World J. Surg. 18, 906-911, 1994

WORLD Journal of

SURGERY 9 1994 by the Soci6t6 internationale de Chirurgie

Necrotizing Pancreatitis: Operating for Life Brian J. Miller, F.R.A.C.S., 1 A l a n H e n d e r s o n , F.R.C.A.P., 2 Russell W. Strong, F.R.C.A.S., 3 G e o r g e A. Fielding, F.R.A.C.S., 4 A n g e l a M. DiMarco, M.B., B.S., 4 Barry S. O'Loughlin, F.R.C.A.S. 4 tDepartment of Surgery, Princess Alexandra Hospital, Brisbane, Queensland 4102, Australia 2Intensive Care Unit, Princess Alexandra Hospital, Brisbane, Queensland 4102, Australia 3Department of Surgery, University of Queensland, Brisbane, Queensland 4006, Australia 4Department of Surgery, Royal Brisbane Hospital, Brisbane, Queensland 4006, Australia Abstract. Infected necrotizing panereatitis is the most fulminant variety of this disease. Colonic involvement and retroperitoneal fasciitis are particularly lethal. The reported mortality is up to 50%. The purpose of this study is to review our combined experience at the Princess Alexandra Hospital and the Royal Brisbane Hospital, Brisbane, to determine whether patient survival was related to a particular etiology, treatment, or complication. All patients treated since 1986 with infected pancreatitis who required surgical necrosectomy and then ventilation in the intensive care unit (ICU) were studied. There were 48 patients so managed. The median age of survivors was 52 years, and for those who died it was 64 years (p = 0.001). The etiology was gallstones in 22 and alcoholism in 12. Of the alcoholics, 11 survived and 1 died. Of the patients with gallstones, 13 survived and 9 died. There was an overall mortality of 31%. Survivors were in hospital for a median of 73 days, whereas deaths occurred after a median of 35 days (p = 0.04). Seven patients underwent hemofiltration; five survived, and two died. N-Acetylcysteine has been used in four patients, of whom three survived and one died. The abdomen was left open in 38 patients and kept closed in 10. Although Ranson's criteria at admission to the ICU did not predict survival, it was found that the median APACHE II score in survivors was significantly lower than in those who died (p = 0.025). However the need for colectomy or the finding of retroperitoneal fasciitis in seven patients caused a significantly higher mortality, which was not predicted by Ranson's criteria or APACHE II scores (p = 0.007). Death was due to overwhelming sepsis in most cases, although 47% of patients who died had also suffered major bleeding or fistulas. Nonparametric, box plot analysis shows the following trends: (1) Alcohol was not the most common cause of necrotizing pancreatitis, nor did it carry the highest mortality. (2) Tissue adjacent to the pancreas progressively necrosed over days or weeks. (3) Low initial APACHE II scores were frequently found in patients who ultimately died with colonic necrosis and retroperitoneal fasciitis. (4) Survivors tended to be treated by open laparostomy sooner, have longer periods in hospital, and be significantly younger. In conclusion, patients do best with early, open, repeated surgical d~bridement of the retroperitoneum for what appears to be an ongoing process.

Infected necrotizing pancreatitis is the most serious form of pancreatitis. The mortality supersedes both edematous and hemorrhagic pancreatitis. Septicemia, hemorrhage, and fistula formation are the most frequent life-threatening complications. Necrotizing pancreatitis is easily distinguished at laparotomy from edematous and hemorrhagic pancreatitis. There is a peripancreCorrespondence to: B.J. Miller, M.D.

atic slough that is black and putty-like in consistency. Associated colonic necrosis carries a mortality of more than 50% [1], and necrotizing retroperitoneal fasciitis was uniformly fatal in one report [2]. The timing and extent of operation for pancreatitis is frequently controversial, and the old adage "too much too early, too little too late" still is relevant. This severe disease is relatively uncommon although it consumes a large proportion of the resources in an intensive care unit (ICU) when considered on a per capita basis [3]. The objective for this study was to determine whether patient survival at the two major Brisbane hospitals had any relation to the etiology and apparent severity of the necrotizing pancreatitis, its treatment, or the complications that supervened. Methods

