should all think about and have discussions on are about that presepsis state. Just like in Ken's house, they have, when you have headache, no two patients are going to look the same before they get sick. You need to think about anatomical location. The mucosal immune response of the lung is different than the mucosal immune response of the gut. This is different than the mucosal immune response of the skin. Timing. We know that antibiotic administration is imperative to be fast. Perhaps even faster is that immune response that is recognizing these pathogens in seconds. Types of infection and understanding the immune response as a whole. As an immunologist, I could talk about the immune system all day, but I want to just bring up some key points that I think we should think about in terms of the immune response. So, when I think about sepsis, I think about the Goldilocks theory of immunology. I don't know if any of you are familiar with this, but if you know the story of Goldilocks and the Three Bears about how the porridge was too hot or the porridge was too cold, your immune system wants to maintain homeostasis. It wants that porridge to be the exact right temperature. And the immune system wants to do the same thing. When our immune system is overreactive, we develop diseases such as hypersensitivity, autoimmunity. And when our immune system is underreactive, we have immunodeficiency diseases. Sepsis occurs in cancer due to cancer-induced immunosuppression or secondary to the chemotherapeutics. Patients who have received a transplant, we purposely immunosuppress them so that they can not reject the organ. And so sepsis is actually both sides of the seesaw because the immune system has overreacted, but also it wants to maintain homeostasis, so it actually suppresses itself. And so when you have to think about the Goldilocks theory, you have to think about where you are in that sepsis timeline to think about how you want to approach this important clinical question. So thinking about the pre-sepsis state, I live in Kentucky. We're one of the nicest states in the country. We actually were voted that way. It's a beautiful state. We have horse racing, we have a lot of bourbon, we have a lot of basketball, and we really love our basketball, and it even shows up on the church signs. So Kentucky's a great place to live. But Kentucky also, we lead the nation in cancer incidence, second in cancer deaths, we are the sixth most obese state, we are first in opioid prescription rate, ninth in drug overdose deaths, second in smoking prevalence, we lead the nation in obstructive lung disorders, we lead the nation in per capita COPD, we are fourth in the nation in deaths from sepsis, we're the fourth poorest state, so social determinants of health, and eighth in diabetes incidence. So our pre-hospital state, our pre-sepsis state may not look like your pre-sepsis state. We have patients who come to our ICU who have never seen a PCP, who have never been vaccinated, we are in the middle of a COVID-19 surge right now, and every single patient last week in our ICU, none of them had received a single COVID-19 vaccine. So we sometimes forget about this pre-hospitalization state of sepsis, because when we are using mice, they are all identical, they're all genetically identical, and they've all had the same housings, the same beds, same diet, and all this kind of stuff. And so I think about, when we're thinking about precision medicine, we need to think about the pre-hospital state of the patient. The other thing I think about is when you read clinical trials, there's always inclusion-exclusion trials. Almost all sepsis and ARDS trials, you have to have an ICU patient with a proper liver function, especially if you're doing investigational drug trials. A MELD greater than 25, or Child's Pew greater than this. I can tell you in Kentucky, that is extremely hard to find. We have, over the last year, we have screened for a clinical trial, we have screened over 300 patients, none of them have had a MELD less than 25. That's due to whether alcohol, hepatitis, or non-alcoholic fatty liver disease. So when you think about how your ICU looks, if you're purposely excluding patients with a liver issue, that may not be representative when you think about clinical trial design and development. But from a scientific standpoint, I understand perfectly, this is an investigational drug, we have to make sure that we're not inducing a liver injury. So it's a hard question, I don't have an answer, it's up for discussion. But I think we often forget about the pre-sepsis state of our patients when we're thinking about the clinical outcomes of sepsis. When the right time and the right place, when we think about anatomical locations, we know, as I said earlier, the immune response is different. For example, I really like this review that was in Lancet Infectious Disease, and they look at the immune response in the lung, blood, bone marrow, liver, lymph node, and spleen. And they actually show that over time, the cell populations are changing, HLA-DR, which is a really important MHC class II molecule in antigen processing and presentation. And they actually say that, look, across the time and across these anatomical locations, there are changes that are tissue specific. So again, when we put all sepsis patients together, whether you're looking at urosepsis, sepsis secondary to pulmonary infections or things like that, we might be not taking into consideration the anatomical locations. Timing is everything. We know that early detection, early screening, early antibiotics or early fluids are all important. And I believe that there's data that for every hour that antibiotic administration is delayed, the chances of worse outcomes are increased. But when you're thinking about the immune response, the innate immune response, our neutrophils, our macrophages, our monocytes, they are recognizing these pathogens and mounting an immune response in seconds to minutes. And so your immune response, and we all would not be sitting here if we did not have an innate immune response, is actually working at the minute and second level. And so I think we need to go back to our foundational immunology knowledge and think about