Descriptions of post pancreaticoduodenectomy (PD) reoperations have largely addressed relaparotomy for early complications such as postpancreatectomy hemorrhage (PPH) and pancreaticoenteric anastomotic leak (PEA) with associated intraabdominal collection [ The incidence and causes of early and late reoperations following PD. Factors predicting the need for early reoperation and its related mortality. The outcome of patients undergoing early and late reoperations.
Five hundred and twenty patients underwent PD between May 1989 and September 2010 at the Department of Surgical Gastroenterology, Sanjay Gandhi Postgraduate Institute of Medical Sciences, a tertiary referral institute in the northern part of India. Data was retrieved from a prospectively maintained database which included variables recorded during the index hospitalization and further readmissions if any. Information about patient follow up was obtained from follow up cards and telephonic follow up interviews.
All pancreaticoduodenectomies at our institute were performed by or under the direct supervision of consultant surgeons. Preoperatively all these patients underwent routine blood counts, liver and renal function tests, abdominal sonography, and an abdominal computed tomography (CT) scan for tumour staging. A side viewing endoscopic examination with biopsy was contemplated in almost all patients as a predominant number of patients who undergo PD at our hospital have periampullary carcinoma. In patients with a negative biopsy and a demonstrable CT scan evidence of a periampullary mass, decision to proceed with PD was taken. Endoscopic ultrasound was selectively used in those patients with a negative biopsy and no evidence of a mass lesion on CT scan. A preoperative endoscopic biliary drainage procedure with stenting was carried out in those patients with cholangitis, high preoperative bilirubin (>15 mg/dL), or poor nutritional status and surgery was then performed 4–6 weeks after stenting. All patients received preoperative antibiotic dose of cefoperazone and sulbactam 2 g and amikacin 500–750 mg at the time of induction. An equal number of patients underwent a pylorus preserving PD or a classical Whipple procedure according to the surgeon’s preference. Pancreatic reconstruction was performed first by an end to end or end to side pancreatico-jejunostomy in 514 patients. Of the remaining, 3 patients underwent pancreatico-gastrostomy and 3 had no pancreatico-enteric reconstruction due to underlying acute pancreatitis and necrosis. Duct to mucosa and pancreatic dunking or invagination was performed equally based on surgeon preference and pancreatic duct stenting was used selectively. This was followed by an end to side hepaticojejunostomy and antecolic duodenojejunostomy or gastrojejunostomy. Nasojejunal tube was used preferentially over feeding jejunostomy as a feeding access. Intraoperative octreotide (100 ug stat) was used selectively in those patients with a soft pancreas and continued for 5 days postoperatively. Abdomen was closed with drainage. A nasogastric tube was placed for gastric decompression. Postoperatively drain fluid and serum amylase levels were estimated on postoperative days 4 and 7.
Reoperations were classified into early and late. Reoperations performed during index hospital admission following PD were classified as early while those reoperations performed any time after the index hospitalization were classified as late. Patients operated for indications unrelated to complications of index surgery (PD), tumour recurrence, or adjuvant radiotherapy were excluded. The reoperation data was retrieved from the database. Patients requiring early reoperations due to complications of index surgery were compared with those who did not need reoperation. The parameters evaluated were demographic factors, clinical presenting symptoms, intra-operative parameters, pathology, and postoperative complications. A univariate analysis was done to determine factors predictive of early reoperation. Chi-square test was used for categorical variables and Mann-Whitney
Between May 1989 and September 2010, 520 patients underwent PD. Of these, 26 patients (5%) underwent PD for benign disease and 494 patients (95%) for malignant disease. The median age was 52 years (range 14–82 years). The in-hospital mortality rate was 8.1 percent (42 of 520), the overall morbidity rate was 62% per cent (322 of 520), and the median hospital stay was 14 days (range 5–112 days). 96 of these 520 patients (18.5%) were reoperated upon. 72 (75%) of these were early, 18 (18.8%) were late reoperations, and 6 patients (6.2%) had both early and late reoperations. For the purpose of analysis, the 6 patients who underwent both early and late reoperation were included in both the groups, thereby accounting for 78 patients
The indications for early reoperation were postpancreatectomy hemorrhage (PPH) in 53 patients (68%), pancreatico-enteric anastomotic leak (PEA) with intra-abdominal collection in 10 (13%), hepatico-jejunostomy (HJ) leak in 3 (3.8%), duodeno-jejunostomy (DJ) leak in 4 (5%), intestinal obstruction in 1 (1.2%), and miscellaneous causes in 7 (9%) such as wound dehiscence
Indications of early reoperations and surgeries performed.