The clinical records of all patients admitted to the ICU at Princess Alexandra Hospital and the Royal Brisbane Hospital between January 1986 and June 1993 with infected necrotizing pancreatitis who required surgical treatment by necrosectomy and mechanical ventilation were examined. In all cases peripancreatic necrosis was present at the initial laparotomy. Further surgery was undertaken as required to remove necrotic retroperitoneal tissue or for local complications. Records were made of the patients' demographics, the etiology of the pancreatitis, the scoring on modified Ranson criteria as published by Blamey et al. [4], and an APACHE II score [5] at admission to the ICU. Of the 11 original prognostic indices described by Ranson, Blarney found that 9 were the most predictive, the base deficit and estimated third space volume being omitted. The methods and frequency of operation were documented. Necrosectomy referred to removal of necrotic pancreas and retroperit0neal peripancreatic slough, rather than a formal pancreatectomy extending beYond the limits of obvious necrosis, within anatomic boundaries. The use of special therapeutic measures such as hemofiltration, N-acetylcysteine, and methylprednisolone was recorded. The number of days spent in hospital, the infecting organisms, and the final outcome were also noted. Patients who reached the ICU with acute pancreatic disease but

Miller et al.: Necrotizing Panereatitis

907

80

75 7O 65 60

A 55 G E so

l

45 4O 35 3O

.-!-

'25

Survivors

Deaths

Fig. 1. Age-related mortality. Notched box and whisker plots show no overlap of notches, indicating a significant difference in age of survivors compared to patients who died, at the 95% confidence level. did not have positive cultures with necrosis were excluded as follows: Five patients died in the ICU with hemorrhagic pancreatitis without undergoing operation. Nine patients with edematous pancreatitis went to the ICU and were operated on but not ddbrided. Thirteen patients had pancreatic pseudocysts; they were treated in the ICU but simply drained. This survey, was performed retrospectively, and patient charts were identified using the ICU computerized data banks at the Royal Brisbane and Princess Alexandra Hospitals. Statistical analysis was done using the SPSS computer package. Results

There were 48 patients in our series, with a male/female ratio of 1.5:1.0. The age range was 28 to 76 years with a median age of 55 years. The mortality among patients under age 55 was 14%, and the mortality for those over age 55 was 46%. The mortality for the entire series was 31%. The median age of the survivors was 52 years, and the median age of those dying was 64 years. On nonparametric analysis using notched box and whisker plots, these figures are significantly different (p = 0.001) (Fig. 1). The etiology in these 48 patients was gallstones in 22, alcoholism in 12, exploration of the common bile duct in 3, endoscopic retrograde cholangiopancreatography (ERCP) in 3, and upper abdominal surgery in 4; in 3 patients the etiology was unclear. There was one case associated with organophosphorus poisoning. We found that gallstones as an etiology carried a higher mortality than alcoholism. Of the 22 patients with gallstones, 13 survived and 9 died. Of the 12 patients with alcoholic necrotizing pancreatitis 11 survived and only 1 died. Overall, 33 patients survived, and 15 died. Nine deaths occurred before 1990, and six during or after 1990. The diagnosis was made by abdominal computed tomographic (CT) scan with oral and intravenous contrast in all but 10 patients, who were diagnosed at laparotomy (Fig. 2). The 33 survivors were in hospital for a median of 73 days, whereas the 15 patients who died did so after a median of 35 days. This difference shows a trend toward longer hospital stays for survivors (p = 0.04).

Fig. 2. Contrast CT scan of the upper abdomen of a recent patient in our series shows the nonenhancing swollen pancreas typical of necrotizing pancreatitis with scattered free gas visible in the peripancreatic tissues. Table 1. Disease severity scores and outcome.

Measurement

Modified Ranson criteria

Survivors (n = 33) Deaths (n = 15)

4.0 4.0

APACHE II score

Median age (years)

11.0 19.0 (p = 0.025)

52 64 (p = 0.001)

