• Open Access

Serum Pepsinogen-A, Canine Pancreatic Lipase Immunoreactivity, and C-Reactive Protein as Prognostic Markers in Dogs with Gastric Dilatation-Volvulus

Authors


  • This study was performed at the Hebrew University Veterinary Teaching Hospital. Canine pepsinogen-A, C-reactive protein, and canine pancreatic-like immunoreactivity were measured at the Gastrointestinal Laboratory, College of Veterinary Medicine and Biomedical Sciences, Texas A&M University.
  • The study was unsupported.
  • Results of this study were presented in part at the Annual Israeli Veterinary Symposium, September 2011, Beit Dagan, Israel.

Corresponding author: I. Israeli, Koret School of Veterinary Medicine, Hebrew University of Jerusalem, PO Box 12, Rehovot, Israel 76100; e-mail: Dr_Inbar@yahoo.com.

Abstract

Background

Pepsinogens are proenzymes secreted by gastric chief cells. In humans, their serum concentrations reflect gastric mucosal morphological and functional status.

Objectives

To evaluate serum canine pepsinogen-A (cPG-A), C-reactive protein (CRP), and canine pancreatic lipase immunoreactivity (cPLI) concentrations in dogs with gastric dilatation-volvulus (GDV).

Animals

Sixty-six dogs presented with GDV and 79 healthy controls.

Methods

Blood was collected prospectively, and records retrospectively reviewed.

Results

Median cPG-A concentration was higher in GDV dogs (median, 397 μg/L; range, 37–5,410) compared to controls (median, cPG-A 304 μg/L; range, 18–848; P = .07). Mortality rate in GDV dogs was 22.7%. In nonsurvivors of GDV, median cPG-A was higher compared to survivors (median, 746 μg/L; range, 128–5,409 versus median, 346; range, 36–1,575, respectively; P = .003). The proportion of dogs with increased cPG-A increased with gastric wall damage score (P = .007). An ROC analysis of cPG-A as a predictor of death showed an area under the curve (AUC) of 0.75, higher than lactate (AUC 0.66), and corresponded to a sensitivity and specificity of 53% and 88%, respectively. CRP was increased in 48 dogs (75%), cPLI was >200 μg/L in 26 dogs (39.4%) and >400 μg/L in 12 dogs (18.2%) but both analytes had no association with outcome.

Conclusions

Presurgical cPG-A concentration was positively and significantly associated with gastric wall lesion severity, but, based on ROC analysis, it was only a moderate outcome predictor. CRP and cPLI were commonly increased in dogs with GDV.

Abbreviations
AUC

area under the curve

CI

confidence interval

cPG-A

canine pepsinogen A

cPLI

canine pancreatic lipase immunoreactivity

CRP

C-reactive protein

FFP

fresh frozen plasma

GDV

gastric dilatation-volvulus

GWDS

gastric wall damage score

OR

odds ratio

pRBC

packed red blood cells

RI

reference interval

ROC

receiver operating characteristics

Gastric dilatation-volvulus (GDV) in dogs is a medical and surgical emergency, commonly involving middle-aged to older, large and giant, deep-chested dogs, with an even sex distribution.[1, 2] It involves gastric dilatation and pyloric clockwise rotation between 90° and 270°. The distended stomach compresses the portal vein and vena cava, thereby decreasing venous return and cardiac output. Consequently, tissue perfusion is decreased, leading to abnormalities in several abdominal organs.[1-3] Increased intragastric pressure leads to collapse of gastric wall capillaries, which along with circulatory failure and rupture of the short gastric vessels, may result in ischemia and gastric wall necrosis.[1-3]

Early studies of canine GDV have reported mortality rates as high as 33–68%.[4-6] However, in more recent studies, mortality rates ranged from 15 to 33%.[6-10] Regardless, in light of this high mortality, research efforts have been made to identify prognostic markers. Although controversy still exists, there are several pre- and postoperative factors that have been shown to be significantly associated with poor outcome. These include prolonged time lags from the last meal to presentation and from the onset of clinical signs to presentation,[4, 7] older age,[4] rectal temperature <38°C,[10] recumbency, and mental status alterations (eg, lethargy, coma) at presentation,[10, 11] increased blood lactate concentration,[2, 12] an absolute change in lactate concentration ≤4 mmol/L after initial treatment,[9] increased serum cardiac troponin concentration,[13] occurrence of hemostatic derangement,[14] disseminated intravascular coagulation (DIC),[7] cardiac arrhythmias,[6, 8] hypotension, peritonitis, sepsis, and acute kidney injury (AKI).[10] Several studies of GDV have associated the need to perform gastrectomy, splenectomy, or their combination with a poor outcome,[4, 6-8] whereas others have failed to document such association.[6, 7] Nonetheless, gastric wall necrosis at surgery has been the most consistent predictor of postoperative complications and death in dogs with GDV, and its presence is associated with an 11-fold higher risk of mortality.[11, 12]

Some of the above factors are nonspecific. Most are either postsurgical markers or represent late complications of GDV (eg, sepsis, peritonitis, DIC, AKI), and thus cannot be used as early prognostic indicators. Plasma gastrin immunoreactivity has been investigated as a potential presurgical prognostic biomarker in canine GDV, but its concentration was not associated with outcome.[15] Therefore, there is still a need for early prognostic indicators in canine GDV.

