- Email: [email protected]
Determinants of complications in pancreaticoduodenectomy
EJSO 33 (2007) 488e492
www.ejso.com
Determinants of complications in pancreaticoduodenectomy B. Topal a,*, R. Aerts a, T. Hendrickx a, S. Fieuws b, F. Penninckx a a
Department of Abdominal Surgery, University Hospital Gasthuisberg, Herestraat 49, 3000 Leuven, Belgium b Department of Biostatistics, University Hospital Gasthuisberg, Herestraat 49, 3000 Leuven, Belgium Accepted 27 October 2006 Available online 4 December 2006
Abstract Aims: The factors determining complications after pancreaticoduodenectomy (PD) have not yet been identified clearly. This retrospective study examined, using reproducible classification systems, the type and severity of complications as well as the factors to predict them. Methods: Between 1998 and 2005 PD was performed in 351 consecutive patients with peri-ampullary tumours. Logistic regression models were used in univariate as well as in corrected, multivariate analyses in order to identify the optimally combined factors related to the occurrence of post-operative complications. Results: Post-operative complication rate was 50.7%, mortality 3.1% and re-operation rate 7.1%. Pancreatic fistula (12%) was responsible for higher mortality (9.5%; p ¼ 0.011) and re-operation (30.9%; p < 0.001) rates. Hospital length of stay (LOS) was ( p < 0.001) longer for patients with post-operative complications (median 21.5 (range 1e128) vs. 14 (7e42) days) or pancreatic fistula (28.5 (8e128) vs. 17 (1e63) days), and related to the severity of complications. Surgeon (Odds ratio [OR] 2.03; confidence interval [CI] 1.20e3.41; p ¼ 0.008), male gender (OR 1.72; CI 1.05e2.81; p ¼ 0.032), and pre-operative hyperbilirubinaemia (OR 1.04; CI 1.001e1.08; p ¼ 0.046) were independent risk factors for post-operative complications. Neither prophylactic octreotide nor pre-operative biliary drainage improved post-operative outcome. Conclusion: Surgeon, male gender, and pre-operative hyperbilirubinaemia determine complication rate following PD. Pancreatic fistula is the most common complication and is associated with increased mortality, re-operation rate and LOS. Ó 2006 Elsevier Ltd. All rights reserved. Keywords: Malignancy; Pancreas; Biliary; Surgery; Complication
Introduction The mortality rate of pancreaticoduodenectomy (PD) has decreased to below 4% in high-volume centres during the last two decades. The complication rate, however, remains high, between 30% and 60%.1e4 One of the most common complications is post-operative pancreatic fistula (POPF) with a reported rate that is highly variable and dependent on its definition, ranging from 2% to more than 20%.5e9 Although POPF is responsible for about one third of all complications, few data are available on other complications. The implementation of reproducible classification systems can provide conclusive assessment and comparison of surgical outcomes among different centres and therapies.10,11
* Corresponding author. Tel.: þ32 16 344265; fax: þ32 16 344832. E-mail address: [email protected] (B. Topal). 0748-7983/$ - see front matter Ó 2006 Elsevier Ltd. All rights reserved. doi:10.1016/j.ejso.2006.10.041
Indeed, there is lack of consensus on the definition and stratification of complications after PD. The factors determining complications after PD have not yet been identified clearly. The aim of the present retrospective study was to evaluate, using reproducible classification systems, the type and severity of complications as well as the variables that can predict them. Patients and methods Patients Between January 1998 and December 2005, 351 consecutive patients (F/M: 158/193; median (range) age 64.4 (31.9e85.3) years) underwent curative pancreaticoduodenectomy for a peri-ampullary tumour. Total pancreatectomy and surgery for chronic pancreatitis were excluded. According to the American Society of Anaesthesiology (ASA) physical status score, 79 patients had an ASA score
B. Topal et al. / EJSO 33 (2007) 488e492
I, 213 ASA score II, 55 ASA III, and 4 ASA IV. On histopathological examination of the resection specimen, the final diagnosis was adenocarcinoma of the pancreas in 141, ampulla in 98, distal bile duct in 34, duodenal cancer in 17, and miscellaneous tumours in 61 patients. Pylorus-preserving PD (PPPD) was performed in 194 patients as compared to classic PD (Whipple’s procedure) in 157. The vast majority of the patients were treated by two surgeons (surgeon 1: 128 PD and 102 PPPD; surgeon 2: 25 PD and 90 PPPD), whereas other surgeons performed 4 PDs and 2 PPPDs. Chemotherapy was used prior to surgery in 10 and radiotherapy in 4 patients. Pre-operative biliary drainage (endobiliary plastic stent 150, external trans-hepatic percutaneous catheter 12, and metallic wall stent 1) was accomplished in 163 patients. Pre-operative serum total bilirubin level after biliary drainage was 24.4 (median; range 2.6e394.3) mmol/L as compared to 13.2 (1.7e665.4) mmol/L in patients (n ¼ 188) without preoperative drainage ( p ¼ 0.06). The serum total bilirubin level in the entire patient population was 17.8 (1.7e665.4) mmol/L. Surgical procedure The surgical technique applied by the two main surgeons was similar. Pancreaticoduodenectomy was performed with various extents of lymph node dissection according to the primary tumour type and cancer stage. Other organs were resected in 34 patients. Portal vein resection was performed in 28 pancreatic cancers, 1 ampullary, 1 bile duct, and 1 other tumour. Absorbable monofilament sutures were used for the reconstruction that first involved a trans-mesocolic end-to-side pancreatico-jejunostomy in two layers, followed by a one-layer end-to-side bilio-enteric anastomosis. A stent catheter was placed through the bilio-enteric anastomosis and exteriorized via the jejunum in 107 patients, a procedure that was mainly done in the early years of the present study. The pancreatic reconstruction was never stented. Finally, the gastric (in PD) or duodenal (in PPPD) reconstruction was achieved with the standard two-layer end-to-side anastomosis. A nasogastric tube was used in patients who underwent a classic PD, while a gastro-cutaneous catheter was placed in PPPD. Two closed suction drains were placed in the vicinity of the biliary and pancreatic anastomoses. Post-operative management All patients were monitored post-operatively in the postanaesthesia care unit during one night and subsequently transferred to the regular care floor. Prophylactic intravenous antibiotics (cefazoline) were started intra-operatively and continued every 8 h for one day. A histamine H2-receptor blocker and a low-molecular-weight heparin were given during the post-operative hospital stay. Intravenous hyperalimentation was not routinely used. Prophylactic octreotide
