Br. J. Surg. 1991, Vol. 78, October, 1230-1 234
J. E. Dominguez-MuAoz*, F. Carballo*, M. J. Garcia?, J. M. de Diegot, L. Rabagog, M. A. Simon7 and J. de la Morena* ACAD Spanish Multicenter Group, *Department of Internal Medicine, University Hospital, Guadalajara, University of Alcala de Henares; ?Department of Gastroenterology, General Hospital, Albacete; $Department of Internal Medicine, Virgen de la Concha' Hospital, Zamora; $Department of Gastroenterology, 'Sever0 Ochoa' Hospital, Legads, Madrid; YDepartment of Gastroenterology, 'San Millan' Hospital, Logrofio, Spain Correspondence to: Dr Fernando Carballo-Alvarez, Servicio de Medicina Interna, Hospital Universitario General, C/Donante de sangre sin, 19002 Guadalajara, Spain
Clinical usefulness of polymorphonuclear elastase in predicting the severity of acute pancreatitis: results of a multicentre study The usefulness and clinical applicability of quantitative plasma polymorphonuclear elastase determinations in the diagnosis of the severity of acute pancreatitis was analysed in a multicentre study and was compared with the usualprognostic systems of Ranson and Osborne et al. The study comprised 182 patients, 1.54 with a mild episode of acute pancreatitis and 28 with a severe episode, defined by the development of major complications or a fatal outcome. In the severe cases neutrophilic elastase reached signijicantly higher values than in mild cases (P < 0.001) by the time the patient was admitted (2-12 h after the onset of the disease), reflecting considerable leucocyte activation. The sensitivity and specijicity of this test are therefore greater than 90per cent, with a positive severity predictive value of almost 80per cent at the time of admission and 97 per cent after 24 h, and a negative predictive value of approximately 98 per cent. In addition to requiring 48 h f o r evaluation, the usual prognostic systems show a sensitivity of 77-8.5 per cent, a speciJicity of 70-77 per cent, a positive predictive value of 40-48 per cent, and a negative predictive value of 92-9.5per cent, clearly lower than those obtained with leucocyte elastase. Polymorphonuclear elastase is therefore a very early and reliable marker in the diagnosis of the severity of acute pancreatitis, in addition to being easily adaptable to the routine of any hospital laboratory.
The phagocytic mononuclear and polymorphonuclear systems are being increasingly implicated in the production of extensive tissue damage in many inflammatory diseases, and may determine the severity of the clinical course in these conditions'-3. The capacity of neutrophils to cause tissue damage has been linked with the great harmful potential of their lysosomal proteases, mainly elastase4. This harmful action is facilitated by the production by the phagocytic cell itself of oxygen-containing metabolites that cause oxidative inactivation of a series of key anti-proteases (essentially a,-protease inhibitor)'^^ in an extracellular environment in which there is little superoxide dismutase or catalase activity6. Furthermore, polymorphonuclear elastase, together with the remaining neutrophilic proteases, is capable of activating cascades of the complement, kinin, coagulation and fibrinolytic systems and arachidonic acid, which are involved in the development of multisystem failure7-l0. This complex mechanism could account for the clinical picture of acute pancreatitis and for the high morbidity and mortality of severe episodes". This attractive hypothesis led us to measure the levels of circulating polymorphonuclear elastase in patients with acute pancreatitis and to compare this test with the prognostic criteria usually used for the prediction of ~everity".'~.
together with a clinical course consistent with acute pancreatitis which could not be explained by any other pathological process. The time between the onset of symptoms and the patient's admission to hospital varied between 2 and 12 h. Severity of pancreatitis was defined according to the development of one or more major local or systemic complications of the disease (Table I ). Local complications were diagnosed by imaging techniques (ultrasonography, computed tomography) and/or surgical findings. Because respiratory alterations which do not require any specific therapeutic management often occur in the course of acute pancreatitis, only an arterial oxygen tension below 60 mmHg (8 kPa) combined with acute changes in clinical signs or on chest radiographs was accepted for the diagnosis of respiratory insufficiency requiring specific treatment (oxygen therapy, assisted respiration, etc.).Criteria for diagnosing other local and systemic complications are those generally accepted'".
