Ablative Surgery for Necrotizing Pancreatitis Lyle G. Henry,
MD, Milwaukee,
Wisconsin
Robert E. Condon, MD, Milwaukee,
Wisconsin
In an effort to reduce mortality, near total pancreatectomy has been proposed in the small number of patients with fulminating pancreatitis unresponsive to heroic nonoperative management [I31. This study investigated pancreatectomy as a treatment of hemorrhagic pancreatitis in a canine model [4] that has these characteristics: (1) The hemodynamic and histologic changes are similar to those in human subjects with hemorrhagic pancreatitis. (2) The disease in untreated animals is highly lethal. (3) Most importantly, resuscitation with fluid improves survival. Material and Methods Forty-four healthy mongrel dogs of both sexes were used; the animals weighed between 12 and 22 kg. Preoperative hematocrit values were 36 to 40 per cent and leukocyte counts were less than 20,006 per mm3. All animals were cage-conditioned and cared for according to accepted standards. After a sixteen hour fast, dogs were anesthetized with an intravenous barbiturate and supported by a Harvard respirator utilizing room air via an endotracheal tube. Operative procedures were carried out under aseptic conditions. The pancreas and duodenum were exposed through a midline abdominal incision. Four animals without induced pancreatitis underwent total pancreatectomy to establish the mortality for this procedure. In the remaining forty dogs, the larger accessory pancreatic duct was isolated and cannulated with a fine polyethylene catheter. The smaller, major pancreatic duct was identified and ligated. Autogenous bile was asFrom the Surgical Service. Veterans AdmInistration Center. Wood, Wisccnstn, and the Divlsbn of Surgery. The Mdkal Cok~e of Wisconsin. Mihwaukee. Wkccdn. Repint requ8sts should be addressed to Robert E. Condon, MD, Vetwans Administration Center. Wood, Wisconsin 53193 Resented at the Sixteenth Annual Meeting of the Society for Surgery of the Alimentary Tract, San Antonb. Texas, May 20-21. 1975
Vohmw 131. January 1978
pirated from the gallbladder and 12 ml mixed immediately with 25,000 units (3 ml) of Tryptar (Armour Pharmaceuticals, lot L62209). The bile-trypsin mixture was forcibly injected into the cannulated pancreatic duct at a pressure exceeding 140 mm Hg. The cannula was removed, the duct ligated, and the abdomen closed. During the operation, 500 ml of Ringer’s lactate was given intravenously to all animals. Animals with induced pancreatitis were assigned randomly to one of three groups. Fifteen animals received no further treatment and are designated untreated controls. The other twenty-five animals were again anesthetized after twenty-four hours and given intravenous fluid resuscitation consisting of 60 ml/kg of Ringer’s lactate [5] plus 500 ml of whole canine blood. Twelve animals had no further surgery and are designated fluidresuscitated controls. The remaining thirteen fluid-resuscitated animals underwent total pancreatectomy; these are designated the pancreatectomy group. The fluid-resuscitated controls and the pancreatectomy group received 500 ml of dextrose in half-normal saline solution intravenously the next day. Animals that underwent pancreatectomy received 3 units of PZI insulin subcutaneously on the day of pancreatectomy and 5 units daily thereafter [6]. Postoperatively, all animals were offered food and water ad libitum. Hematocrit and serum calcium, glucose, and amylase levels were determined daily. Time of death was estimated by body cooling measurement [7]. Animals that survived seven days were considered permanent survivors and were sacrificed. Postmortem examination was performed in all animals.
