- Email: [email protected]
Considerations that lower pancreatoduodenectomy mortality
Considerations That Lower Pancreatoduodenectomy Mortality John W. Braasch, MD, Boston, Massachusetts Bruce N. Gray, MB, BS, FRACS,’ Boston, Massachusetts
Radical pancreatoduodenectomy is under attack by those who point out its high postoperative mortality in some hands and its low rate of cure in treating pancreatic carcinoma. A resective procedure is, however, the only chance for cure in periampullary carcinoma. Preservation of the body and tail of the gland in the Whipple procedure prevents, in most patients, the deficiency states of diabetes and steatorrhea. In the treatment of chronic pancreatitis, this operation is effective in treating right-sided disease with obstruction of the biliary tract and pancreatic duct. Although many improvements have been made in operative technic and postoperative care, pancreatoduodenectomy remains a technically difficult procedure attended by relatively high morbidity and mortality rates. This report describes in detail the Lahey Clinic experience with postoperative complications after pancreatoduodenectomy for both periampullary carcinoma and pancreatitis, covering the eighteen year period from 1957 through 1974. It also attempts to correlate variations in operative technic and patient selection with the complications observed and thereby to suggest safety factors which keep the risk within acceptable limits.
comprised a variety of other neoplasms. (Table I.) Age and Sex Distribution. There were 138 men and 85 women with tumor and 38 men and 17 women with pancreatitis. Ages ranged from twenty-five to seventy-nine years for patients with tumors and from twenty-three to sixty-six years for those with pancreatitis. The mean age for both men and women was in the fifth decade for tumor and in the fourth for pancreatitis. Complications. Postoperative complications developed in 151 (67.7 per cent) of the 223 patients with tumor and in 26 (47.2 per cent) of the 55 patients with pancreatitis. The mean postoperative hospital stay was significantly longer in patients with complications and tumor (28.5 f 0.93 [SEMI days) and pancreatitis (23.9 f 1.75 days) than it was in patients who did not have complications (tumor, 19.6 f 0.8 days; pancreatitis, 16.2 f 0.68 days). Of the 319 complications recorded, wound infection was the most commonly observed in both patient groups. (Table II.) Postoperative pyrexia for which no cause could be found occurred in 13.6 per cent of patients and was probably due, in most instances, to collection of pancreatic and biliary fluid at the operative site. Biliary fistula, although occurring in 13.6 per cent of patients, was seldom a troublesome problem and invariably resolved spontaneously. No association was found between the occurrence of biliary fistula and operative mortality. The low incidence of pancreatic fistula and the absence of acute postoperative pancreatitis in patients being operated
Material and Methods TABLE I Analysis of the case records of 279 patients who underwent pancreatoduodenectomy between the years 1957 through 1974 form the basis of this report. Operative death was defined as that occurring postoperatively in the hospital or within one month after discharge. Complications documented here pertain only to the postoperative period and do not include such late complications as weight loss, malabsorption, or diabetes mellitus. Pathology. In our study were 223 patients with tumor, 55 patients with pancreatitis, and 1 patient with a common duct calculus lodged at the ampulla. Of the patients with tumor, 217 had malignant tumors, 1 had a benign tumor, and 5 had islet cell tumors. Of all tumors 93 per cent were adenocarcinomas of the pancreatic head, amp&a, common bile duct, or duodenum, and the remaining 7 per cent From the Department of Surgery, Lahey Clinic Foundation, Boston, Massachusetts. Reprint requests should be addressed to John W. Braasch, MD, Lahey Clinic Foundation, 605 Commonwealth Avenue, Boston, Massachusetts 02215. Presented at the Annual Meeting of the New England Surgical Society, Whitefield. New Hampshire, September 23-26, 1976. * Present address: Department of Surgery, St. Vincent’s Hospital, Fitzroy, Victoria, Australia 3065.
489
Lesion and Mortality Pancreatoduodenectomy
Lesion _~ Pancreas Adenocarcinoma Pancreatitis Islet cell tumor Cystadenocarcinoma Leiomyosarcoma Ampulla Adenocarcinoma Carcinoid Common bile duct Adenocarcinoma Adenoacanthoma Carcinoid Calculus Duodenum Adenocarcinoma Leiomyosarcoma Adenoma Total
after
Postoperative Deaths
Mortality
82 55 5 2 1
11 4
13.4 7.3
60 1
4
6.7
41 1 1 1
7
17.1
25 3 1 279
8 1
32.0 33.0
35
12.5
No. of Cases
(%I
The Amerlcan Journal of Surgery
Pancreatoduodenectomy
none of the patients with pancreatitis. The higher incidence of jaundice in patients with tumor (67 per cent) compared to that in those with pancreatitis (15 per cent) would appear to be a major factor accounting for the higher incidence of renal failure in the former group. Mortality. Thirty-five of 279 (12.5 per cent) patients died in the postoperative period. Four patients died of cardiac arrhythmia or myocardial infarction in the first two postoperative days. Four patients died of uncontrolled hemorrhage on days 1,9,14, and 19, and one patient died of renal failure on the eleventh postoperative day. Multiple factors contributed to postoperative death in the remaining 26 patients. The complications contributing to death in all 35 patients are shown in Table II.
on for chronic relapsing pancreatitis probably reflect the integrity of the pancreatic anastomosis attainable and the low pancreatic exocrine reserve. Hemorrhage either from the gastrointestinal tract or from the wound occurred in 11.1 per cent of patients and was frequently associated with other complications. Hemorrhage was associated with wound sepsis in 17 of 57 patients (30 per cent) and was associated with pancreatic fistula in 4 of 20 patients (20 per cent). The incidence of hemorrhage was only 10 per cent when a fistula was not present. In 8 of the 20 patients in whom postoperative pancreatic fistula developed, subsequent wound sepsis also developed. Renal failure occurred in 11 patients with tumor but in
TABLE
II
Incidence
of Complications
and Associated Tumor (233 patients)
Mortality Pancreatitis (55 patients)
Total (279 patients)
_
Mortality of Complications*
Complication
No.