Materials and methods
Local complications
Systemic complications
Patients A detailed protocol was applied prospectively to a total of 182 patients
Necrosis Haemorrhage Abscess Pseudocyst Phlegmon
Shock Sepsis Pulmonary insufficiency Cardiac insufficiency Renal insufficiency Severe coagulopathy Pancreatic encephalopathy
diagnosed as having acute pancreatitis over a period of 14 months. The diagnosis was based on the presence of a typical clinical picture together with elevated serum levels of total amylase, pancreatic amylase and lipase, at no less than twice the upper limit of the normal range in each case. Subsequent morphological confirmation was also required,
1230
Methods
All patients underwent the necessary investigations for the prognostic criteria of the New York group" and the Glasgow groupI3, according to which an episode of acute pancreatitis is predicted to be severe on the basis of three or more positive criteria.
Table 1 Major complications of acute pancreatitis considered in the definition of severe disease
0007- I323/9 I / I012%05
6 ' 199I Butterworth-Heinernann Ltd
Polymorphonuclear elastase in acute pancreatitis: J. E. Dominguez-Munoz et al.
Likewise the post-test odds in favour of severity (OddsFI were calculated, which indicate the probability that a patient will present with severe acute pancreatitis, taking 1 as the probability that the episode will be mild and starting from a pre-test OddsF of 0.-’5:l. The optimal cut-off values were calculated from the receiver-operator characteristic (ROC) curves for each test. These show the difkrcnf sensitivity and specificity obtained according to the cut-off value considered to predict severity. The optimal cut-off is the value with which the test shows the best sensitivity/specificity equilibrium in the diagnosis of severity.
Results Patient characteristics
6
a
0
0.8
I
02
0
I
I
0.8
1.0
I
I
0.4 0.6 I - Specificity
loo 250
The mean(s.d.) age of the patients was 60(18.8) years; 94 (51.6 per cent) were men and 88 (48.4 per cent) women. The distribution by aetiology showed a marked predominance of biliary pancreatitis (57.1 per cent) over alcoholic pancreatitis (17.0 per cent), mixed biliary and alcoholic cases (4.4 per cent). other forms (2.8 per cent) and idiopathic pancreatitis (18.7 per cent). Mild episodes were present in 154 patients (84.6 per cent), while the remaining 28 (15.4 per cent) had severe episodes. Overall, seven patients died (3.8 per cent of the total. 25 per cent of the severe episodes). There was no statistically significant correlation between morbidity or mortality and advanced age, sex distribution or associated aetiological factors. Sixteen patients (8.8 per cent) developed local complications, mainly pancreatic necrosis (7.1 per cent), while 26 (14.3 per cent) presented with systemic complications, mainly respiratory failure (10.4 per cent) and sepsis (9.9 per cent). Prognostic criteria
b
02t 01 0
I
I
I
I
I
0 2
04
06
0 8
10
I-Specif icity
Figure 1 Receiver-operator characteristic ( R O C ) curves in diagnosing rhe severiry of acute pancreatitis. a Prognostic systems of Ranson (0) and Osborne (0); b polymorphonuclear elastase on admission (0)and The different cut-offs considered are shown. The optimal after 24 h (0). cut-off point is dejined as the nearest to the upper left corner (sensitivity= I , I-spec$city=O)
Circulating polymorphonuclear elastase was determined quantitatively in plasma samples on admission and on days 1, 2, 3, 5, 7, 10, and 15 by enzyme immunoanalysis (PMN elastase, 2 h version, Merck Immunoassay. Darmstadt, Germany ). This method measures the concentration of polymorphonuclear elastase-a,-protease inhibitor complexes. The normal range for this enzyme, calculated from the analysis of 100 blood samples from donors assumed to be healthy, is 7 4 3 /%/I. Statistical analysis of the quantitative data was by Student’s t-test and for the qualitative data by ,yz test with Yates’ correction in 2 x 2 contingency tables. To quantify the ability of polymorphonuclear elastase and of the multifactorial scoring systems to predict severity, sensitivity (proportion of correctly classified severe episodes), specificity (proportion of correctly classified mild episodes) and overall reliability (proportion of episodes correctly classified according to severity) were determined. The positive and negative severity predictive values were calculated by applying Bayes’ theorem, using a prevalence of 0.2 of severe cases and 0.8 of mild cases, from an epidemiological study conducted in our department (unpublished work). This is a similar distribution of severity to other large series of patients who had not undergone previous election'^.'^. Starting from the initial probability of 20 per cent that an episode of acute pancreatitis will be severe, the positive predictive value defines the probability that a patient with a positive test result will develop a severe attack while the negative predictive value quantifies the probability that a patient with a negative test result will have a mild episode (previous probability 80 per cent).