Results Survival data are recorded in Table I. In the untreated controls the mortality was 66 per cent at seventy-two hours and increased to 93 per cent at seven days. All of these animals exhibited marked pancreatic hemorrhage and necrosis with hemor-
125
Henry
TABLE
and Condon
I
Mortality
from
Pancreatitis
No. of Experimental
Group
Untreated controls Fluid-resuscitated animals Animals receiving pancreatectomy plus fluid resuscitation
Animals
No. of Animals At Three
Days
That
Died
At Seven Days
15 12
10 (66%) 3 (25%)
14 (93%) 3 (25%)
13
4 (39%)
7 (54%)
rhagic peritoneal fluid and saponification of mesenteric fat. Many dogs in this group also showed extensive retroperitoneal hemorrhage. Fluid-resuscitated animals showed improved survival compared with untreated controls (p < 0.05); the mortality was 25 per cent after seventytwo hours and did not increase during the week of observation. Although nine of these twelve animals recovered from induced pancreatitis, survivors were slow to resume normal activity and diet. Postmortem examination revealed pancreatic hemorrhage. In survivors, hemorrhagic fluid was still present in the peritoneal cavity and there was early fibrosis of the pancreas. Animals undergoing pancreatectomy had a 39 per cent mortality at seventy-two hours, which increased to 54 per cent after seven days; survival was significantly better than that in the untreated controls (p < 0.05). Although mortality in this group was greater than that in fluid-resuscitated controls, the difference was not statistically significant (p < 0.10). Survivors in this group recovered faster and usually were eating normally and exercising forty-eight to seventy-two hours after pancreatectomy. Postmortem study of surviving animals demonstrated resolving inflammation in the area of the duodenum but no evidence of peritoneal fluid or acute inflammation. Comments Experimental animal investigations [5,8] have indicated that bradykinin and other vasoactive dilators liberated from damaged or necrotic pancreas are responsible for the severity of acute hemorrhagic pancreatitis. Bradykinin release was found to correlate directly with rising hematocrit levels and the accumulation of peritoneal fluid. Vasoactive peptides increased in concentration until the animal died. Previous experimental studies of pancreatectomy in acute pancreatitis demonstrated increased survival only when resection was per-
126
formed within 30 minutes of the induction of pancreatitis [8,9]. Furthermore, untreated controls had exceedingly short survival times. To our knowledge, models of pancreatitis responsive to fluid resuscitation have not been utilized previously to study the effects of ablative surgery. In the past, clinical reports have indicated that necrotizing pancreatitis is associated with a high mortality. In 1898 Korte [IO] reported on 103 patients, 60 per cent of whom died. Fifty years later, Foster and Ziffren [II] found an 82 per cent mortality in patients with necrotizing pancreatitis. Such figures are widely quoted as characterizing the mortality risk in clinically severe pancreatitis. However, vigorous fluid replacement, early ventilatory support [12], suppression of pancreatic secretion by nasogastric suction, and early treatment of infection [13] have lowered the overall mortality of severe pancreatitis to less than 15 per cent in institutions that treat appreciable numbers of such patients. Nonetheless, there remains a small fraction of patients in whom the clinical course is characterized by progressive deterioration terminating in refractory shock. Methods to identify these high risk patients early in the course of the disease have improved, unfortunately, however, there is no pathognomonic test [14]. In an effort to lower mortality, early pancreatectomy has been advocated in patients with pancreatitis who fail to respond to nonoperative therapy. Watts [x], Khedroo and Casella [2], Frey [3], and Norton and Eiseman [I] have reported encouraging early results in an uncontrolled clinical experience with this treatment. The concept is not new, however, as Hofmann [16] suggested the use of pancreatectomy in 1910. Imrie, Whyte, and Frew [17] and Ranson et al [14] have recently concluded that operative intervention is not of benefit and may be deleterious. Although our animals that survived did have a shortened convalescence and amylase and hematocrit changes that suggested a salutory effect of pancreatectomy, our experimental results do not indicate that pancreatectomy improves survival compared with fluid resuscitation alone.
Summary We have confirmed again that bile and trypsin injected in dogs under high pressure produce highly lethal necrotizing pancreatitis responsive to fluid resuscitation. Animals undergoing pancreatectomy show a reduction in serum amylase levels and hemoconcentration (reflected in hematocrit
The American Journal of Surgery
Necrotizing Pancreatitis
levels) after pancreatectomy, an effect that may be related to removal of the source of vasoactive substances liberated in pancreatitis. Qualitatively, survivors of pancreatectomy exhibit accelerated convalescence. Pancreatectomy, however, does not increase survival once the disease process is established and may be harmful. This experimental study does not support the clinical use of pancreatectomy but rather emphasizes the utility of adequate fluid resuscitation in the treatment of severe pancreatitis.