Lactate concentration at the presentation previously has been described as a useful preoperative prognostic marker in canine GDV.[9, 12, 16] However, it is likely associated only indirectly with the presence and extent of gastric wall necrosis, and is mostly influenced by the severity of systemic hypoperfusion, acidosis, and shock.[12] Therefore, hyperlactatemia may occur in the absence of gastric wall necrosis, and is thus prone to false positive prediction of mortality. In light of these limitations, identifying early, specific, and sensitive biomarkers of gastric wall injury and necrosis will be useful for prognostication of dogs with GDV.

Pepsinogens are inactive proenzymes (zymogens) secreted mainly by gastric chief cells, mostly directly into the gastric lumen. However, approximately 1% of pepsinogens diffuse back into the systemic circulation.[17-20] Serum pepsinogen concentration is considered to reflect the morphological and functional status of the gastric mucosa in human patients, and has long been used as a marker of Helicobacter pylori infections, and for early detection of gastritis and gastric cancer in humans.[18-22] Two major groups of canine pepsinogens (cPG) previously have been purified and partially characterized,[17] and an ELISA for serum cPG-A previously has been analytically validated.[23] However, this marker has yet to be evaluated in dogs with GDV.

The aims of the present prospective study were to measure serum cPG-A concentration in dogs with GDV and to compare them to those measured in healthy controls, and to seek an association between serum cPG-A concentration and surgical findings, morbidity and mortality. Our main hypothesis was that serum cPG-A concentration at presentation would be positively associated with the extent of gastric wall necrosis, as well as with higher morbidity and mortality rates. We also hypothesized that serum C-reactive protein (CRP) and canine pancreatic lipase immunoreactivity (cPLI) concentrations will be associated with disease severity and serve as potential prognostic indicators.

Materials and Methods

Selection of Animals and Collection of Data

Selected dogs were consecutively presented to the Emergency and Critical Care Department at the Hebrew University Veterinary Teaching Hospital (HUVTH), diagnosed with GDV and underwent exploratory celiotomy. GDV was diagnosed based on history, clinical signs, and abdominal radiography, and was surgically confirmed.

Data retrieved from the medical records included signalment, body weight, medical history, the time lag from onset of clinical signs to presentation, clinical signs at presentation, the anesthetic regimen used, and details of blood products administered during hospitalization. Surgical data collected included gastric and splenic anatomic position, assessment of the vitality of the gastric wall, performance of gastrotomy, gastrectomy, invagination and splenectomy, type of gastropexy, occurrence of hemoabdomen, peritonitis, other surgical complications, and intraoperative cardiac arrhythmias.

A gastric wall damage score (GWDS) was assigned to each dog based on the surgical findings, using a scale of 0–2 (0, normal or mildly affected, manifested by normal serosa to mild erythema and hyperemia; 1, moderately affected, manifested by areas of large bruises or purple discoloration, that regained vitality after anatomic repositioning of the stomach; 2, severe damage manifested by dark purple, brown, black or white areas on the serosa, loss of layering or perforation).

Postsurgical data included length of hospitalization, occurrence of postoperative cardiac arrhythmias, hemostatic derangement and other complications (eg, AKI, peritonitis, pancreatitis, sepsis), as well as the final outcome. Hemostatic derangement was defined as concurrent presence of at least 3 of the following: increased (≥25% of the upper reference interval) prothrombin time (PT) and activated thromboplastin time (aPTT), hypofibrinogenemia, hypoantithrombinemia or thrombocytopenia. Nonsurvivors included dogs that died naturally or were euthanized, at the owners' request, during surgery because of a grave prognosis based on the surgeon's assessment of gastric wall condition or later, during hospitalization, because of the deterioration of their clinical condition.

Healthy dogs were used for the establishment of the reference interval (RI) for serum cPG-A, and served as negative controls. These included dogs presented to the HUVTH for elective procedures (eg, routine orthopedic examination, dental prophylaxis, neutering), training guide dogs, and military dogs. These dogs were determined to be healthy based on a normal medical history and physical examination. All were fasted for at least 8 hours before blood collection.