%
No.
%
No.
%
No.
%
Wound and abdominal sepsis Postoperative fever+ Biliary fistula Gastrointestinal tract and wound hemorrhage Pulmonary infection Pancreatic fistula Z Urinary infection Renal failure Unstable diabetes Intestinal fistula Myocardial infarction Septicemia Pancreatitis Delayed gastric emptying Wound dehiscence Urinary retention Congestive heart failure Liver failure Atrial fibrillation Postoperative psychosis Small bowel obstruction Thrombophlebitis Wound hematoma Cholangitis Parotitis Salmonella diarrhea Cerebrovascular accident Acute hepatic necrosis Perforated peptic ulcer Esophagitis Drug reaction Digitalis overdose Pneumothorax
48 30 30 27
21.5 13.4 13.4 12.1
9 8 8 4
16.4 14.5 14.5 7.3
57 38 38 31
20.4 13.6 13.6 11.1
21
37
4 18
11 58
9.9 8.5 5.4 4.9 2.7 2.2 3.1 2.2 2.7 1.8 2.2 1.8 1.3 1.3 0.9 0.4 0.9 0.4 0.9 0.9 0.9 0.9 0.4 0.4 0.4 0.4 0.4 0.4 0.4 0.4 0.4 0.4 0.4
3 1 3 0 2 2 0 1 0 1 0 0 1 0 1 2 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
5.5 1.8 5.5
9.0 7.2 5.4 3.9 2.9 2.5 2.5 2.2 2.2 1.8 1.8 1.4 1.4 1.1 1.1 1.1 0.7 0.7 0.7 0.7 0.7 0.7 0.4 0.4 0.4 0.4 0.4 0.4 0.4 0.4 0.4 0.4 0.4
3 4
12 20
Gout Herpes stomatitis Small bowel necrosis Marginal ulcer
22 19 12 11 6 5 7 5 6 4 5 4 3 3 2 1 2 1 2 2 2 2 1 1 1 1 1 1 1 1 1 1 1
3.6 3.6 1.8 1.8
1.8 1.8 3.6 1.8
25 20 15 11 8 7 7 6 6 5 5 4 4 3 3 3 2 2 2 2 2 2 1 1 1 1 1 1 1 1 1 1 1
73 13 29 71 67 60 25 100
100 100 100
1 1
100 100
* Multiple complications often figured in 1 death. + Postoperative fever refers to those patients with postoperative pyrexia for which no cause could be determined. t Pancreatic fistula refers only to those patients in whom leakage of pancreatic juice was significant enough to require treatment.
volume 133, Apfir 1977
481
Braasch and Gray
TABLE I II
TABLE IV
Use of Stent across Pancreatojejunal Anastomosis”
Type of Stent Short Long
None Duct ligated
No. of Cases
220 11
34 6
Complications
Mortality
__-
No.
%
No.
%
128 6 12 5
58 55 35 83
27 1 5 0
12 9 15 0
* In 14 other
patients, there was no pancreatic anastomosis since the pancreatic duct was ligated or the operation resulted in total pancreatectomy.
The overall mortality has decreased from 17 per cent for the six years from 1957 through 1962 to 10.7 per cent for the six years from 1969 through 1974. Analysis of disease treated and operative mortality (Table I) shows a much higher rate of postoperative deaths for patients with malignant tumors of the duodenum and lower bile duct than for patients with tumors in other periampullary sites and a much lower mortality rate for patients with pancreatitis. Operatiuc Technic. The technic of radical pancreatoduodenectomy at the Lahey Clinic has been modified and standardized over many years and has been previously described in detail [1,2]. However, over the eighteen year period (1957 through 1974), technic has varied somewhat, especially with regard to the management, of the pancreat,ic duct and the use of bilateral truncal vagotomy at the time of resection. In 220 operations, the pancreatic duct was anastomosed to the jejunal mucosa using a short rubber stent to splint the anastomosis. In 11 operations, a much longer stent was used and was brought out through the jejunum and abdominal wall to drain pancreatic secretions externally. In 34 operations, no stent was used although precise mucosa-to-mucosa pancreatojejunostomy was performed. Table III compares the morbidity and mortality of these three ways of dealing with the pancreatic anastomosis. None of the differences observed are significant. In six patients, the pancreatic duct was ligated and no pancreatic anastomosis was const.ructed. The morbidity rate of 83 per cent for this group is significant (p <0.05) when compared with the group having anastomosis. Of the group, 15 patients with tumor and 7 with pancreatitis underwent bilateral truncal vagotomy at the time of resection. Vagotoniy might be expect,ed to decrease the incidence of postoperative gastrointestinal bleeding after resection. However, in two of the patients, postoperative hemorrhage
occurred from a marginal ulcer and a gastric ulcer at the lesser curvature. Effect of Jaundice. The level of preoperative jaundice is correlated with postoperative complications and death in Table IV. Clearly shown is the marked increase in the incidence of liver failure, wound hemorrhage, and postoperative death in those patients with serum hilirubin levels greater than 20 mg/lOO ml. Postoperative Hemorrhage. Postoperative hemorrhage occurred in a total of 31 patients. The severity of bleeding ranged from a transient episode of melena to massive un-
482
Incidence of Complications after Whipple Procedure: Correlation with Serum Bilirubin Level
Serum Bilirubin (mg/lOO ml) 1
l-
9.9
10-19.9 20-30+
No. of Cases
Renal Failure
Wound Hemorrhage
Wound Sepsis
(%I
(%I
(%I
(%I
Mortality
92
2.6
6.5
22.1
13.0
111 48 28
2.6 7.5 9.0
15.0 10.0 22.0
20.4 20.0 26.0
9.7 12.5 22.0
controlled bleeding. Hemorrhage occurred from the gastrointestinal tract in 18 patients, from the wound in 10 patients, and from both sites in 3 patients. The overall mortality for patients with postoperative hemorrhage was 58 per cent. Twelve of the patients were operated on again at least once for uncontrolled bleeding. Eight of these patients died despite reoperation. In the 12 reexplored patients, bleeding was noted in the gastrointestinal tract (gastrojejunal anastoniosis, 4; marginal ulcer, 1; lesser curvature ulcer, 1; undetermined, 1) and in the wound (gastroduodenal artery, 3; pancreatic bed, 2; superior mesenteric vein, 1; portal vein, 1; undetermined, 1). The time at which hemorrhage occurred did not show any particular pattern and varied from within several hours to more than three weeks after operation.