References I. Norton L. Eiseman B: Near total pancreatectomy for hemorrhagic pancreatitis. Am J Surg 127: 191, 1974. 2. Khedroo L, Casella P: Acute hemorrhagic pancreatitis: beneficial effect of primary excision of grossly necrotic pancreatic tissue. 111 Med J 129: 6 I, 1966. 3. Frey C: The operative treatment of pancreatftis. Arch Surg 98: 406, 1969. 4. Pissiotis CA, Condon RE. Nyhus LM: Effect of vasopressin on pancreatic blood flow in acute hemorrhagic pancreatitis. Am J Surg 123: 203, 1972. Ryan J. Moffat J, Thompson A: Role of bradykinin system in acute hemorrhagic pancreatitis. Arch Surg 91: 14, 1965. Markowitz J, et al: Experimental Surgery. Baltimore, Williams & Wilkins, 1964. Lutsky I, Evans WE, Korper KL: Estimating time of death in dogs. Am J Surg Ill: 846, 1966. Hollenburg M. Kobold EE, Pruett R, Thal A: Occurrence of circulating vasoactive substances in human and experimental pancreatttis. Surg Forum 13: 302, 1962. 9. Rittenbury MS, Yarborough D, Egdahl RH: The effect of total pancreatectomy in experimental acute hemorrhagic pancreatitis. Personal communication. IO. Korte W: Die chirurgischen Krankheiten und die Verletzungen des Pankreas. Deutsch Chir vol 14, 1898. Il. Foster P. Ziffren S: Severe acute pancreatitis. Arch Surg 85: 252, 1962. 12. Kellum J, De Meester T, Elkins R. Zuidema G: Respiratory insufficiency secondary to acute pancreatitis. Ann Surg 175: 657. 1972. 13. Kodesch R. Dulont H: Infections complications of acute pancreatitis. Surg Gynecor Obstet 136: 763, 1973. 14. Ranson J, Rifkind K, Roses D, Fink S, Eng K, Spencer F: Prognostic signs and the role of operative management in acute pancreatiiis. Surg Gynecol Obstet 139: 69, 1974. 15. Watts G: Total pancreatectomy for fulminant pancreatitis. Lancet 2: 384, 1963. 16. Hofmann A: Wer den hemorrhagischen infarkt der bauchspeichedreuse. Zentralbl Chir 37: 1153. 1910. 17. lmrie C, Whyte A, Frew E: A pattern of serum amylase concentration in the initial 24 hours after perforation of duodenal ulcers. JR Co/l Surg Edinb 19: 370, 1974.
Discussion Charles F. Frey (Ann Arbor, MI): The authors drew two conclusions about operating on dogs with severe pancreatitis. The dogs that underwent operation and survived did better than those in which the pancreas was not approached surgically. However, there was a greater mortality among dogs subjected to pancreatic resection. Why was that? This question is fundamental.
Vokmw 131, January 1976
NO conclusions can be drawn from the evidence presented as to whether operative or nonoperative management of hemorrhagic pancreatitis is better without knowledge of the cause of death in the dogs subjected to operation. Perhaps some single improvement in postoperative management might have increased the survival of the dogs undergoing operation. John
Ranson
(New York, NY): I would like to em-
phasize here that we are not talking about the carefully planned and prepared delayed resection that Doctor Williams (Am J Surg 131: 114,1976) has described. About four years ago, encouraged by the reports of Hollender, Watts, Khedroo, and others, we evaluated early subtotal pancreatic resection in five patients with severe acute pancreatitis. All were resected within the initial 48 hours after diagnosis and had obvious pancreatic hemorrhage or necrosis at operation. We did not detect any improvement. in the subsequent course of these patients and all eventually died with severe respiratory failure and sepsis. In a subsequent group of two hundred patients, eleven had pancreatitis of comparable severity as estimated by the early objective prognostic signs that we have reported. None of these patients was resected and their early management was by standard methods with the addition of peritoneal dialysis in some cases. Although all but one of these patients had major complications, only three (27 per cent) died. The wide differences in reported clinical experience with early pancreatic resection may be related to technical factors in the care of these patients or to the severity of the pancreatitis. Is it possible to identify any features associated with survival in the resected group? Were these simply the animals with less severe pancreatitis or were there other important factors? Lawrence Norton (Denver, CO): Obviously, there is considerable controversy about the effectiveness of resection for hemorrhagic pancreatitis and I do not want to be in the position of defending the operation against these experiments. My one comment is that our indication for resection in Denver has been a patient who has failed to improve with fluid therapy. I wonder if there might be some modification now of your animal model in which you could compare the effectiveness of resection with peritoneal dialysis or with drainage procedures in those animals in which resuscitation by fluids appears to have failed? R. J. Baker (Chicago, IL): We encountered eight patients with hemorrhagic pancreatitis, proved by aspiration of prune juice fluid in the peritoneal cavity, severe deterioration of their clinical condition in a very short period of time, positive Grey-Turner sign in one, and hyperlipemia in two-thirds. In many, the amylases were not elevated because of destruction of the pancreas. These eight patients were subjected to either near total resection or total pancreatectomy with pancreatoduodenectomy.