Laboratory Tests

Blood samples were collected from GDV dogs at presentation. Whole blood in EDTA was used for a CBC, which was performed within 30 minutes of collection.1 , 2 Whole blood samples for coagulation tests (PT, aPTT, fibrinogen concentration, and antithrombin activity) were collected in 3.2% citrate, centrifuged within 30 minutes of collection, and the separated plasma analyzed immediately.3 , 4 Blood for serum chemistry analysis was allowed to clot, the sera were separated by centrifugation within 1 hour of collection, and either analyzed immediately or refrigerated at 4°C pending analysis, performed within 24 hours of collection.5 , 6 The above-mentioned routine laboratory tests were performed in most, but not all, dogs at presentation depending on the owners' financial constraints and the attending clinicians' judgment. Sera from dogs with GDV and healthy controls were stored at −80°C (for up to 6 years and 9 months, respectively) pending analyses of cPG-A, CRP, and cPLI. Serum cPG-A was measured in all samples using a previously described, in-house ELISA.[23] The RI for cPG-A was defined as the central 95th percent of the range of the healthy control dogs. Serum CRP and cPLI were measured using commercially available ELISA.7 , 8 Blood samples for lactate concentration were collected either in sodium fluoride tubes or as fresh whole blood. The former were centrifuged and the plasma analyzed immediately,5 , 6 whereas the latter were analyzed immediately after collection using a dry chemistry lactatometer that previously has been evaluated in dogs.9 , [24]

Preoperative Care, Anesthetic Protocol, and Postoperative Monitoring

Preoperative treatment was similar in all dogs, and included IV lactated Ringer's solution, lidocaine10 (0.05 mg/kg/min at constant rate infusion IV), and cefazolin11 (25–30 mg/kg IV). Treatment was adjusted individually in each dog based on laboratory test results and the attending clinician's assessment. Dogs were premedicated with pethidine-HCl12 (2–4 mg/kg, IM). Anesthesia was induced with propofol13 (0.5–1.0 mg/kg, IV) and diazepam14 (0.5 mg/kg, IV). Dogs then were intubated, and general anesthesia maintained using isoflurane15 in 100% oxygen (2 L/min). Gastric decompression was accomplished using an oral gastric tube, occasionally following trocarization, using an 18 gauge needle. Postoperatively, dogs recovered in the intensive care unit and were constantly monitored using a continuous ECG. Any occurrences of cardiac arrhythmia or other complications were recorded.

Statistical Analysis

The normality of distribution of continuous parameters was assessed using the Shapiro-Wilk test. Normally and non-normally distributed continuous parameters were compared between groups using Student's t- or Mann-Whitney U-tests, respectively. Fisher's exact or Pearson's χ2-tests of homogeneity were used to compare categorical variables between groups. The correlations between continuous parameters were assessed using Pearson's or Spearman's correlation tests, based on the data distribution.

The association of continuous variables with the outcome was assessed by the receiver operating characteristics (ROC) procedure and its area under the curve (AUC). Cutoff points, with their corresponding sensitivity and specificity for prediction of outcome were selected. An optimal cutoff point was defined as the point associated with the fewest misclassifications. The following definition was used to assess the accuracy of the ROC curve AUC: low (0.5 <AUC <0.7), moderate (0.7 ≤AUC <0.9), and high (0.9 ≤AUC ≤1).[25] Because this study tested several independent hypotheses, the Bonferroni-Holm method for adjustment of alpha was applied to all significant P values, preserving the nominal .05 level of significance, based on the number of comparisons made within each one of the following independent hypothesis groups: (1) GDV group versus the control group (n = 3 comparisons), (2) cPG-A-related hypotheses (n = 7), (3) cPLI-related hypotheses (n = 3), (4) CRP-related hypothesis (n = 3), (5) lactate-related hypotheses (n = 8), and (6) outcome-related hypotheses (n = 10). P-values ≤.05 that remained significant upon adjustment for multiple comparisons are noted as P; those that did not remain significant upon adjustment are noted as P*. A P ≤ .05 was considered statistically significant for all tests applied. All calculations were performed using a statistical software package.16

Results

Signalment and Body Weight

The healthy control dogs included 79 dogs of various breeds, with a median age of 2.5 years (range, from 3 months to 10 years) and a median body weight of 21.7 kg (range, 3.5–37.0 kg), of which 50 were females (7 spayed) and 27 were males (10 castrated). In 2 dogs, sex was not recorded. Sixty-six dogs, of the following breeds, fulfilled the GDV inclusion criteria and were included in the study group: German Shepherd Dog (14 dogs), mixed breed (7), Great Dane and Weimaraner (6 each), Belgian Shepherd (5), Great Pyrenees (4), Saint Bernard and Boxer (3 each), Basset Hound, Samoyed, American Shepherd, Chow-Chow, Newfoundland, and Labrador Retriever (2 each), Golden Retriever, Dog de Bordeaux, Bull Mastiff, Cane Corso, Bernese Mountain Dog, and Akita (1 each). There were 41 males (11 castrated) and 25 females (14 spayed). The median age and body weight were 8.6 years (range, 1–16 years) and 38.7 kg (range, 20–75 kg), respectively. Controls were significantly younger and had a significantly lower median body weight compared to the GDV group (P < .001). There were no group differences in sex distribution.

