Moderate Versus Aggressive Fluids for Acute Pancreatitis

TITLE: Moderate Versus Aggressive Fluid Therapy in Patients with Mild to Moderate Acute Pancreatitis

BACKGROUND AND HYPOTHESES In the United States, acute pancreatitis results in more than 200,000 hospital admissions each year and has an associated mortality of 5%. While protease inhibitors, platelet activating factor inhibitors, and somatostatin have been studied, no pharmacologic agents have been shown to impact the course of acute pancreatitis and the treatment remains supportive. Fluid resuscitation has been recommended based on animal data, however, until recently very limited clinical studies were available to assess its impact. Animal studies using rats, dogs, and pigs suggest that acute pancreatitis compromises splanchnic perfusion and the pancreatic microcirculation. Work by Banks et al demonstrated that hemoconcentration is associated with necrosis and inadequate resuscitation in those with elevated hematocrit was associated with markedly increased rates of necrosis. Blood urea nitrogen and creatinine levels also appear to correlate with mortality in acute pancreatitis.

Nonetheless, recent studies have yielded contradictory results regarding whether fluids should be used judiciously or liberally. Retrospective work by Gardner suggests that early resuscitation is beneficial. Other recent studies however including the work by De Madaria et al suggest that those were received more than 4.1L of fluids during the first 24 hours were more likely to develop organ failure than those who received less. In the only randomized controlled trial of fluids in acute pancreatitis 76 patients with severe pancreatitis, including those with renal failure, were randomized to massive 10-15cc/kg/hr (mean of 545 cc/hour) versus less aggressive resuscitation, 5-10cc/kg/hr. Rates of mechanical ventilation, abdominal compartment syndrome, sepsis, and mortality were higher in the massive resuscitation group In a well done prospective study by Wu et al 40 patients with acute pancreatitis were randomized to goal directed management according to protocol versus standard therapy and also randomized patients lactated ringers (LR) versus normal saline (NS). They were unable to measure a difference between goal directed management and standard therapy arm because the patients were managed very similarly in terms of volume of resuscitation. The rates of systemic inflammatory response syndrome (SIRS) were lower in both groups, 25%, than the predicted level of 50%. It was demonstrated that those who were treated with LR were much less likely to develop systemic inflammatory response syndrome (SIRS) than those with NS. LR will be used in this study based on these results. The current study also employs the system of checkpoints used by Wu et al with the intent to compare moderate versus aggressive fluid resuscitation.

OBJECTIVES AND PURPOSE: Precise fluid management strategy in patients with acute pancreatitis is unclear. The investigators principal aim is to perform a well done randomized trial to assess the hypothesis that aggressive fluids benefits those with mild to moderate acute pancreatitis and help standardize the treatment of these patients. An additional goal of this pilot study will be to determine the effect size of the difference between clinical improvement in those receiving agressive versus moderate fluids with the aim of estimating sample size for larger trials.

STUDY DESIGN: Patients presenting to the Los Angeles County Hospital with mild to moderate pancreatitis are the focus population of this study. Pancreatitis will be diagnosed using 2 of 3 criterion; amylase or lipase greater than 3 times the upper limit of normal, epigastric abdominal pain, and imaging consistent with acute pancreatitis. Available data suggest that in those with severe pancreatitis aggressive fluid rescusctiation may be harmful and these patients will be excluded.Patients will be randomized to aggressive versus moderate fluids within 4 hours of diagnosis of pancreatitis. The inclusion and exclusion criterion will be assessed. Randomization will be performed using a computer program with concealed allocation. The amount of fluids and intervention performed in the emergency department prior to randomization will be recorded but not affect randomization strategy.

Those in the aggressive hydration group will receive a bolus of lactated ringers (LR) of 20cc/kg and LR will be started at 3cc/kg of LR. This protocol was used in a well respected prior randomized trial and was not significantly associated with fluid overload or other adverse consequences. Those in the moderate hydration group will receive a bolus of 10cc/kg and be started on LR at a rate of 1.5cc/kg/hr. Boluses will be given over 30 minutes.

After 8-16 hours both groups will have a hematocrit, Cr, and BUN assessed (which is part of the standard procedure of assessing the CBC and complete metabolic panel) in those with acute pancreatitis.If patients in the moderate fluids group an increased or unchanged hematocrit, Cr, or BUN they will be bolused with 20cc/kg of LR and fluids thereafter adjusted to 3cc/kg/hr of LR. Those in the aggressive group who have decreased hematocrit, Cr, and BUN at this time will have their rate reduced to 1.5cc/kg/hr of LR. Those in the aggressive group who have an increased or unchanged hematocrit, Cr, or BUN will receive a second bolus of 20cc/kg of LR.

