- Email: [email protected]
Evaluation of the International Study Group of Pancreatic Surgery definition of post-pancreatectomy hemorrhage in a high-volume center
Surgical Outcome Research Evaluation of the International Study Group of Pancreatic Surgery definition of post-pancreatectomy hemorrhage in a high-volume center Robert Gr€ utzmann, MD, PhD, Felix R€ uckert, MD, Nele Hippe-Davies, Marius Distler, MD, and Hans-Detlev Saeger, Professor, Dresden, Germany
Background. Although postpancreatectomy hemorrhage (PPH) is observed infrequently after pancreatic surgery, it remains a serious complication with a high rate of mortality. Recently, the International Study Group of Pancreatic Surgery (ISGPS) issued a new definition for PPH. To evaluate and validate this new definition, we analyzed data retrospectively from our center. Methods. Data from 945 patients who underwent pancreatic surgery in our department between October 1993 and December 2009 were identified retrospectively from our prospective database with regard to the occurrences of PPH. We graded the hemorrhages recorded in our database according to the ISGPS consensus definition. We assessed the clinical course, morbidity, mortality, and duration of hospital stay for patients with grade B and C PPHs in comparison with patients who underwent pancreatic resections without hemorrhage. Results. Grade B PPH after pancreatic surgery occurred in 16 patients (1.7%), and grade C PPH occurred in 38 patients (4.0%). Mortality was significantly increased in PPH grades B and C compared with control patients (25.9% vs 2.0%; P < .001) and contributed to nearly one-half of the mortality in the present series. Morbidity was also increased in patients with grade B (76.5%) and C (94.6%) PPH compared with control patients (59.6%; P < .001). Grade B and C PPH correlated significantly with the incidence of grade C postoperative pancreatic fistula (14.8% vs 1.9%), grade C delayed gastric emptying (18.5% vs 4.0%), and wound infection (38.9% vs 13.5%) compared with control patients. Conclusion. This is the first clinical evaluation of the ISGPS PPH definition. Our data indicate that the new definition correlates well with morbidity, mortality, and duration of hospital stay. The definition, therefore, seems suitable for clinical and scientific applications. (Surgery 2012;151:612-20.) From the Department of General, Thoracic and Vascular Surgery, University Hospital Carl Gustav Carus, Technische Universit€ a t Dresden, Dresden, Germany
PANCREATIC SURGERY has improved distinctly in the last 15 years. Mortality rates after pancreatic head resection in high-volume centers have been reduced from 20% to less than 5%1 and this reduced Robert Gr€ utzmann and Felix R€ uckert contributed equally to this work. Accepted for publication September 22, 2011. Reprint requests: Robert Gr€ utzmann, MD, PhD, Department of General, Thoracic and Vascular Surgery, University Hospital Carl Gustav Carus, Technische Universit€at Dresden, Fetscherstrasse 74, 01307 Dresden, Germany. E-mail: Robert.Gruetzmann@ uniklinikum-dresden.de. 0039-6060/$ - see front matter Ó 2012 Mosby, Inc. All rights reserved. doi:10.1016/j.surg.2011.09.039
612 SURGERY
mortality is attributed primarily to technical advances in operative techniques, perioperative care, and management of complications.2 Specific complications after pancreatic surgery include postoperative pancreatic fistula (POPF), delayed gastric emptying (DGE), and postoperative pancreatic hemorrhage (PPH). Although it is rare, PPH is one of the most dreaded complications of pancreatic surgery because of its high mortality. The differences in the frequencies of these complications are used commonly by authors of clinical trials to assess the outcomes and benefits of new techniques. The disparities in postoperative morbidity rates, however, are not only attributed to differences in the operative techniques and patient
Surgery Volume 151, Number 4
care but also to the absence of internationally accepted uniform definitions of complications. International classifications of POPF3 and DGE4 have been validated through clinical trials and now are accepted broadly in the field. Until recently, however, there were no international classifications for PPH. To better assess this complication of pancreatic operations, the International Study Group of Pancreatic Surgery (ISGPS) defined grades of PPHs in 20075 to provide objective and internationally comparable definitions for PPH. This classification was a prerequisite for improving the management of complications and the assessment of the outcomes of clinical trials.6 According to the ISGPS consensus definition, PPH is