489
was started intra-operatively and given for 5 days in 252 patients. Intra-peritoneal drains were removed, usually after post-operative day 5, depending on the drainage output and the fluid amylase level. The biliary stent catheter was removed if the x-ray cholangiography on day 7 did not show any leakage. Patients were evaluated on the outpatient clinic within 6 weeks after discharge. Assessment and statistics Patient and operative data were collected retrospectively. Post-operative complications were classified based on the therapy-oriented severity grading system (TOSGS; grade 1: no need for specific intervention; grade 2: need for drug therapy; grade 3: need for invasive therapy; grade 4: organ dysfunction with ICU stay; grade 5: death),11 and allocated to surgical site (SSC) vs. non-surgical site complications (NSSC). Post-operative pancreatic fistula was defined according to the International Study Group on Pancreatic Fistula (ISGPF) definition i.e. grade A: biochemical fistula without clinical sequelae; grade B: fistula requiring any therapeutic intervention; grade C: fistula with severe clinical sequelae.9 Logistic regression models were used in univariate as well as in corrected, multivariate analyses in order to identify the optimally combined factors related to the presence of postoperative complications. In the multivariate setting various model-building strategies (forward, backward and stepwise selection) were applied to verify the robustness of the obtained conclusions. For each analysis the following potential predictive variables were taken into account: year of surgery, age, gender, ASA score, pre-operative biliary drainage (yes/no), pre-operative serum total bilirubin level (normal <17.1 mmol/L), primary tumour location, tumour diameter, surgeon (surgeon 1/surgeon 2), type of surgery (PD/PPPD), resection of other organs (yes/no), portal vein resection (yes/no), stent catheterisation of the bilio-enteric anastomosis (yes/no), amount of intra-operative blood loss, duration of surgery, intra-operative complication (yes/no), and prophylactic use of octreotide (yes/no). Among these variables significant differences ( p < 0.001) were observed between surgeon 1 vs. surgeon 2 with regard to the type of surgery (classic PD 128 vs. 25; PPPD 102 vs. 90), stent catheterisation of the bilio-enteric anastomosis (102 vs. 1), and prophylactic use of octreotide (131 vs. 115). Analysis of the statistical significance of differences between groups of data was performed using Pearson Chi-square and Wilcoxon/KruskaleWallis test, as appropriate. A p value of 0.05 was considered statistically significant. All analyses were performed using the statistical software SAS (version 9.1). Results The duration of surgery was 250 min (median; range 135e480), with an intra-operative blood loss of 900
B. Topal et al. / EJSO 33 (2007) 488e492
490
(100e5000) ml. Intra-operative complications occurred in 37 patients, i.e. haemorrhage 22 and vascular injury 16. Post-operative complications were observed in 178 patients, with subsequent re-operation in 25 and hospital mortality in 11 patients.
Ampullary cancer was associated with the highest number of grade C pancreatic fistula. No significant differences were found between the severity scores (TOSGS) related to the primary tumour location ( p > 0.12). Predictors of post-operative complications
Type of post-operative complications The cause of death was haemorrhage with POPF in 3, haemorrhage without POPF in 2, septicaemia with POPF in 1, cardiovascular collapse in 4, and pulmonary failure in 1 patient. The mortality rate was significantly ( p ¼ 0.01) higher in patients with POPF (4/42) as compared to those without POPF (7/309). Re-operation was performed because of haemorrhage in 16, intra-peritoneal abscess in 4, peritonitis in 4, and abdominal wall evisceration in 1 patient. Re-operation rate was significantly ( p < 0.001) higher in patients with POPF (13/42) as compared to those without POPF (12/309). The overall median length of hospital stay (LOS) was 17 (1e128) days. The LOS was significantly different ( p < 0.001) between patients with vs. without post-operative complications (21.5 (1e128) vs. 14 (7e42) days), and between patients with vs. without POPF (28.5 (8e128) vs. 17 (1e63) days). Longer LOS was observed with increasing TOSGS-scores ( p < 0.001) i.e. grade 1: 15 (11e25) vs. grade 4: 43 (18e128) days.
In univariate analysis the following factors were significantly related to the presence of post-operative complications: surgeon ( p ¼ 0.005), type of surgery: classic PD vs. PPPD ( p ¼ 0.008), age ( p ¼ 0.029), gender ( p ¼ 0.030), stent catheterisation of the bilio-enteric anastomosis ( p ¼ 0.043), and primary tumour location ( p ¼ 0.044). Surgeon 1 vs. surgeon 2-related morbidity was different in terms of overall post-operative complication rate (130/230 vs. 44/ 115; p ¼ 0.001), and determined by NSSC (53 vs. 12; p ¼ 0.005) and grade A pancreatic fistula (14 vs. 0; p ¼ 0.007). No significantly different mortality, re-operation, or overall pancreatic fistula rates were observed between surgeon 1 vs. surgeon 2. Multivariate analysis identified surgeon 1, male sex, and elevated serum total bilirubin level as significant predictors of post-operative complications (Table 2). Additionally, a set of bivariate logistic regression models was performed to explore in detail whether the effect of the ‘surgeon factor’ could be attributed to other variables. In each logistic regression model containing two predictors (surgeon and any other variable) the effect of surgeon remained significant.
