S who had undergone operations on the pancreas,

Transcription

1 The Whipple Procedure and Other Standard Operative Approaches to Pancreatic Cancer WILLIAM P. LONGMIRE, JR, MD, AND L. WILLIAM TRAVERSO, MDt In a series of 50 patients with cancer of the pancreas who were undergoing resection of the pancreatic head and the duodenum, the average survival was 16.2 months. If the resected margin was free of tumor the average survival was 20.3 months. In 103 less favorable cases treated by biliary bypass, with or without gastroenterostomy, the average survival was only 6.2 months. Although these series are not comparable, pancreaticoduodenal resection for cancer of the pancreas and periampullary area has been utilized in cases without distant metastases in an effort to extend the period of palliation and to achieve occasional long-term survival. Currently available diagnostic techniques may permit earlier diagnosis and improved results. A positive tissue diagnosis is not essential before proceeding with resection, but this aspect of the operation should be fully discussed with the patient. An unfavorable condition of the residual pancreas for anastomosis, or a positive frozen section examination of the pancreas at the line of transection may indicate total pancreatectomy, but complete resection is not used routinely. The incidence of jejunal ulceration, a serious late postoperative complication, may be reduced by utilizing a 60-70% gastric resection or antrectomy and vagotomy. Cancer 47: , A U V ~ in ~ 1908 collected the reports on 11 patients S who had undergone operations on the pancreas, eight of them before the turn of the century. Included was an operation he attributed to Codivilla in which a cancer involving the duodenum and head of the pancreas was resected en bloc with the head of the pancreas, a major portion of the duodenum, and the pyloric end of the stomach. A Roux-Y gastroenterostomy reestablished alimentary continuity and the cholecystoenterostomy was performed with a Murphy button. The patient survived for 24 days but died with progressive cachexia and glycosuria. Sauve also proposed a technique of pancreaticoduodenal resection in two stages, a procedure he had developed from a study of fresh cadaver material but had not performed on a patient. At the first operation a gastroenterostomy was to be created. At the second stage the second portion of the duodenum together with the adjacent Presented at the International Meeting on Pancreatic Cancer, Louisiana State University School of Medicine, New Orleans, Louisiana, March 10-1 I, From the UCLA School of Medicine, Department of Surgery, Los Angeles, California. * Professor of Surgery. t Assistant Professor of Surgery. Supported in part by a grant from the Weingart Foundation. Address for reprints: William P. Longmire, Jr., MD, UCLA School of Medicine. Department of Surgery, Los Angeles. CA Accepted for publication June 16, pancreatic tissue was to be resected and bile flow reestablished by a choledochoenterostomy. He proposed that no attempt be made to anastomose the pancreatic stump but that it be closed and drained. During the early 1900s, reports of limited series of palliative resections of pancreatic tumors continued to be published, some noting operative mortalities as high as 70%. Surgery of the pancreas was given new impetus in 1935 when Whipple, Parsons, and Mullins*l described their successful two-stage radical en bloc resection of the head of the pancreas and the duodenum. A number of technical variations in the method of reconstruction have followed the presentation of their original successful operation. From the results of these modifications and the experiences of the last 45 years, certain general principles of technique and anticipated results have evolved: (1) the pancreaticojejunal anastomosis is the most vulnerable step in the reconstruction. If the stump of the pancreas is closed rather than anastomosed, it should be thoroughly drained because pancreatic leakage is common. If the patient has normal blood glucose levels prior to operation and the pancreas is divided over the superior mesenteric vein, postoperative insulin deficiency is rare but late onset diabetes is a possibility. Some degree of incomplete fat digestion and absorption tends to occur postoperatively, regardless of the method of pancreaticoenterostomy ; however, the clinical symptoms created by this >(/81/0315/1706 $ American Cancer Society 1706