categorized into three grades: A, B, and C. The grading scheme considers the onset of bleeding, the location of the bleeding, and the overall clinical impact (Fig 1). Unfortunately, despite its existence, this classification still lacks clinical validation. Therefore, in the present study, we reviewed our experience with PPH during postoperative courses after pancreatic surgery during the past 15 years to assess the utility of the new ISGPS definition in clinical settings and in clinical trials. During this time, our center performed 945 pancreatic resections. We analyzed the correlation between the new ISGPS consensus definition with clinical course, morbidity, mortality, and duration of stay. METHODS Patients and data collection. Patients who underwent pancreatic surgery between October 1993 and December 2009 in the Department of General, Thoracic and Vascular Surgery in the Carl Gustav Carus University Hospital, Dresden, were entered into a prospective electronic database. The database and medical records for each patient were analysed retrospectively. Patients were contacted by mail or by telephone and asked to participate in our survey. The survey data were complemented by the clinical notes of their physicians and surgeons. Information for deceased patients was obtained from family members or from the general practitioner. The patients were asked about rehospitalization after the initial operation, complications from PPH, and their general condition. Histologic findings were obtained from pathology reports and were classified according to the World Health Organization grading system. Operations. Pylorus-preserving and classical pancreatoduodenectomies and Whipple reconstructions were performed with a single jejunal limb. Pancreatojejunostomies were achieved with an end-to-side ‘‘dunking’’ technique by the use of
Gr€ u tzmann et al 613
two rows of interrupted suturing with absorbable material. Hepaticojejunostomies were performed by the use of a single row of transmural interrupted sutures with absorbable monofilament material (5-0 to 6-0) as described previously.7 For anastomoses involving small bowel (ie, duodeno- or gastrojejunostomy) we use hand-sewn single layer running suture with absorbable monofilamental suture. We performed left resections by suturing over the pancreatic duct with the use of nonabsorbable monofilament material. The parenchyma was closed with the use of absorbable monofilament material. Definitions. On the basis of existing data, we categorized postoperative hemorrhage bleedings according to the ISGPS consensus definition.5 The different grades were defined according to the time of onset of the hemorrhage, the location of the bleeding, the clinical impact and clinical condition of the patient, and the diagnostic and therapeutic consequences (Fig 1). An ‘‘early hemorrhage’’ was defined as hemorrhage within the first 24 hours. Mild bleeding was characterized as a small- or medium-volume blood loss (drop of hemoglobin concentration of <3 g/dL) with no or minimal clinical impairment, no need for invasive intervention (reoperation or interventional angiography), and successful conservative treatment (fluid resuscitation and blood transfusion of 2 to 3 units packed red blood cells). Severe bleeding was defined as a larger volume blood loss (decrease in hemoglobin concentration of >3 g/dL) and potentially lifethreatening clinical impairment with tachycardia, hypotension, and/or oliguria; treatment involved the need for blood transfusion (>3 units packed red blood cells) and/or invasive treatment (reoperation or interventional angiography). The database did not contain grade A PPH patients because minor episodes of hemorrhage that had no clinical impact were often not recorded. Mortality was defined as in-hospital and/or death within 30 days of the index operation. Wound infection, urinary tract infection, pancreatitis, pneumonia, and cholangitis were included as complications when there was the need for antibiotic treatment and/or when hospital stay was prolonged because of these complications. These complications, as well as grades B and C POPFs, grades B and C DGE, and anastomotic leaks were subsumed as ‘‘overall morbidity.’’ DGE and POPF were classified according to the definitions provided by the International Study Group on Pancreatic Fistula3 and the ISGPS.4 ‘‘Anastomotic leaks’’ were assessed either during relaparotomy or by medical imaging. For the purposes of this manuscript, ‘‘pancreatic surgery’’
614 Gr€ u tzmann et al
Surgery April 2012
Fig 1. Overview of the consensus definition for PPH, including grading and diagnostic and therapeutic consequences (adapted from Wente et al5).