Site of post-operative complications Prophylactic octreotide Post-operative SSC were encountered in 122 patients, i.e. pancreatic fistula 42, haemorrhage 32, intra-peritoneal abscess 25, wound infection 18, biliary fistula 15, lymphatic leakage 6, bowel fistula 2, and evisceration 1. Post-operative NSSC occurred in 56 patients, i.e. pulmonary infection 23, central venous catheter sepsis 15, cardiovascular 13, pulmonary failure 6, endocrine complication 4, urinary infection 3, and miscellaneous 3. Both the incidence and severity of POPF were similar after classic PD (19/157) and PPPD (23/194). The relation between the primary tumour location and the type and severity of post-operative complications is illustrated in Table 1.
Surgical outcome was comparable in patients treated with (n ¼ 252) or without (n ¼ 99) prophylactic octreotide i.e. post-operative complication (121 vs. 57; p ¼ 0.11), POPF (34 vs. 8; p ¼ 0.16), SSC (88 vs. 34; p ¼ 0.92), mortality (8 vs. 3; p ¼ 0.94), re-operation (21 vs. 4; p ¼ 0.16) rate and LOS (17 (2e128) vs. 18 (1e63) days; p ¼ 0.59). Pre-operative biliary drainage Pre-operative biliary drainage was considered adequate in case a normal serum total bilirubin level was reached
Table 1 Primary tumour location and type of post-operative complications following pancreaticoduodenectomy in 351 patients Ampulla (n ¼ 98)
Bile duct (n ¼ 34)
Pancreas (n ¼ 141)
Duodenum (n ¼ 17)
Miscellaneous (n ¼ 61)
p value
Site of complication Overall SSC NSSC
42 35 10
21 16 5
78 46 37
11 8 4
26 17 11
0.08 0.29 0.035
POPF Overall Grade A Grade B Grade C
17 5 3 9
6 0 4 2
11 6 3 2
2 1 1 0
6 2 2 2
0.18 0.74 0.11 0.046
SSC, surgical site complication; NSSC, non-surgical site complication; POPF, post-operative pancreatic fistula.
B. Topal et al. / EJSO 33 (2007) 488e492 Table 2 Multivariate analysis of factors related to post-operative complications
Surgeon Gender (male) Bilirubin
p value
Odds ratio
95% CI
0.008 0.032 0.046
2.028 1.716 1.041
1.205e3.413 1.047e2.813 1.001e1.082
CI, confidence interval.
(n ¼ 66), and inadequate in case the bilirubin level remained above 17.1 mmol/L (n ¼ 97). In patients with inadequate drainage the pre-operative serum bilirubin level was significantly ( p < 0.001) lower as compared to jaundiced patients without drainage (n ¼ 80), i.e. 48.7 (17.6e394.3) mmol/L vs. 140.3 (17.8e665.4) mmol/L. Surgical outcome of all patients with pre-operative biliary drainage (n ¼ 163) was comparable with that of jaundiced patients without drainage i.e. post-operative complication 86 vs. 40 ( p ¼ 0.69), POPF 21 vs. 9 ( p ¼ 0.72), and LOS 17 (1e128) vs. 16 (6e52) days ( p ¼ 0.26). Also the outcome between patients with adequate drainage vs. non-jaundiced patients without drainage, and between patients with inadequate drainage vs. jaundiced patients without drainage was similar ( p > 0.17). Discussion The overall post-operative complication rate was 50.7% with pancreatic fistula (12%) as the most common adverse event in the present study. Mortality was 3.1% and the re-operation rate 7.1%. Although few studies report a zero to 1% mortality rate, these figures are comparable with most published data, wherein the importance of institutional expertise and surgical volume has been highlighted.1e4,12e15 Type and severity of post-operative complications The use of the TOSGS classification system showed that the vast majority of the patients in the current study needed drug therapy and/or invasive intervention. In addition to this classification system, post-operative complications were allocated to either SSC or NSSC that were observed in 34.7% and 15.9% of the patients, respectively. According to the ISGPF definition, the severity of pancreatic fistula was equally distributed. Pancreaticoduodenectomy for ampullary cancer was associated with the highest POPF grade C as compared to other peri-ampullary tumours. Although pancreatic texture is hard to describe objectively, these findings might suggest soft pancreatic texture (generally found in patients with non-pancreatic cancer) to be more at risk for POPF. In the present study delayed gastric emptying was not considered as a complication per se, since the authors believe delayed gastric emptying is hard to define and well reflected by the length of post-operative hospital stay. Indeed, the LOS was
491
significantly different between patients with vs. without post-operative complications. Comparable post-operative mortality has been reported in patients with and those without pancreatic fistula, although the mean LOS was found to be longer and the rates of certain complications to be higher in patients with POPF.5 Indeed, the final post-operative outcome following PD is also strongly dependent on the performance or quality of care in non-surgical departments within the same institution, such as the intensive care unit, invasive radiology, as well as on the expertise of the medical and nursing personnel on the wards. In the present study, however, postoperative mortality and re-operation rates were significantly higher in patients with pancreatic fistula. Also, the LOS was significantly longer for patients with post-operative complications or pancreatic fistula, and clearly related to the severity of complications as scored according to the TOSGS. Predictors of post-operative complications Age, extended resections, and soft pancreatic texture have been reported as the most prominent independent risk factors for mortality and post-operative complications following PD in high-volume institutions. Other potential risk factors are impaired renal function, coronary heart disease, main pancreatic duct diameter less than 3 mm, need for blood transfusion, resection of other organs, and absence of preoperative biliary drainage.5,16e18 In the present study surgeon, male patient and pre-operative serum total bilirubin level were found as independent risk factors for post-operative complications. Neither resection of other organs nor portal vein reconstruction at the time of PD influenced post-operative morbidity or mortality. As reported by other high-volume institutions, the use of prophylactic octreotide did not improve surgical outcome