| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Original Article |
Department of Surgery, Duke University Medical Center, Box 3118, Durham, North Carolina 27710
Correspondence: Address correspondence and reprint requests to: Douglas S. Tyler, MD; E-mail: tyler002{at}acpub.duke.edu
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
Methods: A retrospective analysis was conducted of 79 patients with periampullary malignancies who received 5-fluorouracil–based neo-CRT followed by PD.
Results: There was no difference in mortality between PD after neo-CRT (3.8%) and conventional PD for either malignant (4.5%) or benign (2.2%) disease. Focusing only on patients with malignancy, the neo-CRT group had a significantly lower pancreatic leak rate than the conventional group (10% vs. 43%; P < .001). Intra-abdominal abscesses were less common in the neo-CRT group (8.8% vs. 21%; P = .019), and there was one (1.2%) amylase-rich abscess in neo-CRT group, compared with eight (12%) in the conventional group. In addition, two patients in the conventional group died of leak-associated sepsis, compared with none in the neo-CRT group. Multivariate analysis revealed that neoadjuvant chemoradiation (odds ratio, .15) was the most significant factor associated with a reduced risk of pancreatic leak.
Conclusions: Neo-CRT does not increase the mortality or morbidity of PD. In contrast, neo-CRT was associated with a marked reduction in the incidence of pancreatic leak, as well as leak-associated morbidity and mortality.
Key Words: Pancreaticoduodenectomy • Neoadjuvant therapy • Morbidity • Mortality • Pancreatic leak
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
Although the operative morbidity and mortality from a PD continue to decline, it remains a formidable challenge for patients to recover without incident. Advancements in surgeon experience, technique, technology, and perioperative care have decreased the mortality rate to <5% in most centers.7–9 The morbidity associated with PD is reported with wide variability in the literature, and scant information exists regarding PD morbidity in the setting of neoadjuvant therapy. The objective of this study was to determine the effect of neo-CRT on the operative morbidity and mortality of PD and to compare it with the conventional PD (without neo-adjuvant therapy) approach.
|
METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
The treatment protocol for neo-CRT for periampullary cancer in this institution has been previously described by White et al.2 Briefly, the chemotherapy was 5-fluorouracil based. External beam radiotherapy was delivered at an initial dose of 4500 cGy in 180-cGy fractions, with a boost dose of 540 cGy to the tumor bed. After neo-CRT, patients were given a 3- to 4-week convalescence for recovery of blood counts and nutrition. Patients then had a restaging computed tomographic scan and were reclassified with respect to resectability. Resectability was determined by cross-sectional imaging criteria: (1) there was no extrapancreatic disease, (2) there was no direct invasion of the superior mesenteric artery or celiac artery, and (3) the portal vein and superior mesenteric vein were patent. All patients had a cytological or pathologic diagnosis before treatment. PD was then performed on the candidates who met radiologically resectable criteria. Staging laparoscopy was performed to rule out peritoneal seeding or liver metastasis. The standard operative procedure included resection of pancreatic head, duodenum, and distal stomach. The pancreatic neck and bile duct margin were evaluated by frozen section. The retro-peritoneal margin was marked by a stitch for further pathologic examination. Pancreatic reconstruction was then performed to the proximal jejunum with either a duct-to-mucosa or invagination method according to surgeon preference. Biliary and gastric anastomoses were then performed. Feeding jejunostomy was also placed at the initial operation.
Definitions
Surgical Complications
Delayed gastric emptying: radiological or scinti-graphic evidence of delayed gastric emptying or postoperative nasogastric tube decompression >2 weeks.
Pancreatic leak: amylase level of any drainage fluid that was three times the value of serum amylase. Amylase-rich intra-abdominal abscess was classified in both the pancreatic leak and intra-abdominal abscess categories.
Intra-abdominal abscess: culture-positive purulent drainage obtained from percutaneous drainage or reoperative intervention. Amylase-rich abscess was classified in both the pancreatic leak and intra-abdominal abscess categories.
Sterile fluid collection: culture-negative intra-abdominal fluid collection.