includes the following procedures according to the ISGPS definition: pylorus-preserving or classic pancreatoduodenectomy, pancreatic left resection, duodenum-preserving pancreatic head resection, central pancreatectomy, and total pancreatectomy. Diagnoses included ‘‘malignant tumors,’’ ‘‘chronic pancreatitis,’’ and other benign diseases, which were categorized under ‘‘benign tumors’’ (like benign duodenal disease, ampullary hyperplasia and others). Statistical analysis. Values in ratio scales are expressed as the means ± SD. The analysis was performed by use of the v2 test and Student t-test, and a P-value <.05 was considered significant. Statistical computations were performed with Excel (Microsoft, Redmond, WA) and PASW Statistics 18.0 for Windows (SPSS, Chicago, IL, USA). RESULTS Patient cohort. From October 1993 to December 2009, 945 patients underwent pancreatic surgery in our department. The patient group consisted of 544 men (57.6%) and 401 women (42.4%) and had a mean age of 57.9 (±12.9) years. The following procedures were performed in these patients: the classic pancreatoduodenectomy (PD) (262 patients, 27.6%), pylorus-preserving pancreatoduodenectomy (476 patients, 50.4%), total pancreatectomy (13 patients, 1.4%), a left resection (151 patients, 16.0%), central
pancreatectomy (32 patients, 3.4%), and duodenumpreserving head resection (11 patients, 1.2%). A total of 538 cases of malignant neoplasms (56.9%), 149 cases of benign neoplasms (15.8%), and 258 cases of chronic pancreatitis (27.3%) were observed in the patient group (Table I). Prevalence of PPH and grading. Of the 945 patients, 54 developed PPH (5.7%). Of these, 16 were classified as grade B PPH (1.7%), and 38 were classified as grade C PPH (4.0%). Of the 16 patients with grade B PPH, 10 (62.5%) had intraluminal bleeding with a mean onset of bleeding 204 hours postoperatively (±131 hours), and 6 (37.5%) had extraluminal bleeding with a mean onset of bleeding 24 hours postoperatively (±129 hours). Of the 38 patients with grade C PPH, 16 (42.1%) had intraluminal bleeding with a mean onset of bleeding of 216 hours postoperatively (±260 hours). Twenty-two patients with PPH grade C (57.9%) had extraluminal bleeding with a mean onset of bleeding of 144 hours postoperatively (±261 hours; Fig 2). No correlation was observed between the incidence of PPH and patient age, sex, or diagnosis (Table II). In addition, we did not observe any correlation of PPH with the duration of the operation or the amount of intraoperative blood loss (data not shown). We could show a tendency towards a greater incidence of PPH in classic PD and PPPD compared with the
Gr€ u tzmann et al 615
Surgery Volume 151, Number 4
Table I. Characteristics of the patient cohort Number of patients (%) n
Operation PD PPPD Total pancreatectomy Left resection Central pancreatectomy DPHR Total
Malignant neoplasm
262 476 13 151 32 11 945
162 294 12 59 11
(61.8) (61.8) (92.3) (39.1) (34.4) 0 538 (56.9)
Benign neoplasm 15 56 1 54 20 3 149
(5.7) (11.8) (7.7) (35.7) (62.5) (27.3) (15.8)
Chronic pancreatitis
Mortality
85 (32.5) 126 (26.5) 0 38 (25.2) 1 (3.1) 8 (72.7) 258 (27.3)
10 (3.8) 19 (3.9) 0 3 (2.0) 0 0 32 (3.4)
Shown are absolute numbers with percentages in parentheses. DPHR, Duodenum-preserving head resection; PD, classic pancreatoduodenectomy; PPPD, pylorus-preserving pancreatoduodenectomy.