Comments
Radical pancreatoduodenectomy is an effective method of treatment for carcinoma of the lower bile duct, duodenum, and ampulla of Vater. In a recent report from the Lahey Clinic [3] the five year survival rate was 41.3 per cent for lesions of the duodenum, 32 per cent for lesions of the ampulla, and 25 per cent for lesions of the common bile duct. Although postoperative morbidity and mortality rates associated with pancreatoduodenectomy are formidable, the relatively good chance of cure for lesions other than of the pancreatic head overrides the risk of postoperative complications. Previous reports [4-61 have noted a much higher operative mortality than in this series. The mortality for 1,169 cases reported by Fish and Cleveland [4J was 23.7 per cent. The high mortality associated with pancreatoduodenectomy has been used as one factor in favor of total pancreatic resection for periampullary carcinoma. This is not substantiated by analysis of this series of patients. The mortality and complication rates compare favorably with those of a series of 64 patients undergoing total pancreatectomy reported by Pliam and ReMine [7]. The relatively low mortality rate reported here is in part a function of familiarity with the operation and of particular attention paid to operative technic,
The American Journal 01 Surgery
Pancreatoduodenectomy
especially the pancreatojejunal anastomosis. Accurate anastomosis of the main pancreatic duct to the jejunal mucosa has been advocated at this institution to diminish the chance of leakage of pancreatic juice and to preserve some exocrine pancreatic function. Wound sepsis, hemorrhage, and renal failure were the major commonly occurring complications that were asiociated with a high mortality. Similar findings have previously been recorded by others [3,6]. The use of two varieties of stents to splint the anastomosis has not resulted in a statistically significant decrease in the incidence of postoperative fever, wound sepsis, and mortality although there is a suggestion that the use of a long stent would prove statistically beneficial with a larger experience. The suggestion has been offered that pancreatic anastomoses do not stay open and that either oversewing of the end of the pancreas [8] or total pancreatectomy [9] is a safer alternative. Our experience with the former is 6 cases with an 83 per cent morbidity (0 mortality) rate and is too limited for comment. We have determined the serum amylase content of the juice from the jejunum near the pancreatic anastomosis in four patients and found it to contain very high concentrations. Furthermore, two patients with pancreatitis have been reexplored after the Whipple procedure and the anastomoses have been found to be widely patent. Certainly many long-term survivors without pancreatitis do not have clinical steatorrhea. The proponents of total pancreatectomy for malignancy have not as yet proved their case of improved long-term survival. It seems appropriate, therefore, to continue use of the pancreatic anastomosis in most cases. Bilateral truncal vagotomy has not significantly altered postoperative complications. If an adequate amount of the stomach is removed it appears to be unnecessary to add vagotomy to the resection. The high incidence of serious complications in severely jaundiced patients suggests that initial biliary decompression followed at a later date by radical resection has a place in the management of those patients in whom the serum bilirubin level exceeds 20 mg/lOO ml. A staged procedure is mandatory in patients with sepsis in the right upper quadrant after previous exploration. Postoperative hemorrhage is an ominous occurrence with a high mortality. Reexploration and suture of the bleeding point is often followed by further fatal hemorrhage. Attempts at resuturing leaking pancreatic anastomoses at that time are futile. Removal of the rest of the pancreas to control pancreatic fistulas is extremely hazardous. It is possible that continuous copious irrigation of the operative region by means of inlying catheters can prevent hemorvolulna 193. April 1@77
rhage in pancreatic fistulas with sepsis in these particularly high-risk patients. The high mortality associated with lesions of the duodenum and to a lesser extent with common bile duct lesions may be due to the fact that the soft consistency of the pancreas and small size of the main pancreatic duct are more difficult to handle than the firm glands with dilated ducts seen in carcinoma of the pancreatic head or ampulla and chronic pancreatitis. Application of total pancreatectomy to these selected patients would result in a lower postoperative mortality rate. Summary
To aid in case selection for pancreatoduodenectomy and to gain information on the technical management of this operation and its complications, records of 279 patients who were treated for neoplasm or pancreatitis by this procedure between the years 1957 and 1975 were reviewed. The overall operative mortality was 12.5 per cent and was 10.7 per cent for the years 1969 through 1974. The use of vagotomy did not prevent postoperative bleeding from the stomach, and the use of a stent did not make a statistically significant difference in morbidity or mortality. Postoperative hemorrhage is an ominous complication and is best treated conservatively until blood loss cannbt be replaced. Preoperative serum bilirubin levels above 20 mg/lOO ml indicate a two-stage operative procedure as does the presence of right upper quadrant sepsis. The resection of malignant disease of the duodenum and lower bile duct is followed by a high mortality and requires total pancreatectomy if a satisfactory pancreatojejunostomy cannot be constructed. References 1. Warren KW, Braasch JW, Thum CW: Diagnosis and surgical treatment of carcinoma of the pancreas. Curr Probl Surg Chicago, Year Book Medical, 1966, p 3. 2. Braasch JW, Gray BN: Technique of radical pancreatoduodenectomy with consideration of hepatic arterial relationships. Surg C/in North Am 56: 631, 1976. 3. Warren KW, Choe DS, Plaza J, et al: Results of radical resection for periampullary cancer. Ann Surg 161: 534, 1975. 4. Fish JC, Cleveland BR: Pancreaticoduodenectomy for periampullary carcinoma. Analysis of 36 cases. Ann Surg 159: 469, 1964. 5. Hicks RE, Brooks JR: Total pancreatectomy for ductal carcinoma. Surg Gynecol Obstet 133: 16, 1971. 6. Mongit JJ, Judd ES, Gage RP: Radical pancreatoduodenectomy: a 22-year experience with complications, mortality rate, and survival rate. Ann Surg 160: 711, 1964. 7. Pliam MB, ReMine WH: Further evaluation of total pancreatectomy. Arch Surg 110: 506, 1975. 6. GoldsmIth HS, Ghosh BC, Huvos AG: Ligation versus implantation of the pancreatic duct after pancreaticoduodenectomy. Surg GynecolObstet 132: 67, 1971.