Eight patients had isolated major intraabdominal bleeds, of whom six survived and two died. All of the four patients whose course was complicated with gastrointestinal fistulas alone survived. Nine patients had both major bleeding and fistulas, and of these patients only three survived. The cause of death was most often overwhelming sepsis. Only one patient died directly as a result of uncontrollable hemorrhage, although 47% of those dying had experienced episodes of major bleeding and fistulas. Five patients required colectomy during the course of peripancreatic ddbridement; and three of them died. Three patients with retroperitoneal fasciitis, including one who underwent colectomy, all died. Colectomy and retroperitoneal fasciitis were associated with a high probability of dying (p = 0.007). The Ranson criteria and A P A C H E II scores were estimated on the day of entry to the ICU. Median values for these disease severity scores and their correlation with age and mortality are shown in Table 1. Although indicating the severity of the illness in both groups, the Ranson criteria did not significantly differentiate between patients who ultimately died and those who survived, using Mann-Whitney U nonparametric testing, whereas there was a statistical difference found for the A P A C H E II scores of the same two groups (p = 0.025). Similarly, older age significantly correlated with a fatal outcome (p = 0.001). Except in one case, A P A C H E II scores below 15 at admission among the 15 patients

908

who died were found only in those who later sustained virtually irretrievable complications, such as colonic necrosis and retroperitoneal fasciitis. Conversely, those who ultimately survived with initially high APACHE II scores did not suffer these particular complications later. Most patients had many operations. The range was 2 to 45, with a median of seven abdominal operations per patient. The abdomen was ultimately left open in 38 patients. Nine of ten patients in whom the abdomen was kept dosed had low initial A P A C H E II scores and survived. When APACHE II scores were high (> 15), there was a trend toward improved survival with early open laparostomy: The survivors' abdomens were left open after a median of 12 days, whereas in those who died there was a median 18-day period to open abdomen. The first operation was done a median of 7 days after admission. Sepsis was a universal problem by definition in this study, and multiple organisms were cultured including most frequently Escherichia coli and Klebsiella. In addition, multiply resistant Staphylococcus aureus (MRSA) was present in 23 of 48 patients. Thirteen patients were infected with fungus, eight of whom survived; five died. Hemofiltration has been used since 1990 for seven patients: Five have survived, and two have died. N-Acetylcysteine has been used in four patients, of whom three survived and one died. Methylprednisolone has been used in i g doses intravenously in a few patients recently at admission to the ICU as well as preoperatively. No firm conclusions are yet possible regarding these three adjunctive measures. Discussion

This study included only infected pancreatic necrosis, a condition recognized in the Atlanta classification to be three times more likely to be fatal than sterile necrosis [6]. The CT criteria for significant pancreatic necrosis have been established as nonenhancement involving more than 30% of the gland, although in 20% of our series the diagnosis was made at laparotomy [7]. Transcutaneous needle aspiration and culture has been determined as a safe, accurate means of diagnosing infected necrosis [6]. In an earlier study from Queensland, Fielding emphasized the danger of pancreatitis induced by gallstones or operation. Repeated d6bridement of necrotic pancreas and retroperitoneal slough was advocated, as was marsupialization [8]. The therapeutic options for this serious condition revolve around whether to leave the patient open with or without mesh support or to close the abdomen. Should the necrotic process be dealt with by repeated reoperation and d6bridement with packing or by initial d~bridement and then continuous irrigation? What is the place of hemofiltration, N-acetylcysteine, and methylprednisolone? Another major alternative is to wait until the necrotic process has localized in one portion of the pancreas and then approach it through or below the twelfth rib for localized d6bridement. This tactic was reported recently in 20 patients with a mortality rate of 20% [9]. In our series it was done twice, and both patients did well. However, the question arises as to whether the disease in these cases was initially less generalized than in other patients. Our impression was that the necrotic process was not static and appeared to evolve over a period of days or weeks. It is important to make the distinction between necrosectomy and pancreatec-