127
Henry and Condon
This operation has proved to be one of the most difficult operations in the abdomen. There is generally a combination of blunt and sharp dissection, which is largely a matter of scooping out necrotic pancreas. The bleeding is always fierce. Our patients averaged ten or eleven transfusions during the operation, and probably the only things we really accomplished were to remove necrotic pancreas and to drain the biliary tree. Six of our eight patients died, very promptly, after operation. One recovered after an extremely stormy course and one patient had a pancreatic abscess necessitating a secondary operation; fortunately he survived. Incidentally, when we operate on patients with acute pancreatitis by mistake, we feel that draining the biliary tree has some advantage in terms of future study by cholangiogram if not. from the drainage. In acutely ill patients, experience has indicated that urgent operation for pseudocyst is well established. Patients with pancreatic abscess or in whom massive intraabdominal or gastrointestinal hemorrhage develops during the course of pancreatitis must undergo surgical treatment. In general, it has been our experience in patients with hemorrhagic pancreatitis that supportive measures, including peritoneal dialysis, still remain the most reliable and most likely to result in survival. I would like to ask Doctor Henry whether he has studied respiratory function? Acute respiratory disease in the human with pancreatitis has been well documented and it would be interesting to know whether any of the deaths in the experimental model could have been a result of respiratory insufficiency in the face of experimental pancreatitis. M. C. Geokas (Martinez, CA): I would like to emphasize the dangers of extrapolating the results of animal experimentation to human disease, and I would make the following points. First, the pancreas in the dog differs substantially from the pancreas in man. The canine pancreas is more accessible for extirpation. Only the tail of the pancreas in the dog presents some difficulties in contradistinction to the human pancreas, the extirpation of which, in acute hemorrhagic pancreatitis, constitutes a major insult comparable to total colectomy in the patient with toxic megacolon. Second, the dogs at the time of induction of pancreatitis were absolutely healthy, whereas a patient with hemorrhagic pancreatitis usually is an alcoholic who has
128.
been drinking for five or ten years. Most of these patients have got fatty infiltration of their livers and also are prone to infection at time of admission because of exudation of protein-enzyme rich fluid from the pancreas. Third, in such patients we have the difficulty of establishing absolutely and definitely the diagnosis at the time of admission, mainly the diagnosis of acute necrotizing pancreatitis. We do need to use some of the parameters and develop a consensus so we have a definitive diagnosis of hemorrhagic pancreatitis in the first 24 hours of admission. One of the tests that might be useful is the detection of methemalbumin in the abdominal fluid aspirated at the time of admission. Most of the patients with necrotizing pancreatitis have very high levels of amylase in the abdominal fluid and also high levels of methemalbumin. In most of these patients methemalbumin is also present in the peripheral blood. Lyle G. Henry (closing): In response to Doctor Frey’s comment, there was no effort to individualize the treatment of these animals, as that would add innumerable numbers of variables. The amounts of abdominal fluid found in animals not undergoing resection was massive. I am sure that some of the deaths were caused by hypovolemia. Another possible cause was sepsis, especially in those animals dying after the period of three days; but, again, no antibiotics were administered, avoiding another variable. Finally, in answer to Doctors Frey and Baker, in many animals dying of pancreatitis the lungs were heavy and frothy, showing evidence of pulmonary edema and possibly what may correlate to respiratory distress in these animals; no objective measurements were made of respiratory function in these animals. Doctor Ransom, in the few laboratory tests performed, there were no factors that we could identify concerning prognosis in terms of animal survival. In response to Doctor Norton’s comment, no other forms of therapy were evaluated. Peritoneal lavage has improved survival in experimental models and indeed in small human series as well. There is no attempt here at extrapolating the data to man. I think it is a canine model that may provide some insight in terms of design of future clinical experiments if they are to be undertaken.
