The myth of large-volume resuscitation in acute pancreatitis

Introduction

Severe pancreatitis causes fluid extravasation from the vasculature, sometimes causing shock. Traditionally this has been managed by administration of large volumes of crystalloid. For example, the 2013 American College of Gastroenterology Guideline recommended providing 250-500 ml/hour of crystalloid for the first 12-24 hours of hospitalization. They recommended targeting fluid resuscitation to achieve dilution of laboratory values, stating that “the goal to decrease hematocrit and blood urea nitrogen and maintain a normal creatinine during the first day of hospitalization cannot be overemphasized.” There has been the notion that providing large volumes of fluid early in the course might lessen the ultimate severity of the pancreatitis.

Currently there is increasing awareness about the potential harm of excess fluid administration. Although it was traditionally thought that fluid is good for the kidneys, volume overload may impair renal function due to reduced venous outflow causing poor perfusion and renal edema (Prowle 2013). Especially pertinent in pancreatitis, volume overload increases the risk of intra-abdominal compartment syndrome and pulmonary edema.

Although some are extremely enthusiastic about fluid administration, ultimately fluid is neither goodnor evil but rather it is a double-edged sword. The challenge is to develop a rational, evidence-based resuscitation strategy utilizing the ideal amount of fluid to achieve adequate tissue perfusion without causing harm.

Theory: The paradox of volume management in acute pancreatitis

Fluid therapy for acute pancreatitis is extremely frustrating. Pancreatitis increases capillary permeability. Therefore, patients are often intravascularly depleted and thus volume-responsive. Administering volume usually causes hemodynamic improvements, that are unfortunately short-lived because the fluid rapidly leaks out of the vasculature. Therefore, fluid administration may cause a temporary improvement in hemodynamics at the cost of worsening volume overload.

A central question is how aggressively to restore intravascular volume during the initial phase of resuscitation. Attempting to fully restore intravascular volume may cause a greater elevation of intravascular pressures, encouraging fluid to extravasate out of the vasculature. Therefore, it is possible that the best approach is a more gradual fluid resuscitation with the goal of maintaining a low yet adequateintravascular volume. It is possible that mild intravascular hypovolemia may be a compensatory mechanism in severe pancreatitis, and rendering the patient intravascularly euvolemic could actually be dangerous.

Evidence: Prospective Randomized Controlled Trials (RCTs)

The only RCTs available were performed in China at a single medical center. These studies have many limitations, including the use of Chinese herbal medications, hydroxyethyl starch, somatostatin, and varying ratios of crystalloid:colloid. Nonetheless these RCTs are the best human data available.

Mao 2009 randomized 76 patients with severe acute pancreatitis to a rapid fluid expansion group (Group I, initial fluid infusion rate 10-15 ml/kg/hr) or a controlled fluid expansion group (Group II, initial fluid infusion rate 5-10 ml/kg/hr). For patients in the controlled fluid expansion group, vasopressors were given early to maintain blood pressure while patients underwent gradual fluid expansion. Both groups ended up receiving the same amount of total fluid over the first three days, but the rapid expansion group received more fluid initially and sequestered more fluid (5.4 liters net positive versus 4.2 liters; see table below). Patients in the rapid fluid expansion group had statistically significant increases in mechanical ventilation (34% vs. 26%), abdominal compartment syndrome (26% vs. 13%), sepsis (23% vs 15%), and death (75% vs. 64%).

Hemoconcentration correlates with poor outcomes, and it has been proposed that sufficient fluid to reverse hemoconcentration might, therefore, improve outcomes. Mao 2010 randomized 115 patients with severe acute pancreatitis to a rapid hemodilution group (target hematocrit <35% within 48 hours) or slow hemodilution group (target hematocrit >35% within 48 hours). Patients in the rapid hemodilution group received more fluid at admission and during the first hospital day (table below). The rapid hemodilution group had significantly higher rates of sepsis (79% vs. 58%), higher APACHE II scores, lower creatinine clearance, and increased mortality (34% vs. 15%). Therefore, a strategy involving early aggressive volume resuscitation was again shown to worsen outcomes.

Evidence: Correlational Studies

Gardner 2009 and Gardner 2011are the most commonly cited studies to support early aggressive fluid resuscitation. Both studies were retrospective with the same design. The pattern of fluid administration during the first 72 hours of hospitalization was analyzed, and patients were divided into an “early resuscitation” group (patients who received >33% of the total fluid within the first 24 hours) or a “late resuscitation” group (patients who received <33% of the total fluid within the first 24 hours). Outcomes were compared between the two groups, with the implication that the late resuscitation group had received insufficient fluid during the first 24 hours.

These two studies came to conflicting conclusions. Gardner 2009 revealed an improvement in mortality among patients with severe pancreatitis in the early resuscitation group. In contrast, Gardner 2011 found no difference in organ dysfunction or mortality in patients with severe pancreatitis in the early resuscitation group (early resuscitation was associated with improvement in organ function and length of stay only in patients with mild pancreatitis). This difference may reflect that patients in the late resuscitation group received more fluid in Gardner 2011compared to Gardner 2009:

Other correlational studies have demonstrated worse outcomes among patients receiving large-volume resuscitation. de Madaria 2011 found that patients receiving >4.1 liters within 24 hours had higher rates of respiratory failure, renal failure, and intra-abdominal fluid collections. Eckerwall 2006 found that patients receiving >4 liters of fluid in the first 24 hours had increased risk of respiratory complications. It is impossible to determine if patients do worse due to fluid administration, if more fluid is administered because patients have more severe disease, or both. Ultimately it must be emphasized that all of these correlational studies are hypothesis generating only and cannot prove causality.