463
Braasch and Gray
9. Brooks JR, Culebras JM: Cancer of the pancreas: palliative operation, Whipple procedure, or total pancreatectomy? Am J Surg 131: 516, 1976.
Discussion John R. Brooks (Boston, MA): First, Doctor Braasch’s complication rate after the Whipple procedure is high at 67 per cent. He is correct in placing the blame for this upon leaks occurring most commonly at the pancreatojejunostomy, but also the choledochojejunal anastomosis can be a fault. I agree that a mucosa-to-mucosa anastomosis is important. It is true that an indurated pancreas (as in chronic pancreatitis) or a similarly blocked pancreas by tumor is easier to suture and less likely to leak than the pancreas that is associated with an ampullary lesion or duodenal lesion in which the duct of Wirsung has not been totally blocked and the pancreatic body is not firm and indurated. The other major cause of complication is gastrointestinal bleeding. We believe that vagotomy should be performed to minimize the chance of peptic ulceration. Second, the 10 per cent mortality he presented is excellent. His survival figures are also good for lesions around the ampulla and around the lower end of the common duct. I would simply like to ask Doctor Braasch what his five year survival figures are for lesions in.the pancreatic head. Third, debatably, I question the need for a Whipple procedure for pancreatitis. Obviously, none of us has a consistently successful procedure for chronic pancreatitis, but I would be more inclined to do an internal drainage procedure or subtotal distal pancreatectomy rather than the more radical Whipple procedure, which I notice in the manuscript carried a 7 per cent mortality. Finally, Doctor Braasch has lumped together pancreatic head lesions with ampullary, duodenal, and common duct lesions. Survival figures are not comparable, as we all know. As many of you are aware, my recent philosophy has been to employ the Whipple procedure for the periampullary lesions, namely the ampullary, duodenal, and lower end of the common duct-type tumors, but to use total pancreatectomy for lesions of the head of the pancreas or the body and the tail. Mortality figures for the totals are 11per cent in the Mayo Clinic series and 12 per cent in ours, not much different than the 10.7 per cent that Doctor Braasch has shown us here today for the Whipple procedure. George L. Nardi (Boston, MA): Unfortunately pancreatic cancers have one of the lowest operability rates. Only approximately 12 per cent of pancreatic cancers are found to be localized at the time of operation. The newer developments such as ultrasound, pancreatic scanning, and selective angiography have not been as successful in uncovering early pancreatic cancers as we had hoped. The newest technic of computerized body scanning may be of help, but we are always limited by the time the patient actually presents himself. At one time a high operative mortality was another discouraging factor along with a low survival rate. During the past decade, however, numerous series with little or no operative mortality have been accomplished and at the
Massachusetts General Hospital we have been able to increase our five year survival rate for adenocarcinoma of the head of the pancreas from a figure of 5 or 6 per cent to 15 per cent. I certainly agree with Doctor Braasch’s policy of performing preliminary biliary drainage in those patients with high bilirubin levels. This should be a tube drainage, however, and not cholecystojejunostomy. I have not done routine vagotomies in these operations but have depended on an adequate gastric resection to eliminate the possibility of marginal ulcer. Our Radiation Therapy Department has reviewed the mortality after pancreatic resection over the past ten years and they find that in more than 50 per cent of these patients death was due to local recurrence within the abdomen rather than distal metastases. As a result, we have been supplementing our resective efforts with implantation of iodine 125 seeds in the resected stump after frozen section shows no residual tumor. In addition, we follow this treatment with external beam radiation. John W. Braasch (closing): With regard to Doctor Brooks’ question concerning our five year survival, I believe it is similar to that at the Massachusetts General Hospital, roughly 15 per cent. I would admit that all of these 15 per cent probably are not cured, since we have had a number of them show up with more tumor after the five year interval. I think we must bear in mind that it is not always possible to recognize a malignant mass in the head of the pancreas and to differentiate it from the more curable and more favorable carcinomas of the lower duct or ampulla in which about a third survive five years. Therefore, all pancreatic head lesions cannot categorically be rejected for resection. However, I would agree that the indication for the operation is questionable when the lesion is clearly within the head of the pancreas, unless one subscribes to the theory that the Whipple operation is worthwhile as a palliative procedure. We will look forward to further follow-up in Doctor Brooks’ series of total pancreatectomies, especially when a sufficient number have been followed until death or for five years. I would agree that the concentration of this operation in selected hands is able to reduce the mortality for the operation. It is quite an exacting technical procedure. In the two-stage procedure for high preoperative bilirubin levels, the first stage should be the implantation of a T tube as noted by Doctor Nardi and not the construction of anastomosis. Relative to the pancreatic anastomosis, of basic importance is keeping the activated pancreatic juice away from the raw areas and vessels in the posterior part of the abdomen. This might be accomplished by diverting the juice externally by an intraductal tube through the anastomosis, by separating the biliary and pancreatic anastomoses as suggested by Doctor Nardi, or by irrigating the posterior parieties by suitable irrigating and suction tubes. If a leakproof anastomosis could always be constructed, these would not be necessary. Unfortunately such is not the case.