this when we're thinking about precision medicine. And the enemy, again, the immune response to these types of pathogens is very different. How your immune response recognizes and fights off intracellular versus extracellular pathogens is very different. And so going back to the immune response, which I will fully admit is complex. We have to think about all of these confounding variables in the immune response. The acute phase response, which is driven by the liver. Again, we use CRP as a clinical lab. CRP is made by the liver in response to interleukin-6. But our patients who have liver issues, their CRP may be different. And so can we use CRP as a clinical variable or biomarker of infection? Fever, there's been clinical trials looking at permissive pyrexia in the immune response. Immunologists like to make things interesting. We're now discovering even more cells. There's a lot of research literature recently about mate cells. How many of you know what a mate cell is? All right, I like some immunology in the room. It stands for a mucosal associated invariant T cell. They're found in mucosal sites and they have a really important immune response. And so should we start to think beyond a classical or conventional T cell? Should we start to look at gamma delta T cells? Should we look at mate cells? What about ILCs or innate lymphocyte cells? Complement, we have complement activation. The adaptive immune response, T cells and B cells. Immune training, there's now data about the innate immune response, which we have thought for years had no concept of innate immunological memory. We now understand there's what used to be called of endotoxin tolerance is now actually known as innate training or innate immune memory. And coagulation and the microbiome. All of these, which people probably think about in their own unique niche or whatever they are interested in, but they all work together as part of a very complex immune system that I think needs to be taken into consideration when we're talking about precision sepsis trials. And some of the best evidence for the microbiome comes back from the mice. If you actually do experiments with mice that come from Jackson, which is a vendor, versus mice that come from Charles River or Taconic, they're three different vendors, your results are going to be completely different. They can be all genetically the same. You can bring them to your university. They can be treated the same, but it turns out those mice have very different microbiota because one of these mice from Taconic is actually lacking short filamentous bacteria in their microbiome. And so now people actually encourage the use of buying mice from pet stores because mice who are actually outbred, you can catch them in a field, field mice, or pet store mice actually have microbiomes that mimic a human microbiome much better. And so there's actually some talks here at this conference about the microbiome in sepsis. And so a lot of the research in sepsis and a lot of those clinical trials have been targeted at cytokines. TNF is a big one. We've known for years that TNF, let's target TNF. The first clinical trials of blocking TNF were woefully ineffective and in fact did some harm. And when fortunately during the COVID-19 pandemic, cytokines kind of got like this reemergence. People all said, oh, well cytokines are the driving force of everything. These cytokines are so important. Let's start blocking cytokines. Well, at University of Kentucky, we have a pulmonary critical care biobank. And so we collect blood, BAL, and exhaled breath condensate on patients in the ICU. And what we did is we immunophenotyped patients in our ICU that had COVID-19 or had sepsis from bacterial causes. I was even specific that these were bacterial sepsis patients. And you can see from this heat map, I won't go onto it. The only cytokine that was actually higher in COVID-19 patients was soluble CD30 and CXCL5, which is a chemoattractive. And going back to that COVID-19 literature, you can look at all of these cytokines that were actually trialed in clinical trials. Anakinra, Dupixent, Tocilizumab, Dupixent for IL-13. All of these were actually trialed, but these levels were actually higher in sepsis. And I'm not the first person to show this. This has been published by other groups, Carolyn Calfee, Michael Mathay. Other groups have actually said that, look, we started to target cytokines in COVID-19, but wait a second, we know that cytokine levels are actually higher in sepsis, and we should start to think about that. One of my favorite reviews that came out of the pandemic was this article by Jason Mock at UNC that said, wait a second, IL-6, maybe not be the holy grail that we think about. And so one of the drugs that did emerge from the pandemic was a drug called Baricitinib. Those of you who are clinicians, I'm sure you may have used it or know somebody who has. Baricitinib doesn't target a single cytokine or a single cytokine receptor, like Tocilizumab targets the IL-6 receptor, Adalutamab targets TNF, the cytokine. Baricitinib targets the JAK-STAT pathway, where these cytokines converge on signaling. Now, if you go to clinicaltrials.gov right now, well, at least when I did this search a couple of weeks ago, there's not a single trial looking at Baricitinib in sepsis. So should we go back and say, if it does have clinical efficacy in COVID-19, should we try maybe pancytokine inhibitors in sepsis? I don't know. Again, I'm not advocating for that. I'm saying maybe we should think about it. And part of the issue with precision medicine in sepsis is that when we do our classical randomized controlled trial design, we are looking at this heterogeneous group of patients that Dr. Remy alluded to, lumpers. Let's take everybody with sepsis and lump them together. But we know that that was, as I said, that preclinical state, the immune response could be different. And so these RCTs are meant to design, to look for what's called an average treatment effect. Did these patients with sepsis, did they get better? Did they have improved outcomes? Did they have improved survival? Less ICU length of stay? With that major assumption that this treatment effect will be for all. But we know that people are heterogeneous. It's