Acute Lipotoxicity Regulates Severity of Biliary Acute Pancreatitis without Affecting Its Initiation

Abstract

Obese patients have worse outcomes during acute pancreatitis (AP). Previous animal models of AP have found worse outcomes in obese rodents who may have a baseline proinflammatory state. Our aim was to study the role of acute lipolytic generation of fatty acids on local severity and systemic complications of AP. Human postpancreatitis necrotic collections were analyzed for unsaturated fatty acids (UFAs) and saturated fatty acids. A model of biliary AP was designed to replicate the human variables by intraductal injection of the triglyceride glyceryl trilinoleate alone or with the chemically distinct lipase inhibitors orlistat or cetilistat. Parameters of AP etiology and outcomes of local and systemic severity were measured. Patients with postpancreatitis necrotic collections were obese, and 13 of 15 had biliary AP. Postpancreatitis necrotic collections were enriched in UFAs. Intraductal glyceryl trilinoleate with or without the lipase inhibitors resulted in oil red O–positive areas, resembling intrapancreatic fat. Both lipase inhibitors reduced the glyceryl trilinoleate–induced increase in serum lipase, UFAs, pancreatic necrosis, serum inflammatory markers, systemic injury, and mortality but not serum alanine aminotransferase, bilirubin, or amylase. We conclude that UFAs are enriched in human necrotic collections and acute UFA generation via lipolysis worsens pancreatic necrosis, systemic inflammation, and injury associated with severe AP. Inhibition of lipolysis reduces UFA generation and improves these outcomes of AP without interfering with its induction.

The mystique of acute pancreatitis (AP) lies in its diverse origins, unpredictable course, and outcomes, ranging from resolution with minimal care to being a debilitating, protracted, and potentially lethal condition despite intensive care and complex interventions to manage its complications. The course AP takes seems unrelated to the origin in most cases, with differences in the predominant origin of AP reported in studies from different countries.^1–5 However, studies have repeatedly reported a higher body mass index (BMI) or obesity to be associated with severe AP (SAP).^1–8 SAP may result from severe pancreatic necrosis, in which >30% of the pancreas is necrosed,^9,10 or from persistent or multisystem organ failure, such as respiratory and renal failure. Obese patients have been reported to be more prone to both these types of complications of AP.¹⁻⁸

In contrast to the clinical scenario, conventional animal models of AP differ in the initiating factor used, and the severity associated with these has been attributed to the inciting stimulus^11–13 or species in which the model has been executed.^12–15 For example, rat intraductal bile salt–induced pancreatitis has been classified as severe in contrast to the caerulein model, which is mild.^12,13 Interestingly, caerulein-induced AP is milder in rats than in mice, which have more pancreatic necrosis, and thus mouse caerulein pancreatitis is classified as severe.^14,15 However, in both these cases, the pancreas returns to normal a few days after cessation of the insult, with no residual necrotic areas or organ failure. On the basis of such models, a potential target is regarded as therapeutically relevant if it plays a role in mechanistically dissimilar models of AP. An example of this is phosphatidylinositol 3-kinases and associated trypsin generation,^11,16,17 which we and others have previously found to be relevant to AP of different causes.^11,16,17

This discord (ie, the lack of association of outcomes to cause as noted clinically) and how animal models are interpreted have resulted in serious discrepancies between what is predicted to be beneficial in animal models of AP and the success of such interventions in clinical trials. The failure of serine protease and trypsin inhibition to improve outcomes of AP in >70 clinical trials performed during the last 5 decades is a classic example.^18–27

Recently, the mechanistic proof of obesity being a modifier of AP outcomes has emerged, with the same model being mild in lean mice and severe in obese mice, associated with an exaggerated inflammatory response and mortality.²⁸ Our recent studies have found that lipolysis of visceral fat in obese mice may contribute to this severity.²⁹ However, obesity is also associated with a baseline proinflammatory state,^30–32 and because fatty acids (FAs) are proinflammatory,^29,33,34 it has yet to be decided whether short-term generation of FAs by the lipolysis of visceral fat or the preexistent inflammatory state associated with obesity determines the severity of AP in these models.

We therefore analyzed human postpancreatitis necrotic collections (PPNCs) for the nature of FAs in them. We also noted the most common cause of AP in our patients. Because biliary AP was the most common type of AP and unsaturated FAs (UFAs) were abundant in PPNCs, we studied whether their acute lipolytic generation in rats, which are otherwise normal, results in the severe outcomes noted in SAP and whether inhibition of such lipolysis, using 2 distinct lipase inhibitors separately, alters the initiation of AP or the parameters of its severity. Interestingly, we realize that the beneficial effect of lipase inhibition, which decreases the generation of UFAs, is independent of the initiation of biliary AP. These findings have relevance to how we design and interpret animal models of AP in the context of human disease.

