My name is Eithan Pelton from Intermountain Medical Center and the University of Utah. Today I'm going to be talking about fluid management in respiratory failure. These are my disclosures. Just as a reminder and an orientation as we begin the discussion, let's take a quick look at the basic ARDS pathophysiology. ARDS is a syndrome in which innate immune cell mediated damage of the allergic. Endothelial and epithelial barriers leads to accumulation of protein-rich edema of fluid within the interstitium and alveolus. In the healthy alveolus, hydrostatic pressure driving fluid into the alveolus is opposed and exceeded by oncotic pressure, which draws fluid back into the pulmonary capillaries for drainage away. By contrast, in the ARDS alveolus, we have loss of the reflection coefficient due to increased permeability of the alveolar capillary barrier resulting from inflammation effects on both the vascular endothelial and the alveolar epithelium. There's also a component of decreased plasma oncotic pressure and potentially, as we'll discuss, increased hydrostatic pressure deriving from increased preload relating to fluid resuscitation. In this case series from Japan, which compared extravascular lung water calculated from lung weight for over 1,600 autopsy cases with diffuse alveolar damage to over 1,600 patients with normal lungs, we saw that the extravascular lung water is substantially higher. Interstitial edema also disrupts organ function and healing well beyond the lung. As a result, we have something of a Goldilocks problem. From hemostasis in the acutely ill patient with respiratory failure, these patients often have decreased intravascular volume and sometimes decreased total volume status. One of our goals during resuscitation and management of these patients in the initial phase is to restore intravascular volume status and thereby restore end organ perfusion. But we often, in doing so, even as we achieve adequate intravascular volume, may end up with a substantially increased total body volume status. As the patient begins to recover and recruit food from the extravascular spaces back into the intravascular spaces, we may end up with a situation where the patient's both total body volume overloaded and overloaded in the intravascular space. We can consider at that point de-resuscitation to try and get the patient back to homeostasis. Numerous observational studies have demonstrated an association between increased mortality and positive fluid balance in patients with critical illness. In this study of 212 intubated patients with ARDS and septic shock from Mayo Clinic and Washington University, they looked at patients who all survived at least seven days and then evaluated their fluid management during that initial period and the association with mortality. They categorized patients into four groups based on adequate or inadequate initial fluids and conservative or liberal weight fluid management, and found that patients who achieved both adequate and conservative, adequate and initial fluid management, and conservative weight fluid management had better outcomes than patients who had inadequate initial fluids and liberal weight fluid resuscitation and differential stratification within the intermediate groups. As with all observational studies, there's a substantial risk for indication bias. The possibility that instead of differences in fluid management leading to the outcomes, the difference in fluid management are just a signal of the real underlying difference in patient condition and care that are the real cause of differences in outcomes. As we think about how we might avoid fluid overload nevertheless in patients with ARDS, we have to think about the initial resuscitation phase as well as the de-resuscitation phase. The challenge with the initial fluid resuscitation phase, of course, is that it is difficult to assess adequate fluid resuscitation and whether an individual patient should get additional fluid. In this secondary analysis of the FACT trial, which we'll discuss more in a moment, this was a convenient sample of ARDS patients who had a pulmonary artery catheter and got at least one protocol-directed fluid or albumin bolus. They found that among 127 patients who had 569 fluid boluses, that fluid boluses did increase wedge pressures as well as the CVP, but that the associated changes in the mean arterial pressure as well as urine output and cardiac index were much more moderate. While they were statistically significant, the average response was clinically insignificant. In fact, only if you consider a cardiac index increase of at least 15 percent of a response to a fluid bolus, only 23 percent of fluid boluses were associated with resulted in an increased cardiac index or were fluid responsive. It's also important to consider that a large proportion of fluids administered in the ICU are in fact not resuscitation fluids. In this study from a 45-bed general ICU involving over 14,000 adult patients, just six percent of fluids were given during the ICU stay were resuscitation fluids and medications and maintenance fluids made up a much larger overall fraction among the potentially modifiable fluid administrations. Nevertheless, as we think about this gold electron of how to optimally resuscitate a patient, we strive to reach the point which we can achieve adequate intravascular volume sets while avoiding or minimizing the amount of total body volume overload we administer. But we also can consider whether we really need to restore full intravascular volume sets, full homeostasis, and whether a more conservative initial fluid resuscitation