World J. Surg. Vol. 18, No. 6, Nov./Dec. 1994

tomy. It is often difficult to determine what is pancreas and if it is necrotic. The chances of identifying and controlling the necrotic process thus appears to be better by repeated exploratory operation via anterior laparotomy [10]. We found that early and repetitive d~bridement of the process until extension appeared to have ceased, followed by closure of the abdomen over irrigating drains, appeared to be the most effective method of treatment. We used dilute sodium hypochlorite to soak large packs, which were placed in the bed of the pancreas and then changed every 1 or 2 days in the operating theater until the area was clean and granulating. After this phase the abdomen was closed with liberal drainage. The use of sodium hypochlorite, although effective in our experience, could be considered controversial on the basis of its similarity to the active metabolites of the free radical superoxide anion, which generate shock [11]. Wittmann et al. compared primary closure of the abdomen to zippers, slide fasteners, and Velcro analog for temporary abdominal closure, with a result favoring Velcro [12]. Staples have been advocated for temporary closure of polypropylene Marlex mesh [13]. Beger et al. found that single operative blunt finger d6bridement of necrotic pancreas followed by primary abdominal closure and copious irrigation with 8 liters of slightly hyperosmotic dialysis fluid per day for 25 days kept the mortality down to only 14% in 95 patients. However, only 50% were infected, and 25% required further surgery [14]. Although major extraintestinal bleeding episodes and gastrointestinal fistulas led to many acute management problems, they were the direct cause of death in only one of our patients. Death usually occurred as a result of overwhelming sepsis in our experience, associated in a substantial proportion of cases with prior episodes of bleeding or fistula. Wilson et al. studied the broad spectrum of mild and severe pancreatitis patients with daily APACHE II scoring and found that this method stratified their outcome accurately into deaths, complications, or uneventful recovery [15]. The two complications reported to carry a high mortality (i.e., colonic necrosis and retroperitoneal fasciitis [1, 2]) were associated with a low initial A P A C H E II score in our series, which is further evidence of the progressive nature of this disorder. We used hemofiltration during the first 24 hours of fulminant pancreatitis with evident multiorgan failure. The hemofiltration is venous and pump-driven with a hemocompatible filter that does not require systemic heparin. The membranes remove inflammatory medium-sized molecules (<10,000 daltons). Although our experience is small and the treatment should be regarded as experimental, we believe that it improves early survival and prevents multiorgan failure. There is support for hemofiltration with evidence that wound fluid contains factors that markedly impair the host response to sepsis [16]. The early use of this technique also reportedly reduced fluid overload and adult respiratory distress syndrome (ARDS). However, local complications of necrosis, such as sepsis, bleeding, and fistula, are not mitigated by it [17]. N-Acetylcysteine is an antioxidant with possible benefit for patients with ARDS [18]. Antioxidant therapy is directed toward the ablation of free radical generation, such as superoxide anions and the hydroxyl radical species, by regenerating reduced glutathione within the cell [19]. Many other exogenous antioxidants are under trial currently, such as allopurinol, superoxide dismutase,

Miller et al.: Necrotizing Pancreatitis

mannitol, and lazaroids [11]. Their use in patients with necrotizing pancreatitis is encouraging but not yet proved in our experience [3]. There is ongoing work on the role of highly toxic cytokines, such as interleukin 1 and tumor necrosis factor (TNF), in pancreatitis. Preliminary strategies for blocking these two substances with various anticytokines, including monoclonal antibodies to TNF have been proposed, but no definite clinical benefit has yet been demonstrated [20]. A routine policy of early ERCP for patients who require surgery for local pancreatic complications has been suggested by Neoptolemos [21]. It has not been our practice to date. It appears controversial for these mortally ill patients in whom gallstones are not always the etiology and where the damage to the pancreas is already done [22]. In conclusion, it was found that alcoholism was not the most common cause of infected necrotizing pancreatitis, nor did alcoholism as an etiology carry the highest mortality. Tissue adjacent to the pancreas appeared to undergo progressive necrosis, and it was not a static disease. Major hemorrhage and gastrointestinal fistula were often troublesome to manage but were rarely directly related to the demise of patients. Death occurred most often on the basis of uncontrollable sepsis. Colonic necrosis and retroperitoneal fasciitis carried a particularly high mortality and were associated with low initial disease severity scores. The survivors in this series were younger and remained in hospital longer than the patients who died. Hemofiltration and N-acetylcysteine are promising adjuncts but remained unproved in our series. Our best results were with early, open, repeated retroperitoneal ddbridement. Using this technique, fully 69% of patients can survive this otherwise lethal condition. R6sum6