Materials and Methods

Human Pancreatic Fluid Collections

Pancreatic necrosis fluids and corresponding information on age, sex, BMI, primary diagnosis, duration of disease at the time of intervention, and fluid source were collected as a part of a pancreatic waste fluid protocol between December 2010 and June 2012 at the University of Pittsburgh Medical Center, Pittsburgh, Pennsylvania, and at the Washington University Medical Center, St. Louis, Missouri. The diagnosis of pancreatic necrosis was made based on clinical and imaging criteria.³⁵ After collection, all samples were transported immediately or shipped overnight on ice to the laboratory. Freezing was avoided at this step to minimize repeated freeze and thaw as would occur at the time of aliquoting and before analysis. Moreover, because these collections had accumulated at body temperature during at least 4 weeks, a single freeze would unlikely improve preservation. The samples were spun at 300 × g for 5 minutes, and the supernatants were sonicated, aliquoted, and frozen at −80°C until they were analyzed later after a single thaw. The Biospecimen Reporting for Improved Study Quality criteria³⁶ are fulfilled by these details. Approval was obtained from the institutional review boards at the University of Pittsburgh Medical Center and at Washington University Medical Center and the Committee for Oversight of Research Involving the Dead at the University of Pittsburgh Medical Center.

Animals and Animal Procedures

Male, 250- to 300-g Wistar rats (Charles River Laboratories, Wilmington, MA) were used after a minimum of 2 days of acclimatization. They were fed standard laboratory chow and allowed to drink ad libitum until the night before the surgery at which time they were fasted. Housing was at temperatures ranging from 21°C to 25°C, with a 12-hour light-dark cycle. Animals were anesthetized with ketamine-xylazine with sterile precautions. Then 50 μL/100 g of body weight of glyceryl trilinoleate (GTL) (Sigma, St. Louis, MO) alone or with 50 mg/mL of orlistat dissolved in GTL (Cayman Chemical, Ann Arbor, MI; GTLO group) or 25 mg/mL of cetilistat dissolved in GTL (Jinan Wedo Industrial Co., Ltd., Jinan City, China; GTLC group) was injected into the pancreatic duct of rats. This was followed by ligating the biliopancreatic duct just proximal to its entry into the duodenum. The GTL dose (5% to 10% of pancreas volume, based on the rat pancreas weighing approximately 0.5 to 1 g/100 g of body weight^37,38) was chosen to be in the range of intrapancreatic fat (23.4% ± 4.3%) and percentage fat necrosis (12.6% ± 3.4%) noted histologically in patients with SAP.²⁹ Fat necrosis contributes to most pancreatic parenchymal necrosis during SAP in humans.^29,39 Preliminary studies found that duct ligation alone without infusion resulted in mild biliary AP at 1 day evidenced by a transient increase in serum amylase and lipase (more than fourfold above normal), serum alanine transaminase (ALT; > 300 U/L), bilirubin (>3 mg/dL), and pancreatic edema (79% versus 73% in controls), all with a P < 0.01 versus controls but with no mortality during 5 days. The intraductal route was preferred over the intraparenchymal route to avoid hemorrhage. Postoperatively, the animals were administered buprenorphine for pain control, cefazolin to prevent infections, and 10 mL of saline subcutaneously daily. They were followed up for 5 days (survival) or sacrificed at the time near mortality (moribund). In separate studies, the GTLO and GTLC groups were also electively sacrificed the morning after duct ligation, by which time there was 100% mortality in the GTL group. Serum and pancreas tissue were harvested to study levels of cytokines and lipotoxic mediators and for morphologic analysis as described below. There were 8 to 10 animals in each group. All experiments were approved by the Institutional Animal Care and Use Committee of the University of Pittsburgh.

Acinar Harvest and in Vitro Assays

Pancreatic acini were harvested^40–42 and preincubated with 50 μmol/L orlistat²⁹ or cetilistat, after which linoleic acid (LA)²⁹ or GTL was added, followed by incubation for 4 hours. At the end of which, cell death was measured by lactate dehydrogenase (LDH) leakage²⁹ or glycerol generation was measured²⁹ to quantify GTL hydrolysis. The methods are described in the articles referenced.

Cytokine Assays

As previously described,²⁹ cytokine assays were performed on serum samples using the fluorescence-based capture sandwich immunoassay (Luminex) on samples without severe hemolysis. The MILLIPLEX MAP Rat Cytokine-Chemokine Magnetic beads panel from Millipore (Billerica, MA) was used for this. The samples were analyzed at the Luminex Core Facility of the University of Pittsburgh Cancer Institute.