The American Journal of Surgery
MD, Milwaukee,
Wisconsin
Robert E. Condon, MD, Milwaukee,
Wisconsin
In an effort to reduce mortality, near total pancreatectomy has been proposed in the small number of patients with fulminating pancreatitis unresponsive to heroic nonoperative management [I31. This study investigated pancreatectomy as a treatment of hemorrhagic pancreatitis in a canine model [4] that has these characteristics: (1) The hemodynamic and histologic changes are similar to those in human subjects with hemorrhagic pancreatitis. (2) The disease in untreated animals is highly lethal. (3) Most importantly, resuscitation with fluid improves survival. Material and Methods Forty-four healthy mongrel dogs of both sexes were used; the animals weighed between 12 and 22 kg. Preoperative hematocrit values were 36 to 40 per cent and leukocyte counts were less than 20,006 per mm3. All animals were cage-conditioned and cared for according to accepted standards. After a sixteen hour fast, dogs were anesthetized with an intravenous barbiturate and supported by a Harvard respirator utilizing room air via an endotracheal tube. Operative procedures were carried out under aseptic conditions. The pancreas and duodenum were exposed through a midline abdominal incision. Four animals without induced pancreatitis underwent total pancreatectomy to establish the mortality for this procedure. In the remaining forty dogs, the larger accessory pancreatic duct was isolated and cannulated with a fine polyethylene catheter. The smaller, major pancreatic duct was identified and ligated. Autogenous bile was asFrom the Surgical Service. Veterans AdmInistration Center. Wood, Wisccnstn, and the Divlsbn of Surgery. The Mdkal Cok~e of Wisconsin. Mihwaukee. Wkccdn. Repint requ8sts should be addressed to Robert E. Condon, MD, Vetwans Administration Center. Wood, Wisconsin 53193 Resented at the Sixteenth Annual Meeting of the Society for Surgery of the Alimentary Tract, San Antonb. Texas, May 20-21. 1975
Vohmw 131. January 1978
pirated from the gallbladder and 12 ml mixed immediately with 25,000 units (3 ml) of Tryptar (Armour Pharmaceuticals, lot L62209). The bile-trypsin mixture was forcibly injected into the cannulated pancreatic duct at a pressure exceeding 140 mm Hg. The cannula was removed, the duct ligated, and the abdomen closed. During the operation, 500 ml of Ringer’s lactate was given intravenously to all animals. Animals with induced pancreatitis were assigned randomly to one of three groups. Fifteen animals received no further treatment and are designated untreated controls. The other twenty-five animals were again anesthetized after twenty-four hours and given intravenous fluid resuscitation consisting of 60 ml/kg of Ringer’s lactate [5] plus 500 ml of whole canine blood. Twelve animals had no further surgery and are designated fluidresuscitated controls. The remaining thirteen fluid-resuscitated animals underwent total pancreatectomy; these are designated the pancreatectomy group. The fluid-resuscitated controls and the pancreatectomy group received 500 ml of dextrose in half-normal saline solution intravenously the next day. Animals that underwent pancreatectomy received 3 units of PZI insulin subcutaneously on the day of pancreatectomy and 5 units daily thereafter [6]. Postoperatively, all animals were offered food and water ad libitum. Hematocrit and serum calcium, glucose, and amylase levels were determined daily. Time of death was estimated by body cooling measurement [7]. Animals that survived seven days were considered permanent survivors and were sacrificed. Postmortem examination was performed in all animals.
