Read Core Topics in General & Emergency Surgery: Companion to Specialist Surgical Practice Online
Authors: Simon Paterson-Brown MBBS MPhil MS FRCS
Core Topics in General & Emergency Surgery: Companion to Specialist Surgical Practice (40 page)
Following diagnosis, usually confirmed biochemically, transabdominal US should be undertaken in all patients with acute pancreatitis within 24 hours to determine the presence or absence of gallstones. For those patients in whom alcohol is thought to be the causative agent, an accurate alcohol history must be taken, with corroborative information being obtained from other sources as necessary. In patients where alcohol is effectively excluded as a causative agent and no other biochemical cause is identified, it is reasonable to repeat transabdominal US. Endoscopic ultrasound should be considered for patients without an identified aetiology as this may demonstrate biliary microlithiasis (
Fig. 8.11
), in which case cholecystectomy should be considered.
Figure 8.11
Endoscopic ultrasound detection of gallbladder microlithiasis in a patient with recurrent pancreatitis. Transabdominal ultrasound was normal. Following laparoscopic cholecystectomy there have been no further episodes of pancreatitis.
Courtesy of Dr Ian Penman, Edinburgh
General guidelines
Guidelines for the initial management of acute pancreatitis have been published by the 1997 Santorini consensus conference,
50
the British Society of Gastroenterology,
51
the American College of Gastroenterology,
52
the Japanese Society of Abdominal Emergency Medicine
55,
56
and the International Association of Pancreatology.
57
Their recommendations are broadly similar and are described below.
As in other acute abdominal emergencies, initial therapy is aimed at adequate resuscitation (see also
Chapter 16
). Provision of oxygen to maintain arterial oxygen saturation, intravenous fluid therapy and adequate analgesia constitute the mainstays of therapy. In addition, the correction of metabolic abnormalities such as hyperglycaemia or hypocalcaemia may require administration of intravenous insulin or calcium. Patients with acute pancreatitis should be started on some form of thromboprophylaxis (see also
Chapter 14
). Antacid therapy with H
2
antagonists, proton-pump inhibitors or other gastroprotective agents may be commenced as prophylaxis against upper gastrointestinal haemorrhage in those patients with severe disease.
Following resuscitation, patients should be categorised into either prognostically mild or severe disease, allowing decisions to be taken regarding the degree of monitoring, supportive care and intervention appropriate for each patient. Subjective clinical assessment of prognosis is inaccurate and a validated prognostic scoring system should be used. A number of systems currently exist and can be divided as follows:
It should be noted that the Ranson score is based on a North American population with alcohol as the predominant aetiological agent, whereas the Glasgow score is designed for use in a typical British population of gallstone-predominant disease. Furthermore, practical confusion can result from the fact that there are at least three versions of the ‘Glasgow’ score: the original publication in 1978
58
and subsequent modifications in 1981
59
and 1984.
60
An APACHE II score of 9 or more has been validated for predicting prognostic severity in acute pancreatitis,
68
although a number of clinical trials have used lower cut-off points in order to predict severe disease, resulting in lower positive predictive values. Although an APACHE II score can be generated soon after admission, unlike the Ranson and Glasgow systems, it has the disadvantage of requiring collation of a sizeable number of variables. Comparative studies of the commonly used scoring systems have reported that no single system is superior to the others.
69
Accuracy to predict prognosis is typically 75–80%; however, a recent modification of the APACHE II system, which includes a clinical assessment of obesity (APACHE-O score), has been suggested to further improve predictive accuracy
70
and accuracy rates as high as 95% have been reported using computerised artificial neural networks.
71
Finally, it should be noted that the currently used prognostic systems are a one-off or static assessment and serial or dynamic assessment may provide a more accurate determination of outcome. Indeed, it would appear that it is not the presence of organ dysfunction at presentation that is important in determining outcome, rather it is the response to initial resuscitation. Several authors have demonstrated that clinical outcome is worst in patients with persistent organ dysfunction, as manifested by worsening or static organ dysfunction scores following initial resuscitation, compared with patients whose organ dysfunction scores improve.
72
–
74
As already mentioned, US should be performed in order to confirm or exclude the presence of gallstones. Following this, all patients with prognostically ‘severe’ acute pancreatitis should undergo contrast-enhanced CT between the third and tenth days of admission to determine the presence of pancreatic necrosis.