Of the 53 patients undergoing reoperation for PPH, 41 patients (77.3%) had late bleeds (>24 hours) and 32 patients (60.4%) had extra luminal bleeds. The commonest surgery for PPH was suture ligation of the pancreatic cut surface bleeder which was done in 23 (43%) patients. Nearly 10% of patients operated for PPH had a negative laparotomy as no active source of bleed was identified. The median time to surgery in patients being reoperated for PPH was 5 days (range 0–59 days). 10 of these 53 patients (18.8%) presented with rebleed following first relaparotomy. Among these 10 patients, 4 patients required a second relaparotomy and in 6 patients angiographic embolization was done for gastroduodenal artery
Changes over time.
1989–2000 | 2000–2005 | 2006–2010 | |
---|---|---|---|
Number of pancreaticoduodenectomies | 160 | 132 | 228 |
Incidence of PPH | 20% (32/160) | 24.2% (32/132) | 13.1% (30/228) |
reoperation rate | 17.5% (28/160) | 21.2% (28/132) | 9.6% (22/228) |
Indications for reoperation | |||
(i) PPH | 20 | 21 | 12 |
(ii) PJ leak with intra-abdominal collection | 4 | 4 | 2 |
(iii) HJ leak | 1 | 2 | |
(iv) DJ/GJ leak | 2 | 2 | |
(v) Miscellaneous | 2 | 2 | 4 |
Overall in-hospital mortality | 11.9% (19/160) | 9% (12/132) | 4.8% (11/228) |
In-hospital mortality following early reoperation | 42.8% (12/28) | 28.5% (8/28) | 27.2% (6/22) |
In-hospital mortality in patients not undergoing early reoperation | 5.3% (7/132) | 3.8% (4/104) | 2.4% (5/206) |
PPH: post pancreatectomy hemorrhage; PJ: pancreaticojejunostomy; HJ: hepaticojejunostomy; DJ: duodenojejunostomy; GJ: gastrojejunostomy.
Factors predictive of early reoperation on univariate analysis were preoperative factors such as longer duration of jaundice (>3 months) (
Univariate analysis of factors predicting the need for early reoperation.
Parameters | Early reoperations ( |
Not reoperated early ( |
|
---|---|---|---|
Age in years (median) | 52 | 51.5 | 0.850 |
Gender (M/F) | 53/25 | 318/124 | 0.498 |
Duration of jaundice > 3 months | 21 (27%) | 69 (15.6%) | 0.051 |
Comorbidities | 17 (21.8%) | 111 (25.1%) | 0.572 |
Preoperative hemoglobin (median) | 11.0 | 11.2 | 0.425 |
Preoperative albumin |
3.5 | 3.5 | 0.643 |
Total bilirubin > 10 mg% | 18 (23.1%) | 51 (11.5%) | 0.010 |
Preoperative biliary drainage | 41 (52.6%) |
276 (62.4%) | 0.103 |
Duration of surgery (hours) | 7.35 | 7.0 | 0.096 |
Blood loss (mL) (median) | 750 mL | 500 mL | 0.001 |
Blood transfusion | 50 (64.1%) | 213 (48.2%) | 0.010 |
Malignancy (94%) | Malignancy (93%) | 0.844 | |
Pathology | Ampullary Ca (74%) | Ampullary Ca (71%) | |
Benign (6%) | Benign (7%) | ||
PPH | 67.9% | 9.3% | 0.000 |
PEA leak | 34.6% | 14.3% | 0.001 |
HJ leak | 24.3% | 6.3% | 0.000 |
DJ/GJ leak | 14.1% | 2.5% | 0.000 |
DGE | 21.8% | 10.2% | 0.007 |
Intraabdominal collection | 43.6% | 10.2% | 0.000 |
ARF | 11.5% | 1.6% | 0.000 |
Septicemia | 32% | 6.3% | 0.000 |
Postoperative hospital stay (Mean) | 25.5 days | 13 days | 0.000 |
In-hospital mortality | 26 (33.1%) | 16 (3.6%) | 0.000 |
PPH: post pancreatectomy hemorrhage; PEA: pancreaticoenteric anastamosis; HJ: hepaticojejunostomy; DJ: duodenojejunostomy; GJ: gastrojejunostomy; DGE: delayed gastric emptying; ARF: acute renal failure.