Evaluation of Pancreatic Necrosis and Special Stains

Whole pancreas H&E-stained sections were examined by a trained pathologist (S.N.) blinded to the sample, as described previously.^29,40 Briefly, all parenchymal areas were imaged with a ×4 objective and photographed. Necrotic area and total parenchymal area were measured in pixels, and percentage of total area necrosed was calculated for each pancreas. Oil red O staining was performed on pancreatic cryosections as described previously.²⁹ For pancreatic sections from autopsies, the slides of patients were procured and stained with H&E or von Kossa as described previously.^29,39 Terminal deoxynucleotidyl transferase-mediated dUTP nick-end labeling (TUNEL) staining was performed on paraffin sections of the lungs and kidneys as described previously.²⁹

Nonesterified FA Analysis

As previously described, nonesterified FA analysis was performed using gas chromatography²⁹ on samples for which an adequate volume (at least 25 μL) was available. Total UFA amounts were calculated by adding individual C16:1, C18:1, C18:2, and C20:4 FAs.

Serum Analyses

Serum ALT, total bilirubin (on nonhemolyzed samples to avoid interference with the assay), serum amylase, lipase, and blood urea nitrogen (BUN) were measured following the manufacturer’s (Pointe Scientific Inc., Canton, MI) instructions. Tests were performed on a ChemWell-T chemistry analyzer (Awareness Technology, Palm City, FL).

Statistical Analysis

All values unless otherwise specified are reported as means ± SEM. All data were collected as continuous variables. Pairs were compared using the U-test. P < 0.05 was considered to indicate statistical significance. This was adjusted for multiple comparisons when comparing more than two groups.

Results

UFAs Are Higher in Human PPNCs

Patients with PPNCs were obese (BMI, 36 ± 1.8). In 13 of these 15 patients, the cause of PPNCs was biliary AP. Gas chromatographic analysis of the PPNC fluids revealed that UFAs were significantly higher than saturated FAs (SFAs), both in amount (2.7 ± 1.0 mmol/L versus 1.0 ± 0.3 mmol/L, P = 0.033) (Figure 1A) and as a percentage of total (68.5% ± 3.0% versus 31.7% ± 2.9%, P < 0.001) (Figure 1B).

Figure 1.

Lipolysis of intrapancreatic unsaturated triglyceride contributes to parenchymal necrosis. Millimolar concentrations (A) and proportions (as percentage total, B) of SFAs and UFAs in PPNCs from patients in whom UFA in both amounts (P = 0.033) and proportions (P < 0.001) are significantly higher than SFA amounts and proportions. Comparisons between the groups are depicted with box plots showing mean (dashed line), median (solid line), the 25th and 75th percentiles (upper and lower parts of the box plot divided by the median line), the 10th and 90th percentile (whiskers), and the outliers (dots). Serial sections of pancreatic tissue from a patient with AP at the time of autopsy stained for H&E (C) and von Kossa (D). Note the loss of cell outlines and morphologic detail in the parenchyma (signifying parenchymal necrosis; dotted outline) surrounding the fat necrosis (amorphous blue look of adipocytes on H&E and brown staining on von Kossa) and the presence of lesser intense brown staining in the necrosed parenchyma, suggestive of leakage of contents from the fat necrosis. Oil red O–stained pancreatic cryosections from rats belonging to the control (E), GTL (F), GTLO (G), and GTLC (H) groups. Note the accumulation of oil red O–positive staining is associated with loss of cellular detail with GTL but not with GTLO or GTLC, supporting the dependence of this on lipolysis of the triglyceride. Original magnification: ×4 (C and D); ×40 (E–H).

On the basis of i) a high level of UFAs in the PPNCs; ii) histologic evidence that areas of fat necrosis (Fig 1, C and D) have surrounding parenchymal necrosis (areas with loss of cell outline surrounded by the dotted line), which has also been reported previously^29,39,43,44; iii) pancreatic lipases leaking basolaterally into fat in human AP,^45,46 allowing hydrolysis of adipocyte triglyceride; iv) previous studies reporting that UFAs but not SFAs saponify⁴⁷; v) previous evidence that UFAs cause acinar necrosis^29,39; vi) the finding that unhydrolyzed triglyceride within normal adipocytes in the pancreas (Figure 1, C and D) is not surrounded by acinar necrosis; and vii) the patients with PPNCs being obese and having biliary pancreatitis, we designed a model to replicate this pathophysiology.

This model involved intraductal injection of GTL to allow for its lipolysis by lipases. GTL was injected alone or with the lipase inhibitors orlistat or cetilistat in separate animals to study the effect of lipase inhibition on the course of AP. Injection of these lipids resulted in oil red O–positive areas in the pancreas (Figure 1, F–H) simulating the intrapancreatic fat of obesity. In the absence of lipase inhibitors, the red areas had loss of cellular detail suggestive of necrosis (Figure 1F). Both orlistat (Figure 1G) and cetilistat (Figure 1H) caused the accumulation of oil red O–positive droplets; however, cellular details were preserved.