in the current study.19,20 Biliary drainage prior to PD remains controversial with conflicting effects on peri-operative morbidity and mortality.21e23 In the present study, pre-operative serum bilirubin level was significantly lower after biliary drainage as compared to jaundiced patients without drainage. Nevertheless, despite the fact that an elevated serum bilirubin level appeared to be an independent predictor of post-operative complications, pre-operative biliary drainage did not improve the outcome following PD. Therefore, these data argue against pre-operative biliary drainage in jaundiced patients with a peri-ampullary tumour. However, the limitation of the current study is its non-randomized nature. Conclusion Complication rate following pancreaticoduodenectomy is high and determined by surgeon, patient male gender, and pre-operative serum bilirubin level. The vast majority of patients with post-operative complications need drug
492
B. Topal et al. / EJSO 33 (2007) 488e492
or invasive therapy with subsequent increase in length of hospital stay. Because of the diversity of complications following pancreaticoduodenectomy, the implementation of reproducible classification systems is advocated in order to better assess and compare surgical outcome between surgeons and between centres. Patients who develop a pancreatic fistula have a significantly higher mortality and re-operation rate indicating that prevention and adequate treatment of post-operative pancreatic fistula should remain high priority and need further study. Acknowledgements Many thanks to the staff members of the Department of Hepatobiliary and Pancreatic Diseases for including patients in this study, and to the radiologists* for meticulously interpreting the images and their therapeutic interventions: W. Vansteenbergen , C. Verslype , D. Bielen* and D. Vanbeckevoort*. References 1. Birkmeyer LD, Stukel TA, Siewers AE, Goodney PP, Wennberg DE, Lucas FL. Surgeon volume and operative mortality in the United States. N Engl J Med 2003;349:2117–27. 2. Balcom JH, Rattner DW, Warshaw AL, Chang Y, Fernandez-del Castillo C. Ten-year experience with 733 pancreatic resections: changing indications, older patients, and decreasing length of hospitalization. Arch Surg 2001;136:391–8. 3. Yeo CJ, Cameron JL, Sohn TA, et al. Six hundred fifty consecutive pancreaticoduodenectomies in the 1990s: pathology, complications, and outcomes. Ann Surg 1997;226:248–57; discussion 257e60. 4. Suzuki Y, Fujino Y, Ajiki T, et al. No mortality among 100 consecutive pancreaticoduodenectomies in a middle-volume center. World J Surg 2005;29:1409–14. 5. Lin JW, Cameron JL, Yeo CJ, Riall TS, Lillemoe KD. Risk factors and outcomes in postpancreaticoduodenectomy pancreaticocutaneous fistula. J Gastrointest Surg 2004;8:951–9. 6. Popiela T, Kedra B, Sierzega M, Gurda A. Risk factors of pancreatic fistula following pancreaticoduodenectomy for periampullary cancer. Hepatogastroenterology 2004;51:1484–8. 7. Tien YW, Lee PH, Yang CY, Ho MC, Chiu YF. Risk factors of massive bleeding related to pancreatic leak after pancreaticoduodenectomy. J Am Coll Surg 2005;201:554–9.
8. Kazanjian KK, Hines OJ, Eibl G, Reber HA. Management of pancreatic fistulas after pancreaticoduodenectomy: results in 437 consecutive patients. Arch Surg 2005;140:849–54. discussion 854e6. 9. Bassi C, Dervenis C, Butturini G, et al. for the International Study Group on Pancreatic Fistula. Postoperative pancreatic fistula: an international study group (ISGPF) definition. Surgery 2005;138:8–13. 10. Clavien P, Sanabria J, Strasberg S. Proposed classification of complication of surgery with examples of utility in cholecystectomy. Surgery 1992;111:518–26. 11. Dindo D, Demartines N, Clavien PA. Classification of surgical complications: a new proposal with evaluation in a cohort of 6336 patients and results of a survey. Ann Surg 2004;240:214–5. 12. Cameron JL, Riall TS, Coleman J, Belcher KA. One thousand consecutive pancreaticoduodenectomies. Ann Surg 2006;244:10–5. 13. Makary MA, Winter JM, Cameron JL, et al. Pancreaticoduodenectomy in the very elderly. J Gastrointest Surg 2006;10:347–56. 14. de Castro SM, Busch OR, van Gulik TM, Obertop H, Gouma DJ. Incidence and management of pancreatic leakage after pancreatoduodenectomy. Br J Surg 2005;92:1117–23. 15. Aranha GV, Hodul PJ, Creech S, Jacobs W. Zero mortality after 152 consecutive pancreaticoduodenectomies with pancreaticogastrostomy. J Am Coll Surg 2003;197:223–31; discussion 231e2. 16. Gouma DJ, van Geenen RC, van Gulik TM, et al. Rates of complications and death after pancreaticoduodenectomy: risk factors and the impact of hospital volume. Ann Surg 2000;232:786–95. 17. Muscari F, Suc B, Kirzin S, et al. French Associations for Surgical Research. Risk factors for mortality and intra-abdominal complications after pancreatoduodenectomy: multivariate analysis in 300 patients. Surgery 2006;139:591–8. 18. Adam U, Makowiec F, Riediger H, Schareck WD, Benz S, Hopt UT. Risk factors for complications after pancreatic head resection. Am J Surg 2004;187:201–8. 19. Yeo CJ, Cameron JL, Lillemoe KD, et al. Does prophylactic octreotide decrease the rates of pancreatic fistula and other complications after pancreaticoduodenectomy? Results of a prospective randomized placebo-controlled trial. Ann Surg 2000;232:419–29. 20. Li-Ling J, Irving M. Somatostatin and octreotide in the prevention of postoperative pancreatic complications and the treatment of enterocutaneous pancreatic fistulas: a systematic review of randomized controlled trials. Br J Surg 2001;88:190–9. 21. Jagannath P, Dhir V, Shrikhande S, Shah RC, Mullerpatan P, Mohandas KM. Effect of preoperative biliary stenting on immediate outcome after pancreaticoduodenectomy. Br J Surg 2005;92:356–61. 22. Sewnath ME, Birjmohun RS, Rauws EA, Huibregtse K, Obertop H, Gouma DJ. The effect of preoperative biliary drainage on postoperative complications after pancreaticoduodenectomy. J Am Coll Surg 2001;192:726–34. 23. Sohn TA, Yeo CJ, Cameron JL, Pitt HA, Lillemoe KD. Do preoperative biliary stents increase postpancreaticoduodenectomy complications? J Gastrointest Surg 2000;4:258–67; discussion 267e8.
