SCIENTIFIC PAPERS
A Prospective Longitudinal Study of Observation Versus Surgical Intervention in the Management of Necrotizing Pancreatitis Edward L. Bradley III,
MD,
Pancreatic necrosis is now recognized as a principal determinant of survival in acute pancreatitis. However, it is currently unknown how frequently pancreatic necrosis develops in acute pancreatitis, how often pancreatic necrosis becomes secondarily infected, and whether sterile pancreatic necrosis represents an indication for surgery or can be treated by conservative means. In 194 patients with unequivocal acute pancreatitis, pancreatic necrosis developed in 38 (20%)‘) as documented by dynamic pancreatography, and was confirmed by histologic diagnosis at surgery in 28. All patients were prospectively treated by medical means. Patients with pancreatic necrosis who remained persistently febrile underwent fine needle aspiration for bacterial culture. Infected pancreatic necrosis was demonstrated in 27 of the 38 patients (71%) with pancreatic necrosis and was treated by open drainage, yielding a mortality rate of 15%. All 11 patients with demonstrated sterile pancreatic necrosis, including 6 with pulmonary and renal insufficiency, were successfully treated without surgery. Pancreatic necrosis occurs in approximately 20% of patients with acute pancreatitis and is necessary for the development of secondary pancreatic infection. However, pancreatic necrosis by itself, even when accompanied by organ failure, is not an absolute indication for surgery. A trial of medical treatment for all patients with sterile pancreatic necrosis is in order.
From the Department of Surgery, Emory University, Atlanta, Georgia. Requests for reprints should be addressed to Edward L. Bradley III, MD, Department of Surgery, 1968 Peachtree Road, Northwest, Atlanta, Georgia 30309. Presented at the 3 1st Annual Meeting of the Society for Surgery of the Alimentary Tract, San Antonio, Texas, May I5- 16, 1990.
THE AMERICAN
Keith Allen,
MD,
Atlanta.
ceorgia
linicopathologic correlation in acute pancreatitis has recently become possible as a result of proC grammed pancreatic resection for severe pancreatitis in several European centers [I-3]. As an outgrowth of clinical and autopsy studies, both the presence and extent of pancreatic necrosis have become recognized as principal determinants of the severity of the clinical course of acute pancreatitis [4,.5].When present and complicated by secondary infection, pancreatic necrosis virtually proves fatal without surgical intervention [ 1,6,7]. Accordingly, most authorities agree that infected pancreatic necrosis represents an absolute indication for surgery [8-101. However, indications for surgery in patients with noninfected necrosis have not been established. Theoretically, resection of sterile pancreatic necrosis could prevent the secondary systemic effects of necrosis resulting from the release of vasoactive and cytolytic products. Furthermore, removing a necrotic pancreas might reduce the excessive mortality associated with the establishment of secondary infection in the necrotic tissues. Accordingly, a number of surgeons have advocated an aggressive approach to noninfected pancreatic necrosis [I ,2,4,1 l-l 41. Despite such recommendations, however, it is currently unknown how frequently pancreatic necrosis develops in acute pancreatitis, how often pancreatic necrosis becomes secondarily infected, and even whether noninfected necrosis can be safely managed by conservative means. The present report represents the first prospective study designed to address these issues. PATIENTS AND METHODS Between 1986 and 1989,194 patients with unequivocal acute pancreatitis were admitted to 2 hospitals in the Emory University Affiliated Hospitals Program. Acute pancreatitis was considered established when the typical historical and physical features of acute pancreatitis were accompanied by fever, leukocytosis, a serum amylase level twice normal, and either of the following: (1) hyperamylasemia not otherwise explained by the subsequent hospital course, or (2) characteristic findings of acute pancreatitis on sonography or computed tomography (CT). All patients with documented acute pancreatitis were initially treated by supportive measures alone. Supportive treatment consisted primarily of judicious replacement of plasma volume, monitored by pulmonary artery catheterization in severe cases. Pulmonary insufficiency was considered established by a pa02 less than 60 torr on room air, and was treated by intubation and mechanical ventilation. Renal failure was considered established by a serum creatinine level greater than 3 mg/dL with normal
JOURNAL
OF SURGERY
VOLUME
161
JANUARY
1991
19
pancreatic necrosis, with enhancement of head of pancreas (#3 = 85 Hounsfield units [HU]; pancreas:aorta ratio = 72%), lack of enhancement of body and tail of pancreas, and large area of peripancreatic necrosis (#2 = 34 HU; pancreas:aorta ratio = 28%). Sterile pancreatic necrosis was predicted by FNA. The patient survived without surgical intervention despite pulmonary insufficiency and renal failure.
Figure 2. Dynamic pancreatogram demonstrating proximal pancreatic necrosis (#2 = 36 HU; pancreas:aorta ratio = 38% [normal greater than 50%]). The estimated extent of necrosis was 44% as calculated by computer planigraphy. Although dynamic pancreatography is capable of reliably identifying areas of pancreatfc necrosis, FNA culture was required to identify infected necrosis in this patient.
pulmonary artery wedge pressure. Antibiotics were administered to patients remaining persistently febrile during their hospital course. Patients’with severe acute pancreatitis (3 or more Ranson signs in the first 48 hours, or APACHE II scores 20
THE AMERICAN
JOURNAL
OF SURGERY
VOLUME
greater than or equal to 5 at any time during the hospital course) underwent dynamic CT scanning of the pancreas. The specific technique of dynamic pancreatography has been previously described and validated [15]. Dynamic studies were repeated at weekly intervals in any patient who did not improve. Pancreatic necrosis was considered established in patients with areas of pancreatic parenchyma failing to enhance after a large intravenous contrast load (Figure 1). The extent of necrosis was estimated with computer planigraphy by constructing a ratio of the area of decreased pancreatic enhancement to the area of normal perfusion, Patients with documented pancreatic necrosis remaining persistently febrile, or those developing fever later in their hospital course, underwent transcutaneous fine needle aspiration (FNA) for bacterial culture (Figure 2), using the technique described by Gerzof et al [Id. Any patient with demonstrated pancreatic necrosis and a positive FNA culture underwent surgical exploration. The specific surgical techniques employed in patients with demonstrated infected pancreatic necrosis were debridement and open drainage, described in previous publications [8]. Continuing evolution of the technique of open drainage now includes re-debridement every 2 days after initial exploration, closure of the abdominal wound over large-bore lavage catheters placed in the lesser sac whenever granulation tissue replaces retroperitoneal necrosis, then high-volume irrigation (500 ml/hour) of the retroperitoneal cavity using 0.4% sodium oxychlorosene (ClorpactinTM,Guardian Chemical), continuing until the effluent remains clear and the cavity size decreases. Pancreatic necrosis was considered sterile in those 161
JANUARY
1991
_MAI\;AGEMENTOFNECROTIZlNG PANCREATITIS
patients with perfusion defects on dynamic pancreatography pursuing a nonseptic course and in patients with negative FNA cultures. A policy of continued observation and supportive treatment was extended to all patients considered to have noninfected necrosis, and to those patients with acute pancreatitis demonstrated to have normal pancreatic enhancement by dynamic pancreatography.
Comparison of Patients with Necrosis
No. of patients Ranson number (average) Estimated area of pancreatic Pulmonary insufficiency Renal insufficiency
RESULTS
THE AMERICAN
necrosis
Infected
Non-Infected
27 5.1 f 1.8 54f6% 16 (59%) 6 (22%)
11 3.9 f 0.9’ 38 f 4%’ 4 (36%) 2 (18%)
* p 0.05, Fisher’s exact test.