Results Survival data are recorded in Table I. In the untreated controls the mortality was 66 per cent at seventy-two hours and increased to 93 per cent at seven days. All of these animals exhibited marked pancreatic hemorrhage and necrosis with hemor-
125
Henry
TABLE
and Condon
I
Mortality
from
Pancreatitis
No. of Experimental
Group
Untreated controls Fluid-resuscitated animals Animals receiving pancreatectomy plus fluid resuscitation
Animals
No. of Animals At Three
Days
That
Died
At Seven Days
15 12
10 (66%) 3 (25%)
14 (93%) 3 (25%)
13
4 (39%)
7 (54%)
rhagic peritoneal fluid and saponification of mesenteric fat. Many dogs in this group also showed extensive retroperitoneal hemorrhage. Fluid-resuscitated animals showed improved survival compared with untreated controls (p < 0.05); the mortality was 25 per cent after seventytwo hours and did not increase during the week of observation. Although nine of these twelve animals recovered from induced pancreatitis, survivors were slow to resume normal activity and diet. Postmortem examination revealed pancreatic hemorrhage. In survivors, hemorrhagic fluid was still present in the peritoneal cavity and there was early fibrosis of the pancreas. Animals undergoing pancreatectomy had a 39 per cent mortality at seventy-two hours, which increased to 54 per cent after seven days; survival was significantly better than that in the untreated controls (p < 0.05). Although mortality in this group was greater than that in fluid-resuscitated controls, the difference was not statistically significant (p < 0.10). Survivors in this group recovered faster and usually were eating normally and exercising forty-eight to seventy-two hours after pancreatectomy. Postmortem study of surviving animals demonstrated resolving inflammation in the area of the duodenum but no evidence of peritoneal fluid or acute inflammation. Comments Experimental animal investigations [5,8] have indicated that bradykinin and other vasoactive dilators liberated from damaged or necrotic pancreas are responsible for the severity of acute hemorrhagic pancreatitis. Bradykinin release was found to correlate directly with rising hematocrit levels and the accumulation of peritoneal fluid. Vasoactive peptides increased in concentration until the animal died. Previous experimental studies of pancreatectomy in acute pancreatitis demonstrated increased survival only when resection was per-
126
formed within 30 minutes of the induction of pancreatitis [8,9]. Furthermore, untreated controls had exceedingly short survival times. To our knowledge, models of pancreatitis responsive to fluid resuscitation have not been utilized previously to study the effects of ablative surgery. In the past, clinical reports have indicated that necrotizing pancreatitis is associated with a high mortality. In 1898 Korte [IO] reported on 103 patients, 60 per cent of whom died. Fifty years later, Foster and Ziffren [II] found an 82 per cent mortality in patients with necrotizing pancreatitis. Such figures are widely quoted as characterizing the mortality risk in clinically severe pancreatitis. However, vigorous fluid replacement, early ventilatory support [12], suppression of pancreatic secretion by nasogastric suction, and early treatment of infection [13] have lowered the overall mortality of severe pancreatitis to less than 15 per cent in institutions that treat appreciable numbers of such patients. Nonetheless, there remains a small fraction of patients in whom the clinical course is characterized by progressive deterioration terminating in refractory shock. Methods to identify these high risk patients early in the course of the disease have improved, unfortunately, however, there is no pathognomonic test [14]. In an effort to lower mortality, early pancreatectomy has been advocated in patients with pancreatitis who fail to respond to nonoperative therapy. Watts [x], Khedroo and Casella [2], Frey [3], and Norton and Eiseman [I] have reported encouraging early results in an uncontrolled clinical experience with this treatment. The concept is not new, however, as Hofmann [16] suggested the use of pancreatectomy in 1910. Imrie, Whyte, and Frew [17] and Ranson et al [14] have recently concluded that operative intervention is not of benefit and may be deleterious. Although our animals that survived did have a shortened convalescence and amylase and hematocrit changes that suggested a salutory effect of pancreatectomy, our experimental results do not indicate that pancreatectomy improves survival compared with fluid resuscitation alone.
Summary We have confirmed again that bile and trypsin injected in dogs under high pressure produce highly lethal necrotizing pancreatitis responsive to fluid resuscitation. Animals undergoing pancreatectomy show a reduction in serum amylase levels and hemoconcentration (reflected in hematocrit
The American Journal of Surgery
Necrotizing Pancreatitis
levels) after pancreatectomy, an effect that may be related to removal of the source of vasoactive substances liberated in pancreatitis. Qualitatively, survivors of pancreatectomy exhibit accelerated convalescence. Pancreatectomy, however, does not increase survival once the disease process is established and may be harmful. This experimental study does not support the clinical use of pancreatectomy but rather emphasizes the utility of adequate fluid resuscitation in the treatment of severe pancreatitis.
References I. Norton L. Eiseman B: Near total pancreatectomy for hemorrhagic pancreatitis. Am J Surg 127: 191, 1974. 2. Khedroo L, Casella P: Acute hemorrhagic pancreatitis: beneficial effect of primary excision of grossly necrotic pancreatic tissue. 111 Med J 129: 6 I, 1966. 3. Frey C: The operative treatment of pancreatftis. Arch Surg 98: 406, 1969. 4. Pissiotis CA, Condon RE. Nyhus LM: Effect of vasopressin on pancreatic blood flow in acute hemorrhagic pancreatitis. Am J Surg 123: 203, 1972. Ryan J. Moffat J, Thompson A: Role of bradykinin system in acute hemorrhagic pancreatitis. Arch Surg 91: 14, 1965. Markowitz J, et al: Experimental Surgery. Baltimore, Williams & Wilkins, 1964. Lutsky I, Evans WE, Korper KL: Estimating time of death in dogs. Am J Surg Ill: 846, 1966. Hollenburg M. Kobold EE, Pruett R, Thal A: Occurrence of circulating vasoactive substances in human and experimental pancreatttis. Surg Forum 13: 302, 1962. 9. Rittenbury MS, Yarborough D, Egdahl RH: The effect of total pancreatectomy in experimental acute hemorrhagic pancreatitis. Personal communication. IO. Korte W: Die chirurgischen Krankheiten und die Verletzungen des Pankreas. Deutsch Chir vol 14, 1898. Il. Foster P. Ziffren S: Severe acute pancreatitis. Arch Surg 85: 252, 1962. 12. Kellum J, De Meester T, Elkins R. Zuidema G: Respiratory insufficiency secondary to acute pancreatitis. Ann Surg 175: 657. 1972. 13. Kodesch R. Dulont H: Infections complications of acute pancreatitis. Surg Gynecor Obstet 136: 763, 1973. 14. Ranson J, Rifkind K, Roses D, Fink S, Eng K, Spencer F: Prognostic signs and the role of operative management in acute pancreatiiis. Surg Gynecol Obstet 139: 69, 1974. 15. Watts G: Total pancreatectomy for fulminant pancreatitis. Lancet 2: 384, 1963. 16. Hofmann A: Wer den hemorrhagischen infarkt der bauchspeichedreuse. Zentralbl Chir 37: 1153. 1910. 17. lmrie C, Whyte A, Frew E: A pattern of serum amylase concentration in the initial 24 hours after perforation of duodenal ulcers. JR Co/l Surg Edinb 19: 370, 1974.
Discussion Charles F. Frey (Ann Arbor, MI): The authors drew two conclusions about operating on dogs with severe pancreatitis. The dogs that underwent operation and survived did better than those in which the pancreas was not approached surgically. However, there was a greater mortality among dogs subjected to pancreatic resection. Why was that? This question is fundamental.
Vokmw 131, January 1976
NO conclusions can be drawn from the evidence presented as to whether operative or nonoperative management of hemorrhagic pancreatitis is better without knowledge of the cause of death in the dogs subjected to operation. Perhaps some single improvement in postoperative management might have increased the survival of the dogs undergoing operation. John
Ranson
(New York, NY): I would like to em-
phasize here that we are not talking about the carefully planned and prepared delayed resection that Doctor Williams (Am J Surg 131: 114,1976) has described. About four years ago, encouraged by the reports of Hollender, Watts, Khedroo, and others, we evaluated early subtotal pancreatic resection in five patients with severe acute pancreatitis. All were resected within the initial 48 hours after diagnosis and had obvious pancreatic hemorrhage or necrosis at operation. We did not detect any improvement. in the subsequent course of these patients and all eventually died with severe respiratory failure and sepsis. In a subsequent group of two hundred patients, eleven had pancreatitis of comparable severity as estimated by the early objective prognostic signs that we have reported. None of these patients was resected and their early management was by standard methods with the addition of peritoneal dialysis in some cases. Although all but one of these patients had major complications, only three (27 per cent) died. The wide differences in reported clinical experience with early pancreatic resection may be related to technical factors in the care of these patients or to the severity of the pancreatitis. Is it possible to identify any features associated with survival in the resected group? Were these simply the animals with less severe pancreatitis or were there other important factors? Lawrence Norton (Denver, CO): Obviously, there is considerable controversy about the effectiveness of resection for hemorrhagic pancreatitis and I do not want to be in the position of defending the operation against these experiments. My one comment is that our indication for resection in Denver has been a patient who has failed to improve with fluid therapy. I wonder if there might be some modification now of your animal model in which you could compare the effectiveness of resection with peritoneal dialysis or with drainage procedures in those animals in which resuscitation by fluids appears to have failed? R. J. Baker (Chicago, IL): We encountered eight patients with hemorrhagic pancreatitis, proved by aspiration of prune juice fluid in the peritoneal cavity, severe deterioration of their clinical condition in a very short period of time, positive Grey-Turner sign in one, and hyperlipemia in two-thirds. In many, the amylases were not elevated because of destruction of the pancreas. These eight patients were subjected to either near total resection or total pancreatectomy with pancreatoduodenectomy.