51
The majority of patients recover from an episode of acute pancreatitis without complication and require little intervention. However, despite intense biomedical research activity there remains a dearth of effective specific therapies for severe acute pancreatitis. The importance of vigorous volume resuscitation and careful monitoring and treatment for metabolic, respiratory, renal and cardiac complications cannot be overemphasised. However, in addition to this supportive care, the therapeutic options are limited. Several randomised controlled trials have assessed the role of early ERCP and prophylactic antibiotics; however, controversy still persists regarding the relative values of both these therapeutic options. These treatments, together with other potential therapeutic options, are reviewed below.
Early ERCP with endoscopic sphincterotomy (ES) in gallstone-induced acute pancreatitis aims to remove impacted ductal gallstones, thereby eliminating the initiating stimulus and hopefully reducing pancreatic inflammation. Three randomised trials assessing early ERCP have been published. They have produced some conflicting results and so it is worth discussing the results in some detail.
Neoptolemos et al.
75
randomised 121 patients with gallstone-induced acute pancreatitis, 53 of whom had prognostically severe disease, to either early ERCP (
n
= 59) or conventional therapy (
n =
62). Although there was no difference in mortality between the two groups (1 for ERCP vs. 5 for conventional therapy;
P
= 0.23), there was a significant reduction in total complications (17% ERCP vs. 34% conventional;
P
= 0.03) and duration of hospital stay (median 9.5, range 6–36 days ERCP vs. 17.0, range 4–74 days conventional;
P
= 0.035) in those patients with prognostically severe disease. Similarly, Fan et al.
76
randomised 195 patients with acute pancreatitis, of whom 127 had gallstone-induced disease and 81 had prognostically severe disease, to either early ERCP (
n
= 97) or conventional therapy (
n
= 98). Again, there was no significant difference in mortality rates between the two groups (5 for ERCP vs. 9 for conventional therapy;
P
= 0.276). However, there was a significant reduction in the incidence of biliary sepsis following ERCP in those with prognostically severe acute pancreatitis (0% ERCP vs. 20% conventional;
P =
0.008). Moreover, in those patients with gallstones a significant reduction in overall morbidity was observed (16% ERCP vs. 33% conventional;
P
= 0.003).
In contrast, a multicentre trial undertaken by Folsch et al.
77
randomised 238 patients with gallstone-induced acute pancreatitis but without evidence of biliary obstruction to undergo either early ERCP (
n
= 126) or conventional management (
n =
112). In total only 46 patients had prognostically severe disease. In contrast to the previous studies, the mortality rate was higher in the groups undergoing ERCP (10 for ERCP vs. 4 for conventional therapy; odds ratio (OR) 4.57, 95% confidence interval (CI) 0.67–62.7;
P =
0.16). Furthermore, ERCP was associated with an increased rate of respiratory failure (15 for ERCP vs. 5 for conventional, OR 5.16, 95% CI 1.63–22.9;
P =
0.03). Because of the trend towards increased mortality and the significantly increased rate of respiratory failure in those undergoing ERCP, the trial supervisory committee terminated the study early.
The Dutch Acute Pancreatitis Study Group reported an observational multicentre study in eight university medical centres and seven major teaching hospitals involving 153 patients with predicted severe acute biliary pancreatitis without cholangitis who were enrolled in a randomised multicentre probiotic trial and who were prospectively followed. Conservative treatment or ERCP within 72 hours after symptom onset were compared, management being based on the discretion of the treating physician. Patients without and with cholestasis (bilirubin: > 40 μmol/L and/or common bile duct dilatation) were analysed separately. In the 51% of patients with cholestasis, ERCP was associated with fewer complications than conservative treatment (25% vs. 54%,
P
= 0.020; multivariate adjusted OR 0.35, 95% CI 0.13–0.99,
P
= 0.049). Mortality was non-significantly lower after ERCP (6% vs. 15%,
P
= 0.213, multivariate adjusted OR 0.44; 95% CI 0.08–2.28,
P
= 0.330). In patients without cholestasis, ERCP was not associated with reduced complications (45% vs. 41%,
P
= 0.814; multivariate adjusted OR 1.36, 95% CI 0.49–3.76,
P
= 0.554) or mortality (14% vs. 17%,
P
= 0.754; multivariate adjusted OR 0.78, 95% CI 0.19–3.12,
P
= 0.734).
