- Email: [email protected]
Acute pancreatitis after cardiopulmonary bypass
Acute Pancreatitis After Cardiopulrnonary Bypass Gary S. Haas, MD, Boston. Massachusetts A n d r e w L. W a r s h a w , MD, Boston, Massachusetts Willard M. Daggett, MD, Boston, Massachusetts H. T h o m a s Aretz, MD, Boston, Massachusetts
Acute pancreatitis is a reported but poorly understood complication of cardiac surgery. Although it is recognized in less than 1 percent of patients undergoing cardiopulmonary bypass, pancreatitis in this setting frequently occurs in conjunction with renal and other organ failure and runs an aggressive, fatal course [1,5-7]. The diagnosis and treatmen Lof acute pancreatitis after cardiac surgery are difficult. Its symptoms may be atypical and misleading or may be obscured by other complications [5,7]. The usual laboratory characteristics of pancreatitis, including hyperamylasemia, do not appear to be reliable indications of the disease in patients after bypass [8,9]. The diagnosis is often missed until autopsy, but even when it is made antemortem, the mortality rate of affected patients has been higher than in any other recognized form of pancreatitis. It is our purpose to elucidate the spectrum of pancreatic injury, to describe the clinical and laboratory features of patients in whom pancreatitis develops after cardiopulmonary bypass, to examlne risk factors for pancreatitis, and to suggest an approach to effective treatment.
of interstitial fibrosis, stromal fibrosis, and atrophy; (2) acute interstitial by the findings of acinar and stromal edema and an interstitial inflammatory infiltrate; or (3) acute necrotizing by the findingt~of edema, acute inflammatory infiltrate ass~iated with parenchymal fat necrosis, peripancreatic fat necrosis, or both. Pathologic findings were then correlated with cause of death, time of death, associated disease, noncardiac risk factors, and relevant laboratory findings. Patients whose pancreases were autolyzed and those with no specimehs taken for histologic examination were assumed not t(, have pancreatitis for the purposes of this study. Thirty patients were studied prospectivelyafter cardiac surgery to determine the prevalence of subclinical pancreatic h,jury. Serum samples were obtained preoperatively and again on the first, fourth, and seventh days after cardiac surgery for determination of total amylase and pancreatic isoamylase levels (a sensitive and specific index of pancreatic inflammation [I I ]) and of polycytidine-specific ribonuclease (an index of pancreatic cell necrosis
[12,I3]). lsoenzymesof amylase (isoamylases)were separated and quantified by polyacrylamide gel electrophoresis and saccharogenicassay as described by Warshawand Lee Ill]. Ribenuclease was assayed by a modificationof the method of Reddi and Holland [13].
Patients and Methods
Results
The records of 12 patients in whom pancreatit,is was diagnosed after cardiac surgery at the Massachuseits General Hospital between June 1979 and January 1984 were reviewed for physical findings, laboratory data pertinent to the diagnosis, risk factors, clinical course, and therapy. The pathologic findings of the pancreas were reviewed in 138 other patients who died after cardiac surgery and 93 nonsurgical patients who died from acute cardiac disease between December 1975 and April 1981. The pancreatic tissue slides were reviewed by a single pathologist (HTA) without his knowledgeof the clinical history and classified according to Longneeker [10] as (1) chronic by the findings
Clinical pancreatitis after cardiac surgery: Between June 1979 and January 1984, approximate]y 5,400 cardiac procedures were performed. During this time, postoperative pancreatitis was diagnosed antemortem in 12 of these patients (0.2 percent). There were nine male and three female patients who underwent coronary artery bypass (eight patients), coronary artery bypass with aortic valve replacement (two patients), or coronary artery bypass with mitral valve replacement (two patients). Their ages ranged between 50 and 73 years with a mean of 62 years. Preoperative cardiac indexes ranged from 2 to 3.2 liters/rain per m 2 (mean 4- standard deviation, 2.6 .~ 0.2) and ejection fractions from 40 to 60 percent (5Y + 6). None of them manifested signs of cardiogenic shock before operation, but one patient had an intraaortic balloon pump placed to control angina pectoris. Perfusion times ranged from 68 minutes to 208 minutes (mean 134 minutes -2-.46). Three pa-
From the GeneralSurgicalServices. MassachusettsGeneralHospitaland the Department of Surgery, Harvard Medical ~ 1 , Boston,Massachusotts. Requests for reprintsshouldbe addressedto AndrewL. Warshaw,MD, Departmentof Surgery,MassachusettsGeneralHospital,15ParkmanStreet, Suite 336, Boston,Massachusetts02 ! 14. Presentedat the 65LhAnnualMeetingof the New EnglandSurgicalSociety. DlxvllleNotch, New Hampshire,October 12-14, 1984.
508
The AmedeM ~k~ma;.~S u r ~
Pancreatltis After Cardiac Surge,'y
TABLE I
Clinical F e a t u r e s of Necrotizlng PancreaUlls
in Nine Patlenls After Cardlopulmonary Bypass Clinical Finding Early pain and tenderness Abdominal distention and iteus Fever (temperature >101°F) Late abdominal pain and tenderness Gardiogenlc shock in early postoperative course Sudden episodes of late hypotension Leucocytosi,=; (> t2,000 ceUs/mm 3) Unexplained hypocalcemta Early hypera=mylasemla Late hyperar,nylasernia Abdomen explored Deaths
n 0 9 9 9 8 8 9 6 2 4 8 7
% 0 100 100 10(;~ 89 89 100 67 22 44 89 78
tients had hyperlipidemia, one had a histery of alcohol abuse, and two had asymptomatic gallstones. After operation, mild pancreatitis developed in¢ 3 of the 12 patients, The diagnosis was based on abdominal pain, decreased bowel sounds, nausea, and hyperamylasemia. In each case, the symptoms resolved with standard treatment consisting of bowel rest and intravenous fluid support. None of these patients experienced hemodynamic instability during the postoperative period, and their serum total and ionized calcium levels remained within normal limits. Fulminating necrotizing pancreatitis developed in nine patients which was later confirmed by either laparotomy or autopsy. The clinical course of all of these patients was marked by severe hemodynamic instability in the early postoperative period. Eight of them suffered a prolonged low-output state with a cardiac index of less than 2 liters/rain per m 2 and mean aortic pressures less than 80 mm Hg. The ninth patient suffered multiple episodes of ventricular tachyarrh~,thmias with concomitant episodes of hypotension. All nine required both o~-and ~-adrenergic support, and six required intraaortic balloon pump support during this time. The earliest and most common symptoms referable to pancreatitis were abdominal distention and hypoactive bowel sounds (Table I). These were observed in all nine patients within 24 to 48 hours of surgery. Significant abdominal pain and tenderness, however, were not noted until much later in the course of the disease. Prolonged fevers with temperatures above 101°F were present in all patients. Cardiac function improved in all cases, and eventually all patients were weaned from intraaortic balloon pump, inc~tropic support, or both. Despite myocardial recovery, eight of the nine patients were observed to have subsequent episodes of sudden hypotension 3 to 7 days after operation. It. st:: of these patients, serum ionized calcium levels were
Volume 1~.-9,Aprl! 1985
measured during one or more of these hypotension episodes and were found to be less th~n 0.9 raM/liter (normal 1.1 to 1.2 raM/liter) in aH six. Total serum amylase levels were of little value in diagnosing or following progression of the disease in patients with necrotizing pancreatitis. Amylase levels in the early postoperative period were m)rmal in seven of the nine patients and only modestly elevated in two, and they were within normal limits in five of the nine patients when studied "n the later clinical course. Of the four patients in whom serum amylase levels increased in the later clinical course, the levels were less than twice normal in two and could have been attrihuted to concomitant renal dysfunction [141. Leukocytosis (white blood cell count more than 12,000 cells/mm 3) was present in all nine patients with necrotizing pancreatitis. Frequently these pa. tients were thought to be septic, and despite negative blood cultures, six of the nine patients with necrotizing pancreatitis began receiving broad-spectrum antibiotics. The diagnosis of necrotizing pancreatitis was usually not made until very late in its course. Persistence of fever and ileus, the development of late abdominal pain and mild tenderness, and progressive systemic deterioration led to abdominal exploration in eight of the nine patients with necrotizing pancreatitis. Pancreatitis, however, was the leading diagnosis in only two of the eight patients who came to laparotomy and was considered in the differential diagnosis in just two others. Three patients were explored with a leading diagnosis of cholecystitis, two with a leading of peritonitis of unknown cause, and one with a leading diagnosis of ischemic bowel. The operative procedure in the eight patients was mobilization of the pancreas and debridement and drainage, with comcomitant gastrostomy, jejunostomy, and cholecystostomy. Two of these patients survived whereas six eventually died from ongoing pancreatitis, sepsis, and hemorrhage. The ninth died without exploration. Renal failure developed in six and liver failure in four of the nine patients. Autopsy studies a f t e r c a r d i a c s u r g e r y : The pancreatic pathologic characteristics and clinical records of 138 patients who died after cardiac surgery and 93 nonsurgical patients who died from acute cardiac disease between December 1975 and April 1981 were reviewed (Table II). The surgical group consisted of 94 male and 44 female patients who ranged in age from 27 to 85 years (61: 4- 3 years). Fifty-seven had undergone coronary artery bypass grafting, 45 valve replacement, 28 combined Coronary bypass and valve replacement, and 8 other miscellaneous procedures which included cardiopulmonary bypass. Of the 138 patients studied, there was evidence of acute interstitial or necrotizing pancreatitis in 35 specimens (25 percent). In 8 of these, there was only interstitial pancreatitis, but in 27 there was pancreatic and peripancreatic necrosis.
509
Haas et al
TABLE II
Incidence and Pathologic Correlates of Pancreatitis Found at Autopsy
Subgroup
n
interstitlal
Low Output (Prolonged)
Necrotic
Perlop MI
IABP
Acute Hepatic Injury
Renal Failure
Ischemlc Bowel
Splenic Infarct
Sepsis
CholecyMitis
Deaths After Cardiac Surgery (n = 138) Early deaths (< 24 hr postop) With acute pancreatltis Without acute pancreatitis Late deaths (>24 hr postop) With acute pancreatitis Without acute pancreatifis
53
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
4
2
2
0
50%
0
49
0
0
0
47%
33%
85
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
0
0
0
0
0
0
4%
4%
0
0
2%
O
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
31
6
25
87%
84%
45%
87%
50%
22%
55%
16%
3%
54
0
0
72%
63%
43%
52%
11%
5%
41%
2%
0
Nonsurgical Deaths From Cardiac Causes (n = 93) With acute pancreatltis Without acute pancreatltis
9
4
5
100%
33%
. ..
45%
11%
11%
0
11%
11%
84
0
0
33%
19 %
...
14%
4%
4%
0
1%
1%
IABP = intraarterial b~ood pressure; MI = myocardial infarction.
Arterial emboli were noted in two cases, venous thrombi in two cases, a pancreatic abscess in three cases, and a pancreatic pseudocyst was noted in one case. Pancreatitis.was rarely found in patients who died within 24 hours after surgery. Only two cases of acute necrotizing pancreatitis and two cases of interstitial pancreatitis were observed in 53 such patients. The remaining 31 cases of acute pancreatitis were observed in the 85 patients who survived beyond the first postoperative day. The mean survival in this group was 13 :~ 8 days. In 25 of these 31 patients acute necrotizing pancreatitis developed and interstitial panereatitis developed in 6. Twenty-eight of the 31 patients (90 percent) experienced prolonged postoperative hypotension. Fourteen (45 percent) suffered perioperative myocardial infarctions, and 26 (84 percent) required intraaortic balloon insertion. Abnormal renal function developed in 2 (87 percent). Seventeen (54 percent) had evidence of sepsis, 15 (50 percent) had acute hepatic injury, 7 (24 percent) had ischemic bowel, 5 (16 percent) had splenic infarction, and 1 patient (3 percent) had acute cho!ecystitis. In 5 of the 25 patients with acute necrotizing pancreatitis, pancreatitis was considered to be the cause of death. Twenty patients who survived beyond the first postoperative day had evidence of minimal fibrosis of the pancreas consistent with a healed injury. Eight (40 percent) had experienced significant hypotension, 10 (50 percent) required intra-aortic balloon pump support, and 4 suffered perioperative myocardial infarction. Only I0 patients (50 percent) ex-
510
hibited abnormal renal function, 10 (50 percent) were septic, and 2 (10 percent) had acute hepatic injury. Two of these patients were noted to have arterial thrombosis. The mean survival in this group was 35 -A:7 days. Their length of postoperative survival was significantly greater than among patients with evidence of acute pancreatitis. In contrast pancreatic fibrosis was seen in only 2 of 53 patients who died within 24 hours of operation. Thirty-four patients died more than 24 hours after surgery with no pancreatic abnormalities found. The postoperative survival ranged from 2 to 180 days with a mean of 21 4- 6 days. Thirty-one (91 percent) experienced significant postoperative hypotension in the early postoperative period. Nineteen (56 percent) were treated with an intra-aortie balloon pump and 19 had perioperative myocardial infarctions. Abnormal renal function developed in only 18 patients (53 percent). Sepsis occurred in 12 of these patients (35 percent), acute hepatic injury in 4 (13 percent), ischemic bowel in 3 (10 percent), and isehemic interstitial infarction in 1 (3 percent). The latter two groups were similar and were considered together in Table II. Of 93 nonsurgicaI patients who died from cardiac causes, 38 died in low-output states whereas 55 died more suddenly without a significant period of hypoperfusion. Nine of the 38 patients (24 percent) with low output were found to have either acute interstitial pancreatitis (four patients) or acute necrotizing pancreatitis (five patients). There were no cases of acute pancreatitis in the 55 who died without experiencing periods of prolonged hypotension. Of the
The American Journal ¢# Surgery
Pancreatltis After Cardiac Surgery
nine patients in whom acute forms of pancreatitis developed, renal dysfunction occurred in four (44 percent), acute hepatic injury in one (11 percent), ischemic bowel in one (11 pro'cent), and cholecystitis in one (11 percent). Renal failure was present in only 12 of the remaining 84 patients (14 percent), with hepatic injury present in 3 (4 percent), ischemic bowel in 3 (4 percent), splenic infarction in I (1 percent), and acute cholecystitis in ] (1 percent). There was no apparent correlation be~.;wecnpal~creatic changes in any of the~e patient groups with cho!e!ithiasis, preoperative alcoholism, or administration of a specific medication.
o.6~-
0
Comments With the expansion of the field of cardiac surgery, pancreatitis has become recognized as one of its rare but potentially fatal complications [1-9]. Early reports suggested an incidence approaching 1 percent. More recently, Rose et al [5] reported a 0.13 percent incidence of neczotizing pancreatitis, which is consistent with our observation of 0.2 percent. The prevalence of pancreatitis discovered at autopsy after cardiac surgery appears to be much higher than among patients who survive. Feiner [15] has reported 34 cases of pancreatitis from 182 complete autopsies after cardiac surgery. In 29 of these cases (16 percent) there was no apparent cause other than cardiac surgery. Panebianco et al [16] reported eight cases of pancreatitis from 34 autopsies (24 percent) of cardiac surgical patients. One hundred one autopsies after cardiopulmonary bypass at this institution between 1973 and 1975 were reviewed by Warshaw and O'Hara [7] who found 11 (11 percent) cases of major pancreatic injury. Furthermore, that figure increased to 19 percent when only patients who survived beyond the first postoperative day were considered. In the present study, we found evidence of acute pancreatitis in 35 of 138 patients (25 percent) who died after cardiac surgery. There are several possible explanations for the higher prevalenc~ of acute pancreatitis in the autopsy study as compared with the clinical study. First, acute pancreatitis may be more likely to occur in Vohano149, AI~! 1985
P~ s,
ill
0.2 1
w=
~0.8~POST-OP{D'4YI] 0.4 ~----" 0.2
Prospective study of subclinical pancreatic injury: Thirty consecutive unselected patients had
serum levels of pancreatic isoamylase and ribonuclease measured preoperatively and on the first, fourth, and seventh postoperative days. Eight of the 30 patients (27 percent) had an increase of pancreatic isoamylase on the first postoperative day (Figure 1). The levels returned to normal by the fourth day in four of these patients. In four of the eight ~1o ~ ~ per.. cent), levels of serum ribonuclease also increased significantly. The increase characteristically appeared on the fourth postoperative day and remained elevated for 3 days. No clinical manifestations suggestive of pancreatitis developed in any of these patients.
P,
~ POST-OP(DAY4) ~
o.4 0.2 0
................
x_ , l
0.6 0.4 0.2 Origin 10
20
30
40
50
60
70
FRACTION NUMBER Figure 1. Serial determinations of amylase isoenzymes in arep. resenfative pat/ent after cardlopulmonary bypass. P1 and P= are the pri~clpal pancreatic Isoamytase levels and $1 Is the principal nonp~ncreatlc Isoamylese leve£ Elevations of both P1 and P2 suggest pancreatic Injury. OD = optical denslly.
patients who have multiple complications after cardiac surgery and are therefore more likely to die and be autopsied. Second, acute pancreatitis, if it occurs after cardiac surgery, may contribute to a fatal course. Third, because acute pancreatitis is very difficult to diagnose after bypass, its clinical incidence may be underestimated. There is evidence that the pancreas, like the k~dney, may be very sensitive to the effects of hypoperfusion. Experimental hypotension in dogs has been shown to cause a marked reduction in pancreatic blood flow, which in turn is associated with injury to the acinar cells, including lysosomal swelling, vacuolization, and the release of both proteolytic enzymes and a myocardial depressant factor [17]. Furthermore, the potential injurious effects of pancreatic ischemia have been emphasized by Popper et al [18], who showed in dogs whose pancreatic ducts had been ligated, that short periods of ischemia (15 minutes) could convert mild pancreatic edema to severe acute hemorrhagic pancreatitis. Broe et al [19] demon8trated the production of severe edema in the isolated dog pancreas subjected to 2 hours of total ischemia. Anderson [20] also concluded that ischemia may be 511
Haas et al
an important factor in the progression of pancreatic edema to necrotizing pancreatitis. Pfeffer et al [21] showed in dogs that disruption of capillary blood flow by injection of microspheres into the pancreatic arteries caused pancreatitis. Conversely, improvement of pancreatic blood flow by postganglionic sympathetic denervation was shown by Wright and Goodhead [22] to reverse or prevent an otherwise lethal pancreatic injury in dogs. Low molecular weight dextran has also been shown to protect the pancreatic microcirculation against pancreatitis during experimental ischemia [23]. In the clinical setting, the development of acute pancreatitis has been correlated with hypoperfusion. Tilney et al, [24] described 18 patients in whom renal failure developed, presumably due to hypoperfusion, after operation for ruptured abdominal aortic aneurysms. Of these patients, pancreatitis developed in nine (50 percent). Wsrshaw and O'Hara [7] found a similar high correlation between pancreatitis and acute renal tubular necrosis in patients who died after experiencing oligemic and lmnoligemic shock. In their prospective study of 13 patients who were found to have hyperamylasemia after a period of shock, the elevated serum amylase levels in all 13 was of pancreatic origin. In 4 of the 13 patients, clinical evidence of acute pancreatitis developed. The hypothesis that the form of pancreatitis under consideration is caused by an ischemic injury to the pancreas is supported by several of our observations. All nine patients who developed clinically manifest severe necrotizing pancreatitis had experienced prolonged hypoperfusion during the early postoperative period, whereas the three patients with mild pancreatitis did not have comparable antecedent episodes of hypotension. Acute pancreatitis was found most commonly at autopsy in patients who had experienced significant postoperative hypoperfusion and survived beyond the first postoperative day. Pancreatitis in these patients correlated highly with other organ system injuries known to result from low output, such as renal failure and acute infarction of the liver, bowel, or spleen. The autopsy data showed pancreatitis was rarely evident within the first 24 hours after surgery, whereas 36 percent of those who died later had pancreatitis. Among those nonsurgical patients who died from acute cardiac events, there were no cases of acute pancreatitis in those who died suddenly, but 9 of 38 patients (24 percent) who died after a period of cardiogenic shock had acute interstitial or necrotizing pancreatitis. Although acute pancreatitis can produce hypotension, itseems clear that in the present circumstances pancreatitis wa~ the effect rather than the cause. In addition to postoperative cardiovascular instability,there isalso evidence that cardiopulmonary bypass m a y predispose to pancreatitis. Serum amyhse levels and renal amylase to creatinine clearance
5~2
ratios have been shown to increase to a significantly greater extent after thoracic procedures using cardiop'dmonary bypass than after thoracic procedures not requiring cardiopulmonary byp ~ass[8,?.5"].Murray et al [26] showed that use of a pulsatile perfusion system rather than a nonpulsatile system, led to fewer elevations of the amylase to creatinine clearance ratio after cardiopulmonary bypass, and concluded that nonpulsatile perfusion may be a factor in pancreatic injury. In addition, hypothermia has been reported to cause pancreatic necrosis and may be a factor in the patient who has been cooled during cardiopulmonary bypass [27]. Atheromatous or cholesterol emboli may be broken off during cannulation and cross-clamping of the aorta and embolize to the pancreas [28,29]. Venous sludging, which occurs with extracorporeal circulation, has also been reported to injure the pancreas and may further add to the detrimental effects of cardiopulmonary bypass in the pancreaas [20]. Perhaps these factors during bypass cause a mild and relatively insignificant pancreatic lesion. This thesis is supported by our finding of elevated pancreatic isoamylase levels in a fourth of the asymptomatic patients studied after uneventful cardiac surgery. Increases in the serum ribcnuclease level in half of these same patients suggest that there may be some associated pancreatic cell death [13].If the pancreas is sensitized by this injury, a further period of hypoperfusion added during the postoperative period seems to convert the process to necrotizing pancreatitis, m u c h as in the experiments of Popper et al [18]. What we are describing is a spectrum of pancreatic injury demonstrable after cardiac surgery. At one end is a minimal phenomenon, detectable only by relatively sensitive biochemical tests in 25 percent of patients. At the other end is a highly lethal necrotizing pancreatitis, recognized during life in fewer than 1 percent of patients (less than 0.2 percent in our clinical series), but an estimable cause of morbidity and mortality in cardiac surgical patients, being present in 25 percent of those who died after surgery and the principal cause of death in 4 percent~ In between axe those cases that are either clinicaily mild and respond well to conservative treatment or subclinical and heal spontaneously without being diagnosed. These cases may correspond to those classified as chronic fibrosis and healing when found incidentally at autopsy. The recognition of pancreatitis in the patient after cardiac surgery is difficult, partly because so much else may be going on to obscure the abdominal problems and partly because the clinical and laboratory manifestations of this form of pancreatitis are unlike those of the common form. Distention, ileus, and leukcmytosis are typical of this syndrome whereas pain, tenderness, and hyperamylasemia are not. It
Th* ~
~
~ S~nr
Pancreatltis After Cardiac Surgery
may be that ischemic pancreatitis after cardiac surgery produces a relatively pure necrotic injury, free of the primary inflammatory effects of activated enzymes and the abdominal symptoms and signs they produce so characteristically in other forms of pancreatitis. It has become our practice to place pancreatitis high on the list of potential diagnoses in any patient with serious abdominal complaints after heart surgery and to follow such patients with ultrasonography and computerized tomography to increase the identification of pancreatitis. The treatment of pancreatitis after bypass remains difficult and unsettled. Clearly, some patients have resolution of the injury spontaneously or with conservative treatment. We believe that patients who are unstable or deteriorating require laparotomy and aggressive debridement and drainage [30]. Our few survivors were those in whom the most extensive operations were performed to remove devitalized tissues and provide egress for the products of pancreatitis. For the future, prevention of the pancreatic injury is paramount. Using serum isoamylase and ribonuclease as described herein, we have begun a trial of low molecular weight dextran to see if the pancreas can be protected from ischemic injury during cardiopulmonary bypass.
Summary We have described a spectrum of pancreatic surgery after cardiopulmonary bypass. At one end is a subclinical lesion which was manifested only by elevations in serum isoamylase levels (27 percent of patients) and increased ribonuclease levels (13 percent of patients) in asymptomatic patients followed after cardiac surgery. At the other end is a severe and often lethal necrotizing pancreatitis. Acute necrotizing pancreatitis was found at autopsy in 25 percent of 138 patients who died after cardiac surgery, and it correlated strongly with low output, acute tubular necrosis, and infarction of the liver, spleen, or bowel. It was the principal cause of death in 4 percent of these patients. In addition, 24 percent of 38 nonsurgical patients who died from cardiac failure and hypoperfusion had acute pancreatitis at autopsy, whereas acute pancreatitis was not observed in 55 nonsurgical patients who died without a significant period of low output. Acute pancreatitis was recognized postoperatively in 12 patients (0.2 percent). Three had mild pancreatitis, and all responded well to conservative therapy. In nine patients, fulminant necrotizing pancreatitis developed. Their courses were characterized by significant early postoperative hemodynamic comprpmise, abdominal distention, ileus, fever, and episodes of late vascular instability associated with hypocalcemia. The diagnosis of pancreatitis was usually missed because of the absence of pain, tenderness and hyperamylasemia. The
¥oi~1e 149, Apdl 1985
diagnosis was confirmed at laparotomy in eight patients and at autopsy in one. The only two survivors among the nine with severe cases had aggressive mobilization, debridement, and wide drainage of the necrotic pancreas. We suggest that (1) a mild subclinical injury to the pancreas may occur as a consequence of cardiopulmonary bypass and may progress to severe ischemic necrosis if hypoperfusion follows in the postoperative period, (2) the presentation of necrotizing pancreatitis may be atypical in the cardiac surgical patient and should be considered if nonspecific abdominal symptoms are present, and (3) aggressive debridement and drainage may be the optimal treatment for aggressive forms of this disease. Acknowledgment: We thank Denise Duchaineyfor her technical skills with the isoamylase and ribonuclease determinations and Mary Chasse for preparing the manuscript with diligence and patience.
References I. Harjola P-T, SIItanen P, Appelqulst P, Laustela E. Abdominal complications after open heart surgery. Ann Chit Gynaecol 1968;57:272-4. 2. Horton EH, Murthy SK, Seal RME. Haemorrhagic necrosis of small intestine and acute pancreatltts following open heart surgery. Thorax 1968;23:438-45. 3, Wallwork J, David.senKG. The acute abdomen following cardiopulmonary bypass surgery. Br J Surg 1980;67:410-2. 4. Housman L, Bonchek L, Lambert L, Grunkenmeier G, Starr A. Prognosisof patients after mttral commtssurotomy.J "rhorac Cardlovasc Surg 1977;73:742-5. 5. Rose DM, Ranson JHC, Cunntngham MN Jr, Spencer FC. Patterns of severe pancreatic Injury following cardiopulmonary bypass. Ann Surg 1984;199:168-72. 6. HanksJB, Curtis SE, Hanks FF, et al. Gastrointestinal complications after cardlopulmonary bypass. Surgery 1982;92: 394-9. 7. Warshaw AL, O'Hara PJ. Susceptibility of the pancreas to ischemlc Injury in shock. Ann Surg 1978; 188:197-201. 8. Traverse LW, Fen'ad BT, Buckberg GD, Tompklns RK. Elevated postoperative renat clearance of amylase without pancreaUtls after cardiopulmonery bypass. Am J Surg 1977; 133:298-303. 9. Smith CR, Schwartz S, Amylase:creatinine clearance ratios, serum amylase, and lfpase after operations with cardlopulmonary bypass. Surgery 1983; 194:458-63. 10. Longnecker DF. Pathology and pathogenesls of diseases of the pancreas. Am J Pathol 1982;107:103-21. 11. Warshaw AL, Lee KH. Characteristic alterations of serum Isoenzymasof amylase In disease of liver, pancreas, salivary gland, lung, and genitalia. J Surg Res 1977;22:362-9. 12. Wars;taw AL, Fournler PO. Release of dbonuclease from anoxtc pancreas, Surgery 1984;95:537-41. 13. Warshaw AL, Lee K-H. Serum rlbonuclease elevations and pancreatic necrosis in acute pancreaUtis. Surgery 1979; 86:227-34, 14. Banks PA, Sidl S, Gelman ML, Lee K-H, Warshaw AL. Amylase-creaUnlne clearance ratios and serum amylase Isoenzymes in moderate renal insufficiency. J CItn Gastroenterol 1979;1:331-5. 15. Felner H. Pancreatitls after cardiac surgery, a morphologtc study. Am J Surg 1976;131:684-8. 16. PanebiancoAC, Scott SM, Dart CH Jr, TaksJ'oT, Echegaray I@A. Acute pancreatttis following extracorporeal circulation. Ann
513
Haas et al
Thorac Surg 1970:9:562-8. 17. Lefer AM, Spath JA Jr. Pancreatic hypoperfuslon and the production of a myocardial depressant factor. Ann Surg 1974; 179: 868-76. 18. Poppet HL, Necheles H, Russell KC. Transition of pancreatic edema to pancreatic necrosis. Surg Gynecol Obstet 1978;87:79-82. 19. Broe PJ, Zuidema GO, Cameron JL. The role of tschemia in actde pancreatltis: studies with an isolated perfused canine pancreas. Surgery !982;91:377-82. 20. Anderson MC. Venous stasls tn the transition of edematous pancreatltis to necrosis. JAMA 1963; 183:534-7. 21. Pfeffer RB, Lazarinl A, Robertson A, Sadefi D. Gradations of pancreatttts: edematous through hemmorhagic experimentally produced by controlled injection of microspheres into blood vessels in dogs. Surgery 1962;51:764. 22. Goodhead B, Wright PW. The effect of post ganglionic sympathectomy on the development of hemorrhagic pancreatltis In the dog. Ann Surg 1969;170:951-9. 23. Wright PW, ~ h e a d B. The value of dextran In 1he treatment of experimental pancreatttls. Surgery 1970;67:807-15. 24. Tilney NI, Bailey GL, Morgan AP, Sequential system failure after rupture of abdominal aortic aneurysm. Ann Surg 1973; 178:117-22. 25. Hennlngs B, Jacobsen G. Postoperative amylase secretion: a study following thoracic surgery with and without extracorporeal circulation. Ann Clin Res 1974;6:215-22. 26. Murray WR, Mittra S, MIttra D, Roberts LB, Taylor KM. The amylase creatlnine clearance ratio following cardlopulmonary bypass. J Thorac Cardlovasc Surg 1981;82:248 -53. 27. MacLean D, Morrison J, Grlfflth PD. Acute pancreatltts after accidental hypothermta and hypothermic myxoedema. Br Med J 1273;4:757. 28. Probstetn JG, Ram A, Joshi MB, Blumenthal H. Atheromatous embolization: an etiology of pancreatitis. Arch Surg 1957; 75:566-72. 29. Moidveen-Geronimus M, Merrlam JC Jr. C,holesterol embolism. Circulation 1967;35:946-53. 30. Warshaw AL, Imbembo AL, CIvetta JM, Daggett WM. Surgical intervention in acute necrotizlng pancreatltis. Am J Surg 1974; 127:484-91. Discussion
A l e x a n d e r S. G e h a (New Haven, CT): We have been interested in these pancreatic abnormalities after cardiac surgery for a while, and we do have a prospective study upon which we are embarking t h a t is somewhat similar to yours, Dr. Warshaw. Our observations so far indicate t h a t there are a fair amount of chemical abnormalities encountered, the meaning of which I am not sure. So far, we have been somewhat less impressed than you have with the occurrence of a fulminating necrotizing pancreatitis after cardiac surgery. Dr. Warshaw, I think you indicated that there is a fairly high incidence of chemical abnormalities after cardiac surgery as you encountered them in 27 percent of the patients, and t h a t this includes an elevation of the ribonuclease level in a fairly substantial number. M y question is, how specific is the ribonuclease level for pancreatic cell deaths versus other orgma death, and were these patients with the elevated enzyme levels, exhibiting any hypotension, low cardiac output, or other factors t h a t m a y have predisposed them to this? Are you sure that these are not abnormalities of the enzymes t h a t are observed because of lack of urinary clearance versus overproduction of these enzymes? Have you looked at ,~rine amylase clearance, for instance? W h a t is your recommendation regarding this problem? If abnor-
514
malities in the amylase level and elevated enzyme levels are encountered, what should be done? W h a t is the specificity of the computerized tomographic scan and ultrasonography? Should we perform these tests when we see elevation in the enzyme levels in 27 percent of the patients in todays atmosphere of diagnosis-related groups (DRGs) with so much emphasis being placed on cost containment? A n d r e w L. W a r s h a w (closing): I want to emphasize t h a t what we are describing here is not simply an isolated group of patients with pancreatitis, but an injury to the pancreas that follows an entire spectrum ranging from something that is b~ing missed by current methods to something that is being seen with lethal results at the other end. We are hypothesizing and we are presenting evidence from our isoamylase and ribonuclease studies that there is a sensitizing event during cardiopulmonary bypass, and t h a t the subsequent "detonation of the bomb" come0 with some further event such as postoperative hypotension which further injures the pancreas. We are emphasizing t h a t this form of pancreatitis, when it does occur, is a particularly bad form. I think it's hard to think of another group of panereatitis patients who have an overall mortality from clinically recognized pancreatitis in the 80 percent range. So, we have got to do something different in the future and not simply debride them, although we think that is the best thing at the moment. We are suggesting that this is a different form of pancreatitis than what usually occurs. Usually, the course of events is chemical inflammation followed by, in some patients, necrosis, which marks the particularly difficult cases, which are only 5 to 10 percent of the overall cases. In cardiac patients, necrosis is the initial event; it's the central event; it is one that does not have the same diagnostic clues such as pain, tenderness, and hyperamylaseraia, and only becomes obvious later when the second a r y signs of inflammation of the necrotic tissue set in, or perhaps it is not recognized at all until after the patient has died. Dr. Geha, you asked how specific the isoamylase and ribonuclease levels are? They are highly specific, at least by the methods we have used. Pancreatic isoamylase is produced only by the pancreas; it increases only in the presence of pancreatic injury except in cases of renal dysfunction, so your point about renal clearance is correct. However, in this particular study, the prospective patients all had normal renal function, so what we were looking a t was specifically pancreatic injury. The test is no good in patients with renal failure and is probably, therefore, not useful in the severely affected patients we described in the other portion of the study. The ribonuelease level we measured i3 also relatively specific. There are many kinds of ribonuclease in the body. It. is present in almost every cell t h a t has a nucleus. However, the polycytidine-specific ribonuclease that we have used has been shown in our laboratory studies i n the rat to be released only from the dying pancreas and not from the dying heart, intestine, liver, spleen, kidney, or any other organ t h a t we tested. So we think it is meaningful in this case, again with the proviso t h a t renal function is normal; ribonuclvase levels in the serum are highly dependent upon that. It is therefore useful in a study such as this of prospective normal patients who were not hypotensi"e postoperatively, b u t it is not useful in detectin~ pancrear~is in the critically ill patients.
The A m e ~
~ l
o~ Surgery
Pancreatttis After Cardiac Surgery
The renal clearance rates in the sick patienk~ were often abnormal. The amylase to creatinine clearance ratios and creatinine clearance ratios were frequently elevated and were not useful in any of the diagnostic tests. We do think that computerized tomographic scans are goirig to be helpful, at least in the more advanced cases. We do not yet know how sensitive that modali,'.y is. T h e lesson of all this is simple because we do not have all the answers. We are calling attention to a phenomenon. We think that this form of panereatitis is somewhat different. We think it is somewhat more difficult ~n many
ways both to recognize and to manage when it occurs. We can only exhort you at the moment to be more aware of it than perhaps you have been. We found as our index of suspicion has increased, we have begun to think of the diagnosis more often, we have begun to make it earlier and to order the computerized tomographic scans earlier, and perhaps we will do better preventing it in the future. We are embarking on a prospective double-blind trial of vasoactive agents of various kinds to try to protect the pancreas during this critical period, and we hope that will reduce the incidence. Time will tell.
This discussion section has been abbreviated due to space limitations. The full text can be obtained from the reprints author listed on the title page.
Volume149,Apd11985
515
Acute pancreatitis is a reported but poorly understood complication of cardiac surgery. Although it is recognized in less than 1 percent of patients undergoing cardiopulmonary bypass, pancreatitis in this setting frequently occurs in conjunction with renal and other organ failure and runs an aggressive, fatal course [1,5-7]. The diagnosis and treatmen Lof acute pancreatitis after cardiac surgery are difficult. Its symptoms may be atypical and misleading or may be obscured by other complications [5,7]. The usual laboratory characteristics of pancreatitis, including hyperamylasemia, do not appear to be reliable indications of the disease in patients after bypass [8,9]. The diagnosis is often missed until autopsy, but even when it is made antemortem, the mortality rate of affected patients has been higher than in any other recognized form of pancreatitis. It is our purpose to elucidate the spectrum of pancreatic injury, to describe the clinical and laboratory features of patients in whom pancreatitis develops after cardiopulmonary bypass, to examlne risk factors for pancreatitis, and to suggest an approach to effective treatment.
of interstitial fibrosis, stromal fibrosis, and atrophy; (2) acute interstitial by the findings of acinar and stromal edema and an interstitial inflammatory infiltrate; or (3) acute necrotizing by the findingt~of edema, acute inflammatory infiltrate ass~iated with parenchymal fat necrosis, peripancreatic fat necrosis, or both. Pathologic findings were then correlated with cause of death, time of death, associated disease, noncardiac risk factors, and relevant laboratory findings. Patients whose pancreases were autolyzed and those with no specimehs taken for histologic examination were assumed not t(, have pancreatitis for the purposes of this study. Thirty patients were studied prospectivelyafter cardiac surgery to determine the prevalence of subclinical pancreatic h,jury. Serum samples were obtained preoperatively and again on the first, fourth, and seventh days after cardiac surgery for determination of total amylase and pancreatic isoamylase levels (a sensitive and specific index of pancreatic inflammation [I I ]) and of polycytidine-specific ribonuclease (an index of pancreatic cell necrosis
[12,I3]). lsoenzymesof amylase (isoamylases)were separated and quantified by polyacrylamide gel electrophoresis and saccharogenicassay as described by Warshawand Lee Ill]. Ribenuclease was assayed by a modificationof the method of Reddi and Holland [13].
Patients and Methods
Results
The records of 12 patients in whom pancreatit,is was diagnosed after cardiac surgery at the Massachuseits General Hospital between June 1979 and January 1984 were reviewed for physical findings, laboratory data pertinent to the diagnosis, risk factors, clinical course, and therapy. The pathologic findings of the pancreas were reviewed in 138 other patients who died after cardiac surgery and 93 nonsurgical patients who died from acute cardiac disease between December 1975 and April 1981. The pancreatic tissue slides were reviewed by a single pathologist (HTA) without his knowledgeof the clinical history and classified according to Longneeker [10] as (1) chronic by the findings
Clinical pancreatitis after cardiac surgery: Between June 1979 and January 1984, approximate]y 5,400 cardiac procedures were performed. During this time, postoperative pancreatitis was diagnosed antemortem in 12 of these patients (0.2 percent). There were nine male and three female patients who underwent coronary artery bypass (eight patients), coronary artery bypass with aortic valve replacement (two patients), or coronary artery bypass with mitral valve replacement (two patients). Their ages ranged between 50 and 73 years with a mean of 62 years. Preoperative cardiac indexes ranged from 2 to 3.2 liters/rain per m 2 (mean 4- standard deviation, 2.6 .~ 0.2) and ejection fractions from 40 to 60 percent (5Y + 6). None of them manifested signs of cardiogenic shock before operation, but one patient had an intraaortic balloon pump placed to control angina pectoris. Perfusion times ranged from 68 minutes to 208 minutes (mean 134 minutes -2-.46). Three pa-
From the GeneralSurgicalServices. MassachusettsGeneralHospitaland the Department of Surgery, Harvard Medical ~ 1 , Boston,Massachusotts. Requests for reprintsshouldbe addressedto AndrewL. Warshaw,MD, Departmentof Surgery,MassachusettsGeneralHospital,15ParkmanStreet, Suite 336, Boston,Massachusetts02 ! 14. Presentedat the 65LhAnnualMeetingof the New EnglandSurgicalSociety. DlxvllleNotch, New Hampshire,October 12-14, 1984.
508
The AmedeM ~k~ma;.~S u r ~
Pancreatltis After Cardiac Surge,'y
TABLE I
Clinical F e a t u r e s of Necrotizlng PancreaUlls
in Nine Patlenls After Cardlopulmonary Bypass Clinical Finding Early pain and tenderness Abdominal distention and iteus Fever (temperature >101°F) Late abdominal pain and tenderness Gardiogenlc shock in early postoperative course Sudden episodes of late hypotension Leucocytosi,=; (> t2,000 ceUs/mm 3) Unexplained hypocalcemta Early hypera=mylasemla Late hyperar,nylasernia Abdomen explored Deaths
n 0 9 9 9 8 8 9 6 2 4 8 7
% 0 100 100 10(;~ 89 89 100 67 22 44 89 78
tients had hyperlipidemia, one had a histery of alcohol abuse, and two had asymptomatic gallstones. After operation, mild pancreatitis developed in¢ 3 of the 12 patients, The diagnosis was based on abdominal pain, decreased bowel sounds, nausea, and hyperamylasemia. In each case, the symptoms resolved with standard treatment consisting of bowel rest and intravenous fluid support. None of these patients experienced hemodynamic instability during the postoperative period, and their serum total and ionized calcium levels remained within normal limits. Fulminating necrotizing pancreatitis developed in nine patients which was later confirmed by either laparotomy or autopsy. The clinical course of all of these patients was marked by severe hemodynamic instability in the early postoperative period. Eight of them suffered a prolonged low-output state with a cardiac index of less than 2 liters/rain per m 2 and mean aortic pressures less than 80 mm Hg. The ninth patient suffered multiple episodes of ventricular tachyarrh~,thmias with concomitant episodes of hypotension. All nine required both o~-and ~-adrenergic support, and six required intraaortic balloon pump support during this time. The earliest and most common symptoms referable to pancreatitis were abdominal distention and hypoactive bowel sounds (Table I). These were observed in all nine patients within 24 to 48 hours of surgery. Significant abdominal pain and tenderness, however, were not noted until much later in the course of the disease. Prolonged fevers with temperatures above 101°F were present in all patients. Cardiac function improved in all cases, and eventually all patients were weaned from intraaortic balloon pump, inc~tropic support, or both. Despite myocardial recovery, eight of the nine patients were observed to have subsequent episodes of sudden hypotension 3 to 7 days after operation. It. st:: of these patients, serum ionized calcium levels were
Volume 1~.-9,Aprl! 1985
measured during one or more of these hypotension episodes and were found to be less th~n 0.9 raM/liter (normal 1.1 to 1.2 raM/liter) in aH six. Total serum amylase levels were of little value in diagnosing or following progression of the disease in patients with necrotizing pancreatitis. Amylase levels in the early postoperative period were m)rmal in seven of the nine patients and only modestly elevated in two, and they were within normal limits in five of the nine patients when studied "n the later clinical course. Of the four patients in whom serum amylase levels increased in the later clinical course, the levels were less than twice normal in two and could have been attrihuted to concomitant renal dysfunction [141. Leukocytosis (white blood cell count more than 12,000 cells/mm 3) was present in all nine patients with necrotizing pancreatitis. Frequently these pa. tients were thought to be septic, and despite negative blood cultures, six of the nine patients with necrotizing pancreatitis began receiving broad-spectrum antibiotics. The diagnosis of necrotizing pancreatitis was usually not made until very late in its course. Persistence of fever and ileus, the development of late abdominal pain and mild tenderness, and progressive systemic deterioration led to abdominal exploration in eight of the nine patients with necrotizing pancreatitis. Pancreatitis, however, was the leading diagnosis in only two of the eight patients who came to laparotomy and was considered in the differential diagnosis in just two others. Three patients were explored with a leading diagnosis of cholecystitis, two with a leading of peritonitis of unknown cause, and one with a leading diagnosis of ischemic bowel. The operative procedure in the eight patients was mobilization of the pancreas and debridement and drainage, with comcomitant gastrostomy, jejunostomy, and cholecystostomy. Two of these patients survived whereas six eventually died from ongoing pancreatitis, sepsis, and hemorrhage. The ninth died without exploration. Renal failure developed in six and liver failure in four of the nine patients. Autopsy studies a f t e r c a r d i a c s u r g e r y : The pancreatic pathologic characteristics and clinical records of 138 patients who died after cardiac surgery and 93 nonsurgical patients who died from acute cardiac disease between December 1975 and April 1981 were reviewed (Table II). The surgical group consisted of 94 male and 44 female patients who ranged in age from 27 to 85 years (61: 4- 3 years). Fifty-seven had undergone coronary artery bypass grafting, 45 valve replacement, 28 combined Coronary bypass and valve replacement, and 8 other miscellaneous procedures which included cardiopulmonary bypass. Of the 138 patients studied, there was evidence of acute interstitial or necrotizing pancreatitis in 35 specimens (25 percent). In 8 of these, there was only interstitial pancreatitis, but in 27 there was pancreatic and peripancreatic necrosis.
509
Haas et al
TABLE II
Incidence and Pathologic Correlates of Pancreatitis Found at Autopsy
Subgroup
n
interstitlal
Low Output (Prolonged)
Necrotic
Perlop MI
IABP
Acute Hepatic Injury
Renal Failure
Ischemlc Bowel
Splenic Infarct
Sepsis
CholecyMitis
Deaths After Cardiac Surgery (n = 138) Early deaths (< 24 hr postop) With acute pancreatltis Without acute pancreatitis Late deaths (>24 hr postop) With acute pancreatitis Without acute pancreatifis
53
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
4
2
2
0
50%
0
49
0
0
0
47%
33%
85
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
0
0
0
0
0
0
4%
4%
0
0
2%
O
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
31
6
25
87%
84%
45%
87%
50%
22%
55%
16%
3%
54
0
0
72%
63%
43%
52%
11%
5%
41%
2%
0
Nonsurgical Deaths From Cardiac Causes (n = 93) With acute pancreatltis Without acute pancreatltis
9
4
5
100%
33%
. ..
45%
11%
11%
0
11%
11%
84
0
0
33%
19 %
...
14%
4%
4%
0
1%
1%
IABP = intraarterial b~ood pressure; MI = myocardial infarction.
Arterial emboli were noted in two cases, venous thrombi in two cases, a pancreatic abscess in three cases, and a pancreatic pseudocyst was noted in one case. Pancreatitis.was rarely found in patients who died within 24 hours after surgery. Only two cases of acute necrotizing pancreatitis and two cases of interstitial pancreatitis were observed in 53 such patients. The remaining 31 cases of acute pancreatitis were observed in the 85 patients who survived beyond the first postoperative day. The mean survival in this group was 13 :~ 8 days. In 25 of these 31 patients acute necrotizing pancreatitis developed and interstitial panereatitis developed in 6. Twenty-eight of the 31 patients (90 percent) experienced prolonged postoperative hypotension. Fourteen (45 percent) suffered perioperative myocardial infarctions, and 26 (84 percent) required intraaortic balloon insertion. Abnormal renal function developed in 2 (87 percent). Seventeen (54 percent) had evidence of sepsis, 15 (50 percent) had acute hepatic injury, 7 (24 percent) had ischemic bowel, 5 (16 percent) had splenic infarction, and 1 patient (3 percent) had acute cho!ecystitis. In 5 of the 25 patients with acute necrotizing pancreatitis, pancreatitis was considered to be the cause of death. Twenty patients who survived beyond the first postoperative day had evidence of minimal fibrosis of the pancreas consistent with a healed injury. Eight (40 percent) had experienced significant hypotension, 10 (50 percent) required intra-aortic balloon pump support, and 4 suffered perioperative myocardial infarction. Only I0 patients (50 percent) ex-
510
hibited abnormal renal function, 10 (50 percent) were septic, and 2 (10 percent) had acute hepatic injury. Two of these patients were noted to have arterial thrombosis. The mean survival in this group was 35 -A:7 days. Their length of postoperative survival was significantly greater than among patients with evidence of acute pancreatitis. In contrast pancreatic fibrosis was seen in only 2 of 53 patients who died within 24 hours of operation. Thirty-four patients died more than 24 hours after surgery with no pancreatic abnormalities found. The postoperative survival ranged from 2 to 180 days with a mean of 21 4- 6 days. Thirty-one (91 percent) experienced significant postoperative hypotension in the early postoperative period. Nineteen (56 percent) were treated with an intra-aortie balloon pump and 19 had perioperative myocardial infarctions. Abnormal renal function developed in only 18 patients (53 percent). Sepsis occurred in 12 of these patients (35 percent), acute hepatic injury in 4 (13 percent), ischemic bowel in 3 (10 percent), and isehemic interstitial infarction in 1 (3 percent). The latter two groups were similar and were considered together in Table II. Of 93 nonsurgicaI patients who died from cardiac causes, 38 died in low-output states whereas 55 died more suddenly without a significant period of hypoperfusion. Nine of the 38 patients (24 percent) with low output were found to have either acute interstitial pancreatitis (four patients) or acute necrotizing pancreatitis (five patients). There were no cases of acute pancreatitis in the 55 who died without experiencing periods of prolonged hypotension. Of the
The American Journal ¢# Surgery
Pancreatltis After Cardiac Surgery
nine patients in whom acute forms of pancreatitis developed, renal dysfunction occurred in four (44 percent), acute hepatic injury in one (11 percent), ischemic bowel in one (11 pro'cent), and cholecystitis in one (11 percent). Renal failure was present in only 12 of the remaining 84 patients (14 percent), with hepatic injury present in 3 (4 percent), ischemic bowel in 3 (4 percent), splenic infarction in I (1 percent), and acute cholecystitis in ] (1 percent). There was no apparent correlation be~.;wecnpal~creatic changes in any of the~e patient groups with cho!e!ithiasis, preoperative alcoholism, or administration of a specific medication.
o.6~-
0
Comments With the expansion of the field of cardiac surgery, pancreatitis has become recognized as one of its rare but potentially fatal complications [1-9]. Early reports suggested an incidence approaching 1 percent. More recently, Rose et al [5] reported a 0.13 percent incidence of neczotizing pancreatitis, which is consistent with our observation of 0.2 percent. The prevalence of pancreatitis discovered at autopsy after cardiac surgery appears to be much higher than among patients who survive. Feiner [15] has reported 34 cases of pancreatitis from 182 complete autopsies after cardiac surgery. In 29 of these cases (16 percent) there was no apparent cause other than cardiac surgery. Panebianco et al [16] reported eight cases of pancreatitis from 34 autopsies (24 percent) of cardiac surgical patients. One hundred one autopsies after cardiopulmonary bypass at this institution between 1973 and 1975 were reviewed by Warshaw and O'Hara [7] who found 11 (11 percent) cases of major pancreatic injury. Furthermore, that figure increased to 19 percent when only patients who survived beyond the first postoperative day were considered. In the present study, we found evidence of acute pancreatitis in 35 of 138 patients (25 percent) who died after cardiac surgery. There are several possible explanations for the higher prevalenc~ of acute pancreatitis in the autopsy study as compared with the clinical study. First, acute pancreatitis may be more likely to occur in Vohano149, AI~! 1985
P~ s,
ill
0.2 1
w=
~0.8~POST-OP{D'4YI] 0.4 ~----" 0.2
Prospective study of subclinical pancreatic injury: Thirty consecutive unselected patients had
serum levels of pancreatic isoamylase and ribonuclease measured preoperatively and on the first, fourth, and seventh postoperative days. Eight of the 30 patients (27 percent) had an increase of pancreatic isoamylase on the first postoperative day (Figure 1). The levels returned to normal by the fourth day in four of these patients. In four of the eight ~1o ~ ~ per.. cent), levels of serum ribonuclease also increased significantly. The increase characteristically appeared on the fourth postoperative day and remained elevated for 3 days. No clinical manifestations suggestive of pancreatitis developed in any of these patients.
P,
~ POST-OP(DAY4) ~
o.4 0.2 0
................
x_ , l
0.6 0.4 0.2 Origin 10
20
30
40
50
60
70
FRACTION NUMBER Figure 1. Serial determinations of amylase isoenzymes in arep. resenfative pat/ent after cardlopulmonary bypass. P1 and P= are the pri~clpal pancreatic Isoamytase levels and $1 Is the principal nonp~ncreatlc Isoamylese leve£ Elevations of both P1 and P2 suggest pancreatic Injury. OD = optical denslly.
patients who have multiple complications after cardiac surgery and are therefore more likely to die and be autopsied. Second, acute pancreatitis, if it occurs after cardiac surgery, may contribute to a fatal course. Third, because acute pancreatitis is very difficult to diagnose after bypass, its clinical incidence may be underestimated. There is evidence that the pancreas, like the k~dney, may be very sensitive to the effects of hypoperfusion. Experimental hypotension in dogs has been shown to cause a marked reduction in pancreatic blood flow, which in turn is associated with injury to the acinar cells, including lysosomal swelling, vacuolization, and the release of both proteolytic enzymes and a myocardial depressant factor [17]. Furthermore, the potential injurious effects of pancreatic ischemia have been emphasized by Popper et al [18], who showed in dogs whose pancreatic ducts had been ligated, that short periods of ischemia (15 minutes) could convert mild pancreatic edema to severe acute hemorrhagic pancreatitis. Broe et al [19] demon8trated the production of severe edema in the isolated dog pancreas subjected to 2 hours of total ischemia. Anderson [20] also concluded that ischemia may be 511
Haas et al
an important factor in the progression of pancreatic edema to necrotizing pancreatitis. Pfeffer et al [21] showed in dogs that disruption of capillary blood flow by injection of microspheres into the pancreatic arteries caused pancreatitis. Conversely, improvement of pancreatic blood flow by postganglionic sympathetic denervation was shown by Wright and Goodhead [22] to reverse or prevent an otherwise lethal pancreatic injury in dogs. Low molecular weight dextran has also been shown to protect the pancreatic microcirculation against pancreatitis during experimental ischemia [23]. In the clinical setting, the development of acute pancreatitis has been correlated with hypoperfusion. Tilney et al, [24] described 18 patients in whom renal failure developed, presumably due to hypoperfusion, after operation for ruptured abdominal aortic aneurysms. Of these patients, pancreatitis developed in nine (50 percent). Wsrshaw and O'Hara [7] found a similar high correlation between pancreatitis and acute renal tubular necrosis in patients who died after experiencing oligemic and lmnoligemic shock. In their prospective study of 13 patients who were found to have hyperamylasemia after a period of shock, the elevated serum amylase levels in all 13 was of pancreatic origin. In 4 of the 13 patients, clinical evidence of acute pancreatitis developed. The hypothesis that the form of pancreatitis under consideration is caused by an ischemic injury to the pancreas is supported by several of our observations. All nine patients who developed clinically manifest severe necrotizing pancreatitis had experienced prolonged hypoperfusion during the early postoperative period, whereas the three patients with mild pancreatitis did not have comparable antecedent episodes of hypotension. Acute pancreatitis was found most commonly at autopsy in patients who had experienced significant postoperative hypoperfusion and survived beyond the first postoperative day. Pancreatitis in these patients correlated highly with other organ system injuries known to result from low output, such as renal failure and acute infarction of the liver, bowel, or spleen. The autopsy data showed pancreatitis was rarely evident within the first 24 hours after surgery, whereas 36 percent of those who died later had pancreatitis. Among those nonsurgical patients who died from acute cardiac events, there were no cases of acute pancreatitis in those who died suddenly, but 9 of 38 patients (24 percent) who died after a period of cardiogenic shock had acute interstitial or necrotizing pancreatitis. Although acute pancreatitis can produce hypotension, itseems clear that in the present circumstances pancreatitis wa~ the effect rather than the cause. In addition to postoperative cardiovascular instability,there isalso evidence that cardiopulmonary bypass m a y predispose to pancreatitis. Serum amyhse levels and renal amylase to creatinine clearance
5~2
ratios have been shown to increase to a significantly greater extent after thoracic procedures using cardiop'dmonary bypass than after thoracic procedures not requiring cardiopulmonary byp ~ass[8,?.5"].Murray et al [26] showed that use of a pulsatile perfusion system rather than a nonpulsatile system, led to fewer elevations of the amylase to creatinine clearance ratio after cardiopulmonary bypass, and concluded that nonpulsatile perfusion may be a factor in pancreatic injury. In addition, hypothermia has been reported to cause pancreatic necrosis and may be a factor in the patient who has been cooled during cardiopulmonary bypass [27]. Atheromatous or cholesterol emboli may be broken off during cannulation and cross-clamping of the aorta and embolize to the pancreas [28,29]. Venous sludging, which occurs with extracorporeal circulation, has also been reported to injure the pancreas and may further add to the detrimental effects of cardiopulmonary bypass in the pancreaas [20]. Perhaps these factors during bypass cause a mild and relatively insignificant pancreatic lesion. This thesis is supported by our finding of elevated pancreatic isoamylase levels in a fourth of the asymptomatic patients studied after uneventful cardiac surgery. Increases in the serum ribcnuclease level in half of these same patients suggest that there may be some associated pancreatic cell death [13].If the pancreas is sensitized by this injury, a further period of hypoperfusion added during the postoperative period seems to convert the process to necrotizing pancreatitis, m u c h as in the experiments of Popper et al [18]. What we are describing is a spectrum of pancreatic injury demonstrable after cardiac surgery. At one end is a minimal phenomenon, detectable only by relatively sensitive biochemical tests in 25 percent of patients. At the other end is a highly lethal necrotizing pancreatitis, recognized during life in fewer than 1 percent of patients (less than 0.2 percent in our clinical series), but an estimable cause of morbidity and mortality in cardiac surgical patients, being present in 25 percent of those who died after surgery and the principal cause of death in 4 percent~ In between axe those cases that are either clinicaily mild and respond well to conservative treatment or subclinical and heal spontaneously without being diagnosed. These cases may correspond to those classified as chronic fibrosis and healing when found incidentally at autopsy. The recognition of pancreatitis in the patient after cardiac surgery is difficult, partly because so much else may be going on to obscure the abdominal problems and partly because the clinical and laboratory manifestations of this form of pancreatitis are unlike those of the common form. Distention, ileus, and leukcmytosis are typical of this syndrome whereas pain, tenderness, and hyperamylasemia are not. It
Th* ~
~
~ S~nr
Pancreatltis After Cardiac Surgery
may be that ischemic pancreatitis after cardiac surgery produces a relatively pure necrotic injury, free of the primary inflammatory effects of activated enzymes and the abdominal symptoms and signs they produce so characteristically in other forms of pancreatitis. It has become our practice to place pancreatitis high on the list of potential diagnoses in any patient with serious abdominal complaints after heart surgery and to follow such patients with ultrasonography and computerized tomography to increase the identification of pancreatitis. The treatment of pancreatitis after bypass remains difficult and unsettled. Clearly, some patients have resolution of the injury spontaneously or with conservative treatment. We believe that patients who are unstable or deteriorating require laparotomy and aggressive debridement and drainage [30]. Our few survivors were those in whom the most extensive operations were performed to remove devitalized tissues and provide egress for the products of pancreatitis. For the future, prevention of the pancreatic injury is paramount. Using serum isoamylase and ribonuclease as described herein, we have begun a trial of low molecular weight dextran to see if the pancreas can be protected from ischemic injury during cardiopulmonary bypass.
Summary We have described a spectrum of pancreatic surgery after cardiopulmonary bypass. At one end is a subclinical lesion which was manifested only by elevations in serum isoamylase levels (27 percent of patients) and increased ribonuclease levels (13 percent of patients) in asymptomatic patients followed after cardiac surgery. At the other end is a severe and often lethal necrotizing pancreatitis. Acute necrotizing pancreatitis was found at autopsy in 25 percent of 138 patients who died after cardiac surgery, and it correlated strongly with low output, acute tubular necrosis, and infarction of the liver, spleen, or bowel. It was the principal cause of death in 4 percent of these patients. In addition, 24 percent of 38 nonsurgical patients who died from cardiac failure and hypoperfusion had acute pancreatitis at autopsy, whereas acute pancreatitis was not observed in 55 nonsurgical patients who died without a significant period of low output. Acute pancreatitis was recognized postoperatively in 12 patients (0.2 percent). Three had mild pancreatitis, and all responded well to conservative therapy. In nine patients, fulminant necrotizing pancreatitis developed. Their courses were characterized by significant early postoperative hemodynamic comprpmise, abdominal distention, ileus, fever, and episodes of late vascular instability associated with hypocalcemia. The diagnosis of pancreatitis was usually missed because of the absence of pain, tenderness and hyperamylasemia. The
¥oi~1e 149, Apdl 1985
diagnosis was confirmed at laparotomy in eight patients and at autopsy in one. The only two survivors among the nine with severe cases had aggressive mobilization, debridement, and wide drainage of the necrotic pancreas. We suggest that (1) a mild subclinical injury to the pancreas may occur as a consequence of cardiopulmonary bypass and may progress to severe ischemic necrosis if hypoperfusion follows in the postoperative period, (2) the presentation of necrotizing pancreatitis may be atypical in the cardiac surgical patient and should be considered if nonspecific abdominal symptoms are present, and (3) aggressive debridement and drainage may be the optimal treatment for aggressive forms of this disease. Acknowledgment: We thank Denise Duchaineyfor her technical skills with the isoamylase and ribonuclease determinations and Mary Chasse for preparing the manuscript with diligence and patience.
References I. Harjola P-T, SIItanen P, Appelqulst P, Laustela E. Abdominal complications after open heart surgery. Ann Chit Gynaecol 1968;57:272-4. 2. Horton EH, Murthy SK, Seal RME. Haemorrhagic necrosis of small intestine and acute pancreatltts following open heart surgery. Thorax 1968;23:438-45. 3, Wallwork J, David.senKG. The acute abdomen following cardiopulmonary bypass surgery. Br J Surg 1980;67:410-2. 4. Housman L, Bonchek L, Lambert L, Grunkenmeier G, Starr A. Prognosisof patients after mttral commtssurotomy.J "rhorac Cardlovasc Surg 1977;73:742-5. 5. Rose DM, Ranson JHC, Cunntngham MN Jr, Spencer FC. Patterns of severe pancreatic Injury following cardiopulmonary bypass. Ann Surg 1984;199:168-72. 6. HanksJB, Curtis SE, Hanks FF, et al. Gastrointestinal complications after cardlopulmonary bypass. Surgery 1982;92: 394-9. 7. Warshaw AL, O'Hara PJ. Susceptibility of the pancreas to ischemlc Injury in shock. Ann Surg 1978; 188:197-201. 8. Traverse LW, Fen'ad BT, Buckberg GD, Tompklns RK. Elevated postoperative renat clearance of amylase without pancreaUtls after cardiopulmonery bypass. Am J Surg 1977; 133:298-303. 9. Smith CR, Schwartz S, Amylase:creatinine clearance ratios, serum amylase, and lfpase after operations with cardlopulmonary bypass. Surgery 1983; 194:458-63. 10. Longnecker DF. Pathology and pathogenesls of diseases of the pancreas. Am J Pathol 1982;107:103-21. 11. Warshaw AL, Lee KH. Characteristic alterations of serum Isoenzymasof amylase In disease of liver, pancreas, salivary gland, lung, and genitalia. J Surg Res 1977;22:362-9. 12. Wars;taw AL, Fournler PO. Release of dbonuclease from anoxtc pancreas, Surgery 1984;95:537-41. 13. Warshaw AL, Lee K-H. Serum rlbonuclease elevations and pancreatic necrosis in acute pancreaUtis. Surgery 1979; 86:227-34, 14. Banks PA, Sidl S, Gelman ML, Lee K-H, Warshaw AL. Amylase-creaUnlne clearance ratios and serum amylase Isoenzymes in moderate renal insufficiency. J CItn Gastroenterol 1979;1:331-5. 15. Felner H. Pancreatitls after cardiac surgery, a morphologtc study. Am J Surg 1976;131:684-8. 16. PanebiancoAC, Scott SM, Dart CH Jr, TaksJ'oT, Echegaray I@A. Acute pancreatttis following extracorporeal circulation. Ann
513
Haas et al
Thorac Surg 1970:9:562-8. 17. Lefer AM, Spath JA Jr. Pancreatic hypoperfuslon and the production of a myocardial depressant factor. Ann Surg 1974; 179: 868-76. 18. Poppet HL, Necheles H, Russell KC. Transition of pancreatic edema to pancreatic necrosis. Surg Gynecol Obstet 1978;87:79-82. 19. Broe PJ, Zuidema GO, Cameron JL. The role of tschemia in actde pancreatltis: studies with an isolated perfused canine pancreas. Surgery !982;91:377-82. 20. Anderson MC. Venous stasls tn the transition of edematous pancreatltis to necrosis. JAMA 1963; 183:534-7. 21. Pfeffer RB, Lazarinl A, Robertson A, Sadefi D. Gradations of pancreatttts: edematous through hemmorhagic experimentally produced by controlled injection of microspheres into blood vessels in dogs. Surgery 1962;51:764. 22. Goodhead B, Wright PW. The effect of post ganglionic sympathectomy on the development of hemorrhagic pancreatltis In the dog. Ann Surg 1969;170:951-9. 23. Wright PW, ~ h e a d B. The value of dextran In 1he treatment of experimental pancreatttls. Surgery 1970;67:807-15. 24. Tilney NI, Bailey GL, Morgan AP, Sequential system failure after rupture of abdominal aortic aneurysm. Ann Surg 1973; 178:117-22. 25. Hennlngs B, Jacobsen G. Postoperative amylase secretion: a study following thoracic surgery with and without extracorporeal circulation. Ann Clin Res 1974;6:215-22. 26. Murray WR, Mittra S, MIttra D, Roberts LB, Taylor KM. The amylase creatlnine clearance ratio following cardlopulmonary bypass. J Thorac Cardlovasc Surg 1981;82:248 -53. 27. MacLean D, Morrison J, Grlfflth PD. Acute pancreatltts after accidental hypothermta and hypothermic myxoedema. Br Med J 1273;4:757. 28. Probstetn JG, Ram A, Joshi MB, Blumenthal H. Atheromatous embolization: an etiology of pancreatitis. Arch Surg 1957; 75:566-72. 29. Moidveen-Geronimus M, Merrlam JC Jr. C,holesterol embolism. Circulation 1967;35:946-53. 30. Warshaw AL, Imbembo AL, CIvetta JM, Daggett WM. Surgical intervention in acute necrotizlng pancreatltis. Am J Surg 1974; 127:484-91. Discussion
A l e x a n d e r S. G e h a (New Haven, CT): We have been interested in these pancreatic abnormalities after cardiac surgery for a while, and we do have a prospective study upon which we are embarking t h a t is somewhat similar to yours, Dr. Warshaw. Our observations so far indicate t h a t there are a fair amount of chemical abnormalities encountered, the meaning of which I am not sure. So far, we have been somewhat less impressed than you have with the occurrence of a fulminating necrotizing pancreatitis after cardiac surgery. Dr. Warshaw, I think you indicated that there is a fairly high incidence of chemical abnormalities after cardiac surgery as you encountered them in 27 percent of the patients, and t h a t this includes an elevation of the ribonuclease level in a fairly substantial number. M y question is, how specific is the ribonuclease level for pancreatic cell deaths versus other orgma death, and were these patients with the elevated enzyme levels, exhibiting any hypotension, low cardiac output, or other factors t h a t m a y have predisposed them to this? Are you sure that these are not abnormalities of the enzymes t h a t are observed because of lack of urinary clearance versus overproduction of these enzymes? Have you looked at ,~rine amylase clearance, for instance? W h a t is your recommendation regarding this problem? If abnor-
514
malities in the amylase level and elevated enzyme levels are encountered, what should be done? W h a t is the specificity of the computerized tomographic scan and ultrasonography? Should we perform these tests when we see elevation in the enzyme levels in 27 percent of the patients in todays atmosphere of diagnosis-related groups (DRGs) with so much emphasis being placed on cost containment? A n d r e w L. W a r s h a w (closing): I want to emphasize t h a t what we are describing here is not simply an isolated group of patients with pancreatitis, but an injury to the pancreas that follows an entire spectrum ranging from something that is b~ing missed by current methods to something that is being seen with lethal results at the other end. We are hypothesizing and we are presenting evidence from our isoamylase and ribonuclease studies that there is a sensitizing event during cardiopulmonary bypass, and t h a t the subsequent "detonation of the bomb" come0 with some further event such as postoperative hypotension which further injures the pancreas. We are emphasizing t h a t this form of pancreatitis, when it does occur, is a particularly bad form. I think it's hard to think of another group of panereatitis patients who have an overall mortality from clinically recognized pancreatitis in the 80 percent range. So, we have got to do something different in the future and not simply debride them, although we think that is the best thing at the moment. We are suggesting that this is a different form of pancreatitis than what usually occurs. Usually, the course of events is chemical inflammation followed by, in some patients, necrosis, which marks the particularly difficult cases, which are only 5 to 10 percent of the overall cases. In cardiac patients, necrosis is the initial event; it's the central event; it is one that does not have the same diagnostic clues such as pain, tenderness, and hyperamylaseraia, and only becomes obvious later when the second a r y signs of inflammation of the necrotic tissue set in, or perhaps it is not recognized at all until after the patient has died. Dr. Geha, you asked how specific the isoamylase and ribonuclease levels are? They are highly specific, at least by the methods we have used. Pancreatic isoamylase is produced only by the pancreas; it increases only in the presence of pancreatic injury except in cases of renal dysfunction, so your point about renal clearance is correct. However, in this particular study, the prospective patients all had normal renal function, so what we were looking a t was specifically pancreatic injury. The test is no good in patients with renal failure and is probably, therefore, not useful in the severely affected patients we described in the other portion of the study. The ribonuelease level we measured i3 also relatively specific. There are many kinds of ribonuclease in the body. It. is present in almost every cell t h a t has a nucleus. However, the polycytidine-specific ribonuclease that we have used has been shown in our laboratory studies i n the rat to be released only from the dying pancreas and not from the dying heart, intestine, liver, spleen, kidney, or any other organ t h a t we tested. So we think it is meaningful in this case, again with the proviso t h a t renal function is normal; ribonuclvase levels in the serum are highly dependent upon that. It is therefore useful in a study such as this of prospective normal patients who were not hypotensi"e postoperatively, b u t it is not useful in detectin~ pancrear~is in the critically ill patients.
The A m e ~
~ l
o~ Surgery
Pancreatttis After Cardiac Surgery
The renal clearance rates in the sick patienk~ were often abnormal. The amylase to creatinine clearance ratios and creatinine clearance ratios were frequently elevated and were not useful in any of the diagnostic tests. We do think that computerized tomographic scans are goirig to be helpful, at least in the more advanced cases. We do not yet know how sensitive that modali,'.y is. T h e lesson of all this is simple because we do not have all the answers. We are calling attention to a phenomenon. We think that this form of panereatitis is somewhat different. We think it is somewhat more difficult ~n many
ways both to recognize and to manage when it occurs. We can only exhort you at the moment to be more aware of it than perhaps you have been. We found as our index of suspicion has increased, we have begun to think of the diagnosis more often, we have begun to make it earlier and to order the computerized tomographic scans earlier, and perhaps we will do better preventing it in the future. We are embarking on a prospective double-blind trial of vasoactive agents of various kinds to try to protect the pancreas during this critical period, and we hope that will reduce the incidence. Time will tell.
This discussion section has been abbreviated due to space limitations. The full text can be obtained from the reprints author listed on the title page.
Volume149,Apd11985
515