Okay, welcome everyone, and thank you for the jazz hold music, which was really quite a relief after being downstairs. I don't know what they're playing in the ballroom. Has anyone been down there and you're listening to that music? It's kind of creepy. Okay, my name is Michael Shishady, I'm from the University of Pennsylvania. I have the honor of helping to moderate our STAR research presentations in sepsis. For those of you who came to see Jose Diaz Gomez, he's extremely important and was pulled away to something else, so you'll have to try and find him somewhere out there. I'm going to do what I can to take care of this. We have four abstract presenters. We have an hour for the session, so by my rough calculations, that gives us 15 minutes per presentation. I want to encourage everybody who is in the audience to think of and ask questions at the microphone after our presenters have a chance to speak. And I also just want to comment and congratulate our presenters. The STAR research abstracts constitute the top 64 scored abstracts among circa 2,000 or so that were submitted. So congratulations to our presenters, and we are going to start with Dr. Jane Whitney, who is from Boston Children's, and she is going to speak, and I'm sort of looking at the screen because I'm hoping that there might be a title slide, but I will not get that title slide. So I'm going to say that she's going to speak on biomarkers and association with concurrent and subsequent multi-organ dysfunction in pediatric sepsis. Thank you, everyone, for the opportunity to share this work. My name is Jane Whitney, and I'll be presenting endothelial biomarkers are associated with concurrent and subsequent MODS in pediatric sepsis. I'd like to disclose that this work was in part supported by my T32 from the NIH. Endothelial disruption contributes to the pathobiology of multiple organ dysfunction syndrome, or MODS. Whether early endothelial disruption is associated with the incidence and severity of MODS over time in children with sepsis is unknown. So our aims to address these questions were two. First, to determine if early endothelial disruption was associated with the incidence of concurrent and subsequent MODS in septic children. We hypothesized that early derangements in endothelial biomarkers would be associated with the incidence of concurrent and subsequent MODS. Our second aim was to determine if early endothelial disruption was associated with the severity of concurrent and subsequent MODS in septic children. We hypothesized that early derangements in endothelial biomarkers would be associated with the severity of concurrent and subsequent MODS. Our methods to address these aims was to conduct a secondary analysis of a prospective cohort study, which had enrolled children less than 18 years of age who had sepsis with two or more organ dysfunctions or shock, or children with sepsis-associated ARDS, acute respiratory distress syndrome. They were recruited from an academic pediatric ICU between 2014 and 2017. So some definitions. Sepsis for the purpose of this study was defined according to international pediatric consensus conference criteria, which were in use at the time the first patient was enrolled. MODS was defined as two or more organ dysfunctions defined by the IPSCC criteria. And sepsis-associated ARDS defined patients who met criteria for sepsis by IPSCC criteria and also met ARDS by Berlin criteria in use at the time the first patient was enrolled. And children with sepsis-associated ARDS had to meet criteria for both sepsis and ARDS simultaneously. So the biomarkers of interest were those associated with vascular permeability, including angiopoietin 1, 2, the angiopoietin 2 to 1 ratio, vascular endothelial growth factor, or VEGF, soluble FMS-like tyrosine kinase, or SFLT, which is VEGF's receptor. They also included biomarkers associated with chemotaxis, including macrophage inflammatory protein 1-alpha, or MIP-1-alpha, regulated on activation normal T cell expressed and secreted, or RANTES, and monokine-induced by interferon gamma, or MIG. Also we looked at biomarkers that have been associated with neutrophil adhesion to the vascular endothelium in sepsis, including intracellular adhesion molecule, ICAM, vascular cell adhesion molecule, VCAM, e-selectin, and andecan, and finally biomarkers that have been associated with platelet activation in the context of sepsis, including von Willebrand factor of VWF and thrombomodulin. This timeline, this graphic describes the timeline of our study. So children were diagnosed with sepsis and then were discharged from the hospital later if they survived. Time point 1, T1, occurred either at or within 72 hours of sepsis diagnosis. Time point 2, or T2, occurred 7 to 14 days after T1. Blood was drawn at T1, and this is the blood from which the biomarkers were measured. Both the incidence and severity of MODS were defined at both T1 and T2. Severity was defined using the PLOD score. So concurrent MODS is MODS that occurred at T1. It was measured at T1, and it occurred at the same time that the biomarkers were measured from the blood we drew. Subsequent MODS was defined as the incidence and severity of MODS occurring at T2, which was after T1, after the time the blood for biomarker measurement had been drawn. This is the PLOD 2 score, Pediatric Logistic Organ Dysfunction 2. This is a validated measure that we use to summarize MODS severity. So as you can see across the top, different points are assigned for different severity of organ dysfunction based on sort of age-appropriate norms and expected values. A higher PLOD 2 score indicates worse severity of MODS. So our statistical methods. In order to determine the association between the endothelial biomarkers and the incidence of concurrent and subsequent MODS, we used multivariable logistic regression, adjusting for two potential confounders that we thought could affect the outcome, the exposure, or both without being in the causal pathway. The confounders we considered were one, pulmonary site of infection, and