La pancrdatite ndcrosante et infectde est la forme la plus fulminante de cette maladie. L'atteinte colique et/ou rdtropdritondale sont particulidrement meurtriares. La mortalit6 peut alors atteindre jusqu'~ 50%. Le but de cette 6tude a 6t6 de revoir notre expdrience combinde dans deux Hdpitaux, l'H6pital Princess Alexandra et l'H6pital Royal Brisbane, pour ddterminer si la survie (ou ddcds) 6tait en rapport avec une 6tiologie, un traitement ou une complication particuliers. Nous avons 6tudi6 les dossiers de tousles patients traitds pour une pancrdatite infectde ndcessitant une ndcrosectomie chirurgicale et puis une ventilation en soins intensifs, soit 48 patients. L'~ge mddian 6tait de 52 ans, et pour ceux qui sont ddcddds, de 64 ans (p = 0.001). La cause a 6t6 une lithiase biliaire chez 22 patients et alcoolique chez 12 autres. La survie a 6t6, respectivement, de 13/22 (59%) et de 1/12 (8%). La mortalit6 globale a dt6 de 31%. Les patients qui ont survdcu avaient eu un sdjour hospitalier mddian de 73 jours, alors que les ddcds sont survenus aprbs une mddiane de 35 jours (p = 0.04). Sept patients ont ndcessit6 une hdmofiltration, avec deux ddcds et cinq survies. Parmi les quatre patients qui ont re~u de la N-acdtyl cystdine, trois ont survdcu alors qu'un est ddcdd6. L'abdomen a 6t6 laiss6 ouvert chez 38 patients et ferm6 chez 10. Alors que les critdres de Ranson fi l'admission en soins intensifs n'avaient aucune valeur prddictive, le score mddian d'Apache II 6tait significativement plus bas chez les survivants par rapport ~ ceux qui sont ddcddds (p = 0.025). La ddcouverte d'une atteinte colique ou rdtropdritondale, cependant, non prddites par les scores de Ranson ou d'Apache II particulidrement sombres, 6tait

909

associde avec une mortalit6 significativement plus 61evde (p = 0.007). Le ddcds a 6t6 en rapport avec un sepsis grave dans la plupart des cas, bien que 47% des patients qui sont ddcddds avaient aussi une hdmorragie majeure ou une fistule. Par analyse non paramdtrique, on ddmontre que (1) l'alcool n'dtait pas la cause prddominante de pancrdatite ndcrosante, et sa mortalit6 n'dtait pas la plus 61evde; (2) les tissus adjacents au pancrdas se ndcrosent progressivement sur des jours ou des semaines; (3) un score d'Apache II bas a 6t6 frdquemment retrouv6 chez les patients qui allaient mourir plus tard en raison de fistule colique ou atteinte rdtropdritondale; (4) parmi les survivants, on a pratiqu6 une laparotomie plus tdt, le sdjour hospitalier 6tait plus long et les patients 6taient plus jeunes. En conclusion, un meilleur pronostic est retrouv6 lorsque les malades ayant une pancrdatite ndcrosante et infectde sont traitds t6t, le traitement comprenant un nettoyage ?~ciel ouvert et rdpdt6 du rdtropdritoine. Resumen