Br. J. Surg., Vol. 78, No. 10, October 1991
The patients with severe episodes clearly had a larger number of positive criteria than those with mild episodes, with reference to both Ranson’s system (3.7( 1.20) uersiis l . g ( l . 3 3 ) ) and the modified Glasgow criteria (3.4(1.10) cersus 1.5(1.31 ) ) . I n both cases the difference was statistically significant ( P < 0.001 ). Statistical analysis of the results obtained by progressively increasing the minimum number of criteria required for an episode of pancreatitis to be considered severe from 3-6 yielded the ROC curve shown in Figure l a . This shows how optimal reliability is obtained by considering three criteria as the cut-off, agreeing with the original descriptions. The corresponding results for sensitivity, specificity, positive and negative predictive values and overall reliability obtained with these optimal cut-offs are shown in Table 2. The odds in favour of severity increase from a pre-test value of 0.25:l to the post-test OddsF of 0.66:l according to Ranson’s system and to 0.92: 1 according to Osborne’s system. Neither of the prognostic systems showed statistically significant differences between the patients with a fatal outcome and those with the severe form who survived (x’ =2.853 and x2 =2.751, respectively). Polymorphonuc fear elastase The plasma concentration of polymorphonuclear elastase is far higher from the time of admission in patients whose course proves to be unfavourable (severe disease), with a peak at 24 h
Table 2 Sensitivity, speciJicity, positive and negarire predrcrirr
ru/iic\
and overall reliability of the Ranson and Glusgow proynosrii. .\j’.\iwi,v and polymorphonuclear elastase on admission and afier -14 11. in t h 1 2 diagnosis of severity of acute pancreatitis
Sensitivity Specificity Positive predictive value Negative predictive value Reliability
Ranson
Glasgow
Elastase on admission
Elastiisc at 14 h
76.9 10.3 39.6
84.6 77 47.9
92.6 94 79.4
92.9 99.3 97.1
92.4
95.2
98.1
98.2
71.3
78.2
93.8
96.3
1231
Polymorphonuclear elastase in acute pancreatitis: J. E. Dominguez-Muiioz et al.
Discussion
0
0
800-
0
2
700-
.K
600-
5
g
500-
r
,"0
400-
+
c
gu
300200-
-
a
01
I
1
I
I
I
0
2
4
6
8
I
1
12
14
I
10 Day of evolution
I
16
Figure 2 Plasma concentration (mean +s.e.m.) of polymorphonuclear elastase in patients with severe (upper line) and mild (lower line) acute pancreatitis. ---, Upper normal concentration. Days 0-10, P
J. E. Dominguez-MuAoz*, F. Carballo*, M. J. Garcia?, J. M. de Diegot, L. Rabagog, M. A. Simon7 and J. de la Morena* ACAD Spanish Multicenter Group, *Department of Internal Medicine, University Hospital, Guadalajara, University of Alcala de Henares; ?Department of Gastroenterology, General Hospital, Albacete; $Department of Internal Medicine, Virgen de la Concha' Hospital, Zamora; $Department of Gastroenterology, 'Sever0 Ochoa' Hospital, Legads, Madrid; YDepartment of Gastroenterology, 'San Millan' Hospital, Logrofio, Spain Correspondence to: Dr Fernando Carballo-Alvarez, Servicio de Medicina Interna, Hospital Universitario General, C/Donante de sangre sin, 19002 Guadalajara, Spain
Clinical usefulness of polymorphonuclear elastase in predicting the severity of acute pancreatitis: results of a multicentre study The usefulness and clinical applicability of quantitative plasma polymorphonuclear elastase determinations in the diagnosis of the severity of acute pancreatitis was analysed in a multicentre study and was compared with the usualprognostic systems of Ranson and Osborne et al. The study comprised 182 patients, 1.54 with a mild episode of acute pancreatitis and 28 with a severe episode, defined by the development of major complications or a fatal outcome. In the severe cases neutrophilic elastase reached signijicantly higher values than in mild cases (P < 0.001) by the time the patient was admitted (2-12 h after the onset of the disease), reflecting considerable leucocyte activation. The sensitivity and specijicity of this test are therefore greater than 90per cent, with a positive severity predictive value of almost 80per cent at the time of admission and 97 per cent after 24 h, and a negative predictive value of approximately 98 per cent. In addition to requiring 48 h f o r evaluation, the usual prognostic systems show a sensitivity of 77-8.5 per cent, a speciJicity of 70-77 per cent, a positive predictive value of 40-48 per cent, and a negative predictive value of 92-9.5per cent, clearly lower than those obtained with leucocyte elastase. Polymorphonuclear elastase is therefore a very early and reliable marker in the diagnosis of the severity of acute pancreatitis, in addition to being easily adaptable to the routine of any hospital laboratory.