two, baseline technology dependence. For number two, the way we evaluated the association between endothelial biomarkers and the severity of concurrent and subsequent MODS was by using multivariable ordinal logistic regression, again adjusting for those two possible confounders. This table shows the demographic characteristics of children who either had or did not have MODS at the two time points of interest. You can see they were relatively similar in terms of age, sex, race category. Technology dependence was interesting. The kids who were technology dependent at baseline actually had slightly lower rates of MODS at both time points. And then sepsis source being a pulmonary site of infection was no different between the two groups. So the incidence of concurrent MODS, that is, MODS measured at T1 the same time the blood was drawn, was 78%. Most of our patient population had at least two organ system failures at this time point. And the presence of MODS at T1 was associated with three biomarkers, angiopoietin 2 with an adjusted odds ratio of 2.2, VCAM with an odds ratio of 2.5, and RANTES with an odds ratio of 0.7, indicating lower adjusted odds of MODS. Looking at the incidence of subsequent MODS, 24% of our population met criteria for MODS at T2. And the incidence here was associated with multiple biomarkers we measured. There was a positive association between angiopoietin 2 and the incidence of subsequent MODS with an odds ratio of 3.3. The odds ratio associated with angiopoietin 2 to 1 ratio was 1.1. An odds ratio of 1.6 described the association between both VEGF, VCAM, and the incidence of subsequent MODS. The odds ratio associated with von Willebrand factor was 1.8 and 0.5 for RANTES, indicating lower odds for subsequent MODS with that final biomarker. So now looking at the severity of concurrent MODS, nine out of the 16 biomarkers we measured did show that there was a significant association between T1 biomarkers and the p-LOD score also measured at T1. The negative coefficients you see associated with ANG1 and RANTES showed that there was a negative relationship between the biomarker value and the p-LOD score describing the severity of concurrent MODS. The remainder of the coefficients are positive, suggesting a linear relationship. Looking at severity of subsequent MODS, there is a positive association between ANG2, ANG2 to 1 ratio, von Willebrand factor, and endocan, and the severity of subsequent MODS. And there is a negative coefficient describing the relationship between RANTES and the severity of subsequent MODS. So in conclusion, early derangements in endothelial biomarkers in critically ill septic children were associated with both the incidence and severity of concurrent and subsequent MODS. There's several limitations to this study. First, we were testing our hypothesis in a small cohort of patients that was enriched for MODS by virtue of the inclusion criteria. This was done at a single center. The lack of serial biomarker measurements certainly is a limitation, and one that is encountered commonly in small cohort studies because longitudinal biomarkers can start to have their missingness informed by outcome. For example, patients who do not survive are less likely to get follow-up blood draws for biomarker measurements as are patients who rapidly recover. Finally, the timing of biomarker measurement in our study occurred quite close to the time of sepsis diagnosis but did not coincide, of course, with the onset of sepsis physiology. Future directions are two. First, endothelial biomarkers may identify high-risk phenotypes within this clinically heterogeneous syndrome that many of us care for at the bedside. This may be helpful to clinicians because it may help inform risk stratification. It could be helpful to trialists for predictive enrichment. Second, endothelial biomarkers may help to define distinct biologic mechanisms that's driving the varied sepsis phenotypes we see clinically. This would require serial measurements, I believe, of the biomarkers and the outcomes of interest, and I think would also require extending the type of methodology I've shown here with proteomics, just measuring the biomarkers of the protein in the blood to a multi-omics approach where we take this a few steps further to look at the proteins but also the transcriptomics, the genomics, even the epigenomics driving the biology behind these distinct phenotypes. I'd like to acknowledge my mentors and collaborators, and I'm happy to take any questions. If people want to come up to the microphone, I'm going to ask a question. I'm curious, I don't know if you had a chance to take a look at this, but how well early mods predicted subsequent mods? It seemed to me maybe not that well, because if I read your slides correctly, there was about three quarters of the patients who had early mods and only about a quarter who had subsequent mods. Yeah, that's absolutely correct. This being a pediatric cohort, and a small one, though, I think a representative one, the incidence of early mods was quite high, and one week later, it was only a quarter. So this sort of bimodal distribution has been described in mortality associated with pediatric sepsis, and it goes along with the burden of mods in pediatric sepsis. Patients who are very sick and have very poor outcomes who don't survive pediatric sepsis tend to die very early, or they die late of new and progressive mods, which wasn't prevalent in this sample. Overall, the mortality was about 10%, and many of the patients who were in the early mods group had resolution of mods by the second time point. I'm going to ask one more question. You put up groups of biomarkers at the beginning based on function, endothelial and neutrophil migration and that sort of thing. Did you feel from your results that there was any one group or couple of groups that were more overrepresented in their association with mods compared to any other? I did. The biomarkers that have been coming up most consistently in this study and others I have done are the