Incidence of pancreatic necrosis:
Pancreatic necrosis was demonstrated by dynamic pancreatography in 38 of the 194 patients (20%) admitted with acute pancreatitis. In 28 of these patients, prediction of pancreatic necrosis by dynamic pancreatography was confirmed by histologic findings at surgery. Ten patients did not undergo surgery. Frequency of secondary infection: Twenty-seven of the 38 patients (71%) with predicted pancreatic necrosis subsequently developed secondary infection of the necrotic tissues. In each case, infected pancreatic necrosis was demonstrated by FNA culture and confirmed by operative culture and histologic findings. No pancreatic infection developed in any patient in the absence of pancreatic necrosis. Noninfected necrosis: In the remaining 11 patients with pancreatic necrosis, the clinical course was variable. Secondary pancreatic infection was suspected clinically in 5 of these 11 patients. A persistently febrile course associated with leukocytosis led to FNA culture in this subgroup of patients. In each of these five patients with fever persisting longer than 1 week, a negative FNA culture was followed by eventual spontaneous disappearance of fever and signs of inflammation. In another five patients, signs of inflammation resolved within the first 7 to 10 days, and a nonseptic course subsequently ensued. In view of the clinical improvement shown by these patients, FNA culture was not considered justifiable. In the remaining patient, elective exploration was mistakenly undertaken for a persistent pseudocyst 2 months after discharge from the acute episode. At surgery, the “pseudocyst” was, in fact, sterile pancreatic necrosis. Four of the 11 patients with noninfected necrosis developed adult respiratory distress syndrome and required prolonged mechanical ventilation with positive end-expiratory pressures. Two of these 11 patients developed acute renal failure, but only 1 required long-term dialysis. All patients with noninfected necrosis survived with supportive treatment alone. Infected pancreatic necrosis: Twenty-seven of 38 patients (7 1%) with demonstrated pancreatic necrosis developed secondary infection of the necrotic tissues. Operative cultures were positive in all 27 patients, yielding findings of gram-negative bacteria in 22 patients, Cundida albicans in 3, and Staphylococcus aureus in 2. Four of the 27 patients (15%) undergoing open drainage for infected pancreatic necrosis died. Patients with infected necrosis were significantly more ill than patients with noninfected necrosis (Table I). Furthermore, the estimated area of pancreatic necro-
1
TABLE I
sis was significantly greater in those patients with infected necrosis. For these reasons, the necrosis treatment groups should not be considered comparable. COMMENTS In the absence of clinicopathologic correlation, stratification of severity and clinical management of patients with acute pancreatitis have necessarily been empirical. Multiple parameter systems have been proposed by Ranson et al [17] and others in efforts to predict mortality and morbidity from indirect clinical criteria. A recent comparative evaluation of these systems, however, failed to demonstrate any significant differences in overall accuracy, each having a predictive accuracy of only 60% to 70% [It?]. After the recent demonstrations that pancreatic necrosis is a constant feature of fatal acute pancreatitis [5], and that for the most part, both the presence and extent of pancreatic necrosis dictate the clinical course of acute pancreatitis [4], the management value of clinical scoring systems has waned. Emphasis has shifted from indirect clinical estimations of prognosis to direct morphologic evaluation, reflecting events actually taking place within the pancreas. In 1983, Kivisaari and associates [19] demonstrated that the pancreas failed to enhance on dynamic pancreatography in eight of nine patients with pancreatic necrosis. In a series of animal and human studies, Nuutinen and co-workers 1201have shown that the failure of pancreatic parenchyma to enhance after an intravenous bolus of contrast material in patients with necrotizing pancreatitis is due solely to the absence of microcirculatory flow. Extensive clinical studies of dynamic pancreatography have attested to an overall accuracy rate exceeding 90% in detecting macroscopic pancreatic necrosis and its complications [ 15,21-231. Multiple attempts to identify patients with pancreatic necrosis by means of circulating serum markers have also been reported [24], but clinical substantiation is lacking. One of the most promising, C-reactive protein, was found in a prospective study from our institution to have an overall accuracy of only 52% in predicting which patients would eventually require surgical intervention [25]. Clinical evaluation of other putative serum necrosis markers is urgently needed. The incidence of pancreatic necrosis identified by dynamic pancreatography and surgery in patients with acute pancreatitis has previously been reported to range
JOURNAL
OF SURGERY
VOLUME
161
JANUARY
1991
21
BRADLEY AND ALLEN
from 17% to 20% [9,23]. In this study, we observed that 20% of 194 patients with acute pancreatitis developed demonstrable pancreatic necrosis. On the basis of these combined observations, it is reasonable to propose that macroscopic pancreatic necrosis occurs in approximately one of every five patients with acute pancreatitis. Once pancreatic necrosis has been identified, how should it be managed? Controversies exist regarding the specific indications for surgery in patients with necrotizing pancreatitis, the timing of surgical intervention, and even the type of surgical drainage to be employed. A number of surgeons have advocated resection whenever pancreatic necrosis is suspected or recognized [l-4,1 l-141. However, mortality rates for this aggressive approach are often high, up to 60% in one series [12], Despite clinical “logic” that patients often benefit from removal of necrotic tissues in other organs, no convincing evidence currently exists that mortality and morbidity are improved in patients undergoing resection for sterile pancreatic necrosis. Teerenhovi et al [26] were unable to demonstrate any benefit conferred by pancreatic resection in 84 patients with pancreatic necrosis associated with pre-existing shock, pulmonary insufficiency, or renal failure. Similarly, Smadja and Bismuth [27] concluded that debridement of sterile pancreatic necrosis did not improve the complicated clinical course of these patients. Moreover, the remote risks of major pancreatic resection in these patients have not received sufficient attention. Surgeons often overestimate the amount of pancreatic necrosis at surgery [3,28], leading to removal of histologically viable tissue. Long-term follow-up studies of patients undergoing early pancreatic resection for necrotizing pancreatitis have demonstrated a 40% to 90% incidence of diabetes [ I3,28,29]. The timing of resection is also controversial. Should excision of necrosis be done within the first days of development [1,2,1 I ,I 21, or delayed until later in the course when complications develop [30,31]? Machado et al [30] observed a 65% mortality rate for patients undergoing operation within the first 10 days after the onset of acute pancreatitis, in contrast with an 8% mortality rate for delayed surgery. Since Beger and associates [32] have shown that bacterial invasion of necrotic tissues rarely occurs within the first week, it is possible that arguments concerning the timing of surgical intervention in necrotizing pancreatitis may actually reflect the presence or absence of bacteria in the necrotic tissues. In this prospective study, we have shown that patients with sterile pancreatic necrosis of limited extent can be successfully managed by supportive treatment without the necessity for surgical resection. Moreover, this study contradicts previous opinions that “pancreatic necrosis with organ failure” should be regarded as an absolute indication for surgery [1,2,4]. Limited numbers of patients with documented sterile necrosis and organ failure have been successfully managed in other institutions [9,33]. Still unanswered by any study, however, is whether more extensive sterile necrosis (more than 50% of the gland) can be consistently managed without surgical intervention. 22
THE AMERICAN
JOURNAL
OF SURGERY
VOLUME
