• Open Access

Agreement of the Serum Spec fPL™ and 1,2-O-Dilauryl-Rac-Glycero-3-Glutaric Acid-(6′-Methylresorufin) Ester Lipase Assay for the Determination of Serum Lipase in Cats with Suspicion of Pancreatitis

Authors


  • Results of this study were presented in parts as a poster presentation at the American College of Veterinary Internal Medicine Forum in Seattle, WA, June 2013.

Corresponding author: Dr P.H. Kook, Clinic for Small Animal Internal Medicine, Vetsuisse Faculty, University of Zurich, Winterthurerstrasse 260, Zurich 8057, Switzerland; e-mail: peterhendrikkook@gmail.com.

Abstract

Background

Serum lipase activities measured by catalytic assays are claimed to be of limited utility for diagnosing pancreatitis in cats. The Spec fPL assay currently is believed the most sensitive test; however, studies comparing different lipase assays are lacking. 1,2-o-dilauryl-rac-glycero-3-glutaric acid-(6′-methylresorufin) ester (DGGR) assay for the determination of lipase activity has been evaluated in dogs, but no information is available in cats.

Objectives

To investigate the agreement of DGGR-lipase activity and Spec fPL concentration in cats with clinical signs consistent with pancreatitis.

Animals

Two hundred fifty-one client-owned cats.

Methods

DGGR-lipase activity and Spec fPL concentration measured from the same blood sample in cats undergoing investigation for pancreatitis. The agreement between DGGR-lipase and Spec fPL at different cutoffs was assessed using Cohen's kappa coefficient (κ). Sensitivity and specificity were calculated for 31 cases where pancreatic histopathology was available.

Results

DGGR-lipase (cutoff, 26 U/L) and Spec fPL (cutoff, >5.3 μg/L) had a κ of 0.68 (standard error [SE] 0.046). DGGR-lipase (cutoff, 26 U/L) and Spec fPL (cutoff, >3.5 μg/L) had a κ of 0.60 (SE, 0.05). The maximum κ at a Spec fPL cutoff >5.3 μg/L was found when the DGGR-lipase cutoff was set >34 U/L and calculated as 0.755 (SE, 0.042). Sensitivity and specificity were 48% and 63% for DGGR-lipase (cut-off, 26 U/L) and 57% and 63% for Spec fPL (>5.3 μg/L), respectively.

Conclusions and Clinical Importance

Both lipase assays agreed substantially. DGGR assay seems a useful and cost-efficient method compared to the Spec fPL test.

Abbreviations
CV

coefficient of variation

DGGR

1,2-o-dilauryl-rac-glycero-3-glutaric acid-(6′-methylresorufin) ester

Κ

Cohen's kappa coefficient

SE

standard error

Introduction

Pancreatitis is a common disease in cats and the most common disorder of the exocrine pancreas in cats.[1, 2] Because results of routine blood chemistry can be highly variable and generally not contributory to the diagnosis, feline pancreatitis remains a diagnostic challenge.[3, 4] It is widely believed that traditional catalytic assays for measuring serum lipase activity are unreliable in veterinary medicine because of poor sensitivity and specificity. The literature concordantly conveys the concept that serum lipase activity measurements are of limited utility for diagnosing pancreatitis in cats.[4-6] However, care must be taken about which lipase assay is considered, because different methodologies for the determination of serum lipase have been used in studies on pancreatitis. In 2001, a novel catalytic assay, the so-called 1,2-o-dilauryl-rac-glycero-3-glutaric acid-(6′-methylresorufin) ester (DGGR) for colorimetric determination of serum lipase activity, was introduced. DGGR is cleaved by lipase, resulting in an unstable dicarbonic acid ester, which is spontaneously hydrolyzed under alkaline pH to yield glutaric acid and methylresorufin, a bluish purple chromophore with peak absorption at 580 nm. The rate of methylresorufin formation is directly proportional to the lipase activity in the sample. Bile salts, colipase, and calcium ions are added to provide optimal activity of pancreatic lipase.[7] In 2005, this assay was initially validated in dogs and showed good sensitivity and moderate specificity.[8] No data were published so far in cats with pancreatitis. Since the incorporation of the DGGR-lipase into the routine serum biochemistry panel at our institution, it has been our clinical impression that it is a useful test in the investigation of pancreatic disease in cats.