www.ejso.com
Determinants of complications in pancreaticoduodenectomy B. Topal a,*, R. Aerts a, T. Hendrickx a, S. Fieuws b, F. Penninckx a a
Department of Abdominal Surgery, University Hospital Gasthuisberg, Herestraat 49, 3000 Leuven, Belgium b Department of Biostatistics, University Hospital Gasthuisberg, Herestraat 49, 3000 Leuven, Belgium Accepted 27 October 2006 Available online 4 December 2006
Abstract Aims: The factors determining complications after pancreaticoduodenectomy (PD) have not yet been identified clearly. This retrospective study examined, using reproducible classification systems, the type and severity of complications as well as the factors to predict them. Methods: Between 1998 and 2005 PD was performed in 351 consecutive patients with peri-ampullary tumours. Logistic regression models were used in univariate as well as in corrected, multivariate analyses in order to identify the optimally combined factors related to the occurrence of post-operative complications. Results: Post-operative complication rate was 50.7%, mortality 3.1% and re-operation rate 7.1%. Pancreatic fistula (12%) was responsible for higher mortality (9.5%; p ¼ 0.011) and re-operation (30.9%; p < 0.001) rates. Hospital length of stay (LOS) was ( p < 0.001) longer for patients with post-operative complications (median 21.5 (range 1e128) vs. 14 (7e42) days) or pancreatic fistula (28.5 (8e128) vs. 17 (1e63) days), and related to the severity of complications. Surgeon (Odds ratio [OR] 2.03; confidence interval [CI] 1.20e3.41; p ¼ 0.008), male gender (OR 1.72; CI 1.05e2.81; p ¼ 0.032), and pre-operative hyperbilirubinaemia (OR 1.04; CI 1.001e1.08; p ¼ 0.046) were independent risk factors for post-operative complications. Neither prophylactic octreotide nor pre-operative biliary drainage improved post-operative outcome. Conclusion: Surgeon, male gender, and pre-operative hyperbilirubinaemia determine complication rate following PD. Pancreatic fistula is the most common complication and is associated with increased mortality, re-operation rate and LOS. Ó 2006 Elsevier Ltd. All rights reserved. Keywords: Malignancy; Pancreas; Biliary; Surgery; Complication
Introduction The mortality rate of pancreaticoduodenectomy (PD) has decreased to below 4% in high-volume centres during the last two decades. The complication rate, however, remains high, between 30% and 60%.1e4 One of the most common complications is post-operative pancreatic fistula (POPF) with a reported rate that is highly variable and dependent on its definition, ranging from 2% to more than 20%.5e9 Although POPF is responsible for about one third of all complications, few data are available on other complications. The implementation of reproducible classification systems can provide conclusive assessment and comparison of surgical outcomes among different centres and therapies.10,11
* Corresponding author. Tel.: þ32 16 344265; fax: þ32 16 344832. E-mail address: [email protected] (B. Topal). 0748-7983/$ - see front matter Ó 2006 Elsevier Ltd. All rights reserved. doi:10.1016/j.ejso.2006.10.041
Indeed, there is lack of consensus on the definition and stratification of complications after PD. The factors determining complications after PD have not yet been identified clearly. The aim of the present retrospective study was to evaluate, using reproducible classification systems, the type and severity of complications as well as the variables that can predict them. Patients and methods Patients Between January 1998 and December 2005, 351 consecutive patients (F/M: 158/193; median (range) age 64.4 (31.9e85.3) years) underwent curative pancreaticoduodenectomy for a peri-ampullary tumour. Total pancreatectomy and surgery for chronic pancreatitis were excluded. According to the American Society of Anaesthesiology (ASA) physical status score, 79 patients had an ASA score
B. Topal et al. / EJSO 33 (2007) 488e492
I, 213 ASA score II, 55 ASA III, and 4 ASA IV. On histopathological examination of the resection specimen, the final diagnosis was adenocarcinoma of the pancreas in 141, ampulla in 98, distal bile duct in 34, duodenal cancer in 17, and miscellaneous tumours in 61 patients. Pylorus-preserving PD (PPPD) was performed in 194 patients as compared to classic PD (Whipple’s procedure) in 157. The vast majority of the patients were treated by two surgeons (surgeon 1: 128 PD and 102 PPPD; surgeon 2: 25 PD and 90 PPPD), whereas other surgeons performed 4 PDs and 2 PPPDs. Chemotherapy was used prior to surgery in 10 and radiotherapy in 4 patients. Pre-operative biliary drainage (endobiliary plastic stent 150, external trans-hepatic percutaneous catheter 12, and metallic wall stent 1) was accomplished in 163 patients. Pre-operative serum total bilirubin level after biliary drainage was 24.4 (median; range 2.6e394.3) mmol/L as compared to 13.2 (1.7e665.4) mmol/L in patients (n ¼ 188) without preoperative drainage ( p ¼ 0.06). The serum total bilirubin level in the entire patient population was 17.8 (1.7e665.4) mmol/L. Surgical procedure The surgical technique applied by the two main surgeons was similar. Pancreaticoduodenectomy was performed with various extents of lymph node dissection according to the primary tumour type and cancer stage. Other organs were resected in 34 patients. Portal vein resection was performed in 28 pancreatic cancers, 1 ampullary, 1 bile duct, and 1 other tumour. Absorbable monofilament sutures were used for the reconstruction that first involved a trans-mesocolic end-to-side pancreatico-jejunostomy in two layers, followed by a one-layer end-to-side bilio-enteric anastomosis. A stent catheter was placed through the bilio-enteric anastomosis and exteriorized via the jejunum in 107 patients, a procedure that was mainly done in the early years of the present study. The pancreatic reconstruction was never stented. Finally, the gastric (in PD) or duodenal (in PPPD) reconstruction was achieved with the standard two-layer end-to-side anastomosis. A nasogastric tube was used in patients who underwent a classic PD, while a gastro-cutaneous catheter was placed in PPPD. Two closed suction drains were placed in the vicinity of the biliary and pancreatic anastomoses. Post-operative management All patients were monitored post-operatively in the postanaesthesia care unit during one night and subsequently transferred to the regular care floor. Prophylactic intravenous antibiotics (cefazoline) were started intra-operatively and continued every 8 h for one day. A histamine H2-receptor blocker and a low-molecular-weight heparin were given during the post-operative hospital stay. Intravenous hyperalimentation was not routinely used. Prophylactic octreotide