Outcome, Surgical Results, Complications and Hospitalization

Fifty-one of 66 dogs (77.3%) with GDV survived to discharge, and 15 (22.7%) did not survive, of which 8 dogs died naturally and 7 were euthanized. Median age of survivors (8 years; range, 1–13 years) was lower than that of nonsurvivors (10 years; range, 6–16 years; P* = .01). There were no significant differences in body weight between survivors and nonsurvivors.

Forty-five dogs (68.2%) had a GWDS of 0, whereas 6 (9.1%) had a score of 1, and 15 (22.7%) had a score of 2. Partial gastrectomy was performed in 7 dogs (10.6%), 8 dogs (12.1%) had an invagination performed, 5 dogs (7.6%) underwent gastrotomy, 3 dogs (4.5%) had a complete splenectomy, and 3 dogs (4.5%) underwent partial splenectomy. The mortality rate of dogs that underwent partial gastrectomy or invagination was significantly higher compared to those that did not undergo these procedures (7/15, 46.7% versus 6/49, 12.2%, respectively; odds ratio [OR], 6.29; 95% confidence interval [CI95%], 1.66–23.81; P = .004). Eight of the 15 nonsurvivors (53.3%) had a gastric damage score of 2, whereas 3/15 (20%) had a score of 1, and 4/15 (26.7%) had a score of 0, demonstrating significantly increased mortality with an increase in GWDS (P < .001).

Peritonitis caused by gastric perforation was observed during surgery in 3 dogs (4.5%). Postoperative complications, as assessed by the attending clinicians, included cardiac arrhythmias (20 dogs, 30.3%), AKI, splenic thrombosis, hypoalbuminemia with secondary ascites (2 dogs each, 3%), pancreatitis, and pneumothorax (1 dog each, 1.5%). The presence of cardiac arrhythmias during hospitalization was not associated with survival. Hemostatic derangement was present in 14/66 dogs (21.2%) and was positively associated with cardiac arrhythmias compared to dogs without hemostatic derangement (8/14, 57.1% versus 11/47, 23.4%, respectively; OR, 4.36; CI95%, 1.24–15.32). Similarly, hemostatic derangement was positively associated with mortality rate compared to dogs without hemostatic derangement (7/14, 50% versus 7/47 dogs, 14.9%, respectively; OR, 5.71; CI95%, 1.53–21.77; P = .01).

Laboratory Results

The control group median cPG-A was 304 μg/L (range, 18–848 μg/L) and the RI was set at 105–802 μg/L. The median cPG-A (measured in 65/66 dogs) of the GDV group was 397 μg/L (range, 37–5,410 μg/L), which was not significantly different from controls (P = .07). Serum cPG-A concentration was significantly higher in nonsurvivors (median, 746 μg/L; range, 128–5,410 μg/L) compared with survivors (median, 346 μg/L; range, 36–1,575 μg/L; P = .003; Fig 1). Increased serum cPG-A concentration was recorded in 13 dogs (19.7%) of which 5/45 dogs (11.1%) had a GWDS of 0, 2/6 (33%) had a score of 1, and 6/14 (42.9%) had a score of 2 (Fig 2). The proportion of dogs with abnormally high cPG-A concentration increased significantly with increased GWDS (P = .01). The proportion of dogs that underwent partial gastrectomy or invagination was higher among dogs with increased serum cPG-A concentration compared with dogs with normal cPG-A concentration (6/13, 46.2% versus 8/50, 16.0%; OR, 4.5; CI95%, 1.19–16.96; P* = .02). An ROC analysis of cPG-A as a predictor of death had an AUC of 0.75 (CI95%, 0.60–0.91; Fig 3). All dogs with cPG-A >1,575 μg/L died, yielding 100% specificity with 27% sensitivity. A cPG-A cutoff of 1,090 μg/L corresponded to sensitivity and specificity of 40% and 96%, respectively, whereas the optimal cutoff point (740 μg/L), corresponded to a sensitivity and specificity of 53% and 88%, respectively.

Figure 1.

Distribution of serum canine pepsinogen-A and lactate concentrations at presentation in 66 dogs with gastric dilatation-volvulus between survivors and nonsurvivors. Data are presented as boxes and whiskers. Each box includes the interquartile range, whereas the line within a box represents the median, and the whiskers represent the range, extending to a maximum of 1.5 times the interquartile range.