2 No. 6 OPERATIVE APPROACHES TO PANCREATIC CA ' Lotignzire and Traverso 1707 deficiency vary widely from patient to patient. (2) Vagotomy and antrectomy or 70% gastric resection are likely to be necessary to prevent marginal ulceration if the pylorus and first portion of the duodenum are included in the resection. (3) Bile with or without pancreatic juice should be drained into the jejunum proximal to the gastrojejunostomy. In 1958 PorteP reviewed the experience in pancreatic surgery at the Columbia-Presbyterian Medical Center in New York and categorized it into three periods: The Pioneer Period, starting with Whipple's original resection at that institution in 1935 through 1947, characterized by concern simply for "getting away" with the operation per se. The Radical Period, 1948 through 1952, when more extensive resections, including total pancreatectomy, were performed in an effort to improve the unsatisfactory cure rate for patients with cancer of the pancreas. The Rational Period, 1953 through 1957, was the third and final period when Porter prepared his report. It had become clear that the numerous extensive, risky, and ineffective resections being performed until then achieved only minimal improvement. In this third period. therefore, physicians began to select the most favorable cases for resection while employing palliative procedures for more advanced lesions. Despite the continuing predominance of this selectivity among surgeons reporting pancreatic cancer resections, survival rates after five years remain lowlow enough, indeed, to lead some surgeons to conclude that cancer of the head of the pancreas with biliary obstruction should be treated solely by biliary bypass, and that resection is seldom if ever indicated. One might wonder if this is now a fourth period since Porter's 1958 report, inasmuch as several diagnostic aids are now readily available that had not yet been developed or were infrequently used a decade ago. This could be labeled the "Diagnostic Period." Endoscopy, preoperative pancreatograms, CAT scans, ultrasonography, and aspiration cytology are specific tests that may well affect an earlier diagnosis of pancreatic cancer. It is not possible at present to demonstrate that these aids have significantly shortened the interval between onset of disease and exploration, nor is it certain that, if the time interval were shortened, survival would be improved after resection. There does seem reason to believe that the widespread use of these diagnostic aids enables earlier diagnosis of pancreatic cancer. Advanced lesions are easier to identify and treat by biliary decompression rather than by under- going resection; moreover, the resection of earlier, localized tumors should, it is hoped, improve the survival rate. More time and evaluation of results are essential to learn whether such achievements can be realized. The current problems associated with surgery for cancer of the pancreas are not new; we have been grappling with them for several decades, yet the answers are still not entirely satisfactory. Some of our principal questions are: Are the results of pancreaticoduodenectomy so ineffective that we should not consider resection? If a resection is to be performed, is it essential to secure a positive tissue biopsy'? How much of the pancreas should be resected'? Does the method of reconstruction make a significant difference in the patient's postoperative gastrointestinal function'? Should Resection of the Pancreas and Duodenum be Used in the Treatment of the Head of the Pancreas? In our study of 279 patients with carcinoma of the pancreas and periampullary area,' 67 patients were found to have widespread metastases. Biopsy only was performed. The average survival was 3.6 months after operation. A biliary bypass with or without gastroenterostomy was performed in 103 patients; average survival was 6.2 months. The subsequent addition of gastric decompression to the biliary bypass in eight patients prolonged the average survival to 11 months. Interestingly enough, when both bypasses were done at the same operation (41 patients) the average survival was only 4.8 months. Resection of the head of the pancreas and duodenum was performed in 50 patients whose subsequent average survival was 16.2 months. Clearance of the tumor confirmed by histopathology achieved a patient survival of 20.3 months as compared to 12.9 months when tumor was found in the resected margin. Despite its inadequacies, frozen section examination of the resected margins at the time of the operation is mandatory. The five-year survival rate in these cases was 4%. Although such results cannot be interpreted precisely, they tend to support the view that despite the distressingly low five-year survival of these patients, resection is the only form of treatment that has resulted in long-term survival. Resection with the pancreatic margins free of tumor by microscopic examination triples the period of survival over a simple bypass. Some observers believe this procedure improves the quality of life by relieving pain and improving digestive function during the period of palliation.