Bile leaks: bilious drainage from intraoperatively placed drains or radiographically proven fluid collection requiring percutaneous drainage and demonstrating increased bilirubin levels.
Cholangitis: fever, jaundice, leukocytosis, and culture-positive bile.
Gastrojejunostomy leak: radiographic evidence or operative finding of gastrojejunal anastomotic leak.
Intestinal obstruction: clinical and/or radiographic evidence of mechanical bowel obstruction.
Gastrointestinal bleeding: any Hemoccult (Beckman-Coulter, Fullerton, CA)–positive hematemesis, hematochezia, or melena requiring blood product transfusion or reoperation.
Nongastrointestinal bleeding: intra-abdominal bleeding from the operative field other than that characterized as gastrointestinal.
Wound disruption: nonhealing of fascial layer of wound with or without surgical repair.
Wound infection or delayed healing: discharge from wound >7 days after operation with or without evidence of infection, excluding wound disruption.
Enterocutaneous fistula: a persistent sinus with radiographic evidence of continuity with the gastrointestinal tract.
Chyle leak: lymphatic drainage with increased triglyceride levels and response to treatment with medium-chain triglycerides or total parenteral nutrition/fasting.
Nonsurgical Complications
Clostridium difficile colitis: diarrhea with positive clostridium toxin A requiring antimicrobial therapy.
Urinary tract infection: culture-positive urine with urinalysis-proven pyuria.
Deep venous thrombosis: demonstrated by venous duplex ultrasonography.
Myocardial infarction: increase of serum creatinine kinase and/or troponin levels with or without accompanying electrocardiographic changes.
Arrhythmia: characteristic electrocardiographic abnormality with or without symptoms requiring pharmacological or electrical intervention.
Pneumonia: fever, leukocytosis, culture-positive sputum with leukocytes on Gram stain, and chest radiograph demonstrating focal infiltrates.
Mortality: death during the initial operative hospitalization or within 30 days of discharge.
Reoperation: any subsequent surgical procedure performed for the management of complications from PD.
Readmission: subsequent hospital admission associated with PD within 3 months of initial discharge.
Morbidity: any surgical or nonsurgical complication necessitating a prolonged stay, further diagnostic testing, or therapeutic intervention. Each individual could have more than one complication, but when the morbidity rate was calculated, each patient was taken as with or without complication.
Statistical Analysis
The rates of mortality and each item of morbidity in malignant disease were compared by using the 
2 test with or without Fisher’s exact test. Univariate analysis of factors such as age, sex, diabetes mellitus, coronary artery disease, preoperative plasma total bilirubin level, operative blood loss, tumor location, method of pancreaticojejunostomy, and neo-CRT for operative morbidity and pancreatic leak were assessed with the 2 test with or without Fisher’s exact test for categorical variables and Student’s t-test for continuous variables. Multivariate analysis of these factors was performed with logistic regression software (SAS, version 8.2; SAS Institute, Cary, NC).
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
. The sex, age, underlying comorbidities, operative blood loss, preoperative biliary status, and technique of pancreaticojejunostomy are listed in Table 2. There was not a significant difference in median age, sex, technique of pancreaticojejunostomy, or median blood loss between the neo-CRT and conventional groups. Patients with benign disease were younger, with a median age of 53 years. The number of people with diabetes mellitus was significantly higher in the neo-CRT group when compared with the conventional group (18 vs. 6, respectively; P < .05). A significant difference also existed in the preoperative total plasma bilirubin level (.6 mg/dL in the neoadjuvant group and 1.1 mg/dL in the conventional group; P < .001). This difference correlated with the significantly greater rate of pre-operative biliary drainage in the neo-CRT group in comparison to the conventional group (91% vs. 76%, respectively; P = .01). The number of PDs performed each year stratified by benign disease and treatment group for malignant disease is shown in Fig. 1.
|
|
|
. There was no significant difference in these five variables when the neo-CRT group was compared with the conventional group. Although there was no significant difference in operative mortality between the neo-CRT (3.8%) and conventional (4.5%) groups, the cause of death was different. The three deaths in the neoadjuvant group were all due to nonsurgical complications, such as heart disease or pneumonia. However, two out of three deaths in the conventional group were sequelae of pancreatic leak and sepsis.