In the past, “hemorrhagic” necrotizing pancreatitis was considered to be virtually 100% fatal without surgical intervention [6,7]. However, these reports were composed of both autopsy and surgical material; in the absence of a reliable technique for noninvasive determination of necrosis, these investigators could hardly have known how many patients with pancreatic necrosis recovered spontaneously. Furthermore, these studies made no distinction between sterile and infected pancreatic necrosis. Several groups have now shown a significantly higher mortality rate for patients with infected necrosis compared with those with sterile necrosis [30,34,35]. Secondary infection of pancreatic necrosis is an ominous event, accounting for more than 80% of all deaths from acute pancreatitis [5]. Infection of pancreatic necrosis has been reported to occur in 5% to 10% of all patients with acute pancreatitis [29,36] and in 40% to 60% of patients with sterile pancreatic necrosis [32,33]. However, the current study represents the first prospective study of the incidence of secondary infection of pancreatic necrosis. It is uncertain whether the 71% incidence of secondary infection of pancreatic necrosis demonstrated in the present study reflects a delay in operative culture due to persistent observation, or is due to some other feature of management. In any case, the relatively high frequency with which secondary infection occurs in pancreatic necrosis should serve to alert the clinician to its potential development and attendant complications. Transcutaneous needle aspiration culture has proved to be both remarkably safe and significantly more accurate than any combination of clinical parameters in distinguishing the “toxemia” of pancreatic inflammation from bacterial infection [16,28,37]. No false-negative or false-positive results were noted with FNA culture in the current series. Furthermore, the specific bacteria identilied by FNA culture were confirmed by operative culture in each case. This technique should occupy a prominent role in the arsenal of all physicians treating severe acute pancreatitis. When infected necrosis has been demonstrated, how should it be managed? Most authorities agree that transcutaneous drainage techniques have a high failure rate due to the consistency and particulate nature of pancreatic necrosis, and that infected pancreatic necrosis requires surgical debridement [8-101. Once operative debridement of necrotic tissue has been accomplished, however, considerable controversy surrounds the method of subsequent drainage. Pancreatic infections differ from all other infections the surgeon is called on to treat. Necrosis of retroperitoneal tissues is a continuing process in the presence of pancreatic duct disruption and bacteria. Release of bioactive materials and extension of necrosis continues for as long as 2 weeks after initial debridement [9,24,26], perhaps explaining the less favorable results obtained in patients with infected pancreatic necrosis after conventional closed drainage. Although a few advocates of classical closed drainage remain [38], it is now clear that closed drainage necessitates multiple reoperations in 25% to 161
JANUARY
1991
MANAGEMENTOF NECROTIZING PANCREATITIS
40% of cases [IO], requires repeatedly correct surgical decisions, and results in higher mortality [8,9,35]. By providing for either repetitive or continuous drainage of the persistent retroperitoneal process, both open packing, as originally advocated by our group [a], and continuous lavage of the lesser sac using large-bore catheters [9] directly address the problem of continuing necrosis and reaccumulation of infected material. The 15% mortality rate resulting from open packing in the present study is consistent with our overall mortality rate of 8 deaths in 54 patients. Alternatively, Beger et al [9] reported a 14% mortality rate in 37 patients with infected necrosis after debridement and continuous high-volume lesser sac lavage. Other experiences with lesser sac lavage have not been as favorable, however. Larvin and co-workers [33] reported a mortality rate of 21% in 14 patients employing debridement and high-volume lesser sac lavage for infected necrosis. Nicholson et al [39] described 3 deaths in 11 patients (27% mortality rate) using a similar technique for lesser sac lavage. In a prospective study of 24 patients randomized to lesser sac lavage or conventional sump drainage after operative debridement, Teerenhovi and associates [40] found that patients undergoing lesser sac lavage experienced a significantly higher mortality rate (36%) than those undergoing conventional drainage (17%). In the absence of a prospective randomized trial comparing newer drainage options, the optimal form of surgical drainage after debridement of infected pancreatic necrosis has not been established. Currently, we employ lesser sac lavage for patients with small areas of infected necrosis, and reserve open drainage for those with more extensive involvement. In our opinion, open drainage is particularly necessary in patients with retrocolic extensions of the necrotic process, which necessitate taking down the colic flexures to perform adequate debridement. If the colic flexures must be taken down, the lesser sac necessarily communicates with the peritoneal cavity, and the selectivity of lavage is lost. Our results indicate that a combination of open drainage and delayed lavage, as described in this report, may prove to be the optimal technique for drainage after debridement in patients with more extensive infected pancreatic necrosis.
REFERENCES 1. Kivilaakso E, Fraki 0, Nikki P, et al. Resection of the pancreas for acute fulminant pancreatitis. Surg Gynecol Obstet 1981; 152: 493-8. 2. Hollender LF, Meyer C, Marrie A, et al. Role of surgery in the management of acute pancreatitis. World J Surg 1981; 5: 361-8. 3. Leger L, Chiche B, Louvel A. Pancreatic necrosis and acute pancreatitis. World J Surg 1981; 5: 315-7. 4. Beger HG, Krauzberger W, Bittner R, et al. Results of surgical treatment of necrotizing pancreatitis. World J Surg 1985; 9: 972-9. 5. Renner IG, Savage WT III, Pantoja JL, et al. Death due to acute pancreatitis: a retrospective analysis of 405 autopsy cases. Dig Dis Sci 1985; 30: 1005-18. 6. Gliedman ML, Bolooki H, Rosen HG. Acute pancreatitis. Curr Probl Surg 1970; 1: l-64. 7. Frey CF. Hemorrhagic pancreatitis. Am J Surg 1979; 137: 61622. 8. Bradley EL 111.Management of infected pancreatic necrosis by open drainage. Ann Surg 1987; 206: 542-50.
9. Beger HG, Buchler M, Bittner R, et al. Necrosectomy and postoperative local lavage in patients with necrotizing pancreatitis: results of a prospective clinical trial. World J Surg 1988; 12: 25562. 10. Ranson JHC. The role of surgery in the management of acute pancreatitis. Ann Surg 1990; 211: 382-93. 11. Norton L, Eiseman B. Near total pancreatectomy for hemorrhagic pancreatitis. Am J Surg 1974; 127: 191-5. 12. Alexandre JH, Guerreri MT. Role of total pancreatectomy in the treatment of necrotizing pancreatitis. World J Surg 1981; 5: 369-77. 13. Aldridge MC, Ornstein M, Glazer G, et al. Pancreatic resection for severe acute pancreatitis. Br J Surg 1985; 72: 796-800. 14. Warshaw AL. Acute pancreatitis. In: Cameron JL, ed. Current surgical therapy II. Philadelphia: B.C. Decker Co, 1986: 232-5. 15. Bradley EL III, Murphy F, Ferguson C. Prediction of pancreatic necrosis by dynamic pancreatography. Ann Surg 1989; 210: 4955504. 16. Gerzof SG, Banks PA, Robbins AH, ef al. Early diagnosis of pancreatic infection by computed tomography-guided aspiration. Gastroenterology 1987; 93: 13 15-20. 17. Ranson JHC, Rifkind KM, Roses DF, ef al. Prognostic signs and the role of operative management in acute pancreatitis. Surg Gynecol Obstet 1974; 139: 69-81. 18. Demmy TL, Burch JM, Feliciano DV, et al. Comparison of multiple-parameter prognostic systems in acute pancreatitis. Am J Surg 1988; 156: 492-6. 19. Kivisaari L, Somer K, Standertskijold-Nordenstam CG, et al. Early detection of acute fulminant pancreatitis by contrast enhanced computed tomography. Stand J Gastroenterol 1983; 18: 39-41. 20. Nuutinen P, Kivisaari L, Schroder T. Contrast-enhanced computed tomography and microangiography of the pancreas in acute human hemorrhagic/necrotizing pancreatitis. Pancreas 1988; 3: 53-60. 21. Maier W. Early objective diagnosis and staging of acute pancreatitis by contrast-enhanced computed tomography. In: Beger HG, Buchler M, eds. Acute pancreatitis. Berlin: Springer-Verlag, 1987: 132-40. 22. Larvin M, Chalmer AG, McMahon MJ. A technique of dynamic CT angiography for the precise identification of pancreatic necrosis. Gastroenterology 1988; 94: 251-6. 23. Baithazar EJ, Robinson DL, Megibow AJ, ef al. Acute pancreatitis: Value of CT in establishing prognosis. Radiology 1990, 174: 331-6. 24. Allen K, Bradley EL III. Modern management of acute pancreatitis. Gen Surg Rep 1990; 1: 110-26. 25. Ferguson C, Bradley EL III. Can serum markers for pancreatic necrosis be used as indicators for surgery? Am J Surg (in press). 26. Teerenhovi 0, Nordback I, Isolauri J. Influence of pancreatic resection on systemic complications in acute necrotizing pancreatitis. Br J Surg 1988; 75: 793-5. 27. Smadja C, Bismuth H. Pancreatic debridement in acute necrotizing pancreatitis: an obsolete procedure? Br J Surg 1986; 73: 40810. 28. Nordback I, Pessi T, Auvinen 0, et al. Determination of necrosis in necrotizing pancreatitis. Br J Surg 1985; 72: 225-7. 29. Kivilaakso E, Lempinen M, Makelainen A, et al. Pancreatic resection versus peritoneal lavage for acute fulminant pancreatitis. Ann Surg 1984; 199: 426-31. 30. Machado MCC, Bacchella T, Monteiro da Cunha JE, ef al. Surgical treatment of pancreatic necrosis. Dig Dis Sci 1986; 31: 25s. 31. White TT, Heimback DM. Sequestrectomy and hyperalimentation in the treatment of hemorrhagic pancreatitis. Am J Surg 1976; 132: 269-75. 32. Beger HG, Bittner R, Block S, et al. Bacterial contamination of pancreatic necrosis: a prospective clinical study. Gastroenterology 1986; 91: 433-8.