78
Rationalisation of the results of the fully reported trials suggests that early ERCP/ES is of benefit in patients with prognostically severe gallstone-induced acute pancreatitis with evidence of cholangitis or biochemical evidence of obstructive liver function tests (serum bilirubin > 90 μmol/L).
75
–
79
The administration of prophylactic antibiotics in acute pancreatitis is based on the hypothesis that the prevention of infected pancreatic necrosis would improve outcome. Numerous studies have been undertaken to investigate this hypothesis and differing results have led to much debate. Early clinical studies did not show any benefit, possibly due to inclusion of patients with a low risk of infection or because antibiotics with poor pancreatic penetration were used. In the light of these conflicting reports and more recent meta-analyses the details of these studies are worth considering further.
In a multicentre trial published in 1993,
80
74 patients with CT-proven pancreatic necrosis were randomised to receive either imipenem or no initial antibiotic therapy. Pancreatic infection was significantly reduced in the antibiotic-treated group (30% vs. 12%;
P
< 0.001). Furthermore, rates of non-pancreatic infection were also significantly reduced; however, antibiotic therapy had no effect on the development of multiple organ failure, the need for operative intervention or on the mortality rate.
A Finnish study
81
randomised 60 patients with CT-proven pancreatic necrosis to receive either prophylactic intravenous cefuroxime or antibiotics only when there were clinical indications of infection. Although there was no difference in the rates of pancreatic infections between the two groups (30% cefuroxime vs. 40% control), prophylactic cefuroxime significantly reduced the mean number of infectious complications per patient (1.0 vs. 1.8;
P
< 0.01), mainly through a marked fall in the number of urinary tract infections. There was also a significant reduction in mortality in patients treated with cefuroxime (23.3% vs. 3.3%;
P =
0.028), although this was not associated with any difference in local pancreatic infection.
Another study
82
randomised 23 patients with alcohol-induced severe acute pancreatitis to either standard medical therapy or standard medical therapy plus intravenous ceftazidime, amikacin and metronidazole. Antibiotic therapy resulted in a significant reduction in the number of patients with proven infections (0 vs. 7;
P
< 0.03), but there was no significant reduction in mortality rates. Schwarz et al.
83
reported a randomised controlled trial of 26 patients with CT-proven pancreatic necrosis in which the combination of ofloxacin and metronidazole lessened the physiological disturbance found in severe acute pancreatitis, but did not prevent or delay the development of infected pancreatic necrosis.
Subsequently two further studies were published that compared different antibiotic regimens in patients with necrotising pancreatitis. The first of these,
84
a multicentre, randomised controlled trial involving 60 patients with pancreatic necrosis, observed that intravenous imipenem was more effective than perfloxacin at reducing rates of both infected pancreatic necrosis (34% vs. 10%;
P
= 0.034) and extrapancreatic infection (44% vs. 20%;
P
= 0.059), but had no effect on the mortality rate. The second of these studies again examined the effect of imipenem
85
in a randomised trial comparing the prophylactic use of imipenem with the therapeutic use of imipenem in patients with severe acute pancreatitis and CT-proven pancreatic necrosis. Patients enrolled in the prophylactic arm received imipenem from the time of admission, whereas patients in the therapeutic arm only commenced imipenem following the development of signs suggesting the development of infected pancreatic necrosis (pyrexia, leucocytosis, raised CRP, in the absence of another source of infection). There was no significant difference in the number of patients requiring necrosectomy or in the mortality rate between the two groups.
The largest randomised placebo-controlled, double-blind trial on the use of prophylactic antibiotics was published in 2004.
86
This multicentre study randomised 114 patients to receive either ciprofloxacin in combination with metronidazole or placebo. Only patients with a predicted severe attack were included, defined either by an elevated serum C-reactive protein above 150 mg/L or by pancreatic necrosis on CT scan. No beneficial effects with regard to reduction of pancreatic infection, systemic complications or hospital mortality were demonstrated; however, a shift to open antibiotic treatment was more frequent in patients receiving placebo, suggesting a subgroup of patients may warrant antibiotic treatment.
Using a different strategy aimed at eliminating the reservoir from which bacteria arise to colonise pancreatic necrosis, Luiten et al.