La pancreatitis necrotizante infectada es la variedad m~s fulminante de la enfermedad. La afecci6n del colon y la fascitis retroperitoneal son particularmente letales. La mortalidad que informa la literatura llega hasta 50%. E1 propdsito del presente estudio fue revisar nuestra experiencia en los Hospitales Princess Alexandra y Royal Brisbane en Australia, a fin de determinar si la sobrevida de los pacientes apareca relacionada con una determinada etiologia, tratamiento o complicaci6n. Se estudi6 la totalidad de los pacientes tratados a partir de 1986 que presentaban pancreatitis infectada y que requirieron necrosectomia quirfirgica y luego ventilacidn mec~inica en la unidad de cuidado intensivo. E1 Grupo estuvo conformado por 48 pacientes. La edad media de los sobrevivientes rue de 52 afios y la de los que murieron fue de 64 afios (P -- 0.001). De los pacientes con cfilculos biliares, 13 sobrevivieron y 9 murieron. La mortalidad global fue 31%. Los sobrevivientes permanecieron en el hospital por un promedio de 73 dfas, en tanto que las muertes se presentaron luego de un promedio de 35 dias (P = 0.04). Siete pacientes recibieron hemofiltraci6n, 5 sobrevivieron y 2 murieron. Se utiliz6 la N-acetil cisteina en 4 pacientes, de los cuales 3 sobrevivieron y 1 muff& Se dej6 abierto el abdomen (laparostomia) en 38 pacientes y se cerr6 en 10; en tanto que los criterios de Ranson en el momento de la admisi6n a la unidad de cuidado intensivo no fueron predictores de sobrevida, se encontr6 que el indice A P A C H E II fue significativamente mils bajo en los sobreviviventes q u e e n los que murieron (P = 0.025). Sin embargo, la necesidad de colectomia o el hallazgo de fascitis retroperitoneal en 7 pacientes result6 en una mortalidad significativamente mayor, la cual no se pudo predecir pot medio de los criterios de Ranson o de los indices A P A C H E II (P = 0.007). La muerte se debi6 a sepsis fulminante en la mayoria de los casos, aunque 47% de los pacientes que murieron tambidn presentaban sangrado mayor o fistulas. E1 anfilisis estadistico demuestra las siguientes tendencias: 1) El alcohol no rue la causa mils frecuente de fascitis necrotizante y tampoco revisti6 la mayor mortalidad. 2) E1 tejido adyacente al pfincreas desarrolla necrosis progresiva en el curso de dfas o de semanas. 3) Se encontraron bajos indices A P A C H E II iniciales en pacientes que finalmente murieron por necrosis del colon y fascitis retroperitoneal. 4) Los sobrevivientes tendieron a ser tratados mediante laparostomla mils temprana, exhibieron mils largas hospitalizaciones y fueron significativamente mils

910

j6venes. En conclusi6n, los pacientes evolucionan mejor con desbridaciones tempranas, abiertas y repetidas del retroperitoneo de lo que parece ser un proceso progresivo.

World J. Surg. Vol. 18, No. 6, Nov./Dec. 1994

11. 12.

Acknowledgments

We are grateful to Glenda Balderson and Betty Menzies for their expert assistance with the statistical analyses.

13.

References

14.

1. Bouillot, J.L., Alexandre, J.H., Vuong, N.P.: Colonic involvement in acute necrotizing pancreatitis: results of surgical treatment. World J. Surg. 13:84, 1989 2. Woodburn, K.R., Ramsay, G., Gillespie, G., Miller, D.F.: Retroperitoneal necrotizing fasciitis. Br. J. Surg. 79:342, 1992 3. Henderson, A., Miller, B.J., Wright, M.: The resource implications of severe necrotizing pancreatitis treated by necrosectomy. Aust. N.Z.J. Surg. 63:541, 1993 4. Blamey, S.L., Imrie, C.W., O'Neill, J., Gilmour, W.H., Carter, D.C.: Prognostic factors in acute pancreatitis. Gut 25:1340, 1984 5. Knaus, W.A., Draper, E.A., Wagner, D.P., Zimmerman, J.E.: An evaluation of outcome from intensive care in major medical centres. Ann. Intern. Med. 104:410, 1986 6. Bradley, E.L.: A clinically based classification system for acute pancreatitis. Arch. Surg. 128:586, 1993 7. Clavien, P.A., Hauser, H., Meyer, P., Rohner, A.: Value of contrastenhanced CT in the early diagnosis and prognosis of acute pancreatitis: a prospective study of 202 patients. Am. J. Surg. 155:457, 1988 8. Fielding, G.: Severe pancreatitis--still a frequently mortal illness. Aust. N.Z.J. Surg. 57:537, 1987 9. Van Vyve, E.L., Reynaert, M.S., Lengele, B.G., Pringot, J.Th., Otte, J.B., Kestens, P.J.: Retroperitoneal laparostomy: a surgical treatment of pancreatic abscesses after an acute necrotizing pancreatitis. Surgery 111:369, 1992 10. Farthmann, E.H., Sch6ffel, U.: Principles and limitations of operative

Invited Commentary D.C. Carter, M . D . University of Edinburgh, Scotland, U.K.

This review from Brisbane adds useful information to the ongoing discussion about the management of necrotizing pancreatitis and reports a significant number of patients (n = 48) with proved infected necrosis. It is clear that it remains a dangerous form of pancreatitis with a mortality in this group of patients of no less than 31%. The authors appear to classify acute pancreatitis into three major categories: edematous, hemorrhagic, and necrotizing; such grouping remains an area of debate, given that many investigators have adopted the classification used by Beger's group [1] in which acute pancreatitis is classified as edematous or necrotizing, with the possibility of progression to pseudocyst or abscess formation. Gallstones were the leading cause of infective necrosis in the Brisbane series, an observation that supports our own finding that gallstones were the commonest identified cause of fatal acute pancreatitis in a series of 126 Scottish patients [2]. The Brisbane series also stresses the fact that idiopathic and postoperative

15. 16.

