It's really my pleasure to be here today and I appreciate that kind introduction. As I get started, I will note a couple of disclosures. First is my research program is funded by the NIGMS for which my institution receives payment and I'm also a member of the Surviving Sepsis Campaign Research Committee and a few of the ideas I'll be presenting today come from that research committee and a forthcoming manuscript will expand upon them a little bit further. It's no news to this audience that sepsis is a huge problem. Sepsis is estimated to have 49,000 incident cases per year and 11 million global deaths or 20% of all global deaths and the CDC estimates that over 350,000 patients per year will die in the hospital in the U.S. from sepsis. Through improved recognition and supportive care, sepsis mortality has improved over time at about 3.3% per year, but septic shock mortality is improving at a much slower rate at only about 0.6% per year and remains about 50%. Even for patients who survive critical illness, they have chronic problems including chronic critical illness in about a third of survivors. For over 20 years now, many thought leaders including Drs. Levy and Evans have been part of the Surviving Sepsis Campaign which has guided the bedside practice of the care of patients with sepsis with release of both bundles and guidelines. Five adult guidelines have been released with the most recent iteration in 2021. This iteration had 93 recommendations and best practice statements, but unfortunately there were only three strong recommendations based on high quality evidence in these guidelines and those include the avoidance of starches for resuscitation, the first line use of norepinephrine over dopamine, and low tidal volume ventilation for sepsis-induced ARDS. The consequence of having few high quality recommendations leads to practice variability and these are data from one of many studies showing practice variability in the treatment of patients with septic shock. These are data from the ATHOS-3 investigators which was a trial that enrolled patients in whom the investigators deemed to have failed first line vasopressors and the patients were randomized to angiotensin II or placebo. But in a follow-up publication, the investigators astutely noted there was vast regional variability in the care of the enrolled patients such that patients in Europe received much higher doses of norepinephrine and in the U.S. we more frequently used vasopressin and corticosteroids. This practice variability can have implications for patient outcomes though. There is one meta-analysis that looked at septic shock trials and simply looked at the control group mortality and found that the point estimate for control group mortality was 39% but the 95% prediction interval ranged from 14% to 72% with 93% of the overall variability due to between-study heterogeneity. Even when the investigators tried to make the groups more homogenous by defining the trials based on entry criteria, based on MAP or vasopressors, the between-trial heterogeneity persisted. I think one of the most important takeaways from this study though was that only 41% of the variability was explainable by patient characteristics or trial entry criteria. And so the remaining 59%, the variability was due to post-randomization treatments with therapies like fluids and vasopressors. Now fluids and vasopressors represent only two of many knowledge gaps for the treatment of these patients with sepsis and septic shock and I'll touch on them a little bit further. I'll start with initial fluid resuscitation volume and I'm hesitant to even wade into this topic knowing how controversial it is. There are entire Congress sessions devoted to this topic so I will not dwell on it by any means. I'll just summarize my thoughts by saying it's a complex issue. The evidence quality is low and we really need high-quality studies to further our knowledge of this area. However, we know that this is controversial but I think it's important to understand how we actually got to this point of the controversy. In 2005, the Surviving Sepsis Campaign released the first resuscitation bundle and in that bundle they recommended the use of at least 20 milliliters per kilo of initial fluid resuscitation and this recommendation was carried forward into the 2008 guidelines. This was a vital component of the initial resuscitation bundle that was rolled out as part of the campaign's performance improvement process and that process was published, the results of that process were published in 2010 and the campaign did amazing things as part of this process including improving compliance and subsequent mortality. Since 2012, though, the guidelines and bundles have recommended at least 30 milliliters per kilo of fluid initially referencing the performance improvement program noting the volume of fluid those patients received and more recently referencing the process ARISE and PROMISE trials noting these patients received about 30 milliliters per kilo at the time of enrollment. However, it's important to realize, though, we've probably entered into a circular pattern where the guidelines are influenced by clinical practice and clinical practice is influenced by the guidelines and so we need some high quality evidence to help us escape this circular pattern because there is no single randomized trial that has ever evaluated one fluid volume versus another to assess the outcomes in septic shock. The best data available are observational. In late last year, there was one observational study that was published that I found quite intriguing from the University of Pittsburgh group. These investigators included patients with community acquired sepsis and collected the volume of fluids the patients were given in the first six hours after presentation and then estimated through modeling what their in-hospital mortality would be after adjustment for severity of illness and confounders. And we might