The American Journal ol Surgary
Radical pancreatoduodenectomy is under attack by those who point out its high postoperative mortality in some hands and its low rate of cure in treating pancreatic carcinoma. A resective procedure is, however, the only chance for cure in periampullary carcinoma. Preservation of the body and tail of the gland in the Whipple procedure prevents, in most patients, the deficiency states of diabetes and steatorrhea. In the treatment of chronic pancreatitis, this operation is effective in treating right-sided disease with obstruction of the biliary tract and pancreatic duct. Although many improvements have been made in operative technic and postoperative care, pancreatoduodenectomy remains a technically difficult procedure attended by relatively high morbidity and mortality rates. This report describes in detail the Lahey Clinic experience with postoperative complications after pancreatoduodenectomy for both periampullary carcinoma and pancreatitis, covering the eighteen year period from 1957 through 1974. It also attempts to correlate variations in operative technic and patient selection with the complications observed and thereby to suggest safety factors which keep the risk within acceptable limits.
comprised a variety of other neoplasms. (Table I.) Age and Sex Distribution. There were 138 men and 85 women with tumor and 38 men and 17 women with pancreatitis. Ages ranged from twenty-five to seventy-nine years for patients with tumors and from twenty-three to sixty-six years for those with pancreatitis. The mean age for both men and women was in the fifth decade for tumor and in the fourth for pancreatitis. Complications. Postoperative complications developed in 151 (67.7 per cent) of the 223 patients with tumor and in 26 (47.2 per cent) of the 55 patients with pancreatitis. The mean postoperative hospital stay was significantly longer in patients with complications and tumor (28.5 f 0.93 [SEMI days) and pancreatitis (23.9 f 1.75 days) than it was in patients who did not have complications (tumor, 19.6 f 0.8 days; pancreatitis, 16.2 f 0.68 days). Of the 319 complications recorded, wound infection was the most commonly observed in both patient groups. (Table II.) Postoperative pyrexia for which no cause could be found occurred in 13.6 per cent of patients and was probably due, in most instances, to collection of pancreatic and biliary fluid at the operative site. Biliary fistula, although occurring in 13.6 per cent of patients, was seldom a troublesome problem and invariably resolved spontaneously. No association was found between the occurrence of biliary fistula and operative mortality. The low incidence of pancreatic fistula and the absence of acute postoperative pancreatitis in patients being operated
Material and Methods TABLE I Analysis of the case records of 279 patients who underwent pancreatoduodenectomy between the years 1957 through 1974 form the basis of this report. Operative death was defined as that occurring postoperatively in the hospital or within one month after discharge. Complications documented here pertain only to the postoperative period and do not include such late complications as weight loss, malabsorption, or diabetes mellitus. Pathology. In our study were 223 patients with tumor, 55 patients with pancreatitis, and 1 patient with a common duct calculus lodged at the ampulla. Of the patients with tumor, 217 had malignant tumors, 1 had a benign tumor, and 5 had islet cell tumors. Of all tumors 93 per cent were adenocarcinomas of the pancreatic head, amp&a, common bile duct, or duodenum, and the remaining 7 per cent From the Department of Surgery, Lahey Clinic Foundation, Boston, Massachusetts. Reprint requests should be addressed to John W. Braasch, MD, Lahey Clinic Foundation, 605 Commonwealth Avenue, Boston, Massachusetts 02215. Presented at the Annual Meeting of the New England Surgical Society, Whitefield. New Hampshire, September 23-26, 1976. * Present address: Department of Surgery, St. Vincent’s Hospital, Fitzroy, Victoria, Australia 3065.
489
Lesion and Mortality Pancreatoduodenectomy
Lesion _~ Pancreas Adenocarcinoma Pancreatitis Islet cell tumor Cystadenocarcinoma Leiomyosarcoma Ampulla Adenocarcinoma Carcinoid Common bile duct Adenocarcinoma Adenoacanthoma Carcinoid Calculus Duodenum Adenocarcinoma Leiomyosarcoma Adenoma Total
after
Postoperative Deaths
Mortality
82 55 5 2 1
11 4
13.4 7.3
60 1
4
6.7
41 1 1 1
7
17.1
25 3 1 279
8 1
32.0 33.0
35
12.5
No. of Cases
(%I
The Amerlcan Journal of Surgery
Pancreatoduodenectomy
none of the patients with pancreatitis. The higher incidence of jaundice in patients with tumor (67 per cent) compared to that in those with pancreatitis (15 per cent) would appear to be a major factor accounting for the higher incidence of renal failure in the former group. Mortality. Thirty-five of 279 (12.5 per cent) patients died in the postoperative period. Four patients died of cardiac arrhythmia or myocardial infarction in the first two postoperative days. Four patients died of uncontrolled hemorrhage on days 1,9,14, and 19, and one patient died of renal failure on the eleventh postoperative day. Multiple factors contributed to postoperative death in the remaining 26 patients. The complications contributing to death in all 35 patients are shown in Table II.
on for chronic relapsing pancreatitis probably reflect the integrity of the pancreatic anastomosis attainable and the low pancreatic exocrine reserve. Hemorrhage either from the gastrointestinal tract or from the wound occurred in 11.1 per cent of patients and was frequently associated with other complications. Hemorrhage was associated with wound sepsis in 17 of 57 patients (30 per cent) and was associated with pancreatic fistula in 4 of 20 patients (20 per cent). The incidence of hemorrhage was only 10 per cent when a fistula was not present. In 8 of the 20 patients in whom postoperative pancreatic fistula developed, subsequent wound sepsis also developed. Renal failure occurred in 11 patients with tumor but in
TABLE
II
Incidence
of Complications
and Associated Tumor (233 patients)
Mortality Pancreatitis (55 patients)
Total (279 patients)
_
Mortality of Complications*
Complication
No.
%
No.
%
No.
%
No.
%
Wound and abdominal sepsis Postoperative fever+ Biliary fistula Gastrointestinal tract and wound hemorrhage Pulmonary infection Pancreatic fistula Z Urinary infection Renal failure Unstable diabetes Intestinal fistula Myocardial infarction Septicemia Pancreatitis Delayed gastric emptying Wound dehiscence Urinary retention Congestive heart failure Liver failure Atrial fibrillation Postoperative psychosis Small bowel obstruction Thrombophlebitis Wound hematoma Cholangitis Parotitis Salmonella diarrhea Cerebrovascular accident Acute hepatic necrosis Perforated peptic ulcer Esophagitis Drug reaction Digitalis overdose Pneumothorax
48 30 30 27
21.5 13.4 13.4 12.1
9 8 8 4
16.4 14.5 14.5 7.3
57 38 38 31
20.4 13.6 13.6 11.1
21
37
4 18
11 58
9.9 8.5 5.4 4.9 2.7 2.2 3.1 2.2 2.7 1.8 2.2 1.8 1.3 1.3 0.9 0.4 0.9 0.4 0.9 0.9 0.9 0.9 0.4 0.4 0.4 0.4 0.4 0.4 0.4 0.4 0.4 0.4 0.4
3 1 3 0 2 2 0 1 0 1 0 0 1 0 1 2 0 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
5.5 1.8 5.5
9.0 7.2 5.4 3.9 2.9 2.5 2.5 2.2 2.2 1.8 1.8 1.4 1.4 1.1 1.1 1.1 0.7 0.7 0.7 0.7 0.7 0.7 0.4 0.4 0.4 0.4 0.4 0.4 0.4 0.4 0.4 0.4 0.4
3 4
12 20
Gout Herpes stomatitis Small bowel necrosis Marginal ulcer
22 19 12 11 6 5 7 5 6 4 5 4 3 3 2 1 2 1 2 2 2 2 1 1 1 1 1 1 1 1 1 1 1
3.6 3.6 1.8 1.8
1.8 1.8 3.6 1.8
25 20 15 11 8 7 7 6 6 5 5 4 4 3 3 3 2 2 2 2 2 2 1 1 1 1 1 1 1 1 1 1 1
73 13 29 71 67 60 25 100
100 100 100
1 1
100 100
* Multiple complications often figured in 1 death. + Postoperative fever refers to those patients with postoperative pyrexia for which no cause could be determined. t Pancreatic fistula refers only to those patients in whom leakage of pancreatic juice was significant enough to require treatment.
volume 133, Apfir 1977
481
Braasch and Gray
TABLE I II
TABLE IV
Use of Stent across Pancreatojejunal Anastomosis”
Type of Stent Short Long
None Duct ligated
No. of Cases
220 11
34 6
Complications
Mortality
__-
No.
%
No.
%
128 6 12 5
58 55 35 83
27 1 5 0
12 9 15 0
* In 14 other
patients, there was no pancreatic anastomosis since the pancreatic duct was ligated or the operation resulted in total pancreatectomy.
The overall mortality has decreased from 17 per cent for the six years from 1957 through 1962 to 10.7 per cent for the six years from 1969 through 1974. Analysis of disease treated and operative mortality (Table I) shows a much higher rate of postoperative deaths for patients with malignant tumors of the duodenum and lower bile duct than for patients with tumors in other periampullary sites and a much lower mortality rate for patients with pancreatitis. Operatiuc Technic. The technic of radical pancreatoduodenectomy at the Lahey Clinic has been modified and standardized over many years and has been previously described in detail [1,2]. However, over the eighteen year period (1957 through 1974), technic has varied somewhat, especially with regard to the management, of the pancreat,ic duct and the use of bilateral truncal vagotomy at the time of resection. In 220 operations, the pancreatic duct was anastomosed to the jejunal mucosa using a short rubber stent to splint the anastomosis. In 11 operations, a much longer stent was used and was brought out through the jejunum and abdominal wall to drain pancreatic secretions externally. In 34 operations, no stent was used although precise mucosa-to-mucosa pancreatojejunostomy was performed. Table III compares the morbidity and mortality of these three ways of dealing with the pancreatic anastomosis. None of the differences observed are significant. In six patients, the pancreatic duct was ligated and no pancreatic anastomosis was const.ructed. The morbidity rate of 83 per cent for this group is significant (p <0.05) when compared with the group having anastomosis. Of the group, 15 patients with tumor and 7 with pancreatitis underwent bilateral truncal vagotomy at the time of resection. Vagotoniy might be expect,ed to decrease the incidence of postoperative gastrointestinal bleeding after resection. However, in two of the patients, postoperative hemorrhage
occurred from a marginal ulcer and a gastric ulcer at the lesser curvature. Effect of Jaundice. The level of preoperative jaundice is correlated with postoperative complications and death in Table IV. Clearly shown is the marked increase in the incidence of liver failure, wound hemorrhage, and postoperative death in those patients with serum hilirubin levels greater than 20 mg/lOO ml. Postoperative Hemorrhage. Postoperative hemorrhage occurred in a total of 31 patients. The severity of bleeding ranged from a transient episode of melena to massive un-
482
Incidence of Complications after Whipple Procedure: Correlation with Serum Bilirubin Level
Serum Bilirubin (mg/lOO ml) 1
l-
9.9
10-19.9 20-30+
No. of Cases
Renal Failure
Wound Hemorrhage
Wound Sepsis
(%I
(%I
(%I
(%I
Mortality
92
2.6
6.5
22.1
13.0
111 48 28
2.6 7.5 9.0
15.0 10.0 22.0
20.4 20.0 26.0
9.7 12.5 22.0
controlled bleeding. Hemorrhage occurred from the gastrointestinal tract in 18 patients, from the wound in 10 patients, and from both sites in 3 patients. The overall mortality for patients with postoperative hemorrhage was 58 per cent. Twelve of the patients were operated on again at least once for uncontrolled bleeding. Eight of these patients died despite reoperation. In the 12 reexplored patients, bleeding was noted in the gastrointestinal tract (gastrojejunal anastoniosis, 4; marginal ulcer, 1; lesser curvature ulcer, 1; undetermined, 1) and in the wound (gastroduodenal artery, 3; pancreatic bed, 2; superior mesenteric vein, 1; portal vein, 1; undetermined, 1). The time at which hemorrhage occurred did not show any particular pattern and varied from within several hours to more than three weeks after operation.