Figure 2.

Distribution of serum canine pepsinogen-A and blood lactate concentrations at presentation in 66 dogs with gastric dilatation-volvulus, grouped by gastric wall damage score. Data are presented as boxes and whiskers. Each box includes the interquartile range, whereas the line within a box represents the median, and the whiskers represent the range, extending to a maximum of 1.5 times the interquartile range.

Figure 3.

Receiver operator characteristics curve with its area under the curve of serum canine pepsinogen-A concentration at presentation as a predictor of mortality in 51 survivor dogs and 15 nonsurvivor dogs with gastric dilatation-volvulus.

Hemostatic derangement was more common among dogs with an increased cPG-A concentration compared to those with a cPG-A concentration within RI (6/13, 46.2% versus 8/47, 18.7%, respectively; OR, 4.18; CI95%, 1.14–15.80; P* = .03). The proportion of dogs with increased cPG-A treated with fresh frozen plasma (FFP, 6/13, 46.2%), packed red blood cells (pRBC, 8/13, 61.5%), or both (6/13, 46.2%) was significantly higher compared to dogs with a cPG-A concentration within RI that were treated with FFP (6/47, 12.8%; OR, 5.86; CI95%, 1.46–23.44; P = .01), pRBC (8/47, 17.0; OR, 7.8; CI95%, 2.02–30.14; P = .003) or both (4/47, 8.5%; OR, 9.21; CI95%, 2.06–41.14; P = .004). There was no difference in the proportion of cardiac arrhythmias between dogs with increased and normal cPG-A.

The median serum cPLI concentration, measured in 63/66 dogs with GDV, was 101 μg/L (range, 29–3,380 μg/L; RI, 29–200 μg/L). cPLI >200 μg/L was recorded in 26 dogs (39.4%), and 12 dogs (18.2%) had a serum cPLI concentration >400 μg/L, the currently suggested cutoff value for a diagnosis of pancreatitis. There were no significant cPLI concentration differences between survivors and nonsurvivors, between dogs that underwent partial gastrectomy or invagination and those that did not, between those with or without cardiac arrhythmias, or between dogs with or without hemostatic derangement.

The median CRP concentration, measured in 64/66 dogs, was 23 mg/L (range, 0.1–111.0 mg/L; RI, 0.0–7.6 mg/L) and was above RI in 48 dogs (75%). There were no significant CRP concentration differences between survivors and nonsurvivors, between dogs undergoing invagination or partial gastrectomy and those that did not, dogs with and without cardiac arrhythmias, or dogs with or without hemostatic derangement.

The median lactate concentration, measured in 65/66 dogs, was 45 mg/dL (range, 1–227 mg/dL; RI, 10.5–37.6 mg/dL). Thirty-eight dogs (58.5%) had hyperlactatemia. Lactate concentration was not significantly higher in nonsurvivors (median, 67.0 mg/dL; range, 10.0–227.5 mg/dL) compared with survivors (median, 43.4 mg/dL; range, 0.9–179.3 mg/dL; P = .06; Fig 1). There was no difference in the proportion of survivors among dogs with hyperlactatemia (29/38, 76.3%) and those with normolactatemia (22/27, 81.5%). Lactate concentration was >54 mg/dL (a cutoff previously reported to be associated with poor prognosis in canine GDV[12]) in 26 dogs (40%). The proportion of survivors was not significantly higher among dogs with lactate ≤54 mg/dL (34/39, 87.2%) compared to those in which lactate was >54 mg/dL (17/26, 65.4%; P = .06). Hyperlactatemia was not associated with partial gastrectomy or invagination, using both the 37.6 mg/dL and the 54 mg/dL cutoff values. Nevertheless, hyperlactatemia (lactate concentration >54 mg/dL) was present in 13/45 dogs (28.9%) having a score of 0, 4/6 (66.7%) having a score of 1, and 9/14 (64.3%) having a score of 2 and was, thus, positively associated with the GWDS (P = .02; Fig 2). An ROC analysis of lactate concentration at presentation as a predictor of death yielded an AUC of 0.66 (CI95%,0.49–0.85). The optimal lactate concentration cutoff was 65 mg/dL, corresponding to sensitivity and specificity of 57% and 80%, respectively. Hemostatic derangement also was significantly more common among dogs with a lactate concentration >54 mg/dL compared to those with a lactate concentration ≤54 mg/dL (10/24, 41.7% versus 3/36, 8.3%, respectively; OR, 7.86; CI95%, 1.87–32.95; P = .003). The proportion of dogs with lactate >54 mg/dL treated with pRBC (10/24, 41.7%) was higher compared to dogs with a lactate concentration ≤54 mg/dL (6/26, 23.1%; OR, 3.57; CI95%, 1.08–11.97; P* = .03). There was no significant difference in the use of FFP or concurrent use of both blood products between dogs with lactate concentration >54 mg/dL and those with lactate concentration ≤54 mg/dL. There was no significant difference in the proportion of cardiac arrhythmias between dogs with increased and normal lactate concentration, using both cutoffs of 37.6 and 54 mg/dL.