|
). The neo-CRT group had a lower overall rate of surgical complications in comparison to the conventional group. However, this was not statistically significant (65.8% vs. 77.6%, respectively; P = .11). In comparison with the conventional group, the neo-CRT group did have a significantly lower rate of pancreatic leak (10.1% vs. 43.3%; P < .001) and intra-abdominal abscess (8.8% vs. 20.9%; P < .03).
|
). As a whole, operative morbidity (surgical and nonsurgical) was also not significantly different between the two groups (74.4% vs. 79.1%; P = .53).
|
When tumor location was added into the univariate analysis of pancreatic leak, origination of the tumor at the pancreatic head correlated with a significantly higher leak rate in the conventional group compared with the neo-CRT group (24.1% vs. 4.8%, respectively; P = .006). Origination of the tumor at the ampulla of Vater did not have a significantly demonstrable difference between groups. Within the neo-CRT group, there was a significantly higher rate of leak when the tumor was located in the ampulla in comparison to the pancreatic head (36.4% vs. 4.8%, respectively; P = .001). This also bears out within the conventional group (ampulla, 52.6%; pancreatic head, 24.1%; P = .04; Table 6).
|
).
|
Patients with a pancreatic leak were graded on a clinical severity scale into four grades: grade A, spontaneous healing of amylase-rich fluid with drains inserted during initial PD operation without a need for further procedures; grade B, amylase-rich intra-abdominal abscess or inadequate drainage of a previously placed surgical drain necessitating radiological-guided drainage; grade C, bleeding complication, enterocutaneous fistula, or persistent fistula necessitating operative intervention; and grade D, leak or fistula directly contributing to mortality. Grades C and D were classified as major leaks. Under these definitions, comparison of the neoadjuvant group with the conventional group is shown in Fig. 2. The severity of pancreatic leak was less in the neo-adjuvant group.
|
. It was also observed that a greater proportion of abscesses in the conventional group were amylase rich, whereas most abscesses in the neo-CRT group were amylase poor. It should be noted that amylase data for three patients in the conventional were not available at the time of the study. In addition, coexisting pancreatic leaks were more frequent in the conventional group.
|
.
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
The overall morbidity from a PD remains high, approaching 40% to 50%. Leak from the pancreatic anastomosis continues to be a major cause of morbidity from PD and is second only to delayed gastric emptying in terms of causes of morbidity. A variety of techniques for pancreatic anastomosis have been described and debated in the literature in an effort to curb the leak rate.6,10–15 These include duct-to-mucosa versus invagination pancreaticojejunostomy, pancreaticogastrostomy, and the utility of stenting and Octreotide (Novartis, East Hanover, NJ). It has been reported in previous studies that the technique of duct-to-mucosa anastomosis is superior to the invagination method, and this is becoming the preferred method for many pancreatic surgeons.10 Nevertheless, pancreatic leak/fistula rates remain highly variable (1.6%–25%). This wide variability may be the direct result of the lack of a uniformly accepted definition of a pancreatic leak or fistula. Definitions include drainage of fluid with an amylase concentration greater than three times normal, radiographic demonstration, and >50 mL of amylase-rich fluid per day through the surgically placed drains on or after postoperative day 10.9,14–25 There is also a subtle difference between a leak and a fistula that in most studies is overlooked, and the terms are used interchangeably. The lack of standardized definitions for leaks and fistulas makes comparisons within the literature challenging. Table 10 lists some of the more common definitions of pancreatic leak and fistula used in reported series and the respective leak/fistula rate reported. In this series, a liberal definition of pancreatic leak was chosen to allow for a more thorough analysis between the neo-CRT PD and conventional PD groups. In our retrospective design, we defined pancreatic leak as an amylase level of any drainage or aspirated fluid that was three times the serum level. Use of a more inclusive rather than an exclusive or more clinically relevant definition permitted for scrutiny of our data and highlighted areas in which further investigations may be warranted.
|
Ampullary cancer has been shown to be associated with a higher pancreatic fistula rate than pancreatic head cancer.7,32,33 In one series that defined pancreatic leak as drainage of >30 mL of amylase-rich fluid from intraoperatively placed closed-suction drains after the fifth postoperative day, a pancreatic fistula rate of 21.8% was observed for ampullary cancer and 6% for pancreatic head cancer.32 Another series reported similar results, with a pancreatic fistula rate of 25% for ampullary cancer and 14% for PD overall.33 Although the definition of pancreatic leak was different in this study, the results are comparable.