3 ~ 1708 CANCER Mrrrcli 15 Sirpplernetit 1981 Vol. 47 T~RLE. 1, Results of Treatment of Periampullary Carcinoma: Collected Series 3-yr 5-yr No. of Operative survey survey Author cases mortality (P) (P) Carcinoma bile duct Sato Carcinoma ampulla Sat0 Carcinoma pancreas Sat0 Carcinoma duodenum Sato I20* I I t (2 yr) (2 yr) (2 yr) (2 yr) 43.8 * Duodenum and ampulla. t See ampulla (above). t One additional patient had a total pancreatectorny and is alive at 52 months Another argument for resection, and a strong one, is that it is frequently difficult if not impossible to distinguish carcinoma of the intrapancreatic portion of the bile duct from carcinoma of the pancreas. Problems may also arise in differentiating tumor of the duodenum that has invaded the head of the pancreas from tumor of the pancreas that has invaded the duodenum. Moreover, some small occult ampullary tumors may be mistakenly diagnosed as pancreatic carcinoma. Survival with these other periampullary tumors is better than with pancreatic cancer. In a previous report,4 the five-year survival was 24% for patients with ampullary tumors, % for patients with tumors of the distal bile duct, and four or five patients were alive following a Whipple resection for carcinoma of the duodenum from seven months to 4.75 years.,"j et al.,i9 and others1"*1'~'3 have reported similar or better results (Table 1). Therefore, based on limited but occasional long-term survival and with the opportunity to resect other more favorable but indistinguishable periampullary tumors, persistent efforts are made to identify the localized periampullary tumor, including any in the head of the 7 pancreas, and to proceed with resection. It is hoped that present diagnostic methods will prove effective in earlier indentification of localized resectable lesions. Is Positive Tissue Diagnosis Essential Before Proceeding with Resection? If the premise is acceptable that certain cases of periampullary cancers (including cancer of the pancreas) should be treated by resection, the first question at operation (assuming that no preoperative tissue or cellular diagnosis has been made) is to identify the jaundice-producing mass in the head of the pancreas. In nonalcoholic patients 50 years of age or older with a mass in the head of the pancreas producing persistent jaundice, most surgeons proceed with resection after a diligent though limited attempt to obtain a positive tissue diagnosis. Lacking this, it is essential to exclude an impacted stone, penetrating ulcer, or benign tumor of the duct or ampulla. On the other hand, when pain is a prominent symptom, a Whipple resection may be the preferred treatment for certain patients with benign chronic pancreatitis. In younger patients with some degree of alcoholic history, a precise diagnosis is far more compelling. Patients with large obvious cancers will not do as well as those with small occult tumor confined within the substance of the gland. SatoL3 examined patients for evidence of macroscopic invasion of the pancreatic capsule. Three of nine patients who were free of capsular invasion survived more than three years but all nine patients with capsular invasion died within two years. Survival was also affected by lymph node involvement. Two of six patients without lymph node involvement survived more than three years, but only one of 12 patients with positive lymph nodes survived for three years. 1g reported a 16.8% five-year survival rate after resecting cancer of the pancreas when lymph nodes were negative; this dropped to 7.4% when positive nodes were present. Thus it is important to detect and resect the occult early cancer without lymph node metastasis or obvious capsular invasion, yet withhold unnecessary radical operations for benign disease. In a previous report of pancreatic cancer operationsy a positive tissue diagnosis was available in 3 1 of 46 cases prior to resection. In four cases a false-negative biopsy had been obtained and disregarded. Obviously, the small, early intraglandular tumor with negative nodes offers the best chance for cure with resection, but it is also the most difficult upon which to perform an accurate biopsy. This diagnostic dilemma is carefully discussed with the patient before the operation and specific written permission is obtained to proceed on the basis of all