The mortality rate of dogs with lactate concentration ≤54 mg/dL and cPG-A concentration ≤802 μg/L was 6.1%, and was significantly (P = .002) lower compared to dogs with lactate concentration >54 mg/dL (34.6%) and cPG-A >802 μg/L (53.8%; Fig 4). There were no significant correlations between lactate, CRP, cPLI, and cPG-A concentrations in dogs with GDV.

Figure 4.

Mortality rates of dogs diagnosed with gastric dilatation-volvulus that had, at presentation, either lactate and pepsinogen within reference range, only hyperlactatemia, or only increased serum canine pepsinogen-A.

Discussion

Pepsinogens are produced and stored in the gastric mucosa, mainly in chief cells, and thus can be used as markers of gastric health in human patients.[17-20, 22] Several studies have investigated the association of pepsinogen concentrations with diseases in veterinary medicine. In cattle, serum pepsinogen concentration was used to diagnose and monitor treatment of gastrointestinal nematode infections (eg, Ostertagia osteragi and Cooperia oncophora),[26-29] as well as to diagnose abomasal displacement and volvulus.[30, 31] Results of studies of the latter have been conflicting. Whereas 1 study documented increased serum pepsinogen concentration in cows with abomasal displacement, another has documented decreased concentrations, lasting over a week postdisplacement.[30, 31] Serum pepsinogen concentration also has been used for the diagnosis of gastric ulcerations and Hyostrongylus rubidus infections in pigs,[21, 32] and in foals with gastric and duodenal ulcers.[33] Serum cPG-A previously was measured in dogs.[17, 23] However, it has not been used as a marker for gastric disease in this species. It was our hypothesis that events leading to gastric wall damage, such as in GDV, will lead to increased serum cPG-A concentration in dogs.

The present study shows that serum cPG-A concentration is increased in 20% of dogs with GDV, and this increase was associated with disease severity and outcome. Specifically, dogs with increased serum cPG-A concentration had a higher occurrence of gastric necrosis (thus requiring partial gastrectomy or invagination), hemostatic derangement (thus requiring more blood product transfusions), and a higher mortality rate compared to dogs in which cPG-A was within RI. Although this study demonstrates that increased serum cPG-A concentration at presentation in dogs with GDV is associated with gastric wall necrosis, it cannot be determined whether increased cPG-A concentration results from leakage from necrotic cells into the gastric lumen followed by reabsorption, to increased gastric chief cell permeability and therefore leakage into the bloodstream or to a combination of these. Because only 13/66 dogs had increased serum cPG-A concentrations, and because this analysis was conducted in a limited number of dogs, these results should be interpreted with caution, and this test needs further evaluation. Nevertheless, the present significant association between serum cPG-A and GWDS is encouraging, and thus serum cPG-A may serve as a useful presurgical prognostic indicator once testing becomes widely available.

This study did not explore all previously reported prognostic indicators in dogs with GDV. Nonetheless, in agreement with previous studies, the present results show that presence of gastric wall damage, based on gross morphology during surgery, are significantly associated with mortality.

Blood lactate concentration at presentation currently is considered the most useful presurgical prognostic predictor in canine GDV.[9, 12, 16] However, it is only indirectly related to the extent of gastric wall necrosis, and is mostly influenced by the extent of systemic hypoperfusion and the presence of shock and anaerobic metabolism. A previous study of 102 dogs with GDV found that a plasma lactate concentration >54 mg/dL at presentation was associated with gastric wall necrosis, and could be used as a good outcome predictor, with a specificity and sensitivity of 88% and 61%, respectively.[12] In contrast, 2 more recent studies of canine GDV failed to demonstrate any association between hyperlactatemia at presentation and gross gastric wall damage or survival.[9, 16] These studies did show, however, that changes in lactate concentration, before and after IV fluid therapy, served as a better prognostic tool, rather than the absolute concentration at presentation. A decrease of ≥42–50% in plasma lactate concentration compared with its initial concentration was considered a good prognostic sign.[9, 16] Nonetheless, this information has limited usefulness as an early prognostic indicator because it is unavailable at presentation.