group of biomarkers associated with endothelial or venule leak, for example, the angiopoetins, and then also the ones associated with neutrophil migration, including the sort of ICAM, VCAM adhesion molecules. So those ones keep coming up time and time again, and I think those are the groups that have the strongest signals in this study as well. Okay. Thank you very much. Thank you. Okay. Next, we would like to welcome Amanda Ward from Geisinger Medical Center, who is going to present on the Constitution of the Pediatric End Organ Dysfunction Score, a Sepsis Predictive Model. That's OK. I understand. We will be moving on from Amanda Ward's presentation right now to if we can actually queue up the presentation that is going to be given by Dr. Gretchen Satcha on the effect of lactate catecholamine dose and timing in patients with septic shock on vasopressin. Thank you for the introduction, and thank you everyone for coming out to hear this today. So as he said, we're going to be talking about the results of a project that I've been working on over the past year. Arginine vasopressin or vasopressin, when exogenously administered, results in V1 receptor-mediated vasoconstriction, V2 receptor-mediated resorption of free water, and in patients with septic shock, acts as a supplement in what is known to be a relatively vasopressin-deficient state. In the 2008 landmark study, the VAST study, which randomized patients to either norepinephrine alone or norepinephrine in combination with vasopressin, they found no overall difference in 28-day mortality comparing the two treatment arms. But they did find some subgroups that might have hinted at some signal of benefit with the use of vasopressin, and that is when patients were initiated on vasopressin when the norepinephrine dose was less than 15 mics per minute, when the lactate was less than 1.4 at vasopressin's initiation, and within 12 hours from vasopressin's initiation. You can see that in all of these three subcategories, there were lower rates of mortality when vasopressin was administered. However, this is more hypothesis-generating in nature because all of these were subgroup analyses, and you can also see that looking at the timing piece within 12 hours did not actually reach statistical significance, and there was also an interaction detected between lactate, the intervention, and the outcome. Regardless of the overall lack of benefit with vasopressin's use, the 2016 Surviving Sepsis Campaign Guidelines still do recommend its use as an adjunctive second-line agent to either increase the patient's MAP or decrease the patient's norepinephrine dose requirements. And as I said, there is some data that might indicate that earlier initiation of vasopressin is of benefit, but this isn't consistent amongst the small randomized controlled trials that have looked at this and some retrospective evaluations, as well as the VAST study that I just mentioned. This leaves us with a glaring question of when should we be starting vasopressin if we are to be utilizing it in our patient population? You can see the average lactate norepinephrine dose and time from shock onset at which VAST and VANISH both started vasopressin in the patients who received it, but we're not really sure in clinical practice where it's being used. My hypothesis from all of this is that starting vasopressin earlier might be of benefit to our patient, which is the current study goal, which was to evaluate the effect of lactate, catecholamine dose, and timing of vasopressin initiation on mortality in patients with septic shock who received vasopressin, with the primary objective being to compare quintiles of all of these three variables on hospital mortality, and then secondarily to conduct a multivariable logistic regression to determine the association of each of these factors on mortality after adjustment for confounders. This was a retrospective multicenter observational database study that evaluated patients admitted to the Cleveland Clinic Hospital if they were admitted between 2011 and 2017. We included patients who were adults, met the CDC adult sepsis event definition, which means they had to have blood cultures, antibiotics, and a lactate greater than 2, and then those receiving continuous infusion norepinephrine as well as adjunctive vasopressin. Patients were excluded if they had either missing data, if vaso was initiated before or at the same time as norepinephrine, or if it was started more than 48 hours after the first catecholamine. We screened over 10,000 patients who met the CDC adult sepsis event definition, and after appropriate exclusion, we were able to include around 1,690 patients, 41% of those survived the encounter and 59% did not survive. Looking at the baseline characteristics, I have listed for you both survivors and non-survivors, and I don't want to focus on the difference between these groups because it's not surprising that survivors were less sick and non-survivors were much sicker, but to talk about the population as a whole, on average, patients were around 63 years of age, split pretty evenly between males and females. Most patients resided in the medical ICU, and most also resided at our main campus facility compared to our regional facilities. But what I want to focus on is what I would look at as our markers of severity of illness. So you can see the lactate at the time of vasopressin initiation was around 5.5, higher than both VAST and VANISH, and the SOFA score as well as APACHE-3 scores are very high indicating the high severity of illness of the included population. Looking at our vasoactive agent characteristics, you can see that pretty split between patients receiving a dose of vaso at 0.03 units per minute and 0.04 units per minute. And then looking at the timing from norepinephrine to vasopressin initiation, you can see on average was around 8 to 9 hours. The norepinephrine dose at the time of vasopressin initiation was around 30 mics per minute. And when all of the catecholamines the patient was receiving was converted