At 20-28 hours the hematocrit, BUN, and Cr will again be assessed and the same measures will be taken based on the levels. On the second day their diet will be advanced to clears if the have a decreased HCT, BUN, Cr and are afebrile and have only minimal epigastric pain. Once patients are tolerating a regular diet without difficulty their fluids will be decreased at the discretion of their treating physicians. Patients will be assessed for symptomatic fliud overload at each of the checkpoints as well as throughout their stay. If they develop fluid overload including pitting edema, ascites, anasacra, pulmonary edema, or dyspnea will have their fluids halted and be managed at the discretion of their treating physicians. Patients who develop oliguria, anuria or hypotension will also be removed from the treatment protocol as these findings signify renal failure and severe pancreatitis which are endpoints. If patients continue to have rising Cr, BUN, or HCT after the second checkpoint, ie at 32-36 hours they will also be removed from the treatment protocol as this signifies that they are refractory to fluid resuscitation and will be managed at the discretion of their treating physicians.

SELECTION AND WITHDRAWAL OF SUBJECTS: See inclusion and exclusion criterion

STRATIFICATION/RANDOMIZATION SCHEME: Stratification factors. Patients will not be stratified prospectively. At the end of the study subgroup analysis will be performed to compare be stratified into 3 groups; the results of those with gallstone pancreatitis, alcoholic pancreatitis, and other pancreatitis due to other origins.There will be a randomization of patients to moderate versus aggressive fluids groups. Group randomization will be balanced. Randomization will be performed by the SC-CTSI Bioinformatics and Bioinformatics Group.

ASSESSMENT OF SAFETY: If patients develop fluid overload including pitting edema, ascites, anasacra, pulmonary edema, or dyspnea will be removed from the treatment protocol and be managed at the discretion of their treating physicians. Adverse events will be reported to the IRB and data safety monitoring board comprised of the principal investigators from medicine/gastroenterology, emergency medicine, and surgery as well as the entire study team. An outside senior physician from gastroenterology or surgery will also participate.

CRITERIA FOR EVALUATION AND ENDPOINT DEFINITIONS: See Primary and Secondary As a pilot project an additional goal of this study will be to measure effect sizes such that larger trials of moderate versus aggeressive fluids may be informed.

STATISTICAL CONSIDERATIONS: Descriptive statistics will be performed for each group, and comparisons made to determine if the groups are balanced on the stratification factor of type of pancreatitis. Patient characteristics will be compared between groups as well to determine if there is imbalance on other possible covariates. We expect that 40% of participants will present with gallstones, 40% with alcoholic pancreatitis, and 20% of the sample will have pancreatitis due to other causes. Approximately, 60% of our sample is expected to be Latino, and the gender ratio is expected to be balanced. Patients will be classified afterward as mild if the hospitalization >3 days in length, moderate if the hospitalization was 3-10 days, and severe if greater than 10 days or if necrosis and complications of pancreatitis develop..

Because this is a pilot study, any variable for which there is a P \< .10 difference between groups will be included in analyses as a possible confounder. For the primary outcomes of evaluating the proportion of patients in each group who improved with assigned treatment, a chi squared analysis will be performed at 8-16 hours and at 20-28 hours to evaluate the number of participants receiving moderate fluids. Logistic regression modeling will be used to test differences in rates controlling for covariates. Survivial analysis will be used to explore the secondary aim of determining if there is a difference in time to tolerating normal diet. Exploration will also be made of the relationship of outcomes with patient characteristics to determine if these factors need to be included in stratification for future studies. All modeling will follow an intent-to-treat model where the primary factor is assigned group, regardless of treatment.

The main goal of this pilot study is to determine effect sizes, so odds ratios and confidence intervals will be reported for use in developing a fully powered study. A sensitivity analyses was peformed using G starPower to determine detectible differences. With the proposed sample size of 60, there would be sufficient power to detect a decrease in the proportion of participants requiring aggressive fluids at 8-16 hours of 17%, versus the expected 50% in patients randomized to moderate treatment (alpha = 0.05, one sided; 1-beta = 0.80).