Comments
Radical pancreatoduodenectomy is an effective method of treatment for carcinoma of the lower bile duct, duodenum, and ampulla of Vater. In a recent report from the Lahey Clinic [3] the five year survival rate was 41.3 per cent for lesions of the duodenum, 32 per cent for lesions of the ampulla, and 25 per cent for lesions of the common bile duct. Although postoperative morbidity and mortality rates associated with pancreatoduodenectomy are formidable, the relatively good chance of cure for lesions other than of the pancreatic head overrides the risk of postoperative complications. Previous reports [4-61 have noted a much higher operative mortality than in this series. The mortality for 1,169 cases reported by Fish and Cleveland [4J was 23.7 per cent. The high mortality associated with pancreatoduodenectomy has been used as one factor in favor of total pancreatic resection for periampullary carcinoma. This is not substantiated by analysis of this series of patients. The mortality and complication rates compare favorably with those of a series of 64 patients undergoing total pancreatectomy reported by Pliam and ReMine [7]. The relatively low mortality rate reported here is in part a function of familiarity with the operation and of particular attention paid to operative technic,
The American Journal 01 Surgery
Pancreatoduodenectomy
especially the pancreatojejunal anastomosis. Accurate anastomosis of the main pancreatic duct to the jejunal mucosa has been advocated at this institution to diminish the chance of leakage of pancreatic juice and to preserve some exocrine pancreatic function. Wound sepsis, hemorrhage, and renal failure were the major commonly occurring complications that were asiociated with a high mortality. Similar findings have previously been recorded by others [3,6]. The use of two varieties of stents to splint the anastomosis has not resulted in a statistically significant decrease in the incidence of postoperative fever, wound sepsis, and mortality although there is a suggestion that the use of a long stent would prove statistically beneficial with a larger experience. The suggestion has been offered that pancreatic anastomoses do not stay open and that either oversewing of the end of the pancreas [8] or total pancreatectomy [9] is a safer alternative. Our experience with the former is 6 cases with an 83 per cent morbidity (0 mortality) rate and is too limited for comment. We have determined the serum amylase content of the juice from the jejunum near the pancreatic anastomosis in four patients and found it to contain very high concentrations. Furthermore, two patients with pancreatitis have been reexplored after the Whipple procedure and the anastomoses have been found to be widely patent. Certainly many long-term survivors without pancreatitis do not have clinical steatorrhea. The proponents of total pancreatectomy for malignancy have not as yet proved their case of improved long-term survival. It seems appropriate, therefore, to continue use of the pancreatic anastomosis in most cases. Bilateral truncal vagotomy has not significantly altered postoperative complications. If an adequate amount of the stomach is removed it appears to be unnecessary to add vagotomy to the resection. The high incidence of serious complications in severely jaundiced patients suggests that initial biliary decompression followed at a later date by radical resection has a place in the management of those patients in whom the serum bilirubin level exceeds 20 mg/lOO ml. A staged procedure is mandatory in patients with sepsis in the right upper quadrant after previous exploration. Postoperative hemorrhage is an ominous occurrence with a high mortality. Reexploration and suture of the bleeding point is often followed by further fatal hemorrhage. Attempts at resuturing leaking pancreatic anastomoses at that time are futile. Removal of the rest of the pancreas to control pancreatic fistulas is extremely hazardous. It is possible that continuous copious irrigation of the operative region by means of inlying catheters can prevent hemorvolulna 193. April 1@77
rhage in pancreatic fistulas with sepsis in these particularly high-risk patients. The high mortality associated with lesions of the duodenum and to a lesser extent with common bile duct lesions may be due to the fact that the soft consistency of the pancreas and small size of the main pancreatic duct are more difficult to handle than the firm glands with dilated ducts seen in carcinoma of the pancreatic head or ampulla and chronic pancreatitis. Application of total pancreatectomy to these selected patients would result in a lower postoperative mortality rate. Summary
To aid in case selection for pancreatoduodenectomy and to gain information on the technical management of this operation and its complications, records of 279 patients who were treated for neoplasm or pancreatitis by this procedure between the years 1957 and 1975 were reviewed. The overall operative mortality was 12.5 per cent and was 10.7 per cent for the years 1969 through 1974. The use of vagotomy did not prevent postoperative bleeding from the stomach, and the use of a stent did not make a statistically significant difference in morbidity or mortality. Postoperative hemorrhage is an ominous complication and is best treated conservatively until blood loss cannbt be replaced. Preoperative serum bilirubin levels above 20 mg/lOO ml indicate a two-stage operative procedure as does the presence of right upper quadrant sepsis. The resection of malignant disease of the duodenum and lower bile duct is followed by a high mortality and requires total pancreatectomy if a satisfactory pancreatojejunostomy cannot be constructed. References 1. Warren KW, Braasch JW, Thum CW: Diagnosis and surgical treatment of carcinoma of the pancreas. Curr Probl Surg Chicago, Year Book Medical, 1966, p 3. 2. Braasch JW, Gray BN: Technique of radical pancreatoduodenectomy with consideration of hepatic arterial relationships. Surg C/in North Am 56: 631, 1976. 3. Warren KW, Choe DS, Plaza J, et al: Results of radical resection for periampullary cancer. Ann Surg 161: 534, 1975. 4. Fish JC, Cleveland BR: Pancreaticoduodenectomy for periampullary carcinoma. Analysis of 36 cases. Ann Surg 159: 469, 1964. 5. Hicks RE, Brooks JR: Total pancreatectomy for ductal carcinoma. Surg Gynecol Obstet 133: 16, 1971. 6. Mongit JJ, Judd ES, Gage RP: Radical pancreatoduodenectomy: a 22-year experience with complications, mortality rate, and survival rate. Ann Surg 160: 711, 1964. 7. Pliam MB, ReMine WH: Further evaluation of total pancreatectomy. Arch Surg 110: 506, 1975. 6. GoldsmIth HS, Ghosh BC, Huvos AG: Ligation versus implantation of the pancreatic duct after pancreaticoduodenectomy. Surg GynecolObstet 132: 67, 1971.