Around the same time, a pancreatic lipase immunoreactivity assay was developed. The early method (fPLI) was based on a radioimmunoassay that used polyclonal antibodies.[9, 10] Serum fPLI concentration was evaluated in 18 cats with histologic evidence of pancreatitis and found to be 54% sensitive for mild pancreatitis and 100% sensitive for moderate- to-severe pancreatitis, whereas the specificity ranged between 67 and 100%.[11] Further modification resulted in the currently available Spec fPL assay designed as an enzyme-linked immunosorbent assay using monoclonal antibodies that was introduced in 2008. Although data on its clinical utility have not yet been published in a peer-reviewed article, an abstract from 2009 reported a sensitivity of 79% and a specificity of 82%.1 Even though the Spec fPL is generally regarded as the most accurate blood test for the diagnosis of pancreatitis in cats,[12] relatively high cost, limited availability and the inability to produce immediate results are its major drawbacks. Furthermore, no study has ever evaluated the performance of the Spec fPL test compared to a more cost-efficient catalytic lipase determination.

Therefore, the aim of this study was to establish precision and linearity of the DGGR-lipase in cats, and to evaluate the agreement of results of the DGGR-lipase and Spec fPL method in cats with suspicion of pancreatitis. The authors hypothesized that both methods of lipase determinations would show a significant degree of agreement.

Material and Methods

Analytical Performance of the DGGR-Lipase Assay

Precision

For within-run precision, feline serum samples were collected and pooled according to DGGR-lipase activity as low (<35 U/L), medium (35–89 U/L), and high (>89 U/L). Twenty tests were performed consecutively from each of the 3 feline serum pools. Day-to-day precision was measured with 2 commercially available quality control serum2 of high and normal lipase activity. Both were analyzed once daily before analyzing patient's samples during a 23-day block. From the results, the size of random error was determined by the calculation of the coefficient of variation (CV) using commercial software.3

Linearity

Linearity of the measurement range was assessed for feline serum. To obtain values below and above the reference intervals, 1 feline sample with high DGGR-lipase activity was diluted with 0.9% saline solution in steps of 10% to obtain a dilution series from 0 to 100% (undiluted). The samples of diluted and undiluted serum were analyzed in duplicate.[13]

Animals and Study Design

From November 2008 to August 2012, all cats with a suspicion of pancreatitis were included. Pancreatitis was suspected when one of the following criteria was present: anorexia, lethargy, vomiting, or abdominal pain, and could not be explained otherwise.[2-4, 14-16] The following variables were included and evaluated: signalment, serum DGGR-lipase activity, Spec fPL concentration, and serum creatinine concentration. Only results from the same blood sample were considered eligible for inclusion into the study. Repeated measurements in the same cats were excluded. DGGR-lipase activity was measured using an in-house assay.4 The reference range for DGGR-lipase (8–26 U/L) had been previously established using 80 apparently healthy cats of various breeds and either sex. Spec fPL was measured by IDEXX Laboratories.5 When available, histopathology reports (ie, biopsies taken during exploratory surgery or necropsy) were included and the diagnoses (ie, acute pancreatitis, chronic pancreatitis, normal pancreas) were available and compared with the results of both lipase assays. Results from histopathology were only included if the time interval between histopathology and lipase determination did not exceed 5 days. Acute pancreatitis was defined as neutrophilic infiltration and necrosis; chronic pancreatitis was defined as lymphocytic infiltration and fibrosis.

Statistical Analysis

Statistical analysis was performed using commercial software.6 Agreements between DGGR-lipase and Spec fPL at various cutoffs were assessed using Cohen's kappa coefficient (κ).[17] Values <0 indicate no agreement, values between 0–0.20 indicate a slight agreement, values between 0.21–0.40 indicate a fair agreement, values between 0.41–0.60 indicate a moderate agreement, values between 0 .61–0.80 indicate a substantial agreement, and values between 0.81–1 indicate an almost perfect agreement.[18] Spearman correlation coefficient between results of both lipase methods also was calculated.