489
was started intra-operatively and given for 5 days in 252 patients. Intra-peritoneal drains were removed, usually after post-operative day 5, depending on the drainage output and the fluid amylase level. The biliary stent catheter was removed if the x-ray cholangiography on day 7 did not show any leakage. Patients were evaluated on the outpatient clinic within 6 weeks after discharge. Assessment and statistics Patient and operative data were collected retrospectively. Post-operative complications were classified based on the therapy-oriented severity grading system (TOSGS; grade 1: no need for specific intervention; grade 2: need for drug therapy; grade 3: need for invasive therapy; grade 4: organ dysfunction with ICU stay; grade 5: death),11 and allocated to surgical site (SSC) vs. non-surgical site complications (NSSC). Post-operative pancreatic fistula was defined according to the International Study Group on Pancreatic Fistula (ISGPF) definition i.e. grade A: biochemical fistula without clinical sequelae; grade B: fistula requiring any therapeutic intervention; grade C: fistula with severe clinical sequelae.9 Logistic regression models were used in univariate as well as in corrected, multivariate analyses in order to identify the optimally combined factors related to the presence of postoperative complications. In the multivariate setting various model-building strategies (forward, backward and stepwise selection) were applied to verify the robustness of the obtained conclusions. For each analysis the following potential predictive variables were taken into account: year of surgery, age, gender, ASA score, pre-operative biliary drainage (yes/no), pre-operative serum total bilirubin level (normal <17.1 mmol/L), primary tumour location, tumour diameter, surgeon (surgeon 1/surgeon 2), type of surgery (PD/PPPD), resection of other organs (yes/no), portal vein resection (yes/no), stent catheterisation of the bilio-enteric anastomosis (yes/no), amount of intra-operative blood loss, duration of surgery, intra-operative complication (yes/no), and prophylactic use of octreotide (yes/no). Among these variables significant differences ( p < 0.001) were observed between surgeon 1 vs. surgeon 2 with regard to the type of surgery (classic PD 128 vs. 25; PPPD 102 vs. 90), stent catheterisation of the bilio-enteric anastomosis (102 vs. 1), and prophylactic use of octreotide (131 vs. 115). Analysis of the statistical significance of differences between groups of data was performed using Pearson Chi-square and Wilcoxon/KruskaleWallis test, as appropriate. A p value of 0.05 was considered statistically significant. All analyses were performed using the statistical software SAS (version 9.1). Results The duration of surgery was 250 min (median; range 135e480), with an intra-operative blood loss of 900
B. Topal et al. / EJSO 33 (2007) 488e492
490
(100e5000) ml. Intra-operative complications occurred in 37 patients, i.e. haemorrhage 22 and vascular injury 16. Post-operative complications were observed in 178 patients, with subsequent re-operation in 25 and hospital mortality in 11 patients.
Ampullary cancer was associated with the highest number of grade C pancreatic fistula. No significant differences were found between the severity scores (TOSGS) related to the primary tumour location ( p > 0.12). Predictors of post-operative complications
Type of post-operative complications The cause of death was haemorrhage with POPF in 3, haemorrhage without POPF in 2, septicaemia with POPF in 1, cardiovascular collapse in 4, and pulmonary failure in 1 patient. The mortality rate was significantly ( p ¼ 0.01) higher in patients with POPF (4/42) as compared to those without POPF (7/309). Re-operation was performed because of haemorrhage in 16, intra-peritoneal abscess in 4, peritonitis in 4, and abdominal wall evisceration in 1 patient. Re-operation rate was significantly ( p < 0.001) higher in patients with POPF (13/42) as compared to those without POPF (12/309). The overall median length of hospital stay (LOS) was 17 (1e128) days. The LOS was significantly different ( p < 0.001) between patients with vs. without post-operative complications (21.5 (1e128) vs. 14 (7e42) days), and between patients with vs. without POPF (28.5 (8e128) vs. 17 (1e63) days). Longer LOS was observed with increasing TOSGS-scores ( p < 0.001) i.e. grade 1: 15 (11e25) vs. grade 4: 43 (18e128) days.
In univariate analysis the following factors were significantly related to the presence of post-operative complications: surgeon ( p ¼ 0.005), type of surgery: classic PD vs. PPPD ( p ¼ 0.008), age ( p ¼ 0.029), gender ( p ¼ 0.030), stent catheterisation of the bilio-enteric anastomosis ( p ¼ 0.043), and primary tumour location ( p ¼ 0.044). Surgeon 1 vs. surgeon 2-related morbidity was different in terms of overall post-operative complication rate (130/230 vs. 44/ 115; p ¼ 0.001), and determined by NSSC (53 vs. 12; p ¼ 0.005) and grade A pancreatic fistula (14 vs. 0; p ¼ 0.007). No significantly different mortality, re-operation, or overall pancreatic fistula rates were observed between surgeon 1 vs. surgeon 2. Multivariate analysis identified surgeon 1, male sex, and elevated serum total bilirubin level as significant predictors of post-operative complications (Table 2). Additionally, a set of bivariate logistic regression models was performed to explore in detail whether the effect of the ‘surgeon factor’ could be attributed to other variables. In each logistic regression model containing two predictors (surgeon and any other variable) the effect of surgeon remained significant.
Site of post-operative complications Prophylactic octreotide Post-operative SSC were encountered in 122 patients, i.e. pancreatic fistula 42, haemorrhage 32, intra-peritoneal abscess 25, wound infection 18, biliary fistula 15, lymphatic leakage 6, bowel fistula 2, and evisceration 1. Post-operative NSSC occurred in 56 patients, i.e. pulmonary infection 23, central venous catheter sepsis 15, cardiovascular 13, pulmonary failure 6, endocrine complication 4, urinary infection 3, and miscellaneous 3. Both the incidence and severity of POPF were similar after classic PD (19/157) and PPPD (23/194). The relation between the primary tumour location and the type and severity of post-operative complications is illustrated in Table 1.