to a catecholamine or a norepinephrine equivalent dose, it was around 32 mics per minute. Again, higher than what we saw with VAST and VANISH when they were initiating vaso. Looking at the results, first focusing on the lactate. So the quintiles that were established based on the spread of the data are listed for you on this side of the table. And focusing on the rates of survivors and non-survivors, what's interesting to see is that the highest rates of survivors were in the lower lactate quintiles. And the highest rates of non-survivors were in the highest lactate quintiles. And we conducted a multivariable logistic regression looking for association with hospital mortality. And all of the factors that we adjusted for are listed at the bottom of the table or right below the table. And what you can see is that there was an increasing association with each increasing lactate quintile. That met significance when we compared lactate quintile 3 to quintile 1, meaning that if you initiated vasopressin when the lactate was greater than 3.2, this was associated with a 50% increase in the odds of mortality compared to if it was started when the lactate was less than 2.1. And again, this increased with each increasing quintile and kind of had the max effect when we compared quintile 5 to quintile 1, meaning there was more than a fourfold increase in the patient's odds of mortality if vaso was started when the lactate was greater than 8.2 compared to if it was started when it was less than 2.1. So overall, what all of these numbers are telling me is that higher lactate at the time of vasopressin initiation was associated with increased hospital mortality, particularly when it was initiated when the lactate exceeded 3.2 compared to if it was initiated when it was less than 2.1. So now looking at the norepinephrine equivalent doses. So table set up similarly as the last slide. Quintiles are listed on the left-hand side of the table. Looking at rates of survivors and non-survivors, again, we saw highest rates of survivors in the lower norepinephrine dose quintiles and highest rates of non-survivors in the higher norepinephrine dose quintiles. Again, we conducted the multivariable logistic regression, and what we found was that it met significance when we compared norepinephrine dose quintile 3 to quintile 1, meaning that if vaso was started when the norepinephrine dose was greater than 20, this was associated with a 70% increase in the odds of mortality compared to if it was started when the norepinephrine dose was less than 16 mics per minute. And again, increased and maximum effect seen when you compared norepinephrine quintile 5 to quintile 1. So almost a threefold increase in the odds of mortality if vaso was started when the norepinephrine dose was greater than 45 compared to if it was less than 16 mics per minute. Again, overall impression from these results, higher norepinephrine dose at the time of vasopressin initiation was associated with increased hospital mortality, particularly when the norepinephrine dose exceeded 20 compared to when it was less than 16 mics per minute. Now finally, the timing results. Again, timing quintiles are listed on the left-hand side of the table. There was not as much of an obvious trend with looking at the rates of survivors and non-survivors in this specific variable. But when we did our multivariable logistic regression, you can see that when we compared timing quintile 5 to quintile 1, there was a significant association, meaning that if you started vasopressin with more than 14 hours from time of norepinephrine initiation, this was associated with a more than twofold odds in hospital mortality in these patients. And again, what this is telling me overall is higher time from norepinephrine or longer time from norepinephrine start until vasopressin initiation was associated with increased hospital mortality, particularly when it was initiated more than 14 hours from norepinephrine initiation compared to within the first one and a half hours. Overall, to conclude, in patients who received vasopressin, initiating vasopressin earlier in the shock course may be beneficial, particularly when the lactate is less than 3.2, the norepinephrine equivalent dose is less than 20, and within 14 hours from norepinephrine initiation. Now all three of these cutoffs also look pretty similar to what VAST found from their subgroup analyses too. And because of all of these factors and the fact that this was a retrospective evaluation, further evaluations of early initiation of vasopressin in patients with septic shock are still warranted. Thank you guys for your attention. I'll take any questions you might have. Thank you very much. It was an excellent presentation. I was just wondering about the exclusion criteria and the logic behind excluding those groups who have concomitant starting dose with both norepinephrine and vasopressin from the beginning. Wouldn't it be more intuitive to include it and would tell you to start from time zero and go until, you know, probably the 20 milliequivalents for the dose for norepinephrine? Yeah, I think that's a really good question too. And the thing that I think about too is at our institution, we often have to rely on our exact timings based on nursing documentation. So one reason that we included this as an exclusion criteria is because if we go down to the minute one, and first and foremost, we can't tell if that patient was transferred in from an outside hospital. So that was really the main reason why we excluded them being like charted as started at the same time. They might have received it somewhere else first. And the other one is we don't necessarily know if that means that maybe the vaso was started a little bit before. So sometimes it gave us a little bit more wiggle room to know vaso was truly started after norepinephrine if we excluded them. But I do agree that that would be maybe something insightful if we had, I can't remember how many patients were excluded because of that, but to look at them would be