Thus, this model could potentially allow for the mixing of pancreatic lipases with triglyceride as would occur with basolateral leakage of lipases during AP^{45,46,48−51} with consequent hydrolysis of adipocyte triglyceride, morphologically seen as intrapancreatic fat necrosis in obese patients with SAP (Figure 1, C and D).^{29,39,44–46} This model also involves pancreatic duct obstruction as a result of ligating the biliopancreatic duct after GTL injection just proximal to the duodenum. Preliminary studies have noted that the sole act of such a duct ligation within 1 day results in a more than threefold increase in serum amylase or lipase levels over controls and elevated ALT and bilirubin levels, which are markers consistent with human biliary AP,⁵²⁻⁵⁵ and no mortality for 5 days (see Materials and Methods).

Lipase Inhibitors Do Not Interfere with the Initiation of AP

We first studied the effect of GTLO or GTLC on the parameters of biliary AP (Figure 2). GTL caused a greater than threefold increase in serum amylase (4273 ± 1017 U/L) above normal (unoperated on controls, 621 ± 31 U/L) (Figure 2A). Serum bilirubin and ALT levels were elevated in the GTL group (Figure 2, B and C). These findings are compatible with the definition of biliary AP.^52–55 Animals from the GTLO and GTLC groups sacrificed on the same day as the ones with GTL-induced pancreatitis (day 1) had similar serum amylase and bilirubin levels compared with the GTL group (Figure 2A) or to the rats that only had ligation of the biliopancreatic duct. The serum ALT level, although significantly less in the GTLO group (5.3- ± 0.8-fold control) compared with the GTL group (33- ± 15-fold control, P = 0.009), remained significantly elevated in the GTLO group compared with controls (52 ± 10 U/L, P = 0.008). The extremely high ALT levels in the GTL group may be related to mortality and an associated shock-liver–like picture. The bilirubin and ALT levels in the GTLO and GTLC groups at 5 days were not significantly different from the levels in rats that only underwent ligation of the biliopancreatic duct. These findings suggest that administration of lipase inhibitors does not interfere with the induction of biliary AP and were confirmed in in vitro studies that found that, although orlistat and cetilistat prevented hydrolysis of GTL (Supplemental Figure S1A), they do not interfere with cell death induced by LA (Supplemental Figure S1B).

Figure 2.

Inhibition of lipolysis does not affect parameters used to diagnose the onset of biliary AP. Serum amylase (A), bilirubin (B), and ALT (C) as measured in controls (CON) and animals with an intraductal infusion of GTL at the time when they were moribund (GTL) and those with an infusion of GTLO and GTLC on day 1 and day 5. Comparisons between the groups are depicted with box plots showing mean (dashed line), median (solid line), the 25th and 75th percentiles (upper and lower parts of the bars), the 10th and 90th percentile (whiskers), and the outliers (dots). Note that there is no difference in serum amylase and bilirubin levels among the GTL, GTLO, and GTLC groups at day 1 and that the ALT level, although lower in the GTLO group compared with the GTL group, is still significantly elevated compared with controls. At day 5 the serum amylase level is significantly lower in the GTLO and GTLC groups (even lower than in controls) consistent with resolved AP, whereas the bilirubin level remains elevated consistent with persistent biliary obstruction. The ALT level was significantly reduced in both groups, as is typically noted in chronic biliary obstruction in humans. ^∗P < 0.05 versus control; ^†P < 0.05 versus the GTL group.

Lipase Inhibition Reduces the Increase in Serum UFAs Associated with GTL-Induced Pancreatitis

We then compared the effect of the lipase inhibitors on serum lipase and serum FAs to study their efficacy in inhibiting lipases in vivo. GTL administration resulted in a large increase in serum lipase over controls (6443 ± 1619 U/L versus 16 ± 15 U/L, P < 0.001), and both the GTLO (244 ± 61 U/L, P = 0.001 versus GTL) and GTLC groups (788 ± 397 U/L, P < 0.035 versus GTL) had significantly reduced lipase levels, which, however, were elevated above controls in these groups at day 1 (Figure 3A). Consistent with the lipolysis of triglyceride and the release of the free FAs, there was an increase in serum UFAs both in concentration (1561 ± 628 μmol/L versus 94 ± 14 μmol/L in controls, P < 0.001) (Figure 3B) and proportion (60.8% ± 1.6% versus 43.9% ± 3.2% in controls, P < 0.001) (Figure 3C), with a corresponding decrease in SFA proportions in the GTL group. The lipase inhibitors orlistat and cetilistat significantly prevented this GTL-induced increase of UFAs in both concentration and proportion (Figure 3, B and C). These findings, along with the in vitro findings that orlistat and cetilistat reduce glycerol generation from GTL incubated with pancreatic acini (Supplementary Figure S1A), support the role of orlistat and cetilistat as lipase inhibitors in this model.