On multivariate analysis using the logistic regression model, preoperative duration of jaundice > 3 months
Multivariate analysis of factors predicting the need for early reoperation.
Parameter |
|
Exp. ( |
95% CI for Exp. ( |
---|---|---|---|
Duration of jaundice > 3 months | 0.019 | 3.532 | 1.23–10.147 |
PPH | 0.000 | 0.101 | 0.052–0.198 |
Intraabdominal collection | 0.020 | 0.426 | 0.200–0.908 |
DJ/GJ leak | 0.041 | 0.307 | 0.099–0.0951 |
PPH: post pancreatectomy hemorrhage; PEA: pancreaticoenteric anastamosis; HJ: hepaticojejunostomy; DJ: duodenojejunostomy; GJ: gastrojejunostomy; DGE: delayed gastric emptying; ARF: acute renal failure.
Of the 26 patients who had postoperative mortality following early reoperation, the underlying cause was PPH in 17 patients, PEA leak and intra-abdominal collection in 6 patients, DJ leak, feeding jejunostomy site leak with peritonitis and acute renal failure in 1 patient each. In the 17 patients who expired following reoperation for PPH, 15 (88%) of them were reoperated for late bleeds (>24 hours following PD). Septic shock with supervening multiorgan failure was the main cause of death in all these patients. Analysis of factors affecting mortality in patients undergoing early reoperation showed that the only significant factor on multivariate analysis was development of postoperative acute renal failure (
Twenty four patients underwent late reoperations (16 males; 8 females, median age 47.5 years (range 20–68 years)). Of these 6 patients had also undergone early reoperations, 4 for PPH and 2 for PEA leak with intra-abdominal collection. The indications for late reoperations were complications of index surgery (group 1)
Indications for late reoperations.
Indications | Number of patients | Surgery performed | Interval between PD and re-operation |
---|---|---|---|
Group 1 | |||
Incisional hernia | 4 | Mesh hernioplasty | 8–68 months |
Pancreatico-jejunostomy stricture | 2 | Revision PJ/PG | 31/36 months |
Adhesive SAIO | 2 | Band release | 16/96 months |
HJ stricture | 1 | Revision HJ | 29 months |
Persistent gastroparesis | 1 | Distal gastrectomy | 26 months |
Enterocutaneous fistula (ECF) | 1 | Repair of ECF | 8 months |
Afferent limb perforation with |
1 | Abscess drainage, external drainage of afferent limb perforation, lavage and FJ | 21 months |
| |||
Group 2 | |||
Peritoneal dissemination with SAIO | 4 | Peritoneal nodule biopsy: 2 |
5–19 months |
Liver metastasis | |||
(i) Metastatic GIST-2 | 3 | Nonanatomical resection: 1 |
12/30 months |
(ii) Ruptured liver metastasis-1 | Lavage and drainage: 1 | 9 months | |
Scar site recurrence | 1 | Wide local excision with mesh repair | 16 months |
| |||
Group 3 | |||
Radiation enteritis, jejunal stricture |
1 | Jejuno-jejunal by pass | 9 months |
Colonic and afferent loop necrosis | 1 | Excision of afferent loop, right hemicolectomy, revision roux-en-Y hepaticojejunostomy | 21 months |
DJ stricture | 1 | Gastrojejunostomy | 128 months |
| |||
Group 4 | |||
Vascular ectasia of jejunum with upper gastrointestinal bleed | 1 | Partial gastrectomy, revision gastrojejunostomy, side to side jejunojejunostomy | 40 months |
SAIO: sub-acute intestinal obstruction; GIST: gastrointestinal stromal tumour; PJ: pancreatico-jejunostomy; PG: pancreatico-gastrostomy; HJ: hepatico-jejunostomy; DJ: duodeno-jejunostomy; FJ: feeding jejunostomy.