To compare the performance of serum lactate and cPG-A concentrations at presentation as early outcome predictors, ROC analyses of both were performed in this study. Although cPG-A was shown to be a moderately accurate outcome predictor, lactate had a lower accuracy as an early outcome predictor (AUC of 0.75 and 0.66, respectively). In addition, cPG-A was more closely associated with the GWDS. Increased concentrations of both analytes were associated with the occurrence of hemostatic derangement, but whereas increased serum cPG-A concentration was significantly associated with transfusion of all blood products, hyperlactatemia (lactate >54 mg/dL) was associated with pRBC transfusion only.

The effect of GDV on the pancreas and its association with pancreatitis in dogs has never been investigated in a large-scale study, to the best of our knowledge. However, pancreatitis has been considered as a potential complication of GDV in dogs.[34] The pancreas may become edematous and inflamed because of hypoperfusion and ischemia resulting from compromised blood supply caused by gastric and splenic displacement and decreased venous return through the portal vein.[1] The diagnosis of pancreatitis in dogs often is challenging because of its variable clinical presentation and limited sensitivity of diagnostic tools, and its occurrence thus might be underestimated.[35]Canine pancreatic lipase immunoreactivity has been shown to be exocrine pancreas-specific, with diagnostic sensitivity of 71–81.8% in pancreatitis.[36, 37] , 17 Recent studies of cPLI in pancreatitis also have shown that the assay has very high specificity, ranging between 97.5 and 100% for diagnosing pancreatitis using the 400 μg/L cutoff point.[37, 38] Concentrations between 200 and 400 μg/L are considered to be nondiagnostic, and such concentrations should be cautiously interpreted on an individual basis.[36, 39] In a previous study of 25 dogs with biopsy-proven gastritis, only a single dog had a serum cPLI concentration above the suggested cutoff value for pancreatitis.[40] However, a recent study of healthy dogs has reported that 5.9% of the measurements of cPLI over a 3-month period were >200 μg/L.[39] This study is the first to demonstrate that pancreatic damage, as judged by a serum cPLI concentration above the 400 μg/L cut-off, is a more common than expected sequela in GDV. Our results show that approximately 58% of the dogs with GDV had increased serum cPLI concentration, and in 18%, serum cPLI concentration was above the currently accepted pancreatitis cutoff. These results suggest that the pancreas is adversely affected during GDV. Nonetheless, most dogs with an increased serum cPLI concentration in this study neither showed overt classical clinical signs of pancreatitis, nor was pancreatitis noted clinically postoperatively, with the exception of a single case. This dog had the highest recorded serum cPLI concentration (3,380 μg/L) in the study, although, during surgery, its pancreas had been assessed to be normal. Based on these observations, and supported by studies suggesting that subclinical pancreatitis may be common in dogs,[39] we believe that pancreatic compromise occurs during GDV; however, it is transient and reversible in most patients, and does not appear to have a clinically relevant impact on the outcome. These findings may be because of the fact that postoperative therapy of dogs with GDV often mirrors the management of pancreatitis (ie, fluid therapy, analgesic therapy, antiemetic therapy, caloric support).

C-reactive protein is an acute phase protein produced by the liver in response to tissue injury caused by various insults, including trauma, inflammation, and neoplasia.[41, 42] Serum concentration of CRP has been shown to be highly variable in healthy dogs, with intraindividual and interindividual variabilities of 115% and 91%, respectively.[39] CRP often is used in dogs as an early, sensitive, albeit nonspecific, biomarker of inflammation.[39] Serum CRP concentration was measured in this study to explore its potential use as a presurgical outcome predictor in dogs with GDV. Indeed, CRP was increased in 75% of the dogs in this study, demonstrating its high sensitivity as an early marker of tissue damage and inflammation. However, it was a poor predictor of gastric wall lesions, secondary complications or mortality. This finding is not surprising, and is consistent with previous studies. A single, early CRP measurement often does not predict the outcome of several acute diseases in dogs, including acute abdomen syndrome,[43] babesiosis,[44] systemic inflammatory response syndrome,[45] and immune-mediated hemolytic anemia.[39, 42, 46-50] In all these conditions, increased CRP concentration was indeed found to be an early, sensitive marker of inflammation, but was not associated with outcome. In dogs, CRP increases as early as 4 hours after tissue injury, and is persistently above the upper limit of the RI as long as inflammation is ongoing. Its half-life is only 19 hours, and thus, its concentrations are expected to decrease early, once inflammation has subsided. In acute inflammatory conditions, such as the aforementioned conditions, and including GDV, increased serum CRP concentration is to be expected in most dogs, because of its high sensitivity to detect inflammation. In contrast, serial CRP measurements over time have been shown to be useful prognostic indicators.[39, 42, 47, 50] For example, survivors were shown to have a significant decrease in CRP concentration compared with nonsurvivors in dogs with immune-mediated hemolytic anemia.[39] In surgically treated dogs with pyometra, a consistently increased CRP concentration was significantly associated with postoperative complications.[39] Additional studies of serial CRP measurements in dogs with GDV are warranted to assess its prognostic usefulness.

