Good afternoon, SCCM. My thanks to Dr. Teichman. Appreciation to SCCM for hosting the presentation of this publication, and welcome and introduction to my colleague Nate Shapiro. We're going to divide this presentation into two parts. We're going to present the background and motivation for conducting the Crystalloid Liberal or Vasopressors early resuscitation study in sepsis, which I'll refer to as CLOVAS, on behalf of the National Heart, Lung, and Blood Institute's Prevention and Early Treatment of Acute Lung Injury Network. It's really an honor to represent a cast of thousands. This particular investigation is the work over seven years of a large number of investigators and coordinators at many sites, 60, many of whom are in the room and with whom we share the opportunity to share our work. The primary question that we had sought to address in CLOVAS was which of a broad spectrum of therapeutic approaches for optimal fluid resuscitation in septic shock affords the best outcome when we look at survival. On the one hand, one end of the spectrum of care is encapsulated in an approach which is a more restrictive approach when it comes to fluids and prioritizes early vasopressors with a priority to prevent progression of both interstitial and macroscopic edema. And there are many studies that associate accumulation of fluid with high fluid balance with high mortality. The other end of the spectrum of usual care focuses on a more liberal approach, fluid loading to augment preload, optimize organ perfusion, improve microcirculatory dysfunction, and in large part represents empiric usual care across the world. Now this question has been addressed at several levels, both in ICU cohorts and emergency room resuscitator patients. And this meta-analysis, looking with the point estimate representing the size of the cohort, in aggregate would suggest a small beneficial, small attributable benefit to strategies that focus on a more restrictive approach. The largest study of which by Mayhoff and others, the classic study, was focused on only ICU patients. So we hypothesized in Clovis that a restrictive approach, vasopressors followed by fluids, compared with a liberal approach, fluids followed by restrictive vasopressors, would result in improvement in 90-day mortality before discharge in patients identified very early on with sepsis-induced hypertension. Clovis was a multicenter, prospective phase 3 randomized controlled trial. It was non-blinded for reasons related to the protocol. And the intervention period was 24 hours of both fluid and vasopressor titration. The study, which was powered around changing 90-day mortality, sought to detect a 4.5 percent absolute difference in a previously estimated 15 percent prevalent mortality rate. And we planned the study with two interim analyses and a final analysis to look at both efficacy and futility. The study was conducted across the 60 hospitals that participated in our Clovis PETAL network, with Mass General Hospital as our clinical coordinating center. And at the onset, we designed the study to really capture the spectrum of usual care presentations of sepsis in adults with a suspected or confirmed infection who had persistent hypotension, which we operationalized as a blood pressure systolic less than 100 or menoterial pressure less than 65, after at least a one-liter fluid challenge, but no more than three liters of crystalloid fluid. And the exclusions thus included a long delay between meeting inclusion criteria, four hours, or since primary admission, persistent hypertension from a non-sepsis cause, or extreme hypovolemas, such as happens in diabetic ketoacidosis. In addition to our primary outcome, we identified several important organ failure process and systems outcomes as secondary analyses. Our protocol was designed with an intention both to separate the two arms between the fluid restrictive group, where once a patient was randomized, all usual care fluid was discontinued. Pressers were then titrated to maintain hemodynamics. And there was an opportunity to capture usual care decisions around rescue fluids, which could be administered by the clinical team either for indication or at their discretion. Similarly, in our liberal protocol, up to 2,000 mils of crystalloid fluid could be given in 1,000 mil boluses with an opportunity to forego the second one liter of loading if there was a clinical reason to do so. The protocol similarly used fluids to achieve target blood pressure, and there were opportunities to use rescue vasopressors with monitoring not pre-specified, but being used according to the usual care practices in a way that allows for personalization of the care at the bedside by the investigator. It's my pleasure now to invite my colleague, Nate Shapiro, to present the findings. Thanks much. So after the study, the trial was stopped for futility without evidence of harm at a planned second interim analysis. At that point, we had enrolled 1,563 patients over just a little under four years. Here's the consort diagram where we screened roughly 12,000 patients. Approximately 5,000 were eligible, 1,563 randomized, of which 782 went to restrictive, and 781 to liberal. As you'll see here, the reasons patients were eligible and not randomized tended mostly to be around the ability to get informed consent from a patient or surrogate or locate a surrogate, as well as for clinicians entering their patient into the trial. The patients enrolled in the study were quite representative of the types of patients that present with sepsis to our hospitals. Approximately an age of roughly 60 years old, 70.7% white, 15.8% black, and 14.5% identified as Hispanic. When we look, this highlights some of the comorbidities we saw in the trial. There was a good distribution of equality between the two groups. About 28% were diabetics, 11.5% chronic heart failure, 4.7% end-stage renal disease. Here we highlight that patients were enrolled quickly. 61 minutes passed typically. On average, that's the median time from meeting criteria to randomization. 