expect a relationship that looks in a U-shaped pattern like this where patients who receive less fluid volume would have higher mortality in a nadir right at 30 milliliters per kilo or so and then higher mortality after that. But in fact what the investigators found was essentially a flat relationship between these two variables and in patients who had heart failure through reduced ejection fraction there was even similar patterns if not an upward sloping line. Now I don't think that this study is definitive by any means. It's observational but at least suggests that there is not some threshold that is magically will transition a patient from dying from septic shock to surviving from septic shock. Many proponents have tried to investigate this question and it's been thought of in many different ways. One of those trials is the Clover's trial. This is the NIH sponsored Pell Network study where patients with sepsis induced hypoperfusion were randomized to a vasopressor preference therapy or a fluid preference therapy after receiving one liter of initial fluid resuscitation. And about a year ago we received the top line results that found that this study didn't detect significant differences in mortality nor safety and I'm excited that later today the full publication will be presented here and simultaneously released. I think the trial results were not the vindication that strong proponents of either side of this controversy were hoping for but they do at least suggest that one therapy is not more harmful in the average patient overall and we'll look forward to that full publication to better understand if there are patient specific groups where a certain treatment strategy may be preferred. That vasopressor first strategy included the use of catecholamine vasopressors. We know that norepinephrine is the recommended first line vasopressor. After norepinephrine I wish I had the I don't know emoji here because I'm not sure we actually know what to do after starting norepinephrine first and this issue was highlighted in 2018 by the surviving sepsis campaign research committee where research priorities were set including answering this question of what is the optimal approach to selection dose titration escalation of vasopressor therapy and despite five years of research and mounting evidence the data are still unclear such that the draft updated research priorities document has a very similar question that will ask what is the best vasopressor approach for treating progressive septic shock. And we're hopeful that this research priorities document will be published later this year. One of the targeted questions in the upcoming document is investigating the role of epinephrine here in the US epinephrine use is on the rise. These are data from nearly 300000 patients from over 450 hospitals here in the US where in 2010 16 percent of patients with septic shock received epinephrine. But in 2017 25 percent of patients received this drug. Now the reason for this is unclear. The study does overlap a time frame during which bedside ultrasound was increasingly used so it's possible that prevalent sepsis induced myocardial dysfunction was increasingly discovered over this time. And in fact that would be the main reason why we would add epinephrine to norepinephrine to utilize the inotropic properties of epinephrine. The epinephrine data are quite unclear though. There are only two trials that have compared epinephrine to norepinephrine as first line therapy. One of the trials found benefit with a point estimate of epinephrine. The other the point estimate was in favor of norepinephrine. Notably both of these trials had wide confidence intervals that overlapped one but in both trials adverse effects were more frequent with epinephrine including tachyarrhythmias and hyperlactateemia. Indirect evidence from a network meta analysis the point estimate was in favor of epinephrine but again wide confidence intervals because epinephrine is widely available in low resource settings. I think it's important to have a trial conducted that's adequately powered to evaluate these two first line therapies. And even in high resource settings these data will be useful because we frequently experience norepinephrine shortages as is currently affecting our colleagues in Japan right now. Epinephrine has a number of additional challenges. As I mentioned in the U.S. it's frequently used as an adjunctive therapy to norepinephrine but there are no trials that are specifically evaluating epinephrine in this role. And in fact using adjunctive inotropes is a bit controversial. It's unclear if epinephrine has a benefit over some other therapies such as dobutamine or any inotrope has a benefit in this particular complication of sepsis. And I think this is also an important question to answer because observational studies have shown an association between using adjunctive inotropes and higher mortality even after adjusting for severity of illness. Another vasopressor that is quite controversial is the use of vasopressin. These are that epinephrine data that I just showed and in that same study vasopressin use also was on the rise potentially rising even quicker than with epinephrine such that about a third of patients in 2017 received vasopressin. Now more vasopressin and epinephrine could reflect clinicians using multiple drugs to treat septic shock instead of just norepinephrine monotherapy. And that is one of the potential advantages of vasopressin is using a non-adrenergic mechanism to cause vasoconstriction. And despite a number of trials it's quite controversial whether vasopressin actually improves mortality in patients with sepsis and septic shock. These are data from the surviving sepsis campaigns internal meta-analysis that combined trials to inform the guideline recommendation. In this meta-analysis 10 trials were included and you can see here the risk ratio favored the addition of vasopressin to catecholamines with a