is possible. The question being, after the initial fluid adult, how much fluid resuscitation should be given? We are beginning to have some evidence to answer this question. The first is from the classic trial, which was a pilot feasibility trial enrolling patients who had already received a 30-milliliter fluid bolus for septic shock. Patients were then randomized to receive restrictive fluid resuscitation versus usual care. In the restrictive group, fluid boluses of 250-500 milliliters were optional in patients who had hypotension that was refractory to norepinephrine modeling, or oliguria, or elevated lactate. In the usual care arm, by contrast, fluid could be given as long as hemodynamic variables improved, including both dynamic and static measures of fluid resuscitation. The findings from, again, this was a pilot trial enrolling just 151 patients with a primary outcome looking at the feasibility and whether they could achieve separation in fluid management. The primary outcome was resuscitation fluid through day 5. They did demonstrate a significant difference with the restrictive group getting about 1.2 liters, less resuscitation fluid during that period. Interestingly, total intake, as we emphasize through day 5 because of the many other components of fluid intake in IC patients, was not different between the groups. But net fluid balance was different between the groups or approach statistical significance with about a 1.1 liter net fluid balance difference between the two groups. Again, this was a pilot trial and there was statistically significant difference in mortality, but numerically there was 33 percent mortality at day 90 in the restrictive group versus 41 percent in the usual care. Fortunately, we are going to have two major trials providing increased data on resuscitation strategies for patients with sepsis, and that often will apply to patients with ARDS. The classic two trial completed enrollment in mid-November and enrolled slightly over 1,500 patients in Dutch ICUs using alternative treatment strategies similar to those discussed for the classic trial. The primary outcome will be 90-day mortality, and we are looking forward to results which weren't yet available at the time this talk was recorded in mid-February. The Clovers trial is a US trial done by the Pedal Network, had planned to enroll 2,230 patients who were hypotensive after initial fluid resuscitation for septic shock. Patients were randomized to receive additional fluids after their initial 1-3 liters, which qualified them for this study up to about five liters with rescue vasopressors only during that additional fluid resuscitation. The conservative treatment arm randomized patients to receive vasopressors, and only rescue fluids for refractory hypotension or other signs of end-organ hypoperfusion. That trial was just terminated in early February with a total enrollment of 1,566 patients. Detailed results are not yet available, but we do know that this study did not detect a significant difference in the primary outcome of 90-day mortality. So now we can think a little bit about what we know and what evidence we have for management of patients during what we can call the de-resuscitation phase. There is substantial variation in practice and opinions around the management of patients during this phase. In this survey of intensivists administered through a number of critical care societies, 524 intensivists responded with a fairly low response rate, but still giving us some sense of the variation in opinions. Intensivists did feel that fluid overload was a potential modifiable contribution to mortality but also felt that it was largely inevitable. There was substantial variation between intensivists in the threshold for initiating de-resuscitation depending on other clinical factors like acute renal failure and concurrent shock, and the fluid removal strategies used as shown here, as well as the safety thresholds to stop fluid removal. To date, the best evidence we have for de-resuscitation in patients with ARDS comes from the FACT trial. As a reminder, I suspect that this is a trial familiar to many of us, but this was a two-by-two factorial trial conducted by the US ARDS network, comparing conservative versus fluid liberal goals for patients with ARDS, as well as a PA catheter versus a central line for management. We'll focus here, of course, on the fluid management arm. All of these patients were mechanically ventilated with ARDS and were enrolled within the first 48 hours after meeting ARDS criteria. The FACT trial used a fairly complex algorithm for management of fluid resuscitation. But the real take-home is that in the conservative arm, they targeted a central venous pressure of less than four, essentially giving all patients Lasix if they exhibited adequate end-organ perfusion if they were urinating and they were not in shock. The liberal group, by contrast, targeted a central venous pressure of 10-14. The liberal strategies patients were, excuse me, in the primary analysis of 60-day mortality, there was no significant difference between the two groups, about 26 versus 28 percent. But by contrast, in the secondary outcomes of both ventilator-free days and ICU-free days, patients in the conservative strategy fared significantly better with a significant and substantial increase in ventilator-free days and ICU-free days. Interestingly, given the fact that renal dysfunction remains a substantial concern with de-resuscitation, the liberal strategy patients were actually numerically more likely to require dialysis through day 60. Admittedly, the FACT trial doesn't