The phagocytic mononuclear and polymorphonuclear systems are being increasingly implicated in the production of extensive tissue damage in many inflammatory diseases, and may determine the severity of the clinical course in these conditions'-3. The capacity of neutrophils to cause tissue damage has been linked with the great harmful potential of their lysosomal proteases, mainly elastase4. This harmful action is facilitated by the production by the phagocytic cell itself of oxygen-containing metabolites that cause oxidative inactivation of a series of key anti-proteases (essentially a,-protease inhibitor)'^^ in an extracellular environment in which there is little superoxide dismutase or catalase activity6. Furthermore, polymorphonuclear elastase, together with the remaining neutrophilic proteases, is capable of activating cascades of the complement, kinin, coagulation and fibrinolytic systems and arachidonic acid, which are involved in the development of multisystem failure7-l0. This complex mechanism could account for the clinical picture of acute pancreatitis and for the high morbidity and mortality of severe episodes". This attractive hypothesis led us to measure the levels of circulating polymorphonuclear elastase in patients with acute pancreatitis and to compare this test with the prognostic criteria usually used for the prediction of ~everity".'~.
together with a clinical course consistent with acute pancreatitis which could not be explained by any other pathological process. The time between the onset of symptoms and the patient's admission to hospital varied between 2 and 12 h. Severity of pancreatitis was defined according to the development of one or more major local or systemic complications of the disease (Table I ). Local complications were diagnosed by imaging techniques (ultrasonography, computed tomography) and/or surgical findings. Because respiratory alterations which do not require any specific therapeutic management often occur in the course of acute pancreatitis, only an arterial oxygen tension below 60 mmHg (8 kPa) combined with acute changes in clinical signs or on chest radiographs was accepted for the diagnosis of respiratory insufficiency requiring specific treatment (oxygen therapy, assisted respiration, etc.).Criteria for diagnosing other local and systemic complications are those generally accepted'".
Materials and methods
Local complications
Systemic complications
Patients A detailed protocol was applied prospectively to a total of 182 patients
Necrosis Haemorrhage Abscess Pseudocyst Phlegmon
Shock Sepsis Pulmonary insufficiency Cardiac insufficiency Renal insufficiency Severe coagulopathy Pancreatic encephalopathy
diagnosed as having acute pancreatitis over a period of 14 months. The diagnosis was based on the presence of a typical clinical picture together with elevated serum levels of total amylase, pancreatic amylase and lipase, at no less than twice the upper limit of the normal range in each case. Subsequent morphological confirmation was also required,
1230
Methods
All patients underwent the necessary investigations for the prognostic criteria of the New York group" and the Glasgow groupI3, according to which an episode of acute pancreatitis is predicted to be severe on the basis of three or more positive criteria.
Table 1 Major complications of acute pancreatitis considered in the definition of severe disease
0007- I323/9 I / I012%05
6 ' 199I Butterworth-Heinernann Ltd
Polymorphonuclear elastase in acute pancreatitis: J. E. Dominguez-Munoz et al.
Likewise the post-test odds in favour of severity (OddsFI were calculated, which indicate the probability that a patient will present with severe acute pancreatitis, taking 1 as the probability that the episode will be mild and starting from a pre-test OddsF of 0.-’5:l. The optimal cut-off values were calculated from the receiver-operator characteristic (ROC) curves for each test. These show the difkrcnf sensitivity and specificity obtained according to the cut-off value considered to predict severity. The optimal cut-off is the value with which the test shows the best sensitivity/specificity equilibrium in the diagnosis of severity.