confidence interval that did not include one. The need for kidney replacement therapy also had a point estimate favoring the addition of vasopressin but a bit wider confidence interval that overlapped one. This meta-analysis though is probably best described as including trials of patients with vasodilatory shock based on the studies included and there is an individual patient data meta-analysis from four trials that specifically enrolled patients with septic shock and in that population of patients with septic shock this meta-analysis did not detect a benefit with vasopressin for mortality but the confidence interval for kidney replacement therapy did not include one. Now these two meta-analyses largely have overlapping confidence intervals. I don't think that they are substantially different but these data at least suggest that maybe patients with septic shock behave differently than the broader group of patients with vasodilatory shock and we need to think about our therapies specifically within septic shock and not in other patient contexts. Some have also hypothesized that the timing of vasopressin initiation is an important factor. This question was investigated in a large observational study published last year where the investigators included patients who all received vasopressin in the setting of septic shock evaluated the norepinephrine dose at the time of vasopressin initiation and again used estimates of what the patient's predicted mortality would be. You can see here on the left that there is a non-linear positive relationship in this study between the norepinephrine dose at vasopressin initiation and higher mortality such that if we compare simple odds ratios compared with a norepinephrine dose of 10 mics per minute when vasopressin was initiated starting vasopressin at a norepinephrine dose of 25 mics per minute was associated with 33 percent higher odds of mortality and at 60 mics per minute over two and a half fold higher mortality. These data though are contrasted by again that individual patient data meta-analysis that looked at the norepinephrine dose at enrollment of patients into trials comparing vasopressin versus catecholamine monotherapy and they didn't detect any indication here of heterogeneity of treatment effect. Now there are some hypotheses for why these two studies had different findings one of which is that the meta-analysis dichotomized the norepinephrine dose which lower statistical power to detect a difference. But I think that one of the more reasonable explanations is that norepinephrine dose alone isn't the end all be all that is associated with vasopressin associated outcomes. In fact that same observational study found that independent of the norepinephrine effect lactate concentration was associated with higher mortality if vasopressin was started at higher lactate concentrations. And so I think probably we need to think about the use of multiple factors together to describe which therapies should be given to certain patients. And this is an idea of identifying treatment responsive sub phenotypes. Now these would be groups of patient features that would suggest that patients would have differential effect for different therapies which is actually the essence of precision medicine where we apply therapies to patients who will have the most benefit. There are a number of sub phenotypes in sepsis that have been identified. Most of them though related to prognosis in sepsis with a number of fantastic publications in that regard. But treatment responsive sub phenotypes are just starting to emerge in the literature. There is one that has been identified for response to corticosteroids and another most recently with response to balanced crystallites as opposed to saline. The use of sub phenotypes and the design of critical care trials has the potential to revolutionize how trials are conducted. One thing to mention about current clinical trials we all know that most patients or most trials with investigating sepsis and septic shock did not detect a benefit. And this is because most trials estimate the average treatment effect for an average patient. And the average patient does not actually exist when you combine all of these patient characteristics. In reality you have some patients who are predisposed to harm some patients who are predisposed to benefit and some right in the middle. And ideally we'd be able to identify these patients with differential response up front and apply therapy to them as opposed to lumping them all together. Now this is going beyond subgroup analyses from trials but actually identifying these groups of patients in advance. And this might look like one of several different ways and this is just one cartoon depicting what it could look like. We could take the heterogeneous group of patients who are considered for enrollment into a trial apply algorithms to predict what their outcome would be whether it be high mortality or low mortality and only enroll the patients who are the most ill the patients with highest mortality. And this is a process called prognostic enrichment. One of those patients who have a high mortality risk again applying algorithms to determine what therapy might be best for those particular patients and then enrolling patients into different arms of a potential adaptive trial that targets the purported underlying biologic mechanism for that sub phenotype. And this is a process called predictive enrichment. These types of trials are currently enrolling. I think they have the huge potential. I think it both could revolutionize how trials are actually conducted and also how care is delivered for us at the bedside for patients with sepsis. I'll just wrap up by telling you what you already knew and that is that there are few areas of high quality evidence for the treatment of patients with sepsis and much future research is needed. Thank you.

sepsis

septic shock

mortality rates

treatment practices

research