This study has several limitations. First, it included a limited number of dogs with GDV, thereby limiting the strength of the statistical analyses. Second, although collection of sera was prospective, the medical records were reviewed retrospectively, and in certain cases, missing data could not be retrieved, further limiting certain analyses. Third, validated, rigid guidelines for assessing gastric wall status during surgery were not available, and therefore, the overall assessment was subjective, probably introducing variability. We attempted to overcome this limitation by use of the GWDS. Fourth, in a small number of dogs, the volume of stored sera was insufficient to run all of the desired tests, thereby somewhat limiting the statistical analyses. Fifth, the control group was not age- and body weight-matched to the GDV group. The association of cPG-A, cPLI, and CRP concentrations with age and body weight in dogs is unclear. Nonetheless, the control group included a relatively large number of dogs to compensate for this discrepancy. Sixth, the controls were fasted for a period of at least 8 hours, whereas GDV occasionally occurs after a meal. Although in people it has been shown that meals do not seem to increase serum PG-A concentration, in dogs, serum cPG-A concentration has been shown to increase significantly postprandially.[23] Nonetheless, the results of this study have shown that serum cPG-A concentration was significantly associated with the GWDS, and this association was unlikely directly affected by a prior meal. Finally, veterinary care before presentation at the HUVTH, in some of the dogs with GDV, might have affected their hemodynamic and gastric wall status, thereby introducing variability.

In conclusion, serum cPG-A concentration at presentation often is increased in dogs with GDV, compared to healthy controls. Furthermore, serum cPG-A concentration was positively associated with the severity of gastric wall lesions and the requirement to perform additional surgical procedures, such as partial gastrectomy or invagination, supporting the hypothesis that its serum concentration reflects gastric wall integrity. In addition, serum cPG-A concentration performed best as a prognostic marker in this study, yet only a moderate outcome predictor based on the AUC of the ROC curve. Despite its high specificity (88%), and because of its limited sensitivity (53%), cPG-A concentration should be evaluated further and cannot be recommended as a screening test or as a single outcome predictor in evaluation of clinical GDV cases at present. However, serum cPG-A and lactate concentrations can be used together for prognostication of dogs with GDV. The present results show that if both analytes are within their reference intervals, mortality rate is approximately 6%, whereas overall mortality rate in the study was approximately 23%. The present study provides a basis for future larger scale, prospective studies assessing the usefulness of cPG-A in dogs with GDV. Both serum cPLI and CRP concentrations were increased in dogs with GDV, but their usefulness as prognostic indicators was poor. The frequent occurrence of increased serum cPLI concentrations above the cutoff value suggested for a diagnosis of pancreatitis should alert clinicians to the possibility that pancreatitis might be a postoperative complication, and perhaps its true occurrence currently is underestimated.

Footnotes

  1. 1

    Advia 120, Siemens Medical Solutions Diagnostics GmbH, formerly Bayer HealthCare GmbH, Erfurt, Germany

  2. 2

    Abacus or Arcus, Diatron, Wien, Austria

  3. 3

    Coagulometric analyzers: ACL 200 and ACL 9000, Instrumentation Laboratories, Milano, Italy

  4. 4

    Fibrometers: Thrombostat, Behnak Elektronic, Norderstedt, Germany or KC-1 micro, Amelung, Lemgo, Germany

  5. 5

    Cobas Integra 400 Plus, Roche, Mannheim, Germany (at 37°C)

  6. 6

    Cobas-Mira, Roche (at 37°C)

  7. 7

    Tri-Delta Diagnostics Inc, Morris Plains, NJ

  8. 8

    Spec cPL, Idexx Laboratories, Westbrook, ME

  9. 9

    Accutrend lactate, Roche, Manheim, Germany

  10. 10

    Lidocaine, Rafa Laboratories, Jerusalem, Israel

  11. 11

    Cefamezin, Teva Pharmaceutical Industries, Petah-Tikva, Israel

  12. 12

    Dolestine, Teva Pharmaceutical Industries

  13. 13

    Lipuro 1%, Braun, Melsungen, Germany

  14. 14

    Assival, Teva Pharmaceutical Industries

  15. 15

    Isoflurane, Nicholas Piramal, Andhra Pradesh, India or Forane, Abbot, Berkshire, UK

  16. 16

    SPSS 15.0 for Windows Statistical Software, SPSS Inc, Chicago, IL

  17. 17

    Steiner J, Broussad J, Mansfield C, et al. Serum canine pancreatic lipase immunoreactivity (cPLI) concentrations in dogs with spontaneous pancreatitis. J Vet Intern Med 2001;15:274 (abstract)

Ancillary