17. 18.

19. 20. 21. 22.

management of intraabdominal infections. World J. Surg. 14:210, 1990 Schiller, H.J., Reilly, P.M., Bulkley, G.B.: Antioxidant therapy. Crit. Care Med. 21:$92, 1993 Wittmann, D.H., Aprahamian, C., Bergstein, J.M.: Ettappenlavage: advanced diffuse peritonitis managed by planned multiple laparotomies utilizing zippers, slide fastener, and Velcro| analogue for temporary abdominal closure. World J. Surg. 14:218, 1990 Fielding, G.A., Lewandowski, R., Askew, A.R., Wall, D.: Stapled Marlex mesh abdominal closure for repeat laparotomy in pancreatic disease. Aust. N.Z.J. Surg. 57:767, 1987 Beger, H.G., Btichler, M., Bittner, R., Block, S., Nevalainen, T., Roscher, R.: Necrosectomy and post-operative local lavage in necrotizing pancreatitis. Br. J. Surg. 75:207, 1988 Wilson, C., Heath, D.I., Imrie, C.W.: Prediction of outcome in acute pancreatitis: a comparative study of APACHE II, clinical assessment and multiple factor scoring systems. Br. L Surg. 77:1260, 1990 Lazarou, S.A., Barbul, A., Wasserkrug, H.L., Efron, G.: The wound is a possible source of post-traumatic immune suppression. In Proceedings 9th Annual Meeting of the Surgical Infection Society, Denver, 1989 Garzia, F., Todor, R., Scalea, T.: Continuous arteriovenous haemofiltration countercurrent dialysis in acute respiratory failure. J. Trauma 31:1277, 1991 Jepson, S., Helevsen, P., Knudsen, P., Bud, I.M., Klausen, N.: Antioxidant treatment with N-acetyl cysteine during adult respiratory distress syndrome: a prospective, randomized, placebo-controlled study. Crit. Care Med. 20:918, 1992 Braganza, J.M., Holmes, A.M., Morton, A.R., Stalley, L., Ku, R., Kishen, R.: Acetylcysteine to treat complications of pancreatitis. Lancet 1:914, 1986 Dinarello, C.A., Gelfand, J.A., Wolff, S.M.: Anticytokine strategies in the treatment of the systemic inflammatory response syndrome. J.A.M.A. 269:1829, 1993 Neoptolemos, J.P.: Endoscopic sphincterotomy in acute gallstone pancreatitis. Br. J. Surg. 80:547, 1993 Proston, G.J., Williamson, R.C.N.: Surgical management of acute pancreatitis. Br. J. Surg. 77:5, 1990

pancreatitis can be particularly dangerous forms of the disease, and the need to consider the diagnosis of pancreatitis in such patients cannot be overemphasized. Three of the Brisbane patients developed necrotizing pancreatitis after endoscopic retrograde cholangiopancreatography (ERCP); and although these data must not be allowed to detract from the immense diagnostic and therapeutic value of the procedure, the profession at large must continue to appreciate that ERCP can have potentially fatal complications and that it must not be embarked on lightly, particularly when defined indications are lacking. I agree wholeheartedly with the emphasis in this paper on distinguishing between pancreatic resection and necrosectomy. There is now evidence that attempts at formal pancreatic resection expose the patient to even greater risks than necrosectomy [3] as well as the risk of removing viable pancreatic tissue and tipping the patient into lifelong pancreatic exocrine or endocrine insufficiency. It is important to stress, however, that pancreatic and peripancreatic necrosis is often an ongoing procedure, and that the story does not end after one necrosectomy. In Brisbane reliance has clearly been placed on repeated surgery to deal with further necrosis, the abdomen being left open in 38 patients to facilitate reexploration. It was interesting to note that the number of operations undergone by the Brisbane patients ranged from 2 to a staggering 45, and that the median number of procedures was 7. While accepting