actually called heterogeneity in treatment effect, HTE. And so being a basketball fan, going back to this quote from Michael Jordan, there is no I in team, but there is in win. And so maybe we need to start thinking about trial design, not as an average treatment effect, but as an individual treatment effect. And so work by the great, impressive Dr. Wong actually has talked about this for years. And he's actually said that sepsis is a syndrome and that we can't start to think about people as a one size fits all. And so when we start to use these words of precision medicine and personalized medicine, I think a lot of people get nomenclature intertwined. But what I think we need to think about is how can we find a treatment that's gonna be best for that individual? And it's hard. And so this was a really great review that he wrote in Nature Reviews Nephrology a few years ago. And he actually talked about this, this need for patient enrichment when we think about clinical trial design. So this review, this clinical paper came out not too long ago. It's called the PROVIDE trial. And what they actually did is they actually tried to, in real time, assign patients treatment. And they used HLADR and ferritin as their biomarkers of sepsis. And what they showed is they said that these patients would benefit from recombinant human interferon gamma to boost the immune response. Whereas these patients would benefit from anakinra or IL-1 blockade. When you read this paper, you'll find that these groups only had an N of 15 patients. So it's a very small trial. But the amount of work it took to actually define who had immunoparalysis, who had macrophage activation-like syndrome, the molecular work, the groundwork that it took, it took a few days. So actually, these patients were diagnosed with sepsis and they actually weren't treated or grouped for about 48 hours later. So I think this is a great story. And I think it's a very fascinating article. And as an immunologist, I loved it because it actually kind of has taken into consideration a lot of these things. But in practicality, I don't think any clinician would say, we need to wait for 48 hours to add a treatment in sepsis. But this group has gone on and they're expanding this. They have this clinical trial. It's called Immunosep, personalized immunotherapy in sepsis. It's currently in Europe. There are no US sites for it, but they want to expand their protocol of either the recombinant human interferon gamma or the anakinra or the IL-1 receptor blockade into a larger subset for feasibility. And so going back to the Goldilocks and how do we define that population that's just right? Maybe we need to start thinking about what we've learned from asthma. So during my PhD and postdoc, I did a lot of work with asthma. And about this time, this is when the whole asthma endotypes came out, the clusters that the SARP, the Severe Asthma Research Program published. And they did a massive profiling of asthma patients. And so this paper recently came out. It says, predicting sepsis severity, the role of endotypes and mechanistic signatures. And this group did exactly what Ken alluded to. They actually took a huge cohort of sepsis patients and they actually clustered them. They called them sepsis, they gave them very complex names. Their naming was not exactly simple and straightforward. They have these big, long acronyms. And it's very interesting. But it was all done at a molecular level, which again takes time, it takes resources, it takes bandwidth to do the sequencing that these patients had. So I think we're on the idea, but how do we make this feasible and applicable in real time in the clinical setting? So one of the questions is, not everybody is gonna have the bandwidth, the capabilities, the machinery, the cost to do advanced immunophenotyping by flow cytometry or sequencing. Can we endotype sepsis based on clinical variables alone? And I know that this is small and hard to see, but this paper came out in JAMA in 2019 where they actually looked at clinical variables such as AST, you can see CRP, heart rate. And so a lot of this work has also been done in ARDS, but looking at routine clinical variables, do we need to go back and look at the data and say, is there a signature of routine clinical variables that we can use to better endotype patients? And maybe that is what we need to think about when we start to lump patients who have a similar clinical presentation together before we design our clinical trials. And also hindsight is 20-20, right? There's been lots of clinical trials that have done in sepsis and people are starting to go backwards and look at that data. And I know this is controversial about steroids and sepsis, but I bring this up because there are two studies that have actually gone back and looked at multiple clinical trials done in the area of steroids and sepsis. The APPROACHES trial, these major landmark trials about looking at corticosteroids and sepsis. And both of them used a data-driven approach, artificial intelligence, machine learning, and they found that in these groups, there was an individual treatment effect had those patients been grouped more similarly. Now, hindsight's 20-20. It's a lot easier to go back and look at clinical trials and say, oh, you should have done that. But in the real time, how would we make this feasible? How can we do this in real time in the clinic? And so moving forward, I think we need to think about making better clinically relevant animal models that are standardized and reproducible. We talk a lot about data reproducibility and clinical trials across sites. We don't do that in the sciences. Sharing reagents, being more open, and having a reproducible basic science that can help us inform better designed clinical trials. We need to appreciate the presepsis state of the patient and the complexities of host-pathogen interaction. And we need accurate, reliable, and rapid ways to standardize immune profiling of patients with sepsis. And so with that, I'm happy to take questions, but I think we're actually gonna have a Q&A session at the end. Thank you.

precision medicine

sepsis

challenges

patient-specific factors

endotypes

clinical trials

advanced technologies