Another observation from this analysis was that the duct-to-mucosa pancreaticojejunostomy had a decreased rate of pancreatic leak in both the conventional and neo-CRT groups, although it was not statistically significant in the neo-CRT group. Collectively, the duct-to-mucosa leak rate was 17%, and the invagination leak rate was 35.9%. It is interesting to note that neo-CRT decreased the pancreatic leak rate for the invagination method from 55.8% to 13.3%. This suggests that neo-CRT may provide some protection from pancreatic leak that is independent of the technique of anastomosis. This may indicate a role for neo-CRT in patients who are at high risk for a leak. For example, this study would imply that patients with a malignancy of the ampulla of Vater may benefit from neo-CRT and a duct-to-mucosa anastomosis.
Univariate analysis and logistic regression revealed that neo-CRT was the most important protective factor from pancreatic leak. Experience with neo-CRT in rectal cancer shows an increase in operative time, blood loss, and pelvic abscess but no increase in the anastomotic leak rate.34 Ishikawa et al.5 were the first to suggest that preoperative radiotherapy may prevent pancreatic fistula formation after PD. In that series, 22 of 76 patients received 50 Gy of external-beam irradiation before surgery, and only 1 (5%) of these patients developed a pancreatic fistula. In comparison, 10 (19%) of the 54 patients who did not receive preoperative radiation developed a pancreatic fistula. Another study compared 46 (42%) of 110 patients who received preoperative 5-fluorouracil–based chemoradiation with 64 patients who did not have preoperative chemoradiation; this study found pancreatic leak rates of 7% and 11%, respectively.6 Hence, there is no consensus as to the beneficial affect of neo-CRT on pancreatic leak or fistula formation.
The pancreatic leak–associated mortality was also less in the neo-CRT group. No deaths were related to the pancreatic leak in the neo-CRT group, whereas two of three deaths in the conventional group were directly related to leak-associated sepsis. Although the sample size is too small to draw any conclusions from, it does follow a trend that neo-CRT offers some protection from the severity of the leak. As described previously, the severity of the pancreatic leak was graded. In the neo-CRT group, all were classified as minor, either grade A or B. By contrast, the conventional group had 18 minor (grade A or B) and 7 major (grade C or D) leaks. Thus, not only did the conventional group have a higher pancreatic leak rate, but also the severity of the leak was considerably greater. This also demonstrates a protective role from pancreatic leak afforded by neo-CRT.
In conclusion, neo-CRT does not increase the surgical mortality or morbidity of PD. It does not alter the postoperative length of stay or readmission rate in comparison to conventional PD without prior chemoradiation. However, the data do show that neo-CRT was associated with a marked reduction in the incidence of pancreatic leak, as well as leak-associated morbidity. Furthermore, the increased risk of pancreatic leak seen with an ampulla of Vater malignancy and the use of an invagination pancreaticojejunostomy was abrogated by the use of neo-CRT. These data highlight the need for a prospective, randomized trial that would evaluate the affect of neo-CRT on the morbidity and mortality of a PD. In the interim, this study demonstrates that neoadjuvant therapy may offer some benefit in decreasing the incidence of complications associated with PD, namely, pancreatic leak and its concomitant morbidities.
Received for publication February 7, 2005. Accepted for publication August 1, 2005.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
This article has been cited by other articles:
 |
|
 |
|
  A. S. Caudle, J. E. Tepper, B. F. Calvo, M. O. Meyers, L. K. Goyal, W. G. Cance, and H. J. Kim Complications Associated with Neoadjuvant Radiotherapy in the Multidisciplinary Treatment of Retroperitoneal Sarcomas Ann. Surg. Oncol., February 1, 2007; 14(2): 577 - 582. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