Results

Analytical Performance of the DGGR-Lipase Assay

Results from the precision study are shown in Table 1. CVs ranged from 1.0 to 2.0% indicating that the DGGR-lipase assay showed very little imprecision at all ranges. All CVs are far below the maximal acceptable imprecision for the measurement of lipase activity in serum.[19] The DGGR-lipase assay demonstrated excellent linearity (Fig 1).

Table 1. Day-to-day and within-run precision of the DGGR-lipase assay for feline serum
Lipase ActivityWithin-Run Precision (n = 20)Day-to-Day Precision (n = 23)
Mean ± SD (U/L)CV (%)Mean ± SD (U/L)CV (%)
Low16 ± 0.32.0Not determinedNot determined
Middle45.2 ± 0.61.451.4 ± 0.51.0
High146.7 ± 2.81.9126.2 ± 2.41.9
Figure 1.

Linearity plot for feline lipase activity measured with the DGGR-lipase assay (linear fit: 0.2432, +1.494x; polynomial fit: −0.07514, +1.602x, 0.003558x2).

Study Population

The study population consisted of 251 cats including 157 male and 94 female cats. Ages ranged from 0 to 20 years (median, 11 years). The breeds included domestic shorthair (n = 170), Persian (n = 11), Siamese (n = 10), British Shorthair (n = 6), Domestic Longhair (n = 6), Maine Coon (n = 6), Birman (n = 4), Norwegian Forest Cat (n = 4), Abyssinian (n = 3), Balinese (n = 3), Burmese (n = 3), Turkish Angora (n = 3), Domestic Shorthair/Siamese (n = 2), Oriental (n = 2), Russian Blue (n = 2), Sibirian (n=2), Chartreux (n = 1), Devon Rex (n = 1), Ragdoll (n = 1), Tonkinese (n = 1), mixed breed (n = 1). Nine cases had no data about breed available.

DGGR-lipase was ≤26 U/L in 125/251 (49.8%) cats and >26 U/L in 126/251 (50.2%) cats. Spec fPL was ≤3.5 μg/L in 107/251 (42.7%) cats, between 3.6 and 5.3 μg/L in 32/251 (12.7%) cats and ≥5.4 μg/L in 112/251 (44.6%) cats (Table 2).

Table 2. Contingency table
 Spec fPL
<3.5 μg/L3.6–5.3 μg/L≥5.4 μg/LTotal
DGGR-lipase
≤26 U/L912113125
>26 U/L161199126
Total10732112251

Agreement between DGGR-Lipase and Spec fPL

When a DGGR-lipase cutoff >26 U/L (above upper reference range) and a Spec fPL cutoff >5.3 μg/L (consistent with pancreatitis7) were chosen, a κ of 0.681 (SE, 0.046) was calculated. When a DGGR-lipase cutoff >26 U/L (above upper reference range) and a Spec fPL cutoff >3.4 μg/L (gray zone value, re-evaluation recommended7) were chosen, a κ of 0.601 (SE, 0.050) was calculated. The maximal value for κ at a DGGR-lipase cutoff >26 U/L was found when the Spec fPL cutoff was set >6.3 μg/L and was calculated as 0.721 (SE, 0.043). The maximal value for κ at Spec fPL cutoff >5.3 μg/L was found when the DGGR-lipase cutoff was set >34 U/L and was calculated as 0.755 (SE, 0.042). The distribution of results of both assays is shown in Figure 2. The Spearman correlation coefficient between both methods was calculated as rho = 0.826 (Fig 2).

Figure 2.

Scatterplot showing the distribution of lipase values of both assays.

Evaluation of Concurrent Azotemia

If only nonazotemic cats with serum creatinine concentrations within the reference range (98–163 μmol/L) were included (n = 169), κ between DGGR-lipase at a cutoff >26 U/L and Spec fPL at a cutoff >5.3 μg/L was 0.715 (SE, 0.053). If only cats with serum creatinine concentration >163 μmol/L were included (n = 46), κ between DGGR-lipase at a cutoff >26 U/L and Spec fPL at a cutoff >5.3 μg/L was 0.605 (SE, 0.125). If only cats with serum creatinine concentration <250 μmol/L were included (n = 202), κ between DGGR-lipase at a cutoff >26 U/L and Spec fPL at a cutoff >5.3 μg/L was 0.694 (SE, 0.050). If only cats with serum creatinine concentration >250 μmol/L were included (n = 13), κ between DGGR-lipase at a cutoff >26 U/L and Spec fPL at a cutoff >5.3 μg/L was 0.755 (SE, 0.228).

Histopathology

In 31 cases, histopathology (3 biopsy, 28 necropsy) was available. The time interval between histopathology and lipase measurements ranged from 0 to 5 days (median, 0 days; mean, 0.7 days). Histopathologically, pancreatitis was diagnosed in 23/31 (74.2%) cats. Acute pancreatitis was found in 4/23 cats, and chronic pancreatitis was found in 19/23. Overall, DGGR-lipase was increased in 11/23 (47.8%) cats with pancreatitis. DGGR-lipase was increased in 4/4 (100%) with acute pancreatitis and 7/19 (36.8%) with chronic pancreatitis. Overall, Spec fPL was increased >5.3 μg/L in 13/23 (56.5%) cats with pancreatitis, and between 3.6 μg/L and 5.3 μg/L in 2/23 (8.7%). Spec fPL was >5.3 μg/L in 4/4 (100%) with acute pancreatitis. Spec fPL was >5.3 μg/L in 9/19 (47.4%) cats and between 3.6 μg/L and 5.3 μg/L in 2/19 (10.5%) with chronic pancreatitis.

DGGR-lipase was increased (27, 28, and 87 U/L) in 3/8 (37.5%) cats without histopathologic evidence for pancreatitis. Spec fPL was increased >5.3 μg/L (6.8, 10.8, and 39 μg/L) in 3/8 (37.5%) cats without histopathologic evidence for pancreatitis.

Calculated sensitivity and specificity (95% confidence intervals [CI]) of DGGR-lipase for all cases of pancreatitis were 48% (CI, 27–69%), and 63% (CI, 26–90%), respectively. Calculated sensitivity and specificity of DGGR-lipase for acute pancreatitis were 100% (CI, 40–100%) and 63% (CI, 26–90%), respectively. Calculated sensitivity and specificity of DGGR-lipase chronic pancreatitis were 37% (CI, 17–61%) and 63% (CI, 26–90%), respectively. Calculated sensitivity and specificity of Spec fPL at a cutoff >5.3 μg/L for all cases of pancreatitis were 57% (CI, 35–76%) and 63% (CI, 26–90%), respectively. Calculated sensitivity and specificity of Spec fPL at a cutoff >5.3 μg/L for acute pancreatitis were 100% (CI, 40–100%) and 63% (CI, 26–90%), respectively. Calculated sensitivity and specificity of Spec fPL at a cutoff >5.3 μg/L for chronic pancreatitis were 47% (CI, 25–71%) and 63% (CI, 26–90%), respectively. Calculated sensitivity and specificity of Spec fPL at a cutoff >3.5 μg/L for all cases of pancreatitis were 65% (CI, 43–83%) and 63% (CI, 26–90%), respectively. Calculated sensitivity and specificity of Spec fPL at a cutoff >3.5 μg/L acute pancreatitis were 100% (CI, 40–100%) and 63% (CI, 26–90%), respectively. Calculated sensitivity and specificity of Spec fPL at a cutoff >3.5 μg/L for chronic pancreatitis were 58% (CI, 34–79%) and 63% (CI, 26–90%), respectively. Sensitivity and specificity also are shown in Table 3.

Table 3. Sensitivity and specificity (95% confidence interval)
 DGGRSpec fPL >3.5 μg/LSpec fPL ≥5.4 μg/L
Sensitivity
Pancreatitis (n = 23)48% (27–69%)65% (43–83%)57% (35–76%)
Acute pancreatitis (n = 4)100% (40–100%)100% (40–100%)100% (40–100%)
Chronic pancreatitis (n = 19)37% (17–61%)58% (34–79%)47% (25–71%)
Specificity (n = 8)63% (26–90%)63% (26–90%)63% (26–90%)

Discussion

In contrast to the widespread belief that traditional catalytic assays are of limited utility in feline pancreatitis, the data of this study show substantial agreement of the DGGR-lipase with Spec fPL, the test commonly believed to be most sensitive. Data on precision and linearity have not been reported yet for the DGGR-lipase assay in cats. Results from the present study illustrate that the DGGR-lipase assay performed very well in cats. The assay is highly precise at all investigated activity levels with CV values far below those reported for the equivalent test in dogs (Spec cPL).[20] Dilutional linearity results demonstrated reliability of the DGGR-lipase assay over the reportable range of 15–152 U/L. Data on the analytical validation of the Spec fPL assay have not yet been reported.