American College of Gastroenterology (ACG) 2013 guidelines are not evidence-based

ACG guidelines recommend that 250-500 ml/hour of isotonic crystalloid be given to all patients without comorbidities for 12-24 hours, in addition to boluses as required to establish hemodynamic stability. In practice this often causes patients to receive 6-12 liters of fluid over the first 24 hours of hospitalization. This amount of volume resuscitation is not supported by any evidence. All available RCT data suggests that larger volumes of crystalloid may cause harm. Even in the Gardner studies, the “early resuscitation” groups didn't receive anywhere near this amount of fluid.

ACG guidelines also recommend aggressive hydration targeted to decrease the blood urea nitrogen level (BUN). Although hemoconcentration correlates with worsened outcomes, available evidence does not support targeted hemodilution. Mao 2010 demonstrated that fluid resuscitation to target a hematocrit <35 within 48 hours increased mortality. Wu 2011 compared a goal-directed fluid resuscitation strategy using BUN to standard therapy, but patients in both groups received similar amounts of fluid and measuring BUN was not beneficial. Targeting volume resuscitation against BUN has no precedent in the resuscitation of other critical illnesses. For a patient who is meeting hemodynamic targets (i.e., adequate blood pressure and urine output), continuing volume resuscitation based on a BUN value is probably dangerous.

**How should patients with acute pancreatitis be resuscitated?**

Patients with severe acute pancreatitis should probably be resuscitated with a balanced approach employing moderate amounts of fluid as well as vasopressors if needed. This situation has many similarities to resuscitation of septic shock, and may be guided by the same principles.

The 2013 guidelines issued by the International Association of Pancreatology and American Pancreatic Association suggested that most patients require about 2.5-4 liters over the first 24 hours, which seems consistent with the above data (note that this correlates with an infusion rate of no higher than 125 ml/hr, assuming the patient receives 1,000ml fluid initially). Clearly fluid administration should be titrated to clinical response, with efforts to use as little as possible. Nonetheless, it is useful to have a rough idea of the amount of fluid which may be beneficial. It must be noted that the quality of existing evidence is low and further studies are needed to reach any firm conclusion (Haydock 2013). This amount of fluid is similar to volumes used in current studies of septic shock (Marik 2014).

There is little known about the ideal targets for resuscitation in pancreatitis, but this is absolutely critical to delivering the appropriate intensity of resuscitation. For example, tachycardia may be misleading because a low-grade tachycardia may persist despite adequate resuscitation. Continuing resuscitation with a goal of normalizing the heart rate could, therefore, lead to excessive fluid administration. One of the most important goals of resuscitation is avoidance of renal failure, so adequate urine output, if present, is a powerful signal that the patient is sufficiently resuscitated. However, if the patient has already progressed to intrinsic renal failure due to acute tubular necrosis, then urine output may remain low regardless of resuscitative efforts and can be misleading. Central venous pressure is unhelpful (Huber 2008, Marik 2013).

Given that aggressive bolusing may promote fluid extravasation, it may be sensible to try to provide fluid in the form of a continuous infusion (to counteract continuous fluid losses from ongoing extravasation)(Hilton 2012). Since fluid administration has limited long-term efficacy, supplementation with norepinephrine may be needed to support blood pressure, noting that this also increases preload due to venoconstriction and may improve renal function (Bellomo 2008).

Wu 2011 performed a prospective RCT of patients with acute pancreatitis randomized to resuscitation with normal saline or lactated Ringers. One day after randomization, patients treated with lactated Ringers had significantly lower rates of C-reactive protein and lower rates of systemic inflammatory response syndrome (SIRS). Although this is only one study, there are many other reasons to prefer lactated Ringers to normal saline. Until more evidence is available, lactated Ringers may be the preferred crystalloid for resuscitation of acute pancreatitis (more information about lactated Ringers in prior post about pH-guided resuscitation).

Conclusions

Although limited, all prospective RCT evidence reveals harm from large-volume resuscitation in acute pancreatitis.

Excessive volume resuscitation increases the risk of respiratory failure, renal failure, abdominal fluid collections, sepsis, abdominal compartment syndrome, and death.

The American Gastrointestinal Association 2013 guidelines recommend large-volume resuscitation with 250-500cc/hr crystalloid targeted to decrease the BUN. These recommendations reflect dogma rather than current evidence.

We agree with recent evidence-based guidelines by the IAP/APA which state that most patients may be treated adequately with a resuscitation of about 2,500-4,000 ml in the first 24 hours (corresponding to roughly 125 ml/hour crystalloid infusion).

Compared to normal saline, Lactated Ringers produces lower cytokine levels and lower rates of SIRS. Lactated ringers may be the preferred crystalloid for resuscitation of acute pancreatitis.

Coauthored with Paul Farkas, senior consultant in Gastroenterology.

Image credits: https://en.wikipedia.org/wiki/Edema#/media/File:Combinpedal.jpg