THE AMERICAN JOURNAL OF SURGERY
VOLUME 161 JANUARY 1991
23
BRADLEY AND ALLEN
33. Larvin M, Chalmers AG, Robinson PJ, et al. Debridement and closed cavity irrigation for the treatment of pancreatic necrosis. Br J Surg 1989; 76: 465-71. 34. Bittner R, Block S, Buchler M, et al. Pancreatic abscess and infected necrosis: Different local septic complications in acute pancreatitis. Dig Dis Sci 1987; 32: 1082-7. 35. Pederzoli P, Bassi C, Elio A, et al. The infected necrosis is a prognostic factor in necrotizing pancreatitis. Gastroenterology 1989; 96: 1389-91. 36. Allardyce BD. Incidence of necrotizing pancreatitis and factors related to mortality. Am J Surg 1987; 154: 295-300. 37. Block S, Maier W, Bittner R, et al. Identification of pancreas necrosis in severe acute pancreatitis: imaging procedure versus clinical staging. Gut 1986; 27: 1035-42. 38. Warshaw AL, Jim G. Improved survival in 45 patients with pancreatic abscess. Ann Surg 1985; 202: 408-15. 39. Nicholson ML, Mortensen NJ, Espiner HJ. Pancreatic abscess: results of prolonged irrigation of the pancreatic bed after surgery. Br J Surg 1988; 75: 88-91. 40. Teerenhovi 0, Nordback I, Eskola J. High volume lesser sac lavage in acute necrotizing pancreatitis. Br J Surg 1989; 76: 370-3.
DISCUSSION Frank Moody
(Houston, TX): When did the infection occur in the pancreatic necrosis? Apparently you did serial CT-guided aspiration cultures, so there is a time course that you might elucidate for us. Are you implying that medical therapy with antibiotics in some way keeps sterile pancreatic necrosis sterile? Andrew Warshaw (Boston, MA): How many of your noninfected patients had fever or other indications for needle aspiration but did not undergo surgery? You have shown that there was a difference in the overall percentage of pancreatic necrosis among those who were infected and those who were not. Beger has made the point that the degree of necrosis correlates with the risk of infection and with the risk of complications. Therefore, what percentage of your noninfected patients with pancreatic necrosis and with no complications had what we might call trivial necrosis? Finally, you had four deaths among the patients with necrosis. That accounts for only about one fourth of all the deaths in your entire pancreatitis series. What was the cause of death in the patients without pancreatic necrosis, who actually represent the majority of the deaths? John Howard (Toledo, OH): There seems to be a generally accepted belief that the CT scan can distinguish necrotic pancreas from non-necrotic pancreas, but I am not sure that that is true. In our follow-up study, utilizing endoscopic retrograde pancreatography (ERCP) 6 to 12 months after recovery, the patients who appeared to have necrosis often had an intact pancreatic ductal system. Our management policy has been to delay operation for a month or more, if possible. In our series of over 40 patients, only 20% had positive cultures at the time of laparotomy, 1 month to 6 weeks later, even though they had white blood cell counts of 30,000/mm3 to 40,000/ mm3. I wonder how much of the toxicity is due to necrotic tissue and how much to secondary infection.
24
THE AMERICAN
JOURNAL
OF SURGERY
VOLUME
John Ranson (New York, NY): Our experience with a much smaller group of 83 patients studied by dynamic CT is very similar. About 20% of these patients had 30% or more of the gland nonperfused, and about 70% of this group became infected. The incidence of infection, though, as Dr. Warshaw has already mentioned, was very clearly related to the extent of pancreatic nonperfusion on CT. Of those patients who had 30% to 50% of the pancreas nonperfused, 50% recovered without infection, whereas only 15% of those patients with 50% or more nonperfused gland became infected. What was the incidence of infection in patients with less than 50% and those with more than 50% of the pancreas nonperfused? In our experience, in those patients who had a normal perfusing gland on CT, the incidence of infection was still about 8%, so that the presence of pancreatic nonperfusion or interpretation of pancreatic nonperfusion on CT is not the only criterion for the development of infection. What was the incidence of infection in the 80% of patients who did not have pancreatic nonperfusion? Henry Pitt (Baltimore, MD): Often the peripancreatic retroperitoneal fat becomes infected and goes on to develop further necrotizing retroperitonitis. Can you correlate the degree of pancreatic necrosis with the degree of peripancreatic necrosis and tell us where these infections developed? Mitchell Fink (Cincinnati, OH): I am concerned whether one of your goals is to document the incidence of secondary infection of pancreatic necrosis. Your use of a persistently febrile course as the gold standard may be a bit misleading. We published a series from Los Angeles of over 100 consecutive patients with pancreatic septic complications after acute pancreatitis. Approximately 20% of these patients did not present with an initially febrile course, so I think that it would be more appropriate to use the FNA culture as the gold standard for indicating the degree of infection. Louis J. Flanchaum (Columbus, OH): Although I certainly agree that infection is an indication for surgery in these patients, I don’t think that your study specifically addressed that issue. My question relates to the role of operative debridement in patients with sterile pancreatic necrosis. Are you suggesting that in the absence of infection, operative debridement is never required? Also, if there is a role for operative debridement for noninfected pancreatic necrosis, what criteria would you suggest that we employ in order to identify which patients might benefit? Stephen B. Vogel (Gainesville, FL): I agree with Dr. Howard that in patients with a severe pancreatic phlegmon, follow-up ERCP will usually demonstrate an intact pancreatic duct. Therefore, even in patients who have what the radiologists call grade IV CT changes of necrotizing pancreatitis, much of what you see in the many
161 JANUARY
1991
MANAGEMEYT OF NECROTIZINC PANCREATITIS
dense changes of the pancreatic phlegmon is probably peripancreatic tissue and not necessarily pancreatic. Do you believe that patients with sterile pancreatic or peripancreatic necrosis can either resolve or evolve to future pseudocysts, but do not necessarily need to undergo an operation? Although many of my patients have been successfully treated by the open packing technique, I have treated a rather large number of patients both expectantly and conservatively, as long as infection does not intervene. In this group of patients, who had severe changes of necrotizing pancreatitis on CT, follow-up CT examinations after resolution of their illness have demonstrated a large number of pseudocysts and other longterm changes, and only a few patients have returned to normal. Some of these patients required pseudocyst drainage either operatively or percutaneously, but were at least spared the extensive open packing technique. Keith Allen (closing): Dr. Moody, none of our patients required more than one FNA. We believe that pancreatic infection occurs clinically some time after the first week, but we cannot comment on the prophylactic value of antibiotics in this situation. If antibiotics are used, it is important to choose an antibiotic that penetrates the blood-pancreas barrier and also gives good susceptibility to the organisms normally found in infected pancreatic necrosis. In this regard, it is interesting that we had two patients who had pure Candida infection in their areas of pancreatic necrosis. Both of these patients died. Five patients had what was thought to be a septic clinical course and underwent FNA, but their cultures yielded negative results. Those patients were all treated medically. Our data do support the contention that the degree of pancreatic necrosis is a significant risk factor in permitting secondary infection to occur. Since we had no patients with sterile pancreatic necrosis greater than 50% we really can’t comment on whether or not a patient
THE AMERICAN
with significantly more necrosis can be treated conservatively. The accumulated world experience with dynamic pancreatography, which includes our own work, now totals approximately 1,000 cases, with published sensitivity and specificity exceeding 90%. Moreover, we know that pancreatic necrosis does not always involve the major ductal system. We do not think that surgery should be delayed in any patient with a positive bacterial aspirate. Dr. Ranson, we had no patients who had less than 50% necrosis who went on to develop a secondary infection. Furthermore, we did not see any examples of infected necrosis in any patient with normal perfusion. Dr. Pitt, we did not have any patients in this series who had infected peripancreatic necrosis without areas of pancreatic necrosis. We agree that it is often difficult to determine pancreatic from peripancreatic necrosis at surgery, and we believe that the dynamic pancreatogram is more accurate in this regard. In designing this study, we chose persistent fever as a criterion for aspiration since it has been our clinical observation that while all patients with fever do not have pancreatic necrosis, all patients with infected necrosis eventually have fever at some time in their course. Dr. Flancbaum, our data show that all 11 patients who had limited sterile pancreatic necrosis, whether associated with organ failure or not, left the hospital without surgical intervention. Since the inception of the open packing technique almost 15 years ago, we have consistently maintained that its only indication is for infected pancreatic necrosis. Now that necrosis and its extent can be recognized by dynamic pancreatography and secondary infection detected by needle aspiration, candidates for open packing are well delineated. One of the principal goals of this paper was to show that sterile pancreatic necrosis is not necessarily an indication for any form of surgery.