4 No. 6 OPERATIVE APPROACHES TO PANCREATIC CA Longmire and Traverso 1709 the clinical and pathologic findings, not necessarily including a positive tissue diagnosis. It is recognized that such a policy occasionally results in an unnecessary resection. On the other hand, it is possible to overlook an opportunity to cure a malignant lesion which, owing to its size and localization, could have been favorably resected. What Should Be the Extent of Pancreatic Resection? In previous cases,4 total pancreatectomy was performed on six patients with an average survival of 26 months, the longest in the series. Two of the six patients were alive at the time of the report, one at two and another at 6.5 years. Pathologic examination of the excised pancreas, however, revealed that the tumor was confined to the head of the pancreas in five of the six specimens. In only one case was tumor present in both the head and body. This finding recalls a previous study in which residual tumor in the pancreas was found in only three out of 14 patients who died and underwent autopsy after a Whipple resection. Despite the superior survival results in this small series of total pancreatectomies, it is difficult to accept this as a routine procedure in all cases of carcinoma of the head of the pancreas. It is preferable to rely on frozen section of the margins of resection to detect residual or multicentric carcinoma. If a multicentric tumor is identified, total pancreatectomy is indicated. The condition of the remaining pancreas may be another indication for total resection. If the duct of Wirsung has its normal limited diameter and delicate wall, and if the remaining gland is composed of soft, friable, nonfibrotic tissue, the risk of fistula from the stump if it is closed, or from the anastomosis if pancreaticojejunostomy is performed, is high. Pancreatic fistulas occurred in 8.5% of patients after partial pancreatectomy for tumor in the 233 patients reported by Braasch and Gray,2 there was a 20% mortality. Of course, this risk is eliminated if total pancreatectomy is performed. New methods of duct occlusion, which will be discussed later, may reduce the risk of fistula from the pancreatic stump and make partial resection less hazardous. If and when the transplantation of autologous islet cells becomes a reliable procedure, total pancreatectomy might well become the preferred procedure for localized bile duct, ampullary, or duodenal malignancies if the normal friability of the pancreatic stump increases the danger of closure or anastomosis. Total resection is a more extensive operation and, combined with the postoperative diabetic state and a possible increase in exocrine deficiency, diminishes the palliation that such operations provide. Techniques of Reconstruction Following the usual resection of the distal stomach, duodenum, and head and neck of the pancreas, the method of reconstruction described here whenever possible includes a direct mucosa-to-mucosa anastomosis of the pancreatic duct to the jejunum. The anastomosis is splinted with a plastic catheter that is usually passed from the duct in about three weeks. If the duct is quite small and thin walled, the entire end of the pancreas is invaginated into the jejunum although it is believed, based on limited observations at reoperation and autopsy, that the duct of Wirsung tends to become scarred and the end seals within a few weeks of such anastomoses. We have abandoned the technique of oversewing the residual pancreatic stump without an anastomosis because pancreatic fistulas and serious related complications frequently develop. These authors have had no experience with pancreatic duct occlusion by the intraductal injection of an acrylate glue such as that noted by Little et 01.~ in cases of chronic pancreatitis or the method reported by Gebhardt and Stoke6 in which the pancreatic duct after pancreaticoduodenectomy is injected with Prolamin, a fast-solidifying alcoholic solution of aminoacids; in this way the pancreatic stump is closed without an anastomosis. The solution is reported to be nonreactive, to produce a high-grade atrophy of the exocrine parenchyma without causing acute pancreatitis, and to reabsorb in about 14 days. This method has been utilized in 43 patients subsequent to partial duodenopancreatectomy for chronic pancreatiti~.~ An end-to-side choledochojejunostomy is established next. The anastomosis may be splinted with a T-tube inserted through the bile duct; the tube is removed in two weeks if all anastomoses are healed. If leakage occurs at the pancreaticojejunal anastomosis, the T-tube is useful in decompressing the upper segment of the jejunum and reducing the flow through the fistula. In Whipple s original operation in 1935, the entire stomach and pylorus were left intact, the duodenum was closed, and a side-to-side gastroenterostomy was performed. This technique was subsequently modified and a limited gastric resection perf~rmed.~ in 1957 noted the development of three jejunal ulcers in 23 patients and suggested that such late marginal ulceration was more common than heretofore believed. However, he did not recommend high gastric resection or vagotomy until ten years later,20 when he reported the occurrence of jejunal ulcers in 7.5% of patients and suggested that half of the stomach be removed or that a vagotomy be performed. Braasch and Gray2 found that truncal vagotomy did not significantly alter