463
Braasch and Gray
9. Brooks JR, Culebras JM: Cancer of the pancreas: palliative operation, Whipple procedure, or total pancreatectomy? Am J Surg 131: 516, 1976.
Discussion John R. Brooks (Boston, MA): First, Doctor Braasch’s complication rate after the Whipple procedure is high at 67 per cent. He is correct in placing the blame for this upon leaks occurring most commonly at the pancreatojejunostomy, but also the choledochojejunal anastomosis can be a fault. I agree that a mucosa-to-mucosa anastomosis is important. It is true that an indurated pancreas (as in chronic pancreatitis) or a similarly blocked pancreas by tumor is easier to suture and less likely to leak than the pancreas that is associated with an ampullary lesion or duodenal lesion in which the duct of Wirsung has not been totally blocked and the pancreatic body is not firm and indurated. The other major cause of complication is gastrointestinal bleeding. We believe that vagotomy should be performed to minimize the chance of peptic ulceration. Second, the 10 per cent mortality he presented is excellent. His survival figures are also good for lesions around the ampulla and around the lower end of the common duct. I would simply like to ask Doctor Braasch what his five year survival figures are for lesions in.the pancreatic head. Third, debatably, I question the need for a Whipple procedure for pancreatitis. Obviously, none of us has a consistently successful procedure for chronic pancreatitis, but I would be more inclined to do an internal drainage procedure or subtotal distal pancreatectomy rather than the more radical Whipple procedure, which I notice in the manuscript carried a 7 per cent mortality. Finally, Doctor Braasch has lumped together pancreatic head lesions with ampullary, duodenal, and common duct lesions. Survival figures are not comparable, as we all know. As many of you are aware, my recent philosophy has been to employ the Whipple procedure for the periampullary lesions, namely the ampullary, duodenal, and lower end of the common duct-type tumors, but to use total pancreatectomy for lesions of the head of the pancreas or the body and the tail. Mortality figures for the totals are 11per cent in the Mayo Clinic series and 12 per cent in ours, not much different than the 10.7 per cent that Doctor Braasch has shown us here today for the Whipple procedure. George L. Nardi (Boston, MA): Unfortunately pancreatic cancers have one of the lowest operability rates. Only approximately 12 per cent of pancreatic cancers are found to be localized at the time of operation. The newer developments such as ultrasound, pancreatic scanning, and selective angiography have not been as successful in uncovering early pancreatic cancers as we had hoped. The newest technic of computerized body scanning may be of help, but we are always limited by the time the patient actually presents himself. At one time a high operative mortality was another discouraging factor along with a low survival rate. During the past decade, however, numerous series with little or no operative mortality have been accomplished and at the
Massachusetts General Hospital we have been able to increase our five year survival rate for adenocarcinoma of the head of the pancreas from a figure of 5 or 6 per cent to 15 per cent. I certainly agree with Doctor Braasch’s policy of performing preliminary biliary drainage in those patients with high bilirubin levels. This should be a tube drainage, however, and not cholecystojejunostomy. I have not done routine vagotomies in these operations but have depended on an adequate gastric resection to eliminate the possibility of marginal ulcer. Our Radiation Therapy Department has reviewed the mortality after pancreatic resection over the past ten years and they find that in more than 50 per cent of these patients death was due to local recurrence within the abdomen rather than distal metastases. As a result, we have been supplementing our resective efforts with implantation of iodine 125 seeds in the resected stump after frozen section shows no residual tumor. In addition, we follow this treatment with external beam radiation. John W. Braasch (closing): With regard to Doctor Brooks’ question concerning our five year survival, I believe it is similar to that at the Massachusetts General Hospital, roughly 15 per cent. I would admit that all of these 15 per cent probably are not cured, since we have had a number of them show up with more tumor after the five year interval. I think we must bear in mind that it is not always possible to recognize a malignant mass in the head of the pancreas and to differentiate it from the more curable and more favorable carcinomas of the lower duct or ampulla in which about a third survive five years. Therefore, all pancreatic head lesions cannot categorically be rejected for resection. However, I would agree that the indication for the operation is questionable when the lesion is clearly within the head of the pancreas, unless one subscribes to the theory that the Whipple operation is worthwhile as a palliative procedure. We will look forward to further follow-up in Doctor Brooks’ series of total pancreatectomies, especially when a sufficient number have been followed until death or for five years. I would agree that the concentration of this operation in selected hands is able to reduce the mortality for the operation. It is quite an exacting technical procedure. In the two-stage procedure for high preoperative bilirubin levels, the first stage should be the implantation of a T tube as noted by Doctor Nardi and not the construction of anastomosis. Relative to the pancreatic anastomosis, of basic importance is keeping the activated pancreatic juice away from the raw areas and vessels in the posterior part of the abdomen. This might be accomplished by diverting the juice externally by an intraductal tube through the anastomosis, by separating the biliary and pancreatic anastomoses as suggested by Doctor Nardi, or by irrigating the posterior parieties by suitable irrigating and suction tubes. If a leakproof anastomosis could always be constructed, these would not be necessary. Unfortunately such is not the case.
The American Journal ol Surgary