interesting. Okay. One quick question about the, did you have any echocardiographic data on these patients, especially, you know, with the left ventricular outflow structure? Yeah. Great question. And my answer is I wish. Unfortunately, we don't. So all of the data we were able to, we have a very robust IT department that was able to pull any discrete data for us. But anything, unfortunately, that's more, at least for us, echo is not discrete data. So we weren't able to pull that information. But I agree. It would be a really good addition. Were you able to associate any map targets that could potentially help to guide earlier initiation of vasopressin? Yeah. That's a good question. We actually, so we looked at map, what the patient's actual maps were. But we didn't go into the nuances of determining what that patient's actual goal or pre-specified or prescribed goal was. And the way that our orders are set up, it wouldn't be as easy for us to be able to capture what they were titrating those agents to specifically. I have a couple of questions. My first question is, you're looking specifically at timing of vasopressin initiation, or what is the lactate at that time, or what is the catecholamine dose? And you say, OK, well, to some extent, hypothesizing that if you're delayed with vasopressin initiation and those are surrogate measures of that, then you have worse outcomes. And then that is actually a causal association. I'm not saying you're saying it's causal, obviously. But that would be the theoretical. One thing I'm wondering is, how much might those things be surrogates for overall slow care, worse care, care that's not following guidelines or evidence or something like this, including things like, did you factor in time to antimicrobial therapy, if you're looking at sepsis or fluid administration or any other things like that? Yeah. I think that's a great complex question. I think that each of these three variables are extremely dynamic. And we don't really know how they interplay with each other, too, to add another layer to it. For other pieces, too, like you were saying with time to antibiotics, we did not actually look at that. We do have that data. But it didn't factor in necessarily to this piece of the puzzle. It is something we could look at. But I'll add to that, too, another thing that we, because I unfortunately don't have a great answer. I think we knew more about how these variables were able to tell us how to treat the patient moving forward. And I think we need an answer to that. But I will also say that we hypothesized that there might be an interaction playing between these three variables, too. We assessed for that and didn't see one. So that's why we felt that we were at least able to successfully evaluate them. But unfortunately, I don't think we are going to be able to, with this data set, at least answer some of those more dynamic pieces. I think your complex question probably answered this, but I was curious about the timing of the lactate and how that correlated, because the time frame between the lactates from starting vasopressin and norepinephrine, the time between each check. Yes, great question. And I'm looking puzzled because I'm trying to rake my brain and remember what my cutoffs were for the lactate. But it really was, we did collect the lactate at the closest one to vasopressin's initiation. So really, if anything, it was within a couple of hours, but not much of a delay outside of the time of vasopressin's initiation, if that answers the question. Do you have any data about whether the presses were initiated peripherally, or did they have to be administered through central line assent? So for this data set, we don't have the ability to be able to answer that. And at the time, we didn't document heavily whether or not the patient was receiving vasopressors peripherally versus central line. So that would be something that would be interesting to look at, but we don't have the capabilities to for this. So the question was fluid restriction and fluid balance, maybe in general, too, whether or not there was difference between survivors and not survivors. I don't know. We do have that data, and I do think that would be interesting as well, and it could be something we could easily look at. Yep, absolutely, yeah. We didn't get the fluid balance piece until after all of these results were finalized. I think the one other, maybe it's a question, maybe it's a comment, because I'm assuming that you don't necessarily have this in your data set. I think if we think about initiating based on a lactate threshold, and you need to get it, you need to start vasopressin before your lactate goes above X, presupposes that within these patients who had a high lactate at the time of vasopressin initiation, that prior to that time point, they had lower lactates. And you have the opportunity to initiate when there's a lower lactate. So it may be helpful to also look backwards to say, hey, did they all start out with a lactate of six, and then they went down to five, in terms of how you could actually operationalize this in practice. Yeah, that's true. That's really fair, because it could also be that these patients presented to us with this lactate, and there wasn't anything we could have intervened on earlier. Yeah, that's a great point. Thank you. Thanks. And thank you. Yes. Thank you. Okay. And last, I would like to welcome up Eric Wald, who is going to be speaking on hydrocortisone, ascorbic acid, and thiamine use. Basically, this is like the total non-controversies in sepsis session, I think, associated with lower mortality in pediatric septic shock. Thank you. Welcome, everyone. Excited to be here, sort of. Anyway. We're excited to have you. We are excited. I'm not controversial at all. And there's no presenter view. That's awesome. Okay. So I have nothing to disclose. This is my hospital. It's a lot warmer here than it is in Chicago. This is where we started. This presentation, you'll find, is a lot less formal than the two that preceded me, although they