The cited evidence for the poor performance of traditional catalytic assays is weak and is based on only a few cases. Duffel et al. reported that 1 cat with necropsy-proven pancreatitis had normal lipase activity.8 It remains unclear how the diagnosis of pancreatitis was made in the other 4 cases with normal lipase results.[21] Kitchell et al. described experimentally-induced pancreatitis in 6 cats, all of which had a significant increase in nephelomtrically measured serum lipase activity.[22] In a frequently cited retrospective study on 40 cases of acute pancreatitis, lipase was measured (method unknown) in 2 cats with acute pancreatic necrosis, and found to be normal in 1 other cat.[3] In a case report only 1 of 3 cats with pancreatitis had increased serum lipase (method unknown).[14] The most frequently cited reference found normal lipase results (method unknown, most likely the 1,2-diglyceride assay based on given reference range) in 8 cats with histopathological evidence of acute pancreatitis.9 A recent publication comparing this widely used commercial 1,2-diglyceride assay with the DGGR-lipase assay found only a poor correlation (rho = 0.25).[23] Based on this evidence, it is likely that the 1,2-diglyceride assay is not useful for diagnosing pancreatitis in cats, and that usage of this assay most probably contributed to the general poor perception of traditional catalytic lipase assays. With the advent of the Spec fPL assay, no further studies have compared its diagnostic value to different lipase tests in feline pancreatitis.

We have noted a striking agreement between the pancreatic-specific lipases (fPLI/Spec fPL) and the in-house DGGR-lipase results, which led to the initiation of this study. We decided to only focus on the comparison of DGGR-lipase versus Spec fPL, because the Spec fPL test represents the currently available assay. In order to correctly analyze the data, we used Cohen's kappa coefficient to assess agreement between the DGGR-lipase assay and the Spec fPL over the commonly used Spearman correlation coefficient. This was performed because correlations assess the degree of relationship in the data set, but the true aim of method comparison is to assess agreement beyond the data set, between the measurements in any realistically conceivable situation. Also, the magnitude of the correlation coefficient strongly depends on selection of samples, and the range of measurements. Although smaller or larger correlations can be obtained, the relationship might still be the same between methods. For these reasons, the use of correlation coefficients for method comparison studies usually is discouraged by statisticians.[24] However, because of its widespread use, the Spearman correlation coefficient also is described. The interpretation of Cohen's kappa values given by Landis and Koch[18] is somewhat arbitrary, but we felt they are useful to the reader to give some guidance in interpreting the kappa values.

When comparing DGGR-lipase at a cutoff >26 U/L to Spec fPL at a cutoff >3.5 μg/L, κ decreased slightly from 0.681 (SE, 0.046) to 0.601 (SE, 0.05). This can either be explained with a larger percentage of false or true positives discriminated by the Spec fPL test at 2 cutoffs. Interestingly, a higher agreement (κ = 0.755) with spec fPL >5.3 μg/L was found for the DGGR-lipase at a cutoff at 34 U/L. Again, this might be because of an increased percentage of false negatives or true negatives discriminated by the DGGR-lipase test. These considerations reflect the limitation of an approach comparing the agreement of 2 methods without actually knowing the true state of disease. It is currently not known which assay yields more accurate results. For further evaluation of both lipase assays, studies comparing results to a defined gold standard are needed. However, there is currently lack of agreement regarding a gold standard against which diagnostic tests should be evaluated. Even pancreatic histopathology has its drawbacks, because distribution of pancreatic inflammation has been shown to be highly localized.[1] Another objection would be that mild histopathologic abnormalities might not be associated with clinical signs, and well-designed clinical studies might serve as a more useful gold standard than histopathologic studies.