Surgical outcome was comparable in patients treated with (n ¼ 252) or without (n ¼ 99) prophylactic octreotide i.e. post-operative complication (121 vs. 57; p ¼ 0.11), POPF (34 vs. 8; p ¼ 0.16), SSC (88 vs. 34; p ¼ 0.92), mortality (8 vs. 3; p ¼ 0.94), re-operation (21 vs. 4; p ¼ 0.16) rate and LOS (17 (2e128) vs. 18 (1e63) days; p ¼ 0.59). Pre-operative biliary drainage Pre-operative biliary drainage was considered adequate in case a normal serum total bilirubin level was reached
Table 1 Primary tumour location and type of post-operative complications following pancreaticoduodenectomy in 351 patients Ampulla (n ¼ 98)
Bile duct (n ¼ 34)
Pancreas (n ¼ 141)
Duodenum (n ¼ 17)
Miscellaneous (n ¼ 61)
p value
Site of complication Overall SSC NSSC
42 35 10
21 16 5
78 46 37
11 8 4
26 17 11
0.08 0.29 0.035
POPF Overall Grade A Grade B Grade C
17 5 3 9
6 0 4 2
11 6 3 2
2 1 1 0
6 2 2 2
0.18 0.74 0.11 0.046
SSC, surgical site complication; NSSC, non-surgical site complication; POPF, post-operative pancreatic fistula.
B. Topal et al. / EJSO 33 (2007) 488e492 Table 2 Multivariate analysis of factors related to post-operative complications
Surgeon Gender (male) Bilirubin
p value
Odds ratio
95% CI
0.008 0.032 0.046
2.028 1.716 1.041
1.205e3.413 1.047e2.813 1.001e1.082
CI, confidence interval.
(n ¼ 66), and inadequate in case the bilirubin level remained above 17.1 mmol/L (n ¼ 97). In patients with inadequate drainage the pre-operative serum bilirubin level was significantly ( p < 0.001) lower as compared to jaundiced patients without drainage (n ¼ 80), i.e. 48.7 (17.6e394.3) mmol/L vs. 140.3 (17.8e665.4) mmol/L. Surgical outcome of all patients with pre-operative biliary drainage (n ¼ 163) was comparable with that of jaundiced patients without drainage i.e. post-operative complication 86 vs. 40 ( p ¼ 0.69), POPF 21 vs. 9 ( p ¼ 0.72), and LOS 17 (1e128) vs. 16 (6e52) days ( p ¼ 0.26). Also the outcome between patients with adequate drainage vs. non-jaundiced patients without drainage, and between patients with inadequate drainage vs. jaundiced patients without drainage was similar ( p > 0.17). Discussion The overall post-operative complication rate was 50.7% with pancreatic fistula (12%) as the most common adverse event in the present study. Mortality was 3.1% and the re-operation rate 7.1%. Although few studies report a zero to 1% mortality rate, these figures are comparable with most published data, wherein the importance of institutional expertise and surgical volume has been highlighted.1e4,12e15 Type and severity of post-operative complications The use of the TOSGS classification system showed that the vast majority of the patients in the current study needed drug therapy and/or invasive intervention. In addition to this classification system, post-operative complications were allocated to either SSC or NSSC that were observed in 34.7% and 15.9% of the patients, respectively. According to the ISGPF definition, the severity of pancreatic fistula was equally distributed. Pancreaticoduodenectomy for ampullary cancer was associated with the highest POPF grade C as compared to other peri-ampullary tumours. Although pancreatic texture is hard to describe objectively, these findings might suggest soft pancreatic texture (generally found in patients with non-pancreatic cancer) to be more at risk for POPF. In the present study delayed gastric emptying was not considered as a complication per se, since the authors believe delayed gastric emptying is hard to define and well reflected by the length of post-operative hospital stay. Indeed, the LOS was
491
significantly different between patients with vs. without post-operative complications. Comparable post-operative mortality has been reported in patients with and those without pancreatic fistula, although the mean LOS was found to be longer and the rates of certain complications to be higher in patients with POPF.5 Indeed, the final post-operative outcome following PD is also strongly dependent on the performance or quality of care in non-surgical departments within the same institution, such as the intensive care unit, invasive radiology, as well as on the expertise of the medical and nursing personnel on the wards. In the present study, however, postoperative mortality and re-operation rates were significantly higher in patients with pancreatic fistula. Also, the LOS was significantly longer for patients with post-operative complications or pancreatic fistula, and clearly related to the severity of complications as scored according to the TOSGS. Predictors of post-operative complications Age, extended resections, and soft pancreatic texture have been reported as the most prominent independent risk factors for mortality and post-operative complications following PD in high-volume institutions. Other potential risk factors are impaired renal function, coronary heart disease, main pancreatic duct diameter less than 3 mm, need for blood transfusion, resection of other organs, and absence of preoperative biliary drainage.5,16e18 In the present study surgeon, male patient and pre-operative serum total bilirubin level were found as independent risk factors for post-operative complications. Neither resection of other organs nor portal vein reconstruction at the time of PD influenced post-operative morbidity or mortality. As reported by other high-volume institutions, the use of prophylactic octreotide did not improve surgical outcome in the current study.19,20 Biliary drainage prior to PD remains controversial with conflicting effects on peri-operative morbidity and mortality.21e23 In the present study, pre-operative serum bilirubin level was significantly lower after biliary drainage as compared to jaundiced patients without drainage. Nevertheless, despite the fact that an elevated serum bilirubin level appeared to be an independent predictor of post-operative complications, pre-operative biliary drainage did not improve the outcome following PD. Therefore, these data argue against pre-operative biliary drainage in jaundiced patients with a peri-ampullary tumour. However, the limitation of the current study is its non-randomized nature. Conclusion Complication rate following pancreaticoduodenectomy is high and determined by surgeon, patient male gender, and pre-operative serum bilirubin level. The vast majority of patients with post-operative complications need drug