Figure 3.

Inhibition of lipolysis is associated with a reduction in serum UFAs. Elevation of serum lipase (A) induced by GTL infusion is reduced in the GTLO and GTLC groups on day 1 (when the moribund animals in the GTL group were sacrificed), consistent with the role of orlistat and cetilistat as lipase inhibitors. This finding was associated with the lipase inhibitors preventing an increase in both the concentration (B) and proportion (C) of UFAs induced by GTL. The lipase levels remained significantly reduced at day 5 and similar to amylase (Figure 2A) returned to near normal. Comparisons between the groups are depicted with box plots showing mean (dashed line), median (solid line), the 25th and 75th percentiles (upper and lower parts of the bars), the 10th and 90th percentile (whiskers), and the outliers (dots). ^∗P < 0.05 versus control; ^†P < 0.05 versus the GTL group.

Lipase Inhibition Reduces GTL-Induced Pancreatic Necrosis

GTL injection resulted in gross evidence of severe hemorrhagic pancreatitis within the first day. Pancreatic hemorrhages were seen mostly in the body and tail (Figure 4A), and the whole pancreas had a necrotic appearance. Histologically, this was seen as sheets of necrotic parenchyma with a few surviving islands. Both orlistat (Figure 4B) and cetilistat (Figure 4C) reduced gross evidence of pancreatic hemorrhages and necrosis. Histological appearance in control, GTL, GTLO and GTLC groups is shown (Figure 4, D–G) with a few scattered hemorrhages in the latter two groups. A total of 72.1% +/− 10.6% of the parenchyma was necrosed in the GTL group (Figure 4H). On histologic quantification of the necrosis, both orlistat and cetilistat significantly reduced the necrotic area at day 1 (1.1% ± 0.3% versus 8.3% ± 6.5%, both with P < 0.001 versus GTL) and at day 5 (2.4% ± 0.3% versus 5.0% ± 1.6%, both with P < 0.001 versus GTL). These findings were no different from rats that only had ligation of the biliopancreatic duct. These findings are consistent with previous studies that report that UFAs cause acinar necrosis,^29,39 orlistat and cetilistat prevent GTL-induced cell death in vitro (Supplementary Figure S1, C and D), and these agents are efficacious in reducing serum lipase and UFA generation (Figure 3).

Figure 4.

Inhibition of lipolysis is associated with a reduction in necrotizing pancreatitis. Gross appearance of the entire pancreas from eight different animals, including the duodenum in the upper part of the image and the spleen in the lower part, showing hemorrhagic pancreatic necrosis (black arrows) in the GTL group (A), which are markedly reduced in the GTLO (B) and GTLC (C) groups. There was only one animal in the GTLC group that had hemorrhagic necrosis restricted to the head of the pancreas; this was the only death in this group. The gross images were collected on day 5 in the GTLO and GTLC groups, and bile staining is visible in some of the pancreata consistent with persistent obstruction from the ligature. Representative images of pancreatic histologic sections obtained on day 1 stained with H&E and photographed with a 2× objective from a control rat (D) and rats treated with GTL (E), GTLO (F), and GTLC (G). Note the sheets of necrotic cells with a few surviving islands (yellow outline) in E, which are reduced in F and G, despite showing edema, inflammatory cells, and a few hemorrhages (yellow arrows) consistent with mild AP. Comparisons between the groups are depicted with box plots quantifying the amount of pancreatic necrosis as mean (dashed line), median (solid line), the 25th and 75th percentiles (upper and lower parts of the bars), the 10th and 90th percentile (whiskers), and the outliers (dots) and showing that GTL-induced pancreatic necrosis is prevented by GTLO and GTLC on day 1 and day 5 (H). ^∗P < 0.05 versus control; ^†P < 0.05 versus the GTL group. Original magnification: ×4 (D–G).

Lipase Inhibition Reduces the Increase in Serum Cytokines Associated with GTL-Induced Pancreatitis

Because we have previously²⁹ found that UFAs up-regulate proinflammatory mediators such as CXCL1/IL-8/keratinocyte chemoattractant (KC) and tumor necrosis factor-α in vitro and that serum levels of IL-6, monocyte chemotactic protein-1, tumor necrosis factor-α, and resistin follow the course of lipotoxicity in a mechanistically distinct model, we examined whether UFA reduction with lipase inhibition is also associated with a reduction in inflammatory mediator levels. Serum levels of IL-1β,⁵⁶ IL-6⁵⁷⁻⁶⁰, IL-8,^57,60,61 and IL-18,^61–63 all of which have been previously reported to be increased in human SAP, were measured on both day 1 and day 5. GTL induced a significant increase in cases (Figure 5). Lipase inhibition resulted in a significant reduction of these at day 1 and day 5, with the exception of cetilistat not reducing IL-1β on day 1 and IL-18 on day 5 (Figure 5, A and D). Notably, the only animal that had died in the cetilistat group had elevated levels of serum IL-18 (see below). In addition, the IL-1β and IL-18 levels in these instances were not significantly different from controls (Figure 5, A and D). Thus, lipase inhibition in vivo is associated with a reduction in the inflammatory response that is associated with the hydrolysis of the unsaturated triglyceride GTL.