PPH Grade C (n=38)
PPH Grade B (n=16)
Intraluminal (n=10) Sentinel Bleeding/ Clinical signs • Hematemesis/ Blood in the nasogastric tube output (n=8) • Black stool (n=1)
Initial endoscopic treatment n= 8 7 successfull 1 recurrent endoscopy (succesfull) Conservative treatment n=1 1 successfull
Extraluminal (n=6) Sentinel Bleeding/ Clinical signs • Hematemesis (n=1) • Blood in the abdominal drain output (n=1) • Hypotensions (n=4)
Relaparotomy n=5 4 successfull 1 with multiple relaparotomies (not successfull)
Intraluminal (n=16)
Extraluminal (n=22)
Sentinel Bleeding/ Clinical signs
Sentinel Bleeding/ Clinical signs
• Hematemesis/ Blood in the nasogastric tube output (n=8) • Black stool (n=7) • Septic shock (n=1)
• Hematemesis (n=4) • Blood in the abdominal drain output (n=8) • Hypotension (n=1) • Drop in Hb (n=6) • Hematochezia (n=2) • Septic shock (n=1)
Initial endoscopic treatment n= 16 4 successfull 10 recurrent endoscopies - 10 relaparotomy (6 successfull) 2 relaparotomies (successfull)
Initial endoscopic treatment n=1 1 relaparotomy neccessary (successfull)
15 Relaparotomies 7 successfull 8 with re-relaparotomie (2 successfull) 2 endoscopic treatments 1 not successfully 1 with relap (not successfull) 3 interventional angiographies 2 successfull 1 relarotomy (not succesfull) 2 percutaneous drainages 1 successfull 1 with relap (successfull)
Mortality 0%
Mortality 17% (n=1)
Mortality 25% (n=4)
Mortality 41% (n=9)
Fig 2. Incidence of grade B and C PPHs in the patient cohort, including the locations and clinical signs of bleeding, and the overall therapeutic consequence and outcome.
other operative procedures. However, this was not statically significant, maybe because of low case numbers of left resections, pancreatectomy, and central pancreatectomy (Table II). Location of the hemorrhage and outcome analysis. The main bleeding sites were the pancreatic cut surfaces, the suture lines at the pancreatojejunostomy site (n = 11), the suture line of the duodenojejunostomy after PPPD, and the suture line of the gastrojejunostomy after the classic PD (n = 10;
Fig 3). Patients with grade B PPH with intraluminal bleeding had no mortality. Clinical signs of the bleeding included hematemesis or blood in the nasogastric tube output in 8 patients, and black stool in one patient. Eight patients were treated successfully with endoscopy, and one patient was treated by transfusion of packed red blood cells. Of the patients in the grade B group with extraluminal bleeding, one patient died (17%). Clinical signs of the bleeding included hematemesis
616 Gr€ u tzmann et al
Surgery April 2012
Table II. Characteristics of the patient cohort No. of patients PPH
Sex Male Age, yr Diagnosis Malignant neoplasm Benign neoplasm Chronic pancreatitis Procedure PD PPPD Total pancreatectomy Left resection Central pancreatectomy DPHR Mortality Portal vein resection ICU stay, days POS, days
Control n = 891
B n = 16
C n = 38
P value
509 (57.2) 58.0 (±12.9)
10 (58.8) 57.6 (±13.6)
25 (67.6) 57.7 (±11.3)
n.s. n.s.
21 (55.3) 8 (21.0) 9 (23.7)
n.s.
7 25 2 4
n.s.
508 (57.1) 138 (15.4) 245 (27.5) 250 441 11 146 32 11 18 88 4.0
(28.1) (49.5) (1.2) (16.4) (3.6) (1.2) (2.0) (9.9) (±5.9)
18.9 (±13.5)
9 (56.3) 3 (18.7) 4 (25.0) 5 (31.2) 10 (62.6) 0 1 (6.2) 0 0 1 (6.2) 4 (25.0) 4.2 (±4.0)*
(18.4) (65.8) (5.3) (10.5) 0 0 13 (34.2) 2 (5.3) 21.2 (±28.3)y
20.5 (±8.7)*
40.2 (±35.0)y
<.001 n.s. n.s.* <.001y n.s.* <.001y
*t-test comparing grade C PPH and controls. yt-test comparing grade C PPH and controls. Total numbers are shown with percentages in parentheses. Statistical analyses were performed by the use of Pearson’s v2 test and Student’s t-test. DPHR, Duodenum-preserving head resection; n.s., not significant; PD, classic pancreatoduodenectomy; POS, postoperative stay; PPPD, pylorus-preserving pancreatoduodenectomy.