were excellent presentations. This is Maddie. You all know Maddie, to a certain extent. I was on service one day, walking into early morning to find that I had inherited this child on my service. She's six, new onset ALL, recent chemo, neutropenic, septic shock, severe enteritis, profound capillary leak, a really sick child. And within a few hours, we had to intubate her because she was just pouring fluid into every extravascular space that she had. And then quickly we realized that we couldn't ventilate her, so we put in an abdominal drain to help us with that. But this was the first time that I ever approached a family, and I asked them that there was a new adult paper recently out that had showed that hydrocortisone and ascorbic acid and thiamine might confer a survival benefit in severe sepsis. Luckily, the mom and dad said yes. And we'll get back to Maddie in a bit. But that was my first time. But before we go any further, I think we need to put this all into a little bit more context because it is somewhat controversial. So I'll give you a quick history of HAT and vitamin C. My purpose is not to judge any of these papers. It's just to present you with the data. I of course did not invent HAT or use it first. That was Paul Merrick, seen there on the right. He wrote a paper in 2017 detailing 47 adults that he treated in an ICU in Virginia. And they found a profound decrease in mortality from about 40% to under 10%, lower organ failure scores, and lower vasopressor duration. And while there was a lot of wow out there about this study, a lot of people were quick to criticize that it was a single center, retrospective, and before and after study. All of these things are true. And when I first heard about the paper, I was very intrigued, and I definitely did my reading on vitamin C. But part of my brain thought maybe this is a little too good to be true, much like this friend here. But I did some research, I did some reading, and I really tried to make myself an expert on vitamin C. By the way, it is not easy to find a hat full of oranges on the internet, but I was able to do that. But what I found is that intuitively this makes sense, right? The levels of ascorbic acid and thiamine are low in critical illness, and this is associated with poor outcomes. And that if you give IV vitamin C and thiamine, it raises plasma tissue and cellular levels to super normal levels at times, and this may attenuate some of the changes we see in septic shock. This is a busy slide, but on the left, you'll see a laundry list of things that ascorbic acid can do preclinically to cofactor in the production of catecholamines, cortisol, and vasopressin. All these things are important. It's an antioxidant, helps out with reactive oxygen species, protects the endothelial barrier, preserves the microcirculation. It may protect against ischemia, reperfusion injury. It helps regulate pro-inflammatory pathways, and it supports immune function and really seems like the perfect drug for sepsis in some ways. And there's also this synergy between hydrocortisone and vitamin C at multiple sites in the inflammatory cascade, and especially at the vascular endothelium. And that was what really intrigued me that day in the ICU is this child's capillary leak was out of control, and it was going to cause a great deal of morbidity and perhaps mortality. After using that cocktail, this child dried up pretty quickly, within 6 to 12 hours, and we were able to wean her ventilator and get her extubated and move her forward without really getting in trouble with any sort of morbidity. And her severe enteritis, which, you know, awful CT scan, she just was able to not perf and not... We didn't have any issue with source control, and that was excellent. The third component of the cocktail is thiamin, and why I think it was used primarily in previous studies to reduce the risk of renal oxalate stone formation, I do believe it's a cofactor in many important energy production pathways of the mitochondria, so it makes sense to use it there. I would encourage you to read all these articles that are listed in this presentation, because they're really sort of educational. And then a couple of years later, we found another signal. This is not a HAT study. This is a vitamin C study. It's a randomized control trial done by Alfa Fowler. It was published in JAMA in 2019, using a much higher dose than was used in previous studies, 200 milligrams per kilogram per day in adults with sepsis and ARDS. And while they didn't find any change in their primary endpoints, which were SOFA scores and biomarkers at 96 hours, although this was only in a live patient, so that's a bit of a survival bias, but we won't talk about that. It was a reduction in 28-day mortality that was pretty compelling, especially within the first 96 hours of therapy. Lastly, last month, vitamins trial came out and tried to poke a hole in the balloon of all the HAT enthusiasts out there. This is a multicenter RCT, HAT versus hydrocortisone and sepsis. And while, ironically, they found lower SOFA scores, they didn't find any change in vasopressor free time or mortality between groups. Some of the criticisms of this study was that they excluded 85 patients who were at risk of imminent death, and those are children, I think, that deserve HAT therapy. They also, it took them about 32 hours, median time, to get the drugs into the patients, because they had to be admitted, they had to be put in the ICU, they had to be randomized, and finally had to get the drugs. That's a really long time. I think HAT works really well if you administer it early in the course, but enough about that. So in May, we developed our own protocol, which is listed here, sorbic acid at 30 mg per kg per dose every 6 hours for 4 days with a max of 1.5 grams, IV hydrocortisone, 50 mg per meter squared per day, every 6 hours at the taper, and IV thiamine, 4 mg per kg per day every 12 hours for 4 days. And after accumulating a cohort of patients over the next year and a half, we decided to do a single center