492
B. Topal et al. / EJSO 33 (2007) 488e492
or invasive therapy with subsequent increase in length of hospital stay. Because of the diversity of complications following pancreaticoduodenectomy, the implementation of reproducible classification systems is advocated in order to better assess and compare surgical outcome between surgeons and between centres. Patients who develop a pancreatic fistula have a significantly higher mortality and re-operation rate indicating that prevention and adequate treatment of post-operative pancreatic fistula should remain high priority and need further study. Acknowledgements Many thanks to the staff members of the Department of Hepatobiliary and Pancreatic Diseases for including patients in this study, and to the radiologists* for meticulously interpreting the images and their therapeutic interventions: W. Vansteenbergen , C. Verslype , D. Bielen* and D. Vanbeckevoort*. References 1. Birkmeyer LD, Stukel TA, Siewers AE, Goodney PP, Wennberg DE, Lucas FL. Surgeon volume and operative mortality in the United States. N Engl J Med 2003;349:2117–27. 2. Balcom JH, Rattner DW, Warshaw AL, Chang Y, Fernandez-del Castillo C. Ten-year experience with 733 pancreatic resections: changing indications, older patients, and decreasing length of hospitalization. Arch Surg 2001;136:391–8. 3. Yeo CJ, Cameron JL, Sohn TA, et al. Six hundred fifty consecutive pancreaticoduodenectomies in the 1990s: pathology, complications, and outcomes. Ann Surg 1997;226:248–57; discussion 257e60. 4. Suzuki Y, Fujino Y, Ajiki T, et al. No mortality among 100 consecutive pancreaticoduodenectomies in a middle-volume center. World J Surg 2005;29:1409–14. 5. Lin JW, Cameron JL, Yeo CJ, Riall TS, Lillemoe KD. Risk factors and outcomes in postpancreaticoduodenectomy pancreaticocutaneous fistula. J Gastrointest Surg 2004;8:951–9. 6. Popiela T, Kedra B, Sierzega M, Gurda A. Risk factors of pancreatic fistula following pancreaticoduodenectomy for periampullary cancer. Hepatogastroenterology 2004;51:1484–8. 7. Tien YW, Lee PH, Yang CY, Ho MC, Chiu YF. Risk factors of massive bleeding related to pancreatic leak after pancreaticoduodenectomy. J Am Coll Surg 2005;201:554–9.
8. Kazanjian KK, Hines OJ, Eibl G, Reber HA. Management of pancreatic fistulas after pancreaticoduodenectomy: results in 437 consecutive patients. Arch Surg 2005;140:849–54. discussion 854e6. 9. Bassi C, Dervenis C, Butturini G, et al. for the International Study Group on Pancreatic Fistula. Postoperative pancreatic fistula: an international study group (ISGPF) definition. Surgery 2005;138:8–13. 10. Clavien P, Sanabria J, Strasberg S. Proposed classification of complication of surgery with examples of utility in cholecystectomy. Surgery 1992;111:518–26. 11. Dindo D, Demartines N, Clavien PA. Classification of surgical complications: a new proposal with evaluation in a cohort of 6336 patients and results of a survey. Ann Surg 2004;240:214–5. 12. Cameron JL, Riall TS, Coleman J, Belcher KA. One thousand consecutive pancreaticoduodenectomies. Ann Surg 2006;244:10–5. 13. Makary MA, Winter JM, Cameron JL, et al. Pancreaticoduodenectomy in the very elderly. J Gastrointest Surg 2006;10:347–56. 14. de Castro SM, Busch OR, van Gulik TM, Obertop H, Gouma DJ. Incidence and management of pancreatic leakage after pancreatoduodenectomy. Br J Surg 2005;92:1117–23. 15. Aranha GV, Hodul PJ, Creech S, Jacobs W. Zero mortality after 152 consecutive pancreaticoduodenectomies with pancreaticogastrostomy. J Am Coll Surg 2003;197:223–31; discussion 231e2. 16. Gouma DJ, van Geenen RC, van Gulik TM, et al. Rates of complications and death after pancreaticoduodenectomy: risk factors and the impact of hospital volume. Ann Surg 2000;232:786–95. 17. Muscari F, Suc B, Kirzin S, et al. French Associations for Surgical Research. Risk factors for mortality and intra-abdominal complications after pancreatoduodenectomy: multivariate analysis in 300 patients. Surgery 2006;139:591–8. 18. Adam U, Makowiec F, Riediger H, Schareck WD, Benz S, Hopt UT. Risk factors for complications after pancreatic head resection. Am J Surg 2004;187:201–8. 19. Yeo CJ, Cameron JL, Lillemoe KD, et al. Does prophylactic octreotide decrease the rates of pancreatic fistula and other complications after pancreaticoduodenectomy? Results of a prospective randomized placebo-controlled trial. Ann Surg 2000;232:419–29. 20. Li-Ling J, Irving M. Somatostatin and octreotide in the prevention of postoperative pancreatic complications and the treatment of enterocutaneous pancreatic fistulas: a systematic review of randomized controlled trials. Br J Surg 2001;88:190–9. 21. Jagannath P, Dhir V, Shrikhande S, Shah RC, Mullerpatan P, Mohandas KM. Effect of preoperative biliary stenting on immediate outcome after pancreaticoduodenectomy. Br J Surg 2005;92:356–61. 22. Sewnath ME, Birjmohun RS, Rauws EA, Huibregtse K, Obertop H, Gouma DJ. The effect of preoperative biliary drainage on postoperative complications after pancreaticoduodenectomy. J Am Coll Surg 2001;192:726–34. 23. Sohn TA, Yeo CJ, Cameron JL, Pitt HA, Lillemoe KD. Do preoperative biliary stents increase postpancreaticoduodenectomy complications? J Gastrointest Surg 2000;4:258–67; discussion 267e8.