Figure 5.

Lipase inhibition results in a reduction of serum inflammatory cytokine levels. Serum levels of IL-1β (A), KC/GRO (B), IL-6 (C), and IL-18 (D) as measured in controls (CON) and animals with an intraductal infusion of GTL at the time when they were moribund (GTL) and those with an infusion of GTLO and GTLC on day 1 and day 5. Comparisons between the groups are depicted with box plots showing mean (dashed line), median (solid line), the 25th and 75th percentiles (upper and lower parts of the bars), the 10th and 90th percentile (whiskers), and the outliers (dots). Note the significant reduction in most of these in lipase inhibitor–treated groups compared with the GTL group. ^∗P < 0.05 versus control; ^†P < 0.05 versus the GTL group.

Lipase Inhibition Prevents Multisystem Organ Failure and Improves Mortality Associated with GTL-Induced Pancreatitis

Rats receiving GTL had 100% mortality within a day (Figure 6A). In rats administered the lipase inhibitors along with GTL, 0 of 10 died in the GTLO group and 1 of 10 died in the GTLC group when followed up for 5 days (Figure 6A). This finding was associated with GTL-induced lung injury, as evidenced by an increase in TUNEL-positive apoptotic cells in the lung (Figure 6B) as occurs in acute respiratory distress syndrome associated with the infusion of the UFA oleic acid.⁶⁴ Compared to controls (Figure 6C), GTL infusion resulted in large number of apoptotic cells in the lungs (Figure 6D). There was a significant reduction of apoptotic cells in the GTLO and GTLC groups (Figure 6, E and F), in both animals electively sacrificed on day 1 and those at the end of the 5-day study. We also measured serum BUN level and renal injury because these factors were a part of the spectrum of multisystem organ failure. GTL resulted in a significant increase in BUN (88 ± 22 mg/dL versus 16 ± 1 mg/dL in controls, P < 0.001, Figure 7A), which was significantly reduced in the GTLO group at day 1 and day 5 (25 ± 5 mg/dL versus 19 ± 1 mg/dL, P < 0.006 versus GTL for both) and was reduced in the GTLC group only at day 5 (51 ± 13 mg/dL at day 1, P = 0.40 versus GTL, and 18 ± 6.mg/dL at day 5, P < 0.001 versus GTL, Figure 7A). The levels in the GTLO or GTLC groups were no different from rats that only had ligation of the biliopancreatic duct. Compared to controls (Figure 7B), GTL resulted in renal tubular injury seen on TUNEL staining (Figure 7C), as noted previously in the IL-12/18 model of SAP in obese mice²⁹ and with the infusion of the UFA oleic acid,⁶⁵ with marked improvement in the GTLO and GTLC groups (Figure 7, D and E). Therefore, inhibition of unsaturated triglyceride lipolysis using two distinct lipase inhibitors during SAP is associated with a reduction in UFA generation and a reduction of lung and renal injury.

Figure 6.

Inhibition of lipolysis improves survival and reduces apoptotic cells in the lungs. Kaplan-Meyer 5-day survival curve of rats infused with GTL (red), GTLO (green), and GTLC (blue) (A). Note that there is 100% mortality in the GTL group within 24 hours, which is significantly reduced in the GTLO and GTLC groups at 5 days (P < 0.001). Comparisons between the groups are depicted with box plots showing mean (dashed line), median (solid line), the 25th and 75th percentiles (upper and lower parts of the bars), the 10th and 90th percentile (whiskers), and the outliers (dots) (B) and representative images of TUNEL-positive cells in the lungs in controls (C) and animals treated with GTL (D), GTLO (E), and GTLC (F). Black arrows point to the numerous apoptotic cells in the GTL group. ^∗P < 0.05 versus control; ^†P < 0.05 versus the GTL group. Original magnification: ×40 (C–F).

Figure 7.