JOURNAL
OF SURGERY
VOLUME
161 JANUARY
1991
25
A Prospective Longitudinal Study of Observation Versus Surgical Intervention in the Management of Necrotizing Pancreatitis Edward L. Bradley III,
MD,
Pancreatic necrosis is now recognized as a principal determinant of survival in acute pancreatitis. However, it is currently unknown how frequently pancreatic necrosis develops in acute pancreatitis, how often pancreatic necrosis becomes secondarily infected, and whether sterile pancreatic necrosis represents an indication for surgery or can be treated by conservative means. In 194 patients with unequivocal acute pancreatitis, pancreatic necrosis developed in 38 (20%)‘) as documented by dynamic pancreatography, and was confirmed by histologic diagnosis at surgery in 28. All patients were prospectively treated by medical means. Patients with pancreatic necrosis who remained persistently febrile underwent fine needle aspiration for bacterial culture. Infected pancreatic necrosis was demonstrated in 27 of the 38 patients (71%) with pancreatic necrosis and was treated by open drainage, yielding a mortality rate of 15%. All 11 patients with demonstrated sterile pancreatic necrosis, including 6 with pulmonary and renal insufficiency, were successfully treated without surgery. Pancreatic necrosis occurs in approximately 20% of patients with acute pancreatitis and is necessary for the development of secondary pancreatic infection. However, pancreatic necrosis by itself, even when accompanied by organ failure, is not an absolute indication for surgery. A trial of medical treatment for all patients with sterile pancreatic necrosis is in order.
From the Department of Surgery, Emory University, Atlanta, Georgia. Requests for reprints should be addressed to Edward L. Bradley III, MD, Department of Surgery, 1968 Peachtree Road, Northwest, Atlanta, Georgia 30309. Presented at the 3 1st Annual Meeting of the Society for Surgery of the Alimentary Tract, San Antonio, Texas, May I5- 16, 1990.
THE AMERICAN
Keith Allen,
MD,
Atlanta.
ceorgia
linicopathologic correlation in acute pancreatitis has recently become possible as a result of proC grammed pancreatic resection for severe pancreatitis in several European centers [I-3]. As an outgrowth of clinical and autopsy studies, both the presence and extent of pancreatic necrosis have become recognized as principal determinants of the severity of the clinical course of acute pancreatitis [4,.5].When present and complicated by secondary infection, pancreatic necrosis virtually proves fatal without surgical intervention [ 1,6,7]. Accordingly, most authorities agree that infected pancreatic necrosis represents an absolute indication for surgery [8-101. However, indications for surgery in patients with noninfected necrosis have not been established. Theoretically, resection of sterile pancreatic necrosis could prevent the secondary systemic effects of necrosis resulting from the release of vasoactive and cytolytic products. Furthermore, removing a necrotic pancreas might reduce the excessive mortality associated with the establishment of secondary infection in the necrotic tissues. Accordingly, a number of surgeons have advocated an aggressive approach to noninfected pancreatic necrosis [I ,2,4,1 l-l 41. Despite such recommendations, however, it is currently unknown how frequently pancreatic necrosis develops in acute pancreatitis, how often pancreatic necrosis becomes secondarily infected, and even whether noninfected necrosis can be safely managed by conservative means. The present report represents the first prospective study designed to address these issues. PATIENTS AND METHODS Between 1986 and 1989,194 patients with unequivocal acute pancreatitis were admitted to 2 hospitals in the Emory University Affiliated Hospitals Program. Acute pancreatitis was considered established when the typical historical and physical features of acute pancreatitis were accompanied by fever, leukocytosis, a serum amylase level twice normal, and either of the following: (1) hyperamylasemia not otherwise explained by the subsequent hospital course, or (2) characteristic findings of acute pancreatitis on sonography or computed tomography (CT). All patients with documented acute pancreatitis were initially treated by supportive measures alone. Supportive treatment consisted primarily of judicious replacement of plasma volume, monitored by pulmonary artery catheterization in severe cases. Pulmonary insufficiency was considered established by a pa02 less than 60 torr on room air, and was treated by intubation and mechanical ventilation. Renal failure was considered established by a serum creatinine level greater than 3 mg/dL with normal
JOURNAL
OF SURGERY
VOLUME
161
JANUARY
1991
19
pancreatic necrosis, with enhancement of head of pancreas (#3 = 85 Hounsfield units [HU]; pancreas:aorta ratio = 72%), lack of enhancement of body and tail of pancreas, and large area of peripancreatic necrosis (#2 = 34 HU; pancreas:aorta ratio = 28%). Sterile pancreatic necrosis was predicted by FNA. The patient survived without surgical intervention despite pulmonary insufficiency and renal failure.
Figure 2. Dynamic pancreatogram demonstrating proximal pancreatic necrosis (#2 = 36 HU; pancreas:aorta ratio = 38% [normal greater than 50%]). The estimated extent of necrosis was 44% as calculated by computer planigraphy. Although dynamic pancreatography is capable of reliably identifying areas of pancreatfc necrosis, FNA culture was required to identify infected necrosis in this patient.
pulmonary artery wedge pressure. Antibiotics were administered to patients remaining persistently febrile during their hospital course. Patients’with severe acute pancreatitis (3 or more Ranson signs in the first 48 hours, or APACHE II scores 20
THE AMERICAN
JOURNAL
OF SURGERY
VOLUME
greater than or equal to 5 at any time during the hospital course) underwent dynamic CT scanning of the pancreas. The specific technique of dynamic pancreatography has been previously described and validated [15]. Dynamic studies were repeated at weekly intervals in any patient who did not improve. Pancreatic necrosis was considered established in patients with areas of pancreatic parenchyma failing to enhance after a large intravenous contrast load (Figure 1). The extent of necrosis was estimated with computer planigraphy by constructing a ratio of the area of decreased pancreatic enhancement to the area of normal perfusion, Patients with documented pancreatic necrosis remaining persistently febrile, or those developing fever later in their hospital course, underwent transcutaneous fine needle aspiration (FNA) for bacterial culture (Figure 2), using the technique described by Gerzof et al [Id. Any patient with demonstrated pancreatic necrosis and a positive FNA culture underwent surgical exploration. The specific surgical techniques employed in patients with demonstrated infected pancreatic necrosis were debridement and open drainage, described in previous publications [8]. Continuing evolution of the technique of open drainage now includes re-debridement every 2 days after initial exploration, closure of the abdominal wound over large-bore lavage catheters placed in the lesser sac whenever granulation tissue replaces retroperitoneal necrosis, then high-volume irrigation (500 ml/hour) of the retroperitoneal cavity using 0.4% sodium oxychlorosene (ClorpactinTM,Guardian Chemical), continuing until the effluent remains clear and the cavity size decreases. Pancreatic necrosis was considered sterile in those 161
JANUARY
1991
_MAI\;AGEMENTOFNECROTIZlNG PANCREATITIS
patients with perfusion defects on dynamic pancreatography pursuing a nonseptic course and in patients with negative FNA cultures. A policy of continued observation and supportive treatment was extended to all patients considered to have noninfected necrosis, and to those patients with acute pancreatitis demonstrated to have normal pancreatic enhancement by dynamic pancreatography.
Comparison of Patients with Necrosis
No. of patients Ranson number (average) Estimated area of pancreatic Pulmonary insufficiency Renal insufficiency
RESULTS
THE AMERICAN
necrosis
Infected
Non-Infected
27 5.1 f 1.8 54f6% 16 (59%) 6 (22%)
11 3.9 f 0.9’ 38 f 4%’ 4 (36%) 2 (18%)
* p 0.05, Fisher’s exact test.