5 1710 CANCER March 15 Supplement 1981 Vol. 47 postoperative complications if an adequate amount of stomach was removed. Two recent reports again emphasize the hazard of marginal ulceration, particularly with total pancreatectomy, and recommend routine vagotomy after total and subtotal pancreatectomy, with certain exceptions. Scott et al. l5 emphasized the ulcerogenic potential of pancreaticoduodenectomy. Of 26 patients undergoing a Whipple resection and 13 in whom a total pancreatectomy were performed, eight developed marginal ulceration at the gastrojejunostomy from one month to six years after operation. Seven of the 17 patients (41%) who did not have a vagotomy with subtotal resection had ulceration. Marginal ulcer also developed in one of four patients who underwent total pancreaticoduodenectomy without vagotomy. In contrast, only one of 22 patients (5%) who had concomitant vagotomy developed a marginal ulcer. Grant and Van Heerden7 in a review of 297 patients who had undergone radical pancreaticoduodenectomy (Whipple procedure, 233 cases) or total pancreatectomy (64 cases) found that in 6% of these patients marginal ulceration occurred from one month to 12 years later, with an average interval from operation to diagnosis of ulcer of 1. years. A 50% gastric resection proved inadequate in preventing ulceration. The chance of an anastomotic ulcer was 6% for a patient surviving five years after a Whipple procedure and 18% for a patient surviving five years after total pancreatectomy. Resection for pancreatitis seemed to predispose to postoperative ulcer more than resection for tumor. The prolonged survival after resection for benign disease also represents a greater risk. Grant and Van Heerden commented that a high gastric resection could leave the patient a nutritional cripple and that vagotomy exposed the patient to the hazards of prolonged operating time, immediate complications, and late postvagotomy syndromes. These sequelae together with the other undesirable postoperative effects of pancreatectomy led them to believe that vagotomy should not be dealt with lightly. Nonetheless, they concluded that for patients undergoing total pancreatectomy, and possibly the Whipple procedure for benign disease, ampullary carcinoma resected for cure or perhaps a low-grade periampullary carcinoma without nodal metastases, the addition of vagotomy should be considered. Moreover, if there were any previous history of peptic ulcer disease, vagotomy should be performed. It has been our custom to remove 60-70% of the stomach with pancreatic resection or to perform a vagotomy. The only late marginal ulcer in our series occurred in a patient who was subsequently diagnosed as having a Zollinger- Ellison tumor. In an attempt to minimize the undesirable gastrointestinal sequelae of pancreaticoduodenectomy for benign disease, these authors began in February 1977 to preserve the pylorus and first portion of the duodenum, making certain that the blood supply of the duodenum was adequate and that the anastomotic suture lines did not interfere with the function of the pylorus.'' To date, 17 such operations have been done with satisfactory early results in all cases. A detailed reevaluation of ten such patients is the subject of a future report. In addition to those with chronic pancreatitis, the procedure has been used for patients with carcinoma of the third portion of the duodenum, ampullary carcinoma, and for a very small localized pancreatic carcinoma located at the ampulla and without lymph node metastases. Inasmuch as the first operation was performed three years ago, the possibility of late peptic ulceration cannot be ruled out, but it is encouraging that gastric acids are within normal limits in all of these patients and there has been no evidence of peptic ulceration. Carcinoma of the Body and Tail of the Pancreas Sixty-seven patients in this series had widespread tumor and only a biopsy was performed. Pain was the most frequent (72%) presenting symptom. None of them was jaundiced. Patients in this group who had a history of pain aggravated by meals and in whom a dilated pancreatic duct could be identified were treated with pancreaticogastrostomy with a T-tube placed in the duct of Wirsung that passed through the stomach and anterior abdominal wall. Some lessening of pain was noted but this result was difficult to evaluate. After another palliative procedure, the injection of 50 cc of 50% alcohol about the coelic axis, it seemed that the pain diminished postoperatively but again it was not possible to evaluate this result. Narcotic pain medication could not be stopped entirely in any patient after operation. Six other patients, who had no jaundice