exactly reflect well current practice because we don't use PA catheters very much anymore, and not all ARDS patients are going to have a central venous catheter. In the FACT-LITE group, we also aren't using dope-butamine and some of the other strategies that were used in FACT. The FACT-LITE protocol has been used heavily through subsequent ARDS and PETAL network trials, and essentially focuses on just managing fluid based on central venous pressure, or even just based on the presence or diuresis in the absence of shock, and does appear to yield similar results in this study by Colin Grissom, in terms of differences in ventilator-free days or ICU-free days versus a liberal strategy. It's important to note that it's possible that one de-resuscitation strategy does not fit all patients. Several groups have looked at potential differential effects of conservative versus fluid liberal strategies and alternative definitions for what are called sub-phenotypes of ARDS. In this re-analysis of the FACT trial, Katie Famous from University of California, San Francisco, applied previously defined definitions for sub-phenotypes based on latent class analysis. Yields essentially two groups of ARDS patients with the sub-phenotype 2, which is being characterized by more inflammatory biomarkers and more hypotension. What we see here is that in the sub-phenotype 1, again, the less inflamed patients, a conservative strategy was associated with lower mortality and higher ventilator-free days. Whereas in the more inflamed group, a conservative strategy was potentially associated with even more mortality benefit, but potentially less ventilator-free days. Again, that number is not statistically significant. But for the mortality effects, there does appear to be a potential differential effect in this exploratory analysis. The question largely remains, when and how should we begin to de-resuscitate patients? We can begin to get some information from the Radar 2 trial by Silversides et al that was recently published online in intensive care medicine. This study enrolled adult ICU patients who had been intubated for 24-48 hours and were expected to remain in the ICU essentially at least for another 24-48 hours until the day after tomorrow. It was a pragmatic trial comparing restrictive early fluid resuscitation as well as active de-resuscitation beginning early versus usual care. Patients were randomized one-to-one to different arms with a conservative arm that included two components. One was stopping conservative initial fluid resuscitation, stopping maintenance fluids, concentrating meds, and minimizing additional IV fluid administration, as well as active de-resuscitation beginning on study day 2. Active de-resuscitation was triggered by fluid overload at at least two plus sites, including with one potentially being the lung, as evidenced by hypoxemia in airspace capacities. Patients who met these criteria and were not in severe shock received a regimen including endopamide and spironolactone, administered orally plus titrated IV LASIX for a goal net negative of 1-3 liters per day, or if the patient was on a renal replacement therapy, an ultrafiltration of 2 liters per day. The study did successfully achieve its primary outcome in the 180 randomized patients. There was not much difference in fluid accumulation from day 1 to day 2 in the patients who were based on the more conservative fluid administration strategies. But once active de-resuscitation was implemented, we see that the cumulative fluid balance rapidly begins to split, and that the day 2 to 3 net fluid balance, which was the primary outcome, was significantly different with an average difference of just about 1 liter between the two groups. Day 3 and day 5 cumulative balances were also statistically significant, with the restrictive group achieving a near even balance by day 5. There were some more adverse events in the conservative group, but this was largely driven by metabolic alkalosis in the group with no difference in new AKI. Mortality was not different between the groups, nor was 180-day mortality. Again, these being secondary outcomes in this pilot trial. However, while the overall study group did not show a significant difference in mortality between the study arms, there was a potential for concern in the subgroup analyses of mortality conducted in patients with sepsis and ARDS. The group with sepsis had statistically higher mortality, and the group with ARDS had numerically higher mortality. While it's important to recognize that these are subgroup analyses of a secondary outcome, the study didn't include criteria for initiation of de-resuscitation that interrogated intravascular volume tests. It's impossible that in patients who are actively inflamed that de-resuscitation may have been started on some patients in these groups who remained intravascularly dry. These outcomes certainly require evaluation in future studies, and may merit addition in future studies of clear evidence of intravascular volume status prior to initiation of de-resuscitation exploration and cautious interpretation in future studies. In summary, for patients with ARDS, dry lungs are happy lungs. Conservative initial fluid resuscitation is feasible, but clinical outcomes remain unclear. We're going to have some really good, high-quality data soon from two major trials. And fluid de-resuscitation is feasible in critically ill patients, but the timing and threshold for initiation and clinical outcomes continue to need better data. Thank you very much.

fluid management

respiratory failure

ARDS

fluid overload

mortality

research