(n = 1), blood in the abdominal drain output (n = 1), and hypotension (n = 4). All patients underwent relaparotomy; however, one patient received endoscopy before the relaparotomy because the bleeding was initially suspected in the upper gastrointestinal tract. Patients with grade C PPH exhibited significantly greater mortality rates, as expected. Of the patients with PPH grade C, 4 of 16 patients with intraluminal bleeding died (25%). The clinical signs of bleeding were mostly hematemesis or blood in the nasogastric tubes output (n = 8) and melena (n = 7). Although all patients in this group underwent initial endoscopic treatment, only four of these patients were treated successfully. Ten patients underwent multiple endoscopies and relaparotomy and exhibited a high rate of mortality (n = 4, 40%). Two patients underwent successful reoperation immediately after the initial endoscopy. In the patients with PPH grade C and extraluminal bleeding, the mortality rate was even greater (n = 9, 41%). The clinical signs of bleeding were not consistent (Fig 2). Relaparotomies were required in 15 patients, 6 of whom died. Because of the complex
underlying causes of hemorrhage in this group, we also performed additional therapeutic interventions in a high percentage of these patients (two endoscopic treatments, three interventional angiographies). Two postoperative pancreatic fistulas in his group were treated by computed tomographyguided respiratory or ultrasound-guided percutaneous drainage. We observed significantly greater overall mortality in patients with postoperative hemorrhage compared with patients without bleeding (25.9% vs 2.0%; P < .001; Table II). Altogether, PPH accounted for 14 of the 32 postoperative deaths in the entire patient cohort. Our data also indicated a significantly greater duration of stay in the intensive care unit (ICU) for the patients with PPH. Patients without postoperative hemorrhage stayed in the ICU for only 4.0 days (±5.9), whereas the patients with PPH remained in the ICU for an average of 14.3 days (±21.3; P <.001). The PPH grades correlated strongly with DGE and POPF. The incidence of grade C DGE was 5.1 times greater in patients with grade C PPH than in the patients without PPH (P < .001). The occurrence of grade C POPF was 11 times greater in
Surgery Volume 151, Number 4
Gr€ u tzmann et al 617
Fig 3. Bleeding sites after pancreatic surgery in our patient cohort. (1) Pancreatic cut surface and suture line of pancreaticojejunostomy sites/jejunum (n = 11). (2) Suture line of duodenojejunostomy after PPPD or gastrojejunostomy after PD (n = 10). (3) Area of resection/diffuse intra-abdominal bleeding (n = 9). (4) Stomach intraluminal bleeding (n = 7). (5) Branches of the superior mesenteric artery (n = 4). (6) Splenic hilum/spleen (n = 3). (7) Common hepatic artery (n = 2). (8) Portal vein (n = 2). (9) Gastroduodenal artery (n = 2). (10) Hepaticojejunostomy (n = 2). (11) Upper mesenterial vein (n = 1). (12) Liver (n = 1). The pie charts show the mortality rates (red section) for the four most frequent bleeding locations, and the size of the diagram correlates with the case number. The bleeding sites in the figure are numbered according to the sequence shown above and are highlighted in magenta.
patients with grade C PPH than in patients without PPH (P < .001). The incidence of relaparotomy was 11.2 times greater in the grade C PPH patient group than in the patients without PPH (P < .001). The grade-dependent rates of other parameters are presented in Table III. Long-term results. Fourteen of the patients with PPH died postoperatively. Of the remaining patients, 16 had died by the time our study was conducted, but 24 patients with PPH were still alive. Follow-up for these patients occurred at 91 (±44) months after operation. Complete responses were obtained from 20 (83%) of the 24 living patients. None of the 6 patients with grade B PPH required additional hospitalization. One patient (16.6%) reported persistent diffuse abdominal pain after PPH. Of the 14 patients with grade C PPH, 3 (21%) patients required addition hospitalisation because of complications, which can be attributed to PPH. Six patients with PPH (43%) needed narcotics because of persistent intra-abdominal pain after PPH (Supplementary data). DISCUSSION The use of different and nonstandardized definitions of PPH after pancreatic surgery5 have increased the difficulty in comparing the results
from clinical trials. The need for an internationally accepted consensus definition of postoperative haemorrhage led to the suggested definition compiled by the ISGPS in 2007.5 The present study is the largest retrospective study to analyze PPH in a high-volume center. Our results show that the ISGPS consensus definition appears to be suitable for both clinical and scientific applications. Grading was achieved by reclassification of data from our prospective data base. The prevalence of grades B and C PPH in the patient cohort using this definition was 5.7%. A weak point of the new PPH grading system is the definition of grade A bleeding that subsumes mild bleedings within the first 24 as ‘‘small or medium volume blood loss (drop of hemoglobin concentration of <3 g/dL) with no or minimal clinical impairment, no need for invasive intervention, and successful conservative treatment.’’ Treatment of patients after operation on intensive care unit may also cause a decrease in Hb levels, resulting in false-positive classifications. Because there is no immediate clinical consequence, these kinds of hemorrhages were also not recorded in our prospective database. However, we found 16 patients with grade B PPH and 38 patients with grade C PPH.