and were operated upon for severe back pain, underwent resection of the distal pancreas. Approximately 50% of the pancreas was removed together with the spleen in five of these patients; and in the remaining patient a 95% distal pancreatectomy was performed with splenectomy. One patient died during the first month. The average survival was eight months with a range of 0-20 months. During operation four patients had obvious metastases that were not removed, and two were deemed to have been cleared of all macroscopic tumor, although at subsequent pathologic examination tumor was found at the resected margin. Summary A review of these 279 patients with cancer of the pancreas and the current reports of others lead to the conclusion that cancers of the pancreas and periampullary tumors without distant metastasis should be

6 No. 6 OPERATIVE APPROACHES TO PANCREATIC CA * Longmire and Traverso 1711 resected. Despite poor long-term results for patients who have had resection of pancreatic cancer, improvement may be possible if earlier diagnosis can be made with currently available procedures. Diligent attempts should be made to rule out benign disease and to obtain a positive biopsy before proceeding with pancreaticoduodenectomy. The patient s permission must be obtained prior to operation so that resection may proceed on the basis of clinical evidence and gross pathologic findings. The place of total pancreatectomy in treating cancer of the pancreas has not been firmly established. Although routine total resection is not currently performed, the procedure is reserved for selected cases. Management of the residual pancreas after partial pancreatectomy is important inasmuch as postoperative pancreatic fistula remains a frequent and serious complication. At present some type of pancreaticojejunal anastomosis is recommended. If the condition of the pancreas is unfavorable for anastomosis, total resection may be necessary. Jejunal ulceration is a serious late postoperative complication and although it is undesirable to add the side effects of vagotomy or high gastric resection of those of an already major rearrangement of the upper gastrointestinal tract, high gastric resection (70%) or vagotomy is indicated with total pancreatectomy and 60-70% gastrectomy or vagotomy with subtotal resection. For certain periampullary tumors, preservation of the pylorus may eliminate the need for these additional procedures. REFERENCES 1. Aston SJ, Longmire WP Jr. Management of the pancreas after pancreaticoduodenectomy. Ann Surg 1973; 179: Braasch JW, Gray BN. Considerations that lower pancreaticoduodenectomy mortality. Am J Surg 1977; 133: Cattell RB, KW. Surgery ofthe Pancreas. Philadelphia: W. B. Saunders Co., JF, Longmire WP Jr. Carcinoma of the pancreas and periampullary region. Ann Surg 1979; 189: Gall FP, Gabhardt Ch. Em neues konzept in der chirurgie der chronischen pankreatitis. Dfsch Med Wochenschr 1979; 104: Gebhardt Ch, Stoke M. Pankreasgang-Okklusion durch injektion liner schnell hartenden aminosaurenlosung. Langenbecks Arch Chir 1978; 346: Grant CS, Van Heerden JA. Anastomotic ulceration following subtotal and total pancreatectomy. Ann Surg 1979; 19O:l Little JM, Lauer C, Hog J. Pancreatic duct obstruction with an acrylate glue: A new method for producing pancreatic exocrine atrophy. Surgery 1977; 81: Longmire WP Jr, Bruckner WL. Periampullary carcinoma. Proc. Sectional Meeting American College of Surgeons-German Surgical Society, Heidelberg, Springer-Verlag, 1968; MongC JJ, Judd ES, Gage RP. Radical pancreaticoduodenectomy: A 22-year experience with the complications, mortality rate, and survival rate. Ann Surg 1964; 160: A, Matsumoto Y, Uchida K, Honjo I. Surgical treatment of cancer of the pancreas and the periampullary region. Ann Surg 1977; Porter MR. Carcinoma of the pancreatico-duodenal area: Operability and choice of procedure. Ann Surg 1958; 148: Sat0 T, Saitoh Y, Noto N, Matsuno S. Follow-up studies of radical resection for periampullary cancer. Ann Surg 1977; 186: Sauve L. Der pancreatectomies et specialement de la pancreatectomie cephalique. Rev Chir 1908; 37: Scott HW, Dean RH, Parker T, Avant G. The role of vagotomy in pancreatico-duodenectomy. Ann Surg 1980; 191: 688-6%. 16. R. Progress in the surgical treatment of pancreatic disease. Am J Surg 1973; 1: Traverso LW, Longmire WP Jr. Preservation of the pylorus in pancreaticoduodenectomy. Surg Gynecol Obstef 1978; 146: KW. Complications of pancreatic surgery. Surg Ciin North Am 1957; 6: KW, Choe DS, Plaza J, Relihan M. Results of radical resection for periampullary cancer. Ann Surg 1975; 181: KW, Veidenheimer MC, Pratt HS. Pancreaticoduodenectomy for periampullary cancer. Surg Clin North Am 1%7; 47: Whipple AO, Parsons WB, Mullens CR. Treatment of carcinoma of ampulla of Vater. Ann Surg 1935; 102: