Clinical Chemistry 43: 2339-2344, 1997;
(Clinical Chemistry. 1997;43:2339-2344.)
© 1997 American Association for Clinical Chemistry, Inc.
Polyclonal pancreatic elastase assay is superior to monoclonal assay for diagnosis of acute pancreatitis
Volker Keima,
Niels Teich,
Andrea Reich,
Fritz Fiedler1 and
Joachim Mössner
1
Institut für Anästhesie, Klinikum Mannheim, Universität Heidelberg, Heidelberg, Germany.
a Author for correspondence. Fax 49 (0) 341 97 12 209.
 |
Abstract
|
|---|
We compared the clinical values for diagnosis of acute pancreatitis of
two commercial assays for pancreatic elastase: an ELISA procedure with
monoclonal antibodies and a RIA technique with polyclonal antibodies.
In 14 patients with acute pancreatitis, serum concentrations of
elastase determined by ELISA (ELISA-elastase) decreased much faster
(half-life 0.4 days) than those of elastase determined by RIA
(RIA-elastase) (2.2 days), amylase (0.8 days), or lipase (0.9 days).
Serum samples from 253 additional patients with abdominal pain (32 of
these with acute pancreatitis) were analyzed. In sera collected up to
48 h after the onset of disease, the ROC curves showed a slightly
higher diagnostic value of RIA-elastase. In samples taken later, at a
sensitivity of 90% the specificity of RIA-elastase was 95%
(ELISA-elastase 40%). We conclude that serum ELISA-elastase is of much
lower clinical value than RIA-elastase for diagnosis of acute
pancreatitis.
|
Introduction
|
|---|
In the past, measurement of numerous serum markers has been
recommended for diagnosis of acute pancreatitis. Among these, nearly
all pancreatic secretory enzymes such as amylase (EC 3.2.1.1), lipase
(EC 3.1.1.3.), trypsin (EC 3.4.21.4.), elastase (EC 3.4.21.36.),
ribonuclease (EC 3.1.27.5.), phospholipase A2 (EC
3.1.1.4.), and carboxypeptidase B (EC 3.4.17.2) can be found
(1)(2)(3)(4)(5)(6)(7). Also, pancreatic proteins without enzymatic
activity such as the pancreatitis-associated protein (8)
or lithostathin (9) have been used for this purpose. For
only a small number of markers, however, was sufficient data available
to appreciate their value for diagnosis of pancreatic diseases
(5). In clinical practice, measurement of amylase or
lipase is prevailing, and ROC curve analysis showed acceptable values
for sensitivity and specificity (5).
Diagnosis of acute pancreatitis by measurement of serum enzymes was
impaired when patients were admitted to the hospital several days after
onset of symptoms. This was mostly due to the short half-life of the
enzymes in serum (10). Pancreatic elastase measured by
polyclonal antibodies in a RIA system seems to be an exception, as this
enzyme remained increased for >1 week after an attack of acute
pancreatitis (11). Therefore, measurement of elastase has
been recommended in the past but not used widely, as the assay was only
available as RIA. Recently, an ELISA technique with monoclonal
antibodies was developed that might be used in smaller laboratories
without availability of radioactive isotopes, or in emergency units.
Our aim was to test and compare the clinical value of this new elastase
ELISA with that of the RIA procedure.
|
Methods
|
|---|
Serum samples were taken daily from 14 patients with acute
pancreatitis (ages 27–74 years, 10 men, 4 women; etiology: 6 biliary
origin, 5 alcoholics, 3 unclear). The pancreatitis was verified by
contrast-enhanced computed tomography (CT) or
ultrasound.1
Commercially available kits from Boehringer Mannheim were used to
measure enzyme activities of amylase (on the basis of hydrolysis of a
p-nitrophenylmaltoheptaoside derivative, cat. no.
1 040 693) and lipase (turbidimetric measurement of triolein
hydrolysis, cat. no. 1 268 449). For elastase assay, an ELISA with
monoclonal antibodies (from Schebo Tech, cat. no. 06) and a RIA with
polyclonal antibodies (from Abbott, cat. no. 1081–22) were used. For
evaluation of diagnostic values the following upper limits of normal
indicated by the manufacturers of the assays were used: amylase, 3.7
µmol/L·s; lipase, 3.1 µmol/L·s; polyclonal elastase assay, 4
µg/L; monoclonal elastase assay, 3.5 µg/L.
An additional 253 individuals (ages 18–92 years, 147 men, 106 women)
presenting with acute abdominal pain at our university hospitals and
treated as inpatients were included into the study. The first serum
sample was taken at admission but not later than 48 h (days 1–2)
after onset of symptoms. Further samples were collected at 48–96 h
(days 3–4) as well as at 96–144 h (days 5–6). Pancreatitis was
diagnosed by CT or ultrasound. In the majority of patients (72%)
ultrasound was used to exclude pancreatitis, whereas in only a minority
of the nonpancreatitis group a CT was performed.
In 568 individuals without evidence of acute or chronic pancreatitis
(normal amylase or lipase in serum, normal pancreas in ultrasound or
CT), serum elastase was measured by the ELISA procedure to establish a
reference range.
Levels of significance were calculated by a nonparametric sign test; at
P <0.05 a statistically significant difference was assumed.
The ROC curves were calculated (12) to evaluate the
diagnostic value of the enzyme measurements. In accordance with the
declaration of Helsinki of 1975 (revised in 1983), informed consent of
the patients was obtained for use of their serum samples for scientific
purposes.
|
Results
|
|---|
The precision of the elastase ELISA or RIA was controlled by
repeated measurements of the same samples on different days and by
multiple measurements of reference samples of low and high value on the
same day. Precision of RIA and ELISA were similar; the RIA data were as
follows: day-to-day reproducibility was 5.7% (at 1.5 µg/L) and 4.0%
(CV) (at 15 µg/L); the CVs of samples measured on the same day were
3.9% (at 5 µg/L) and 5.7% (at 0.5 µg/L). The values were not
influenced by repeated freezing and thawing (data not shown).
In all sera collected from the 14 pancreatitis patients on the first
day of the disease, increased concentrations of elastase measured by
RIA and by ELISA (RIA-elastase and ELISA-elastase, respectively),
amylase, or lipase were found (Fig. 1
A). After 3 days RIA-elastase was still increased in all
patients, whereas in 11 of the 14 samples the values of ELISA-elastase
were already within the normal range. After 1 week RIA-elastase was
increased in all, but ELISA-elastase in none of the samples. On day 1
amylase, lipase, and RIA-elastase were threefold increased in all and
ELISA-elastase in 12 of the 14 patients (Fig. 1B
). RIA-elastase was
threefold increased in half of the samples for >1 week. On day 2
ELISA-elastase was increased threefold above controls in 6 of 14
individuals and on day 3 in only one of 14 samples (Fig. 1B
).
On the first day the serum concentrations of RIA- and ELISA-elastase
were not different (Fig. 2
A). But as already expected from the sensitivity data, the
elastase concentration measured by ELISA decreased much faster and from
the second day on both values were statistically different (Fig. 2A
).
The decreases of amylase and lipase in the course of the disease seemed
to be parallel (Fig. 2A
).
View larger version (13K):
[in this window]
[in a new window]
|
Figure 2. Time course and half-lives of pancreatic enzymes in serum.
(A) Enzyme concentrations or activities in serum were
indicated in fold above normal; for cutoff values see Materials
and Methods. Values are means ± SE, n = 14;
*P <0.05, ***P <0.001, RIA-E vs Eli-E,
respectively. (B) Half-lives were calculated by using the
serum concentrations on the first 4 days after an acute attack of
pancreatitis; individual values and medians of the 14 patients are
shown. ***P <0.001, RIA-E vs Eli-E, A, and L. L, lipase; A,
amylase; RIA-E, RIA-elastase; Eli-E, ELISA-elastase.
|
|
The enzyme concentrations in serum measured on the first 4 days were
used to calculate, in each patient, the respective half-lives. The
individual values as well as the medians are shown in Fig. 2B
. The
half-life of RIA-elastase was 2.2 days, a value significantly higher
(P <0.05) than that of lipase (0.9 days) or amylase (0.8
days), and much higher (P <0.001) than that of
ELISA-elastase (0.4 days).
For further comparison of the clinical value of the two assay systems
we performed elastase measurements in the serum of patients admitted to
the clinic with acute abdominal pain. An acute pancreatitis was found
in 32 of these individuals by CT or ultrasound. The ROC curves
calculated from samples collected up to 48 h after onset of
symptoms showed that the diagnostic quality of the RIA procedure was
slightly higher than that of the ELISA technique (Fig. 3
A): At a sensitivity of 95%, the specificity was near 95%
(ELISA: 92%). With samples collected at 48–96 h the RIA showed much
better results: At a sensitivity of 90% the specificity was 89%
(ELISA-elastase <40%, Fig. 3B
). The difference was even larger with
samples at 96–144 h after onset of pancreatitis: At a sensitivity of
80%, the specificity of RIA-elastase was 80% and that of the
ELISA-elastase <10% (Fig. 3C
).
View larger version (16K):
[in this window]
[in a new window]
|
Figure 3. ROC plots for diagnosis of pancreatitis by ELISA- ( )
and RIA-elastase (•) with samples collected 0–48 h (A),
48–96 h (B), and 96–120 h (C) after onset of
acute abdominal pain.
|
|
The diagnostic accuracy of elastase measurement [(true positives +
true negatives)/all samples] was calculated in relation to different
cutoff concentrations. It was a remarkable finding that the highest
accuracy of ELISA-elastase was reached with a cutoff below the upper
limit of normal (Fig. 4
A), whereas the corresponding cutoff for RIA-elastase was at
threefold normal (Fig. 4B
). As anticipated from the ROC curves, the
diagnostic accuracy of RIA-elastase was much higher than that of
ELISA-elastase in samples collected later than 48 h.
To control the reference range of ELISA-elastase, in a series of 568
patients admitted to our gastroenterological clinic without evidence of
pancreatitis (normal concentrations of pancreatic amylase and lipase,
no evidence of acute pancreatitis in ultrasound or CT), concentrations
of ELISA-elastase were measured in serum. Here we found that 95% of
the values were <1.35 µg/L, 97% were <2.0 µg/L, and 99% were
<2.2 µg/L.
|
Discussion
|
|---|
Radioisotope assays for clinical diagnostics have been
disappearing more or less from the panel of available techniques mainly
because of increasingly complicated regulations for the handling of
radionuclides. In accord with this, the pancreatic elastase assay with
polyclonal antibodies in a RIA system has not been widely used, though
the technique has been shown to have a high sensitivity
(4). By comparing the results of the recently available
elastase ELISA with monoclonal antibodies with those of the RIA
procedure with polyclonal antibodies, we found, surprisingly, that the
latter assay with polyclonal antiserum was much better for diagnosis of
pancreatitis than the ELISA system with monoclonals.
One major bias in this study may be our definition of pancreatitis. We
chose to use as the sole indicator the findings in CT or ultrasound to
assign patients to the pancreatitis group. It is generally assumed that
by this procedure 10–15% of pancreatitis cases are overlooked, but
precise data were not available. As we intended to compare the
diagnostic efficiency of enzyme measurements, we had to use criteria
independent of the results of serum enzymes. As an important advantage,
by this procedure we completely avoided false positives in our
pancreatitis group. On the other hand, pancreatitis must be excluded in
all other patients. This cannot easily be achieved by any simple
procedure, but in a recent publication sonography was known to exhibit
a high negative predictive value of nearly 90% (13). In
this study, the prevalence of pancreatitis patients was even higher
(28%) than in ours (12%). As in the majority of our patients, an
ultrasound investigation was performed so we could be rather sure not
to misclassify a significant number of patients.
Diagnostic accuracy and also ROC curves of RIA-elastase from samples
collected at admission were previously analyzed in a series of papers
(14)(15)(16)(17)(18)(19)(20)(21). They show, corresponding to our findings, a high
sensitivity (>90%) at high specificity (>90%) of elastase, but also
of other pancreatic secretory proteins. In these papers mostly lipase
was favored because of easy assay techniques. ELISA-elastase, however,
has not been evaluated in detail; only a measurement of healthy blood
donors has been performed (22). We can show here that in
samples collected early, RIA and ELISA measurements (i.e., measurements
with polyclonal or monoclonal antibodies) were of rather similar
clinical value, the RIA being slightly superior (Fig. 3A
). However, the
use of the cutoff value of 3.5 µg/L in the ELISA recommended by the
manufacturer of the assay resulted in a much lower sensitivity. This,
together with the accuracy data shown in Fig. 4
, may suggest that the
definition of the reference range of the ELISA may be a problem.
Therefore ELISA-elastase measurements in 568 individuals without
evidence of pancreatic diseases were done. In 97% of the samples
concentrations were <2 µg/L. This was in contrast to the findings of
others (22) in which 97% of control sera concentrations
were <3.5 µg/L. Though this difference was obviously a small one, it
may suggest the use of an upper limit of normal near 1.5–2 µg/L. At
this cutoff the sensitivity of the ELISA-elastase would be close to the
RIA procedure, but only with samples collected up to 48 h.
The main diagnostic problem in pancreatitis, however, was that patients
do not come to the clinic early after onset of pain. Individuals with
chronic alcoholic pancreatitis, in particular, present late, so that
the half-life of the enzymes in serum after an acute attack has to be
taken into account (23)(24)(25). This could lead to normal
enzyme concentrations and, in turn, to overlooking acute pancreatitis
in a significant number of patients (25). When we focused
on samples collected later after onset of pain we could show that
RIA-elastase was persistently increased for >1 week. This was a
finding similar to that reported by Büchler et al.
(11). ROC curves for the late samples have not been
published yet; here we demonstrate that RIA-elastase was of high
accuracy even a week after onset of pain, and at a sensitivity of 80%
the specificity was still near 80%. The corresponding values for
ELISA-elastase were far lower.
One reason for this large difference in clinical value could be the
short half-life of ELISA-elastase (0.4 days) compared with the 2.2 days
for RIA-elastase. At the moment we cannot offer any simple explanation
for this difference. It is not very likely that this has anything to do
with the type of the assay procedure, but seems to be associated with
the antibodies used. In addition, elastase measurement in serum is
difficult, as in its active form the enzyme is bound to
2-macroglobulin (A-2-MG) and
1-antitrypsin (A-1-AT). Whereas the complex with A-1-AT
may be recognized by the polyclonal antibodies, the elastase bound to
A-2-MG cannot be detected (26). It has, however, not been
determined whether this is also true for the monoclonal antibodies used
in the ELISA. The situation may be even more complex during
pancreatitis as both activated and nonactivated proelastase could be
present in serum and recognized differently by the antibodies.
We could further speculate that monoclonal antibodies in the ELISA
system mainly recognize the activated elastase, as antibodies were
raised against the active form, and that the polyclonal antibodies in
the RIA detect both activated and nonactivated forms. This could
explain the high values immediately after onset of pancreatitis, where
activated enzymes could be present in serum. After coupling of elastase
to A-2-MG or A-1-AT the complex will be cleared at a higher rate from
serum. In humans free A-2-MG had a half-life of >100 h, whereas its
half-life was <10 min after formation of complexes with proteases
(27).
Independent from the putative explanation for our findings, we conclude
from the present data that only the polyclonal elastase assay can be
recommended for diagnosis of acute pancreatitis in patients with
abdominal pain. In particular, 2 days or more after onset of pain the
sensitivity of the ELISA system was far too low.
|
Acknowledgments
|
|---|
The skillful technical assistance of Susanne Kistner, Ines
Sommerer, and Ute Kullrich is gratefully acknowledged.
|
Footnotes
|
|---|
Medizinische Klinik und Poliklinik II, Universität Leipzig, Philipp Rosenthal Str. 27, D-04103 Leipzig, Germany.
1 Nonstandard abbreviations: CT, computed tomography; A-2-MG, 2-macroglobulin; and A-1-AT, 1-antitrypsin. 
|
References
|
|---|
-
Elman A, Arneson N, Graham EA. Value of blood amylase estimations in the diagnosis of pancreatic disease: a clinical study. Arch Surg 1929;19:943-967.
-
Warshaw AL, Lee KH. Serum ribonuclease elevations and pancreatic necrosis in acute pancreatitis. Surgery 1979;86:227-234.
[ISI][Medline]
[Order article via Infotrieve]
-
Nordback I, Teerenhovi O, Auvinen O, Koivula T, Thuren T, Kinnunen P, et al. Human pancreatic phospholipase A2 in acute necrotising pancreatitis. Digestion 1989;42:128-134.
[ISI][Medline]
[Order article via Infotrieve]
-
Büchler M, Malfertheiner P, Schädlich H, Nevalainen TJ, Friess H, Beger HG. Role of phospholipase A2 in human acute pancreatitis. Gastroenterology 1989;97:1521-1526.
[ISI][Medline]
[Order article via Infotrieve]
-
Agarwal N, Pitchumoni CS, Sivaprasad AV. Evaluating tests for acute pancreatitis. Am J Gastroenterol 1990;85:356-366.
[ISI][Medline]
[Order article via Infotrieve]
-
Benini L, Bevilacqua D, Brocco G, Pilati S, Bardelli E, Vantini I, Cavallini G. Lipase latex test for acute abdominal pain: comparison with serum lipase, trypsin, elastase and amylase. Ital J Gastroenterol 1992;24:61-64.
[ISI][Medline]
[Order article via Infotrieve]
-
Bennini L, Rizotto P, Vaona B, Sembenini C, Brocco G, Micciolo R, et al. Elastase 1 vers trypsin, lipase, amylase serum levels in pancreatic diseases. Int J Pancreatol 1987;2:361-370.
[ISI][Medline]
[Order article via Infotrieve]
-
Iovanna JL, Keim V, Nordback I, Montalto G, Camarena J, Leboublon C, et al. Serum levels of the pancreatitis-associated protein as indicators of the course of acute pancreatitis. Gastroenterology 1994;106:728-734.
[ISI][Medline]
[Order article via Infotrieve]
-
Schmiegel W, Burchert M, Kalthoff H, Roeder C, Bützow B, Grimm H, et al. Immunochemical characterization and quantitative distribution of pancreatic stone protein in sera and pancreatic secretions in pancreatic disorders. Gastroenterology 1990;99:1421-1430.
[ISI][Medline]
[Order article via Infotrieve]
-
Ventrucci M, Pezzilli R, Naldoni P, Plate L, Baldoni F, Gullo L, Barbara L. Serum pancreatic enzyme behavior during the course of acute pancreatitis. Pancreas 1987;2:506-509.
[ISI][Medline]
[Order article via Infotrieve]
-
Büchler M, Malfertheiner P, Uhl W, Beger HG. Diagnostic and prognostic value of serum elastase 1 in acute pancreatitis. Klin Wochenschr 1986;64:1186-1191.
[ISI][Medline]
[Order article via Infotrieve]
-
Zweig MH, Campell G. Receiver-operating characteristic (ROC) plots: a fundamental evaluation tool in clinical medicine [Review]. Clin Chem 1993;39:561-577.
[Abstract/Free Full Text]
-
Hollerbach S, Ruser J, Ochs A, Frick E, Schölmerich J. Current status of abdominal pancreatic ultrasound. A retrospective analysis of 585 pancreatic ultrasound examinations. Med Klinik 1994;89:7-13.
-
Buamah PK, Cornell C, Cassels-Smith AJ, Skillen AW. Serum human pancreatic elastase 1 levels in pancreatic disease. Clin Chim Acta 1987;166:57-60.
[ISI][Medline]
[Order article via Infotrieve]
-
Flamion B, Delhaye L, Horanyi Z, Demanet H, Quenon M, Van Melsen A, et al. Comparison of elastase 1 with amylase, lipase, and trypsin-like immunoreactivity in the diagnosis of acute pancreatitis. Am J Gastroenterol 1987;82:532-535.
[ISI][Medline]
[Order article via Infotrieve]
-
Gullo L, Ventrucci M, Pezzilli R, Platé L, Naldoni P. Diagnostic value of serum elastase 1 in pancreatic disease. Br J Surg 1987;74:44-47.
[ISI][Medline]
[Order article via Infotrieve]
-
Hakansson HO, Borgström A, Ohlsson K. Human cationic pancreatic elastase in acute pancreatitis. Scand J Lab Invest 1993;53:17-24.
-
Gwodz GP, Steinberg WM, Werner M, Henry HP, Pauley C. Comparative evaluation of the diagnosis of acute pancreatitis based on serum and urine enzyme assays. Clin Chim Acta 1990;187:243-254.
[ISI][Medline]
[Order article via Infotrieve]
-
Malfertheiner P, Büchler M, Stanescu A, Uhl W, Ditschuneit H. Serum elastase 1 in inflammatory pancreatic and gastrointestinal diseases and in renal insufficiency. A comparison with other serum pancreatic enzymes. Int J Pancreatol 1987;2:159-170.
[ISI][Medline]
[Order article via Infotrieve]
-
Ventrucci M, Pezilli R, Montone L, Plate L, Buonamici L, Bergami R, Conci T. Serum pancreatic enzyme assays in acute abdomen: a comparative prospective study. Ital J Gastroenterol 1992;24:115-118.
[ISI][Medline]
[Order article via Infotrieve]
-
Dubick MA, Mar G, Mayer AD, Majumdar AP, McMahon MJ, Geokas MC. Digestive enzymes and protease inhibitors in plasma from patients with acute pancreatitis. Pancreas 1987;2:187-194.
[ISI][Medline]
[Order article via Infotrieve]
-
Scheefers-Borchel U, Scheefers H, Arnold R, Fischer P, Sziegoleit A. Pancreatic elastase 1: parameter for diagnosis of chronic and acute pancreatitis. Lab Med 1992;16:427-432.
-
Clavien PA, Robert J, Meyer P, Borst F, Hauser H, Herrmann F, et al. Acute pancreatitis and normoamylasemia: not an uncommon combination. Ann Surg 1989;210:614-620.
[ISI][Medline]
[Order article via Infotrieve]
-
Lankisch PG, Schirren CA, Kunze E. Undetected fatal acute pancreatitis: why is the disease so frequently overlooked?. Am J Gastroenterol 1991;86:322-326.
[ISI][Medline]
[Order article via Infotrieve]
-
Wilson C, Imrie CW. Deaths from acute pancreatitis: why do we miss the diagnosis so frequently?. Int J Pancreatol 1988;3:273-281.
[ISI][Medline]
[Order article via Infotrieve]
-
Murata A, Ogawa M, Fujimoto K, Kitahara T, Matsuda Y, Kosaki G. Radioimmunoassay of human pancreatic elastase 1: in vitro interaction of human pancreatic elastase 1 with serum protease inhibitors. Enzyme 1983;30:29-37.
[ISI][Medline]
[Order article via Infotrieve]
-
Ohlsson K, Laurell CB. The disappearance of enzyme-inhibitor complexes from the circulation of man. Clin Sci 1976;51:87-92.
The following articles in journals at HighWire Press have cited this article:
|
|
|
|
 
G. B. Johnson, G. J. Brunn, and J. L. Platt
Cutting Edge: An Endogenous Pathway to Systemic Inflammatory Response Syndrome (SIRS)-Like Reactions through Toll-Like Receptor 4
J. Immunol.,
January 1, 2004;
172(1):
20 - 24.
[Abstract]
[Full Text]
[PDF]
|
|
|
Top
Abstract
Introduction
, lipase; 
,
amylase; 
, 96–144 h.