retrospective propensity score matched cohort study of patients with septic shock admitted to our unit between the years of 2014 and February 2019. Patients with septic shock were defined as those admitted to the VICU with suspected or confirmed infection who required vasoactives for the treatment of shock within 24 hours of admission. And you were considered treated with HAT therapies if you got the HAT within 24 hours of the initiation of vasoactives. We analyzed three groups, untreated controls, hydrocortisone-only patients, and HAT therapy patients. We did our propensity matching one-to-one on the variables listed there, age, immunocompromised state, two complex chronic conditions or more, and the following variables in the first 24 hours of VICU admission, peak VIS, peak serum lactate level, the need for mechanical ventilation, and the lowest PF ratio. And we performed two analyses, HAT versus untreated controls and HAT versus hydrocortisone-only patients. Our primary outcome was 30-day mortality. Our secondary outcomes were 90-day mortality, vasoactive ionotrope free days at 30 days and hospital-free days at 30 days. And these are our unmatched patients in the table. 557 patients met inclusion criteria, 64 of which died within 30 days of VICU admission. It's 11.5%. And 75 died within 90 days. Our cohort on the right was a fairly sick cohort in my mind. One-third of them were immunocompromised patients. A hundred percent of them had greater than two complex chronic conditions. 80% of them were on a ventilator with a PF ratio somewhere in the moderate ARDS range at 150. And their VIS score was pretty robust at 19. And this is the meat, obviously, of the presentation, is that our HAT therapy patients had lower 30-day mortality than matched controls, 9% versus 28%, as well as matched hydrocortisone patients, 9% versus 30%. We also had HAT therapy patients also had a lower 90-day mortality than matched controls, 14% versus 35%, as well as matched hydrocortisone patients, 14% versus 37%. All of our secondary outcomes were not different between groups. So in the end, oh, sorry, it's not in the end. These are our Kaplan-Meier curves for survivor probability. And as you can see, the median time of death is listed on the left. And in our control patients, it was a little less than two days. In our hydrocortisone-only patients, it was about two and a half days, and in HAT patients, six days. So we figured that the improvement in mortality seen in our study appears to be primarily associated with reduced early deaths, particularly within the first few days of diagnosis. We know that up to a third of children still die from refractory septic shock, which is kind of hard to believe, with a quarter of those deaths occurring within the first day of diagnosis and a third within three days. So in the end, this is just another piece in the puzzle, and you can choose to believe all the evidence that's piling up or not, but I think there's a signal there. It's pretty compelling. I think it's fairly safe. The safety profile of vitamin C and thiamine are well established. My advice would be to use it as early as possible in the treatment course. Avoid in those rare diseases listed below because of hemolysis. And while in spirit we do wait for well-designed RCTs, I will continue to use a therapy that I think helps our sickest patients. So our real conclusion is that children with septic shock treated with HAT had a significant decrease in mortality when compared to propensity-matched untreated controls in those who got hydrocortisone only. The most likely mechanism for this effect is reduction in early deaths. There may be a more robust benefit in children with multiple sources of oxidative distress, those super-oxidized kids that have sepsis and ARDS, but we recognize that these are preliminary data and further validation is needed. I thank my co-authors. I thank my research support, and I thank you for your attention. I'm happy to take questions. Would anyone like to wade into the discussion? Hello. I would like to ask you a question. Thank you for the presentation. Just one question based on you present and based on your investigations and research. Do you think it's a matter of time and this combination is good for everyone with a syndrome of sepsis, or we have the problem of identification of patients that this therapy can be good for a certain group of septic patients? Thank you. That's a great question. I definitely think that we're sort of searching for the phenotype that this works well on. Certainly I've had children who received HAT therapy that didn't seem to have as profound a response, although typically those are kids that struggled with either source control or other issues that were more chronic. So I think that we will continue to look for that phenotype and choosing the right therapy for the right child, but right now we're using it on everybody with vasoactive refractory septic shock. Really excellent talk. My question is, one of the things you brought up about criticism of the recent vitamin C trial is the timing from when they administered the drug. Did you get the opportunity to look at that in your data where timing mattered and did it matter in your data? Yeah. We didn't look specifically at the timing because it wasn't necessarily a large cohort, but the median time for our kids to get HAT therapy was 11 hours, so it's not perfect, but obviously not too horrible either. There are a lot of people out there who are trying to set cutoffs for, you know, you have to give it under 12 hours, you have to give it under a blank, some arbitrary number, and I don't think anyone knows what that number is, but I do believe that waiting 32 hours is not, it's not really, you're not really investigating early resuscitation with HAT therapy. You're just sort of, you know, like 40% of that cohort in vitamins had already received hydrocortisone, so then they got the HAT, which is fine. I'm not trying to criticize, I'm just saying, like, I think trying to somehow seamlessly put this