Inhibition of lipolysis prevents sustained renal failure and renal tubular injury induced by biliary SAP. Box plots of serum BUN levels in the various groups showing mean (dashed line), median (solid line), the 25th and 75th percentiles (upper and lower parts of the bars), the 10th and 90th percentile (whiskers), and the outliers (dots) (A). Note that the BUN level is significantly increased in the GTL group at the time of mortality. Although the animals in the GTLC group developed transient renal failure at day 1, this normalized by day 5. TUNEL staining of controls (B) and rats treated with GTL (C), GTLO (D), and GTLC (E) produces strong evidence of tubular damage (black outline) in the GTL group, which is prevented in the GTLO and GTLC groups. ^∗P < 0.05 versus control; ^†P < 0.05 versus the GTL group. Original magnification: ×40 (B–E).

Discussion

Using conditions that simulated the most common cause (ie, biliary AP) and risk factor of SAP (ie, obesity-associated increase in intrapancreatic fat and high UFA amounts in our patients) noted in our patients, we found that preventing acute generation of UFAs, which are formed by the hydrolysis of intrapancreatic triglyceride (normally stored within adipocytes) by using two distinct lipase inhibitors, results in a reduction of local necrosis, inflammatory mediator levels, prevention of multisystem organ failure, and improved survival. This, however, occurs without a reduction in markers of biliary AP (ie, an increase in serum ALT, bilirubin, or amylase within 24 hours of onset of the disease), signifying that lipase inhibition does not interfere with the initiation of biliary AP.

We did not inject bile acids or salts into the pancreatic ducts to cause biliary AP and instead used a model of biliopancreatic duct ligation alone or with injection of GTL, GTLO, or GTLC. There are several reasons for this preference: i) the fact that commonly used concentrations (3% to 5%)⁶⁶ of agents such as sodium taurocholate to induce severe biliary AP result in local concentrations ranging from 60 to 100 mmol/L, which are 5- to 100-fold above the critical micellar concentration,⁶⁷ resulting in a detergent effect on cell membranes that they contact, and we are not aware of any human studies to verify the appropriateness of these concentrations; ii) FA toxicity at concentrations much below those noted in the PPNC occurs along with the release of calcium from thapsigargin sensitive intracellular stores²⁹ unlike detergent-induced damage of cell membranes, resulting in extracellular calcium influx; iii) recent expert reviews have questioned the relevance of bile injection to the pathogenesis of biliary AP¹² and argue in favor of the consequences of duct ligation alone; iv) duct ligation alone resulted in criteria fulfilling mild biliary AP⁵²⁻⁵⁵ (see Materials and Methods for details); v) the injection of GTL in amounts equivalent to a 5% to 10% intrapancreatic fat area, which is in the range of intrapancreatic fat (23.4% ± 4.3%) and fat necrosis (12.6% ± 3.4%) noted in obese humans with SAP²⁹, resulted in SAP in our model; and vi) the intraductal route allows for hydrolysis of the triglyceride by pancreatic lipases as occurs in fat necrosis in human AP^29,39,45,46 (Figure 1, C and D) and from the basolateral leakage^{45,46,48,49,68} of lipases during AP causing hydrolysis of adipocyte triglyceride. Therefore, this model is highly relevant to human disease because it fulfills criteria of human biliary AP and the pathophysiologic features resulting from basolateral leakage, mimics increased intrapancreatic fat that occurs in obesity,^29,39,69 and is associated with worse outcomes in patients with AP.^6−8,29,70

Both cetilistat⁷¹ and orlistat are highly lipophilic, which facilitated their dissolution in the GTL without the need of a vehicle. Although cetilistat is a benzoxazinone⁷¹ lipase inhibitor, orlistat reacts with the nucleophilic serine of lipases via its β-lactone moiety, resulting in a covalent complex.^72,73 Both agents improved outcomes in the SAP model studied, along with inhibiting lipase activity and preventing the increase in serum UFAs, supporting the role of lipolytic generation of UFAs in the adverse outcomes of SAP. The advantage of pharmacologic inhibition over currently available genetic knockouts of lipases is the ability of the inhibitors to reduce the activity of all lipases expressed in the pancreas⁷⁴ (Singh Lab, unpublished data). This is beneficial because lipases seem to have redundant roles as evidenced by mice with a deletion of a single pancreatic lipase gene achieving normal adult weight,⁷⁵ whereas dual lipase knockouts⁷⁶ experience embryonic and neonatal mortality, which would also prevent studies in adult mice.