When considering the present histopathologic results, the DGGR-lipase as well as the Spec fPL showed excellent sensitivity for acute pancreatitis. In contrast, the sensitivity of both tests for chronic pancreatitis was fair at best. Even though results still were better for the Spec fPL assay (especially at a cutoff >3.5 μg/L), sensitivities in this range are clinically impractical. Similar results have been reported recently in dogs with chronic pancreatitis.[25] The false-negative results in cats with chronic pancreatitis may indicate a lack of active disease with enzyme activity and release. Another explanation for the low sensitivity might be mild severity of histopathologic changes, as recently also demonstrated in dogs.[26] Unfortunately, severity grading of pancreatic histopathology that could have addressed this point was not available for the present study. Ideally, lipase measurements should be compared to a standardized pancreatic histopathological assessment from freshly procured pancreata, so that the relevance of severity and type (acute versus chronic) of inflammation as well as its parenchymal distribution can be clarified. The same problem holds true for the assessment of specificity, because the histopathological diagnosis pancreatitis might have been missed during necropsy or surgery because of focal disease. It should be noted that most of the histopathologic results in this study come from necropsy reports, indicating a bias toward cats that died or were euthanized.

There is debate whether concurrent azotemia affects catalytic lipase activity results because of decreased glomerular filtration rate. Although κ between groups at different cutoffs for serum creatinine concentrations do slightly differ, these differences may be of negligible relevance considering the larger range of the SE. It appears that both assays are similarly affected by concurrent azotemia, and it could be stated that catalytic lipase activity seems no more influenced by azotemia than Spec fPL. However, the impact of concurrent azotemia cannot be conclusively assessed because only the lipase agreement between azotemic and nonazotemic cats was compared. Also, because there were only few cases (13) with serum creatinine concentration at cutoffs >249 μmol/L, a statistical analysis using higher serum creatinine concentrations cutoffs did not seem reasonable. Nonetheless, we chose to include this information in this study because the effect of renal function on serum lipase concentrations in cats has not been evaluated. For directly assessing the influence of concurrent azotemia on serum lipase activity, studies on experimentally induced azotemia in cats without pancreatic disease theoretically would be required, although ethically not justifiable.

When considering the results of this study, the marked cost difference between both methods should be taken into consideration (currently, the cost of running a Spec fPL is >10 times the cost of the DGGR lipase at the authors' institution). This certainly represents a major disadvantage for both the client and clinician. Another aspect is turnaround time. Results of the DGGR-lipase are available within a very short time frame (20 minutes–1 hour depending on urgency) compared to the Spec fPL.

The results of the present study suggest that both lipase assays agree substantially, and the DGGR-assay seems to be a useful method compared to the Spec fPL test, particularly when considering its cost efficiency. Additional studies evaluating the sensitivity and specificity of both lipase assays compared to a clinical scoring system or a standardized histopathologic gold standard seem warranted.

Acknowledgment

Conflict of Interest: Authors disclose no conflict of interest.

Footnotes

  1. 1

    Forman MA, Shiroma J, Armstrong PJ, Robertson JE, Buch J, (2009). Evaluation of Feline Pancreas-Specific Lipase (Spec fPLTM) for the Diagnosis of Feline Pancreatitis. [ACVIM Abstract 165]. J Vet Intern Med 2009; 23: 733–734 (abstract)

  2. 2

    Precinorm PUC, Precipath PUC, Roche Diagnostics, Rotkreuz, Switzerland

  3. 3

    Microsoft Excel Add in Analyse-it, Analyse-it Software Ltd, Leeds, UK

  4. 4

    Lipase colorimetric for Roche Cobas Integra 800, Roche Diagnostics

  5. 5

    IDEXX GmbH Ludwigsburg, Germany

  6. 6

    IBM SPSS v.20 for Mac OS X, IBM Corporation, New York, NY

  7. 7

    http://vetmed.tamu.edu/gilab/service/assays/pli

  8. 8

    Sigma serum lipase kit, Sigma-Aldrich Corporation, St. Louis, MO

  9. 9

    Parent C, Washabau RJ, Williams DA, et al. Serum trypsin-like immunoreactivity, amylase and lipase in the diagnosis of feline acute pancreatitis. J Vet Intern Med 1995; 9, 194 (abstract)

Ancillary