Incidence of pancreatic necrosis:
Pancreatic necrosis was demonstrated by dynamic pancreatography in 38 of the 194 patients (20%) admitted with acute pancreatitis. In 28 of these patients, prediction of pancreatic necrosis by dynamic pancreatography was confirmed by histologic findings at surgery. Ten patients did not undergo surgery. Frequency of secondary infection: Twenty-seven of the 38 patients (71%) with predicted pancreatic necrosis subsequently developed secondary infection of the necrotic tissues. In each case, infected pancreatic necrosis was demonstrated by FNA culture and confirmed by operative culture and histologic findings. No pancreatic infection developed in any patient in the absence of pancreatic necrosis. Noninfected necrosis: In the remaining 11 patients with pancreatic necrosis, the clinical course was variable. Secondary pancreatic infection was suspected clinically in 5 of these 11 patients. A persistently febrile course associated with leukocytosis led to FNA culture in this subgroup of patients. In each of these five patients with fever persisting longer than 1 week, a negative FNA culture was followed by eventual spontaneous disappearance of fever and signs of inflammation. In another five patients, signs of inflammation resolved within the first 7 to 10 days, and a nonseptic course subsequently ensued. In view of the clinical improvement shown by these patients, FNA culture was not considered justifiable. In the remaining patient, elective exploration was mistakenly undertaken for a persistent pseudocyst 2 months after discharge from the acute episode. At surgery, the “pseudocyst” was, in fact, sterile pancreatic necrosis. Four of the 11 patients with noninfected necrosis developed adult respiratory distress syndrome and required prolonged mechanical ventilation with positive end-expiratory pressures. Two of these 11 patients developed acute renal failure, but only 1 required long-term dialysis. All patients with noninfected necrosis survived with supportive treatment alone. Infected pancreatic necrosis: Twenty-seven of 38 patients (7 1%) with demonstrated pancreatic necrosis developed secondary infection of the necrotic tissues. Operative cultures were positive in all 27 patients, yielding findings of gram-negative bacteria in 22 patients, Cundida albicans in 3, and Staphylococcus aureus in 2. Four of the 27 patients (15%) undergoing open drainage for infected pancreatic necrosis died. Patients with infected necrosis were significantly more ill than patients with noninfected necrosis (Table I). Furthermore, the estimated area of pancreatic necro-
1
TABLE I
sis was significantly greater in those patients with infected necrosis. For these reasons, the necrosis treatment groups should not be considered comparable. COMMENTS In the absence of clinicopathologic correlation, stratification of severity and clinical management of patients with acute pancreatitis have necessarily been empirical. Multiple parameter systems have been proposed by Ranson et al [17] and others in efforts to predict mortality and morbidity from indirect clinical criteria. A recent comparative evaluation of these systems, however, failed to demonstrate any significant differences in overall accuracy, each having a predictive accuracy of only 60% to 70% [It?]. After the recent demonstrations that pancreatic necrosis is a constant feature of fatal acute pancreatitis [5], and that for the most part, both the presence and extent of pancreatic necrosis dictate the clinical course of acute pancreatitis [4], the management value of clinical scoring systems has waned. Emphasis has shifted from indirect clinical estimations of prognosis to direct morphologic evaluation, reflecting events actually taking place within the pancreas. In 1983, Kivisaari and associates [19] demonstrated that the pancreas failed to enhance on dynamic pancreatography in eight of nine patients with pancreatic necrosis. In a series of animal and human studies, Nuutinen and co-workers 1201have shown that the failure of pancreatic parenchyma to enhance after an intravenous bolus of contrast material in patients with necrotizing pancreatitis is due solely to the absence of microcirculatory flow. Extensive clinical studies of dynamic pancreatography have attested to an overall accuracy rate exceeding 90% in detecting macroscopic pancreatic necrosis and its complications [ 15,21-231. Multiple attempts to identify patients with pancreatic necrosis by means of circulating serum markers have also been reported [24], but clinical substantiation is lacking. One of the most promising, C-reactive protein, was found in a prospective study from our institution to have an overall accuracy of only 52% in predicting which patients would eventually require surgical intervention [25]. Clinical evaluation of other putative serum necrosis markers is urgently needed. The incidence of pancreatic necrosis identified by dynamic pancreatography and surgery in patients with acute pancreatitis has previously been reported to range
JOURNAL
OF SURGERY
VOLUME
161
JANUARY
1991
21
BRADLEY AND ALLEN
from 17% to 20% [9,23]. In this study, we observed that 20% of 194 patients with acute pancreatitis developed demonstrable pancreatic necrosis. On the basis of these combined observations, it is reasonable to propose that macroscopic pancreatic necrosis occurs in approximately one of every five patients with acute pancreatitis. Once pancreatic necrosis has been identified, how should it be managed? Controversies exist regarding the specific indications for surgery in patients with necrotizing pancreatitis, the timing of surgical intervention, and even the type of surgical drainage to be employed. A number of surgeons have advocated resection whenever pancreatic necrosis is suspected or recognized [l-4,1 l-141. However, mortality rates for this aggressive approach are often high, up to 60% in one series [12], Despite clinical “logic” that patients often benefit from removal of necrotic tissues in other organs, no convincing evidence currently exists that mortality and morbidity are improved in patients undergoing resection for sterile pancreatic necrosis. Teerenhovi et al [26] were unable to demonstrate any benefit conferred by pancreatic resection in 84 patients with pancreatic necrosis associated with pre-existing shock, pulmonary insufficiency, or renal failure. Similarly, Smadja and Bismuth [27] concluded that debridement of sterile pancreatic necrosis did not improve the complicated clinical course of these patients. Moreover, the remote risks of major pancreatic resection in these patients have not received sufficient attention. Surgeons often overestimate the amount of pancreatic necrosis at surgery [3,28], leading to removal of histologically viable tissue. Long-term follow-up studies of patients undergoing early pancreatic resection for necrotizing pancreatitis have demonstrated a 40% to 90% incidence of diabetes [ I3,28,29]. The timing of resection is also controversial. Should excision of necrosis be done within the first days of development [1,2,1 I ,I 21, or delayed until later in the course when complications develop [30,31]? Machado et al [30] observed a 65% mortality rate for patients undergoing operation within the first 10 days after the onset of acute pancreatitis, in contrast with an 8% mortality rate for delayed surgery. Since Beger and associates [32] have shown that bacterial invasion of necrotic tissues rarely occurs within the first week, it is possible that arguments concerning the timing of surgical intervention in necrotizing pancreatitis may actually reflect the presence or absence of bacteria in the necrotic tissues. In this prospective study, we have shown that patients with sterile pancreatic necrosis of limited extent can be successfully managed by supportive treatment without the necessity for surgical resection. Moreover, this study contradicts previous opinions that “pancreatic necrosis with organ failure” should be regarded as an absolute indication for surgery [1,2,4]. Limited numbers of patients with documented sterile necrosis and organ failure have been successfully managed in other institutions [9,33]. Still unanswered by any study, however, is whether more extensive sterile necrosis (more than 50% of the gland) can be consistently managed without surgical intervention. 22
THE AMERICAN
JOURNAL
OF SURGERY
VOLUME
In the past, “hemorrhagic” necrotizing pancreatitis was considered to be virtually 100% fatal without surgical intervention [6,7]. However, these reports were composed of both autopsy and surgical material; in the absence of a reliable technique for noninvasive determination of necrosis, these investigators could hardly have known how many patients with pancreatic necrosis recovered spontaneously. Furthermore, these studies made no distinction between sterile and infected pancreatic necrosis. Several groups have now shown a significantly higher mortality rate for patients with infected necrosis compared with those with sterile necrosis [30,34,35]. Secondary infection of pancreatic necrosis is an ominous event, accounting for more than 80% of all deaths from acute pancreatitis [5]. Infection of pancreatic necrosis has been reported to occur in 5% to 10% of all patients with acute pancreatitis [29,36] and in 40% to 60% of patients with sterile pancreatic necrosis [32,33]. However, the current study represents the first prospective study of the incidence of secondary infection of pancreatic necrosis. It is uncertain whether the 71% incidence of secondary infection of pancreatic necrosis demonstrated in the present study reflects a delay in operative culture due to persistent observation, or is due to some other feature of management. In any case, the relatively high frequency with which secondary infection occurs in pancreatic necrosis should serve to alert the clinician to its potential development and attendant complications. Transcutaneous needle aspiration culture has proved to be both remarkably safe and significantly more accurate than any combination of clinical parameters in distinguishing the “toxemia” of pancreatic inflammation from bacterial infection [16,28,37]. No false-negative or false-positive results were noted with FNA culture in the current series. Furthermore, the specific bacteria identilied by FNA culture were confirmed by operative culture in each case. This technique should occupy a prominent role in the arsenal of all physicians treating severe acute pancreatitis. When infected necrosis has been demonstrated, how should it be managed? Most authorities agree that transcutaneous drainage techniques have a high failure rate due to the consistency and particulate nature of pancreatic necrosis, and that infected pancreatic necrosis requires surgical debridement [8-101. Once operative debridement of necrotic tissue has been accomplished, however, considerable controversy surrounds the method of subsequent drainage. Pancreatic infections differ from all other infections the surgeon is called on to treat. Necrosis of retroperitoneal tissues is a continuing process in the presence of pancreatic duct disruption and bacteria. Release of bioactive materials and extension of necrosis continues for as long as 2 weeks after initial debridement [9,24,26], perhaps explaining the less favorable results obtained in patients with infected pancreatic necrosis after conventional closed drainage. Although a few advocates of classical closed drainage remain [38], it is now clear that closed drainage necessitates multiple reoperations in 25% to 161
JANUARY
1991
MANAGEMENTOF NECROTIZING PANCREATITIS
40% of cases [IO], requires repeatedly correct surgical decisions, and results in higher mortality [8,9,35]. By providing for either repetitive or continuous drainage of the persistent retroperitoneal process, both open packing, as originally advocated by our group [a], and continuous lavage of the lesser sac using large-bore catheters [9] directly address the problem of continuing necrosis and reaccumulation of infected material. The 15% mortality rate resulting from open packing in the present study is consistent with our overall mortality rate of 8 deaths in 54 patients. Alternatively, Beger et al [9] reported a 14% mortality rate in 37 patients with infected necrosis after debridement and continuous high-volume lesser sac lavage. Other experiences with lesser sac lavage have not been as favorable, however. Larvin and co-workers [33] reported a mortality rate of 21% in 14 patients employing debridement and high-volume lesser sac lavage for infected necrosis. Nicholson et al [39] described 3 deaths in 11 patients (27% mortality rate) using a similar technique for lesser sac lavage. In a prospective study of 24 patients randomized to lesser sac lavage or conventional sump drainage after operative debridement, Teerenhovi and associates [40] found that patients undergoing lesser sac lavage experienced a significantly higher mortality rate (36%) than those undergoing conventional drainage (17%). In the absence of a prospective randomized trial comparing newer drainage options, the optimal form of surgical drainage after debridement of infected pancreatic necrosis has not been established. Currently, we employ lesser sac lavage for patients with small areas of infected necrosis, and reserve open drainage for those with more extensive involvement. In our opinion, open drainage is particularly necessary in patients with retrocolic extensions of the necrotic process, which necessitate taking down the colic flexures to perform adequate debridement. If the colic flexures must be taken down, the lesser sac necessarily communicates with the peritoneal cavity, and the selectivity of lavage is lost. Our results indicate that a combination of open drainage and delayed lavage, as described in this report, may prove to be the optimal technique for drainage after debridement in patients with more extensive infected pancreatic necrosis.
REFERENCES 1. Kivilaakso E, Fraki 0, Nikki P, et al. Resection of the pancreas for acute fulminant pancreatitis. Surg Gynecol Obstet 1981; 152: 493-8. 2. Hollender LF, Meyer C, Marrie A, et al. Role of surgery in the management of acute pancreatitis. World J Surg 1981; 5: 361-8. 3. Leger L, Chiche B, Louvel A. Pancreatic necrosis and acute pancreatitis. World J Surg 1981; 5: 315-7. 4. Beger HG, Krauzberger W, Bittner R, et al. Results of surgical treatment of necrotizing pancreatitis. World J Surg 1985; 9: 972-9. 5. Renner IG, Savage WT III, Pantoja JL, et al. Death due to acute pancreatitis: a retrospective analysis of 405 autopsy cases. Dig Dis Sci 1985; 30: 1005-18. 6. Gliedman ML, Bolooki H, Rosen HG. Acute pancreatitis. Curr Probl Surg 1970; 1: l-64. 7. Frey CF. Hemorrhagic pancreatitis. Am J Surg 1979; 137: 61622. 8. Bradley EL 111.Management of infected pancreatic necrosis by open drainage. Ann Surg 1987; 206: 542-50.
9. Beger HG, Buchler M, Bittner R, et al. Necrosectomy and postoperative local lavage in patients with necrotizing pancreatitis: results of a prospective clinical trial. World J Surg 1988; 12: 25562. 10. Ranson JHC. The role of surgery in the management of acute pancreatitis. Ann Surg 1990; 211: 382-93. 11. Norton L, Eiseman B. Near total pancreatectomy for hemorrhagic pancreatitis. Am J Surg 1974; 127: 191-5. 12. Alexandre JH, Guerreri MT. Role of total pancreatectomy in the treatment of necrotizing pancreatitis. World J Surg 1981; 5: 369-77. 13. Aldridge MC, Ornstein M, Glazer G, et al. Pancreatic resection for severe acute pancreatitis. Br J Surg 1985; 72: 796-800. 14. Warshaw AL. Acute pancreatitis. In: Cameron JL, ed. Current surgical therapy II. Philadelphia: B.C. Decker Co, 1986: 232-5. 15. Bradley EL III, Murphy F, Ferguson C. Prediction of pancreatic necrosis by dynamic pancreatography. Ann Surg 1989; 210: 4955504. 16. Gerzof SG, Banks PA, Robbins AH, ef al. Early diagnosis of pancreatic infection by computed tomography-guided aspiration. Gastroenterology 1987; 93: 13 15-20. 17. Ranson JHC, Rifkind KM, Roses DF, ef al. Prognostic signs and the role of operative management in acute pancreatitis. Surg Gynecol Obstet 1974; 139: 69-81. 18. Demmy TL, Burch JM, Feliciano DV, et al. Comparison of multiple-parameter prognostic systems in acute pancreatitis. Am J Surg 1988; 156: 492-6. 19. Kivisaari L, Somer K, Standertskijold-Nordenstam CG, et al. Early detection of acute fulminant pancreatitis by contrast enhanced computed tomography. Stand J Gastroenterol 1983; 18: 39-41. 20. Nuutinen P, Kivisaari L, Schroder T. Contrast-enhanced computed tomography and microangiography of the pancreas in acute human hemorrhagic/necrotizing pancreatitis. Pancreas 1988; 3: 53-60. 21. Maier W. Early objective diagnosis and staging of acute pancreatitis by contrast-enhanced computed tomography. In: Beger HG, Buchler M, eds. Acute pancreatitis. Berlin: Springer-Verlag, 1987: 132-40. 22. Larvin M, Chalmer AG, McMahon MJ. A technique of dynamic CT angiography for the precise identification of pancreatic necrosis. Gastroenterology 1988; 94: 251-6. 23. Baithazar EJ, Robinson DL, Megibow AJ, ef al. Acute pancreatitis: Value of CT in establishing prognosis. Radiology 1990, 174: 331-6. 24. Allen K, Bradley EL III. Modern management of acute pancreatitis. Gen Surg Rep 1990; 1: 110-26. 25. Ferguson C, Bradley EL III. Can serum markers for pancreatic necrosis be used as indicators for surgery? Am J Surg (in press). 26. Teerenhovi 0, Nordback I, Isolauri J. Influence of pancreatic resection on systemic complications in acute necrotizing pancreatitis. Br J Surg 1988; 75: 793-5. 27. Smadja C, Bismuth H. Pancreatic debridement in acute necrotizing pancreatitis: an obsolete procedure? Br J Surg 1986; 73: 40810. 28. Nordback I, Pessi T, Auvinen 0, et al. Determination of necrosis in necrotizing pancreatitis. Br J Surg 1985; 72: 225-7. 29. Kivilaakso E, Lempinen M, Makelainen A, et al. Pancreatic resection versus peritoneal lavage for acute fulminant pancreatitis. Ann Surg 1984; 199: 426-31. 30. Machado MCC, Bacchella T, Monteiro da Cunha JE, ef al. Surgical treatment of pancreatic necrosis. Dig Dis Sci 1986; 31: 25s. 31. White TT, Heimback DM. Sequestrectomy and hyperalimentation in the treatment of hemorrhagic pancreatitis. Am J Surg 1976; 132: 269-75. 32. Beger HG, Bittner R, Block S, et al. Bacterial contamination of pancreatic necrosis: a prospective clinical study. Gastroenterology 1986; 91: 433-8.
THE AMERICAN JOURNAL OF SURGERY
VOLUME 161 JANUARY 1991
23
BRADLEY AND ALLEN
33. Larvin M, Chalmers AG, Robinson PJ, et al. Debridement and closed cavity irrigation for the treatment of pancreatic necrosis. Br J Surg 1989; 76: 465-71. 34. Bittner R, Block S, Buchler M, et al. Pancreatic abscess and infected necrosis: Different local septic complications in acute pancreatitis. Dig Dis Sci 1987; 32: 1082-7. 35. Pederzoli P, Bassi C, Elio A, et al. The infected necrosis is a prognostic factor in necrotizing pancreatitis. Gastroenterology 1989; 96: 1389-91. 36. Allardyce BD. Incidence of necrotizing pancreatitis and factors related to mortality. Am J Surg 1987; 154: 295-300. 37. Block S, Maier W, Bittner R, et al. Identification of pancreas necrosis in severe acute pancreatitis: imaging procedure versus clinical staging. Gut 1986; 27: 1035-42. 38. Warshaw AL, Jim G. Improved survival in 45 patients with pancreatic abscess. Ann Surg 1985; 202: 408-15. 39. Nicholson ML, Mortensen NJ, Espiner HJ. Pancreatic abscess: results of prolonged irrigation of the pancreatic bed after surgery. Br J Surg 1988; 75: 88-91. 40. Teerenhovi 0, Nordback I, Eskola J. High volume lesser sac lavage in acute necrotizing pancreatitis. Br J Surg 1989; 76: 370-3.
DISCUSSION Frank Moody
(Houston, TX): When did the infection occur in the pancreatic necrosis? Apparently you did serial CT-guided aspiration cultures, so there is a time course that you might elucidate for us. Are you implying that medical therapy with antibiotics in some way keeps sterile pancreatic necrosis sterile? Andrew Warshaw (Boston, MA): How many of your noninfected patients had fever or other indications for needle aspiration but did not undergo surgery? You have shown that there was a difference in the overall percentage of pancreatic necrosis among those who were infected and those who were not. Beger has made the point that the degree of necrosis correlates with the risk of infection and with the risk of complications. Therefore, what percentage of your noninfected patients with pancreatic necrosis and with no complications had what we might call trivial necrosis? Finally, you had four deaths among the patients with necrosis. That accounts for only about one fourth of all the deaths in your entire pancreatitis series. What was the cause of death in the patients without pancreatic necrosis, who actually represent the majority of the deaths? John Howard (Toledo, OH): There seems to be a generally accepted belief that the CT scan can distinguish necrotic pancreas from non-necrotic pancreas, but I am not sure that that is true. In our follow-up study, utilizing endoscopic retrograde pancreatography (ERCP) 6 to 12 months after recovery, the patients who appeared to have necrosis often had an intact pancreatic ductal system. Our management policy has been to delay operation for a month or more, if possible. In our series of over 40 patients, only 20% had positive cultures at the time of laparotomy, 1 month to 6 weeks later, even though they had white blood cell counts of 30,000/mm3 to 40,000/ mm3. I wonder how much of the toxicity is due to necrotic tissue and how much to secondary infection.
24
THE AMERICAN
JOURNAL
OF SURGERY
VOLUME
John Ranson (New York, NY): Our experience with a much smaller group of 83 patients studied by dynamic CT is very similar. About 20% of these patients had 30% or more of the gland nonperfused, and about 70% of this group became infected. The incidence of infection, though, as Dr. Warshaw has already mentioned, was very clearly related to the extent of pancreatic nonperfusion on CT. Of those patients who had 30% to 50% of the pancreas nonperfused, 50% recovered without infection, whereas only 15% of those patients with 50% or more nonperfused gland became infected. What was the incidence of infection in patients with less than 50% and those with more than 50% of the pancreas nonperfused? In our experience, in those patients who had a normal perfusing gland on CT, the incidence of infection was still about 8%, so that the presence of pancreatic nonperfusion or interpretation of pancreatic nonperfusion on CT is not the only criterion for the development of infection. What was the incidence of infection in the 80% of patients who did not have pancreatic nonperfusion? Henry Pitt (Baltimore, MD): Often the peripancreatic retroperitoneal fat becomes infected and goes on to develop further necrotizing retroperitonitis. Can you correlate the degree of pancreatic necrosis with the degree of peripancreatic necrosis and tell us where these infections developed? Mitchell Fink (Cincinnati, OH): I am concerned whether one of your goals is to document the incidence of secondary infection of pancreatic necrosis. Your use of a persistently febrile course as the gold standard may be a bit misleading. We published a series from Los Angeles of over 100 consecutive patients with pancreatic septic complications after acute pancreatitis. Approximately 20% of these patients did not present with an initially febrile course, so I think that it would be more appropriate to use the FNA culture as the gold standard for indicating the degree of infection. Louis J. Flanchaum (Columbus, OH): Although I certainly agree that infection is an indication for surgery in these patients, I don’t think that your study specifically addressed that issue. My question relates to the role of operative debridement in patients with sterile pancreatic necrosis. Are you suggesting that in the absence of infection, operative debridement is never required? Also, if there is a role for operative debridement for noninfected pancreatic necrosis, what criteria would you suggest that we employ in order to identify which patients might benefit? Stephen B. Vogel (Gainesville, FL): I agree with Dr. Howard that in patients with a severe pancreatic phlegmon, follow-up ERCP will usually demonstrate an intact pancreatic duct. Therefore, even in patients who have what the radiologists call grade IV CT changes of necrotizing pancreatitis, much of what you see in the many
161 JANUARY
1991
MANAGEMEYT OF NECROTIZINC PANCREATITIS
dense changes of the pancreatic phlegmon is probably peripancreatic tissue and not necessarily pancreatic. Do you believe that patients with sterile pancreatic or peripancreatic necrosis can either resolve or evolve to future pseudocysts, but do not necessarily need to undergo an operation? Although many of my patients have been successfully treated by the open packing technique, I have treated a rather large number of patients both expectantly and conservatively, as long as infection does not intervene. In this group of patients, who had severe changes of necrotizing pancreatitis on CT, follow-up CT examinations after resolution of their illness have demonstrated a large number of pseudocysts and other longterm changes, and only a few patients have returned to normal. Some of these patients required pseudocyst drainage either operatively or percutaneously, but were at least spared the extensive open packing technique. Keith Allen (closing): Dr. Moody, none of our patients required more than one FNA. We believe that pancreatic infection occurs clinically some time after the first week, but we cannot comment on the prophylactic value of antibiotics in this situation. If antibiotics are used, it is important to choose an antibiotic that penetrates the blood-pancreas barrier and also gives good susceptibility to the organisms normally found in infected pancreatic necrosis. In this regard, it is interesting that we had two patients who had pure Candida infection in their areas of pancreatic necrosis. Both of these patients died. Five patients had what was thought to be a septic clinical course and underwent FNA, but their cultures yielded negative results. Those patients were all treated medically. Our data do support the contention that the degree of pancreatic necrosis is a significant risk factor in permitting secondary infection to occur. Since we had no patients with sterile pancreatic necrosis greater than 50% we really can’t comment on whether or not a patient
THE AMERICAN
with significantly more necrosis can be treated conservatively. The accumulated world experience with dynamic pancreatography, which includes our own work, now totals approximately 1,000 cases, with published sensitivity and specificity exceeding 90%. Moreover, we know that pancreatic necrosis does not always involve the major ductal system. We do not think that surgery should be delayed in any patient with a positive bacterial aspirate. Dr. Ranson, we had no patients who had less than 50% necrosis who went on to develop a secondary infection. Furthermore, we did not see any examples of infected necrosis in any patient with normal perfusion. Dr. Pitt, we did not have any patients in this series who had infected peripancreatic necrosis without areas of pancreatic necrosis. We agree that it is often difficult to determine pancreatic from peripancreatic necrosis at surgery, and we believe that the dynamic pancreatogram is more accurate in this regard. In designing this study, we chose persistent fever as a criterion for aspiration since it has been our clinical observation that while all patients with fever do not have pancreatic necrosis, all patients with infected necrosis eventually have fever at some time in their course. Dr. Flancbaum, our data show that all 11 patients who had limited sterile pancreatic necrosis, whether associated with organ failure or not, left the hospital without surgical intervention. Since the inception of the open packing technique almost 15 years ago, we have consistently maintained that its only indication is for infected pancreatic necrosis. Now that necrosis and its extent can be recognized by dynamic pancreatography and secondary infection detected by needle aspiration, candidates for open packing are well delineated. One of the principal goals of this paper was to show that sterile pancreatic necrosis is not necessarily an indication for any form of surgery.
JOURNAL
OF SURGERY
VOLUME
161 JANUARY
1991
25