92% of the patients were in the emergency department at the time of randomization, and the median amount of fluids was strikingly similar between the two groups at 2,050 cc. So this slide now shows us or asks the question, did we perform a clinical trial? Did we have a group of patients who had a restrictive approach and compare them to a group of patients who had a fluid liberal approach? And I would submit that these data support that we did in fact have a clinical trial. If we look, a lot of the separation occurred over the first six hours where we see 2,300 cc's in the liberal group compared to 500 cc's in the restrictive group for a difference of 1,800 cc's more fluid in the liberal group. That persisted out to 2,134 cc's difference in the 24-hour period that the protocol occurred. When we look at the other side with use of vasopressors, 59% of patients in the restrictive group received vasopressors as compared to 37% in the liberal for a 21.7% difference. The time from randomization to first vasopressor was shorter in the restrictive group, and the duration of vasopressor use was longer also in the restrictive group. So taking another look at this, just because I think this is an important part to understand and interpret the trial, again, pre-randomization, the fluids were the same, and then we have about a two-liter separation that occurs over six hours that persisted, or there's a little more separation over 24 hours. And when you incorporate pre-randomization fluid up until 24 hours after randomization, you see it's 5,400 cc's compared to 3,300. On the other side, the restrictive group received vasopressors more often, earlier, and longer. So at the end of the day, in setting up our hypothesis of a fluid-centric group, we saw that in the liberal where there was more fluids with less and later vasopressors compared to a restrictive group, which was vasopressor-centric. The primary outcome, 14% mortality for death before discharge home by day 90 in the restrictive group compared to 14.9% in the liberal group, difference of 0.9%, p-value of 0.61, no difference between the groups. When we look at the Kaplan-Meier, we'll see these lines are strikingly close. So at the end of the day, there was no difference. When we look at secondary outcomes, organ support free days, ventilator free days, hospital free days, I'm showing a couple of them here. The rest are obviously in the paper, but no difference. Other secondary outcomes, renal function, organ function, ARDS onset, no difference. New intubation and serious adverse events, also no difference in these important safety outcomes. There was a number of pre-specified subgroups that we had identified for analysis, and there was, the summary is no difference across any pre-specified subgroup. I'll highlight two of them. One is chronic heart failure, where the point estimate was strikingly close to zero. And end-stage renal disease, where the point estimate was further away, but the confidence interval still crossed zero in the 73 patients in this trial. I'd also like to share some other metrics, or we'd like to share some other metrics, which was ICU utilization. This is a post-hoc analysis, but we felt it warranted presentation, which is 8% more patients in the restrictive group ended up in the ICU during the protocol period, and that was sustained over the first seven days. There was also higher central venous catheter insertion in the restrictive group. Another part of our trial, and while this wasn't part of the randomization process, we identified in the trial that the use of vasopressors was becoming more common, vasopressors administrated through peripheral catheters was becoming more common in clinical practice. So we actually incorporated this as a protocol allowable action in the trial, and we included it in the consent process. So in total, 500 patients in our trial received vasopressor infusion through a peripheral catheter. There were three identified complications, of which all three events resolved without intervention and did not have any clinical consequences. So 0.6% event rate, and the events were self-limited and had no clinical confidence. Perhaps this data is in support of this practice. So a couple items, there's a lot to unpack in this trial. A couple things to unpack as people interpret it, I'll mention a few. First is the results applied to the type of patients enrolled in the trial, patients identified early, patients who are not at the extreme of volume status, in other words, if you're volume overloaded, you weren't in the trial, and if you're severely volume deplete, you wouldn't have been in the trial. And our patients received about two liters on average of medium fluid before entering the trial. I'd also like to call out, as Dr. Douglas highlighted, that there was a degree of personalized care in the trial, because we all felt this was important as we designed the protocol to incorporate in such a protocol. So there was enough flexibility that you had a vasopressor-centric approach or a fluid-centric approach, but you weren't locked in to only use one modality or the other. Use of hemodynamic monitoring was allowed, but it wasn't commonly used, and it wasn't really part of the protocol or part of what was studied. It's possible there's a future direction for hemodynamic monitoring that wasn't addressed here. In our study design, we had a hypothesis of fluid-liberal versus fluid-restrictive, which was vasopressors early and less fluids. Two arms in the trial, and that's what we tested. We considered but decided not to include an unstructured or wow-type arm, as we thought that there'd be issues with generalizability, and so this is what we postulated was the best design for our trial to answer our hypothesis, and that's why we went with that design choice. It's also possible there's subgroups out there that we didn't test. Maybe there's endotypes. Endotypes identified through more advanced analytic techniques, which may or may not include advanced microbiology testing or advanced protein assay testing or RNA testing, et cetera. There could be subgroups out there who may benefit from one strategy or the other, but we did not identify them in this trial, and so putting it together, maybe future efforts need to look in other areas in order to improve mortality and sepsis beyond how to manipulate vasopressors and fluids. So in conclusion, the fluid-restrictive strategy, as compared to the fluid-liberal strategy used in this trial, did not result in lower or higher mortality prior to discharge home before day 90 among patients with sepsis-induced hypotension. We performed a trial where we had two strategies, fluid-liberal, fluid-restrictive. The data show that we were able to effectuate those strategies, and at the end of the day, both seemed to address hypotension, and neither led to a higher or lower mortality. It was the same. Thank you very much.

Crystalloid Liberal

Vasopressors early resuscitation

sepsis

fluid resuscitation

septic shock