into your ERs and get that sort of buy-in from everybody would be important in terms of, you know, because these kids are often transported in from outside centers, they sit in the ER for long periods of time, it is very challenging to get it into them early, but I'm not sure what the cutoff is, but I think earlier, the better. I'm Sarma Valmuri, I mostly deal with the adult side of this. I'm curious, very brave, thank you for presenting. Have you thought about measuring vitamin C excretion levels in urine on these patients, and then, you know, prospectively as you go along, cross-referencing your patients who are getting vitamin C with phenotypical markers of inflammation and sepsis to be able to sort of say, okay, you know what, we have a little bit, maybe an explanation, like, do you have a research trial that you're going to start, or? Well, there's a lot of projects that are in the works right now. That's not one of them, but I'm happy to sit down with you and talk further about it. This is not tag, you're it. No, no, I'm saying that to me sounds fascinating. I think that would, I mean, I think one of my jobs in the next few years is to try to characterize why certain kids seem to respond more robustly than others, and try to figure that out. So, thank you. Just quick question about the study design. Was it a retrospective study? Yeah. Did you set up your protocol ahead of time, depending on the adult data, or did you have anything in the pediatric? So when I first treated that child, I was using, I set up, she was large enough that I could use, you know, roughly the same vitamin C dose that they had used in the chest paper. But after more people were interested in how it was working, and our group is fairly open-minded, at least half of them are, I would say. So, I think that after it was used more often in the unit, there was a quick sort of idea that we needed to sit down and formalize a process to how much we're going to be giving each kid, and sat down with pharmacy and other people to sort of make sure that we were on the right track. So, originally, it was just sort of extrapolation and me being at the bedside and wanting a kid to do better, and I don't like to lose. So that was, that's sort of how it took off. I'm going to make a comment and then ask a question. My comment is, I have to applaud you for going through and trying to apply something that's been a big focus in the adult literature, and then get a large cohort, a relatively speaking large cohort of pediatric patients, and then also including a comparison group for hydrocortisone alone. I think that that's very helpful for interpreting the data. I think you already got the question, which was one of my questions about essentially what do we know, or what do you know about hat use at your institution? I think that you did comment in the abstract about, you know, there was a time point at which you introduced this. My first question is just, were there any other effects of time if you were using in the control group patients who were enrolled earlier before anyone was using hat? Were there any kind of temporal trends, or were you able to account for any temporal trends in your analysis? That would be a great question for my big data guy who's not in the audience, but I'm not aware. I mean, we were able to look at certain variables very well that he had collected, but I don't believe that looking at timing like that was available to us at that time. My other question, and that's the kind of thing that theoretically if you account for time in the models that you have, you may be able to account for that. The one other thing I was curious about, if you were able to explain that, you know, the absolute difference in mortality, the point estimate for that, at least, was pretty high. I realize small numbers, and so not high confidence in those. But when you have roughly a third the mortality, can you explain to us how that didn't translate into increased vasoactive free days or increased hospital free days among those who received the hat therapy? I mean, I think that's how we arrived at the hypothesis that this was reduced early deaths, and I was surprised to sort of look in the literature and see that, you know, a good chunk of kids still die from refractory shock, although we see it all the time, but it seemed like if we could get them over that first hump and not have them die, then they were still ended up being a very sick cohort. They still had the same amount of organ failure days. They still had the same amount of ventilator time for the most part. All those things that we looked at weren't any different. It just seems like if you can get them to survive, they're still sick, and they still have plenty of morbidities that we have to face, but surviving is kind of key. So I think that's. No disagreement there. Yeah. I think that we were struck because we didn't see any of it. I would assume that there would be some organ failure signal like in Paul Merrick's paper, but there wasn't. Yeah, I think one interesting way, if you have the data to look at it, would be even because vasoactive free days is just, that's just a yes, no for each day. But if you were able to look at vasoactive doses, if that's your hypothesis that, you know, these, the children, at least in the early days, just had totally run away, you know, high, high dose, you could not rescue them. And were you able to spare doses in the early going when the ones who didn't receive the therapy ended up dying? That would be an interesting way to at least get at some of that question. That's a great idea. Thank you. Okay. Thank you very much. Again. And I want to thank everybody for showing up for our star research presentations in sepsis.

retrospective study

HAT therapy

pediatric septic shock

30-day mortality

90-day mortality

vasoactive ionotrope free days

hospital-free days

untreated controls

hydrocortisone-only patients

Sepsis

Pediatrics

Shock

Lactate

Presentation

Select

Professional

Associate

2020

Advanced

Sepsis Resources