87
evaluated the role of selective gut decontamination. In a multicentre trial, 102 patients with prognostically severe acute pancreatitis were randomised to either conventional therapy or conventional therapy supplemented with selective gut decontamination using oral and rectal colistin, amphotericin and norfloxacin. In addition, patients in the selective gut decontamination group received intravenous cefotaxime until aerobic Gram-negative bacteria were eliminated from the oral cavity and rectum. Selective gut decontamination reduced rates of pancreatic infection compared with the control group (38% vs. 18%;
P =
0.03). Although there was a reduction in mortality from 35% in the control group to 22% in the selective decontamination group, it did not reach significance (
P
= 0.19). However, when allowing for differences in disease severity, multivariate analysis suggested a significant survival benefit following gut decontamination (
P
= 0.048). Although the results of this trial appear promising, selective gut decontamination is not regarded as a standard for the prevention of pancreatic infection because of the difficulties of drug administration in the intensive care setting.
The inconsistencies among these trials may result from the relatively small number of patients per study and the limited statistical power to detect such differences. There is also considerable heterogeneity in patient selection and the antibiotics used. Other differences include variety in the non-antibiotic treatments, such as fluid resuscitation, enteral nutrition and timing of surgical intervention. Although many published guidelines recommend the use of antibiotics in patients with severe acute pancreatitis, none of the published randomised trials is in itself of sufficient power to mandate antibiotic prophylaxis.
In an attempt to strengthen the evidence regarding the use of antibiotic prophylaxis in acute pancreatitis, several meta-analyses have been performed. In 1998, Golub et al.
88
undertook a meta-analysis of all trials that had been published up to that point and reported that in patients with severe disease, the risk of death fell from 18.2% in patients not receiving antibiotic prophylaxis to 5.3% in those receiving prophylaxis (log odds ratio − 0.32 to − 2.44;
P
= 0.008). Similarly, Sharma and Howden
89
undertook a meta-analysis of three of the trials utilising broad-spectrum antibiotics in patients with severe disease. They suggested that antibiotic prophylaxis resulted in a 12.3% (95% CI 2.7–22%) absolute risk reduction in mortality rate, with eight (95% CI 5–37) patients needing treatment to prevent one death.
More recent meta-analyses, however, do not support the use of prophylactic antibiotics in patients with severe acute pancreatitis. In 2006, Mazaki et al.
90
analysed six trials and concluded that prophylactic antibiotics were not associated with a statistically significant reduction in infected necrosis (relative risk 0.77, CI 0.54–1.12;
P =
0.173), non-pancreatic infections (relative risk 0.71, CI 0.32–1.58), surgical intervention (relative risk 0.78, CI 0.55–1.11) or mortality (relative risk 0.78, CI 0.44–1.39), but were associated with a significant reduction in hospital stay. In 2007, de Vries et al.
91
undertook a meta-analysis on six trials and reported that prophylactic antibiotics had no significant effect on infection of pancreatic necrosis (absolute risk reduction 0.055, CI − 0.084 to 0.194) or mortality (absolute risk reduction 0.058, CI − 0.017 to 0.134).
Furthermore, there is concern that the use of potent antibiotics may increase the rate of fungal-associated infected pancreatic necrosis and in turn adversely affect outcome. The use of antibiotic therapy has been shown to alter the organisms involved in the development of infected pancreatic necrosis,
92
and it has been suggested that higher rates of fungal infection have been reported in recent studies.
93
There is currently debate as to whether fungal-associated infected pancreatic necrosis is associated with increased mortality; however, a recent randomised trial of antifungal therapy in patients with severe acute pancreatitis showed a definite reduction in the incidence of fungal infection in the treatment group.
94
In the light of all these conflicting studies, it is clear that further trials are needed to clarify the role of antibiotic and antifungal prophylaxis in acute pancreatitis. At present, a pragmatic approach is that routine administration of prophylactic antibiotics to all patients with predicted severe acute pancreatitis is not indicated, but should be considered in those patients with evidence of pancreatic necrosis who appear septic (with leucocytosis, fever and/or organ failure). If blood and other cultures are subsequently found to be negative and no source of infection is identified, antibiotics should be discontinued. If positive microbiological cultures are obtained, appropriate antibiotics should be continued based on microbiological sensitivities.
88
–
91