Results Patient characteristics
6
a
0
0.8
I
02
0
I
I
0.8
1.0
I
I
0.4 0.6 I - Specificity
loo 250
The mean(s.d.) age of the patients was 60(18.8) years; 94 (51.6 per cent) were men and 88 (48.4 per cent) women. The distribution by aetiology showed a marked predominance of biliary pancreatitis (57.1 per cent) over alcoholic pancreatitis (17.0 per cent), mixed biliary and alcoholic cases (4.4 per cent). other forms (2.8 per cent) and idiopathic pancreatitis (18.7 per cent). Mild episodes were present in 154 patients (84.6 per cent), while the remaining 28 (15.4 per cent) had severe episodes. Overall, seven patients died (3.8 per cent of the total. 25 per cent of the severe episodes). There was no statistically significant correlation between morbidity or mortality and advanced age, sex distribution or associated aetiological factors. Sixteen patients (8.8 per cent) developed local complications, mainly pancreatic necrosis (7.1 per cent), while 26 (14.3 per cent) presented with systemic complications, mainly respiratory failure (10.4 per cent) and sepsis (9.9 per cent). Prognostic criteria
b
02t 01 0
I
I
I
I
I
0 2
04
06
0 8
10
I-Specif icity
Figure 1 Receiver-operator characteristic ( R O C ) curves in diagnosing rhe severiry of acute pancreatitis. a Prognostic systems of Ranson (0) and Osborne (0); b polymorphonuclear elastase on admission (0)and The different cut-offs considered are shown. The optimal after 24 h (0). cut-off point is dejined as the nearest to the upper left corner (sensitivity= I , I-spec$city=O)
Circulating polymorphonuclear elastase was determined quantitatively in plasma samples on admission and on days 1, 2, 3, 5, 7, 10, and 15 by enzyme immunoanalysis (PMN elastase, 2 h version, Merck Immunoassay. Darmstadt, Germany ). This method measures the concentration of polymorphonuclear elastase-a,-protease inhibitor complexes. The normal range for this enzyme, calculated from the analysis of 100 blood samples from donors assumed to be healthy, is 7 4 3 /%/I. Statistical analysis of the quantitative data was by Student’s t-test and for the qualitative data by ,yz test with Yates’ correction in 2 x 2 contingency tables. To quantify the ability of polymorphonuclear elastase and of the multifactorial scoring systems to predict severity, sensitivity (proportion of correctly classified severe episodes), specificity (proportion of correctly classified mild episodes) and overall reliability (proportion of episodes correctly classified according to severity) were determined. The positive and negative severity predictive values were calculated by applying Bayes’ theorem, using a prevalence of 0.2 of severe cases and 0.8 of mild cases, from an epidemiological study conducted in our department (unpublished work). This is a similar distribution of severity to other large series of patients who had not undergone previous election'^.'^. Starting from the initial probability of 20 per cent that an episode of acute pancreatitis will be severe, the positive predictive value defines the probability that a patient with a positive test result will develop a severe attack while the negative predictive value quantifies the probability that a patient with a negative test result will have a mild episode (previous probability 80 per cent).
Br. J. Surg., Vol. 78, No. 10, October 1991
The patients with severe episodes clearly had a larger number of positive criteria than those with mild episodes, with reference to both Ranson’s system (3.7( 1.20) uersiis l . g ( l . 3 3 ) ) and the modified Glasgow criteria (3.4(1.10) cersus 1.5(1.31 ) ) . I n both cases the difference was statistically significant ( P < 0.001 ). Statistical analysis of the results obtained by progressively increasing the minimum number of criteria required for an episode of pancreatitis to be considered severe from 3-6 yielded the ROC curve shown in Figure l a . This shows how optimal reliability is obtained by considering three criteria as the cut-off, agreeing with the original descriptions. The corresponding results for sensitivity, specificity, positive and negative predictive values and overall reliability obtained with these optimal cut-offs are shown in Table 2. The odds in favour of severity increase from a pre-test value of 0.25:l to the post-test OddsF of 0.66:l according to Ranson’s system and to 0.92: 1 according to Osborne’s system. Neither of the prognostic systems showed statistically significant differences between the patients with a fatal outcome and those with the severe form who survived (x’ =2.853 and x2 =2.751, respectively). Polymorphonuc fear elastase The plasma concentration of polymorphonuclear elastase is far higher from the time of admission in patients whose course proves to be unfavourable (severe disease), with a peak at 24 h
Table 2 Sensitivity, speciJicity, positive and negarire predrcrirr
ru/iic\
and overall reliability of the Ranson and Glusgow proynosrii. .\j’.\iwi,v and polymorphonuclear elastase on admission and afier -14 11. in t h 1 2 diagnosis of severity of acute pancreatitis
Sensitivity Specificity Positive predictive value Negative predictive value Reliability
Ranson
Glasgow
Elastase on admission
Elastiisc at 14 h
76.9 10.3 39.6
84.6 77 47.9
92.6 94 79.4
92.9 99.3 97.1
92.4
95.2
98.1
98.2
71.3
78.2
93.8
96.3
1231
Polymorphonuclear elastase in acute pancreatitis: J. E. Dominguez-Muiioz et al.
Discussion
0
0
800-
0
2
700-
.K
600-
5
g
500-
r
,"0
400-
+
c
gu
300200-
-
a
01
I
1
I
I
I
0
2
4
6
8
I
1
12
14
I
10 Day of evolution
I
16
Figure 2 Plasma concentration (mean +s.e.m.) of polymorphonuclear elastase in patients with severe (upper line) and mild (lower line) acute pancreatitis. ---, Upper normal concentration. Days 0-10, P