To analyze the impact of reoperation on survival, patients with a minimum follow up of 3 years were included and Kaplan Meier survival curves were generated. 3 year survival of patients undergoing early reoperations
Kaplan Meier survival curve. Comparison of 3 year survival of patients undergoing early reoperation versus those patients not requiring reoperation.
Experience from high volume tertiary care centers around the world has shown a significant decrease in mortality following PD over the last couple of decades. Despite a significant decrease in postoperative mortality, PD is still associated with a fairly high postoperative morbidity as reported by various centers in the range of 30–60% [
Reoperation rates in series dealing with pancreatic head resection have varied from 4–11% [
PPH is one of the grave complications following PD and occurs in 2–20% of patients as reported by various series [
PPH was the predominant cause of early reoperation in our subset of patients similar to that reported by the other series [
The second commonest cause of reoperation after PPH was PEA leak with intra-abdominal collection. Majority of these collections can be managed successfully by percutaneously placed drains under image guidance [
Bilioenteric anastomotic leak is very uncommon following PD and the management is usually conservative [
Various surgical series have looked into the factors predictive of the occurrence of complications following PD. Some of them are presence of associated medical risk factors, need for preoperative biliary drainage, texture of remnant pancreas, and size of pancreatic duct [
Overall mortality following PD significantly increases after early reoperation, and is in the range of 13–60% as shown by Standop et al. [
Due to improved long-term survival of patients with nonpancreatic periampullary carcinoma as compared to pancreatic cancer, quite a few patients present on long-term follow up with complications related to the index surgery or disease recurrence which may be amenable to surgical intervention. With the increasing use of adjuvant radiotherapy to gain better local control and thereby improve disease-free survival, patients may require intervention for complications of the same. Excluding patients with malignant pancreatic neuroendocrine tumours, this issue has been very sparsely addressed in the literature with most of them being occasional case reports or short case series [
The indications for late reoperations are varied. Patients undergoing second surgery for complications related to recurrence of primary disease following PD other than those with malignant neuroendocrine tumours usually have a relatively poor outcome even after resectional surgery for local or distant recurrences as shown by Nakano et al. and Fujii et al. [
Early reoperation had no impact on 3 year survival of patients in our series. This is in contrast to that reported by Yeo et al. in the 1990s, in which they showed that in addition to the tumour pathologic characteristics, the addition of reoperation had a negative impact on long-term survival of patients undergoing PD for periampullary carcinoma [
In this study, 18.5% of patients were reoperated following PD. The two main indications for early reoperation were PPH (68%) and PEA leak with associated intra-abdominal collection (13%). Reoperation for PPH is usually indicated in patients with early presentation of PPH (<24 hours) or delayed presentation of PPH where angiographic embolization is not feasible or successful. Early reoperation when used judiciously in conjunction with arterial coil embolization continues to be an important tool in the armamentarium for the management of PPH in the present era. Although in-hospital mortality in this subset of patients was high (33.3%), this is largely due to the associated sepsis rather than insult of the reoperation. Early reoperation did not have a bearing on long-term survival
This paper was presented as a poster at the joint conference of the International Association of Pancreatology and the Indian Pancreas Club, February 10th–13th, 2011, Cochin, India.