127
Henry and Condon
This operation has proved to be one of the most difficult operations in the abdomen. There is generally a combination of blunt and sharp dissection, which is largely a matter of scooping out necrotic pancreas. The bleeding is always fierce. Our patients averaged ten or eleven transfusions during the operation, and probably the only things we really accomplished were to remove necrotic pancreas and to drain the biliary tree. Six of our eight patients died, very promptly, after operation. One recovered after an extremely stormy course and one patient had a pancreatic abscess necessitating a secondary operation; fortunately he survived. Incidentally, when we operate on patients with acute pancreatitis by mistake, we feel that draining the biliary tree has some advantage in terms of future study by cholangiogram if not. from the drainage. In acutely ill patients, experience has indicated that urgent operation for pseudocyst is well established. Patients with pancreatic abscess or in whom massive intraabdominal or gastrointestinal hemorrhage develops during the course of pancreatitis must undergo surgical treatment. In general, it has been our experience in patients with hemorrhagic pancreatitis that supportive measures, including peritoneal dialysis, still remain the most reliable and most likely to result in survival. I would like to ask Doctor Henry whether he has studied respiratory function? Acute respiratory disease in the human with pancreatitis has been well documented and it would be interesting to know whether any of the deaths in the experimental model could have been a result of respiratory insufficiency in the face of experimental pancreatitis. M. C. Geokas (Martinez, CA): I would like to emphasize the dangers of extrapolating the results of animal experimentation to human disease, and I would make the following points. First, the pancreas in the dog differs substantially from the pancreas in man. The canine pancreas is more accessible for extirpation. Only the tail of the pancreas in the dog presents some difficulties in contradistinction to the human pancreas, the extirpation of which, in acute hemorrhagic pancreatitis, constitutes a major insult comparable to total colectomy in the patient with toxic megacolon. Second, the dogs at the time of induction of pancreatitis were absolutely healthy, whereas a patient with hemorrhagic pancreatitis usually is an alcoholic who has
128.
been drinking for five or ten years. Most of these patients have got fatty infiltration of their livers and also are prone to infection at time of admission because of exudation of protein-enzyme rich fluid from the pancreas. Third, in such patients we have the difficulty of establishing absolutely and definitely the diagnosis at the time of admission, mainly the diagnosis of acute necrotizing pancreatitis. We do need to use some of the parameters and develop a consensus so we have a definitive diagnosis of hemorrhagic pancreatitis in the first 24 hours of admission. One of the tests that might be useful is the detection of methemalbumin in the abdominal fluid aspirated at the time of admission. Most of the patients with necrotizing pancreatitis have very high levels of amylase in the abdominal fluid and also high levels of methemalbumin. In most of these patients methemalbumin is also present in the peripheral blood. Lyle G. Henry (closing): In response to Doctor Frey’s comment, there was no effort to individualize the treatment of these animals, as that would add innumerable numbers of variables. The amounts of abdominal fluid found in animals not undergoing resection was massive. I am sure that some of the deaths were caused by hypovolemia. Another possible cause was sepsis, especially in those animals dying after the period of three days; but, again, no antibiotics were administered, avoiding another variable. Finally, in answer to Doctors Frey and Baker, in many animals dying of pancreatitis the lungs were heavy and frothy, showing evidence of pulmonary edema and possibly what may correlate to respiratory distress in these animals; no objective measurements were made of respiratory function in these animals. Doctor Ransom, in the few laboratory tests performed, there were no factors that we could identify concerning prognosis in terms of animal survival. In response to Doctor Norton’s comment, no other forms of therapy were evaluated. Peritoneal lavage has improved survival in experimental models and indeed in small human series as well. There is no attempt here at extrapolating the data to man. I think it is a canine model that may provide some insight in terms of design of future clinical experiments if they are to be undertaken.
The American Journal of Surgery