We have previously found UFAs to induce cell death and to up-regulate inflammatory mediators, including CXCL1, CXCL2, and tumor necrosis factor-α.²⁹ This dual role has relevance to the end points noted in this study based on the fact that both lipase inhibitors prevent large areas of pancreatic necrosis and massive increase of cytokines induced by GTL. In addition, GTL-induced lung injury and renal failure are also prevented by lipase inhibition. Although the increase in necrosis is likely due to these formed UFAs inhibiting mitochondrial complexes I and V and consequently reducing ATP levels,²⁹ it remains unclear whether it is the UFAs themselves or the inflammatory cytokines generated that result in worse systemic outcomes and mortality. Interestingly, the published literature supports the finding that lipotoxicity results in these conditions, as is noted in oleic acid–induced acute lung⁶⁴ and renal injury, during which the levels of IL-1β, IL-6,⁷⁷ and tumor necrosis factor-α are also increased. The role these cytokines may play remains to be explored, with recent studies reporting them to play a protective role as evidenced by IL-6⁷⁸ improving lung inflammation and IL-8 improving survival in hemorrhagic shock.⁷⁹ In addition, studies looking at patients with a high BMI having SAP have found this to be independent of the levels of cytokines, such as IL-1β and IL-6.¹

In contrast to this study, where we found the benefits of targeting a modifier of AP outcomes (ie, the lipolysis of intrapancreatic fat), the relevance of a target to improving outcomes in AP has classically been based on revealing that the target plays a role in multiple models initiated by different insults. Examples include trials studying the efficacy of serine protease and trypsin inhibitors, such as gabexate mesylate,^22,80 aprotinin,^26,80 and nafamostat,^24,81 in improving AP outcome, based on that rationale that intrapancreatic active trypsin is noted during mechanistically dissimilar AP models, such as caerulein, choline deficient ethionine supplemented, and taurocholate-induced AP.^11,16,17 However, there is no clear evidence that trypsin inhibition improved AP outcomes in clinical trials.^19,21 Further proof that targeting an initiator of AP, such as trypsin, may not affect outcomes, such as necrosis or mortality, comes from the observation that patients with hereditary pancreatitis due to trypsinogen gene mutations, such as PRSS1 (resulting in increased intracellular amounts of the active enzyme), do not experience SAP.⁸² Last but not least, the severity of an AP attack is typically unrelated to its cause, but obesity has been repeatedly reported to be associated with adverse outcomes.^6–8,29,70

The clinical part of this study may be limited by its small size, the fact that the predominant cause was biliary AP, the lack of data on the reason for intervention on the PPNCs, the identity of which collections were infected, and the changes that may have taken place during the period that the collection formed (because the earliest time to collection of the samples from onset of AP was 4 weeks). Although the small size could result in a type 2 error and us falsely estimating the absolute amount of UFAs to be higher than the amount of SFAs (2.7 ± 1.0 mmol/L versus 1.0 ± 0.3 mmol/L, P = 0.033, Figure 1A), the facts that UFAs formed a significantly higher percentage (68.5% ± 3.0% versus 31.7% ± 2.9% for SFAs, P < 0.001) of the overall FA content (3.8 ± 1.3 mmol/L) in the PPNCs and that the UFAs necrose acinar cells at 300 μmol/L in contrast to the SFAs (which do not even at 1.2 mmol/L)²⁹ minimize the significance of this error. We cannot explain our patient population being enriched in biliary AP or justify the lack of data regarding which of the collections were infected. However, infections would not change our conclusions because we do not dispute the clinical tenet that infected collections must be treated. Moreover, UFA formation would occur primarily from the hydrolysis of visceral adipocyte triglyceride by lipases, which are produced by the exocrine pancreas and not infections. Although potential mediators and their concentrations may have changed over the time the collections accumulated, a period of >4 weeks from the onset of AP to the time of intervention is consistent with the current guidelines for such interventions.^10,34,83

In summary, using a model of biliary AP relevant to our patients, we found that pharmacologic inhibition of lipolysis using two distinct lipase inhibitors reduces the generation of UFAs, which are normally enriched in pancreatic necrosis. This inhibition results in ameliorating pancreatic necrosis, the increase in serum cytokines, and lung and renal injury without interfering with the increase in markers of biliary AP. Therefore, targeting acute lipotoxicity improves outcomes in severe biliary acute pancreatitis without affecting its initiation.

Supplemental Data

Supplemental Figure 1

Effect of 50 μmol/L cetilistat and orlistat on glycerol generated by the hydrolysis of 300 μmol/L GTL added to the medium of pancreatic acini during 4 hours (A). A total of 212 ± 42 μmol/L glycerol was generated over this time from 300 μmol/L GTL (depicted as maximal or 100%), which was reduced significantly by both the agents (^∗P < 0.05 versus 300 μmol/L GTL). Effect of 50 mmol/L orlistat or cetilistat on LDH leakage during 4 hours induced by 300 mmol/L LA added to the pancreatic acinar medium (B). Effect of 50 μmol/L orlistat (C) and cetilistat (D) on LDH leakage during 4 hours induced by different concentrations of GTL added to the pancreatic acinar medium (^∗P < 0.05 versus corresponding GTL concentration). ^∗P < 0.05 versus control. Neither orlistat nor cetilistat affected LA-induced LDH leakage.