618 Gr€ u tzmann et al
Surgery April 2012
Table III. Characteristics of the patient cohort No. of patients (%) PPH Control n = 891 Morbidity* Bleeding Pancreatitis H-J leak P-J leak Pneumonia Urinary tract infect DGE Grade A Grade B Grade C POPF Grade A Grade B Grade C Postoperative cholangitis Wound infection Relaparotomy
534 (60.0) 0 48 (5.4) 21 (2.4) 51 (5.7) 37 (4.2) 52 (5.8)
B n = 16 13 16 1 1 2 2
(81.2) (100) (6.2) (6.2) (12.5) (12.5) 0
186 (20.9) 67 (7.5) 36 (4.0)
3 (18.7) 3 (18.7) 2 (12.5)
100 47 17 14 120 59
4 (25.0) 2 (12.5) 0 0 5 (31.2) 6 (37.5)
(11.2) (5.3) (1.9) (1.6) (13.5) (6.6)
C n = 38
P value
(92.1) (100) (23.7) (15.8) (34.2) (28.9) (15.8)
<.001 <.001 <.001 <.001 <.001 <.001 .043
8 (21.0) 5 (13.2) 8 (21.0)
<.001
35 38 9 6 13 11 6
10 4 8 2 16 28
(26.3) (10.5) (21.0) (5.3) (42.1) (73.7)
<.001
n.s. <.001 <.001
*Morbidity included wound infections, urinary tract infections, grade B and C POPF, grade B and C DGE, pancreatitis, anastomosis insufficiencies, pneumonia, and cholangitis. Statistical analyses were performed using Pearson’s chi-squared test. DGE, Delayed gastric emptying; H-J, hepaticojejunostomy; P-J, pancreatojejunostomy; PPH, postpancreatectomy haemorrhage; POPF, postoperative pancreatic fistula.
Previous groups analyzing postoperative hemorrhage after pancreatic surgery have reported occurrences ranging between 2.5% and 9.3%.8-13 The differences in these numbers are mostly attributable to the different definitions of PPH used. One other study in which authors used the ISGPS definition demonstrated that 10.4% of their patient cohort developed PPH.14 In our study, patients with grade B PPH had greater incidences of intraluminal bleeding, and grade C PPH patients exhibited greater occurrences of extraluminal bleeding. Early extraluminal bleeding in grade B patients was often the result of inadequate hemostasis at the time of operation. Bleeding in this location is relatively easy to manage and has a relatively low mortality rate. Extraluminal bleeding in grade C patients correlated with anastomotic leaks at the bilio and pancreatico-enteric anastomoses and were located mainly in the retroperitoneal operative field or peripancreatic vessels. Grade C intraluminal bleedings usually occurred from the pancreatic parenchyma at the site of gland transaction. Contrary to PPH that occurred early postoperatively (first 24 hours), these types of PPH,
particularly extraluminal bleeding, are much more difficult to handle because of the underlying pathogenesis and have greater mortality rates. The aggressive therapeutic efforts required to treat these PPH are mirrored by the associated morbidity. Although 6.7% of patients without PPH underwent relaparotomy, 38% of patients with grade B PPH and 74% of patients with grade C PPH required relaparotomy, this difference was statistically significant. A suggested diagnostic and therapeutic algorithm is depicted in Fig 4. In the present study, other complications, including wound infection, urinary tract infection, grades B and C POPFs, grades B and C DGE, pancreatitis, anastomosis leaks, pneumonia, and cholangitis, were also associated with PPH. Sixty percent of patients without PPH exhibited complications, whereas 81% and 92% of the patients with grades B and C PPH developed complications; indeed, nearly all types of complications were more common in patients with grades B and C PPH. One of the most important findings was that PPH contributed to nearly one-half of the mortality in the present study. The mortality rate in patients with grades B and C PPH was 3.1 and 17.1 times greater,
Gr€ u tzmann et al 619
Surgery Volume 151, Number 4
Fig 4. Proposed algorithm for diagnosis and treatment of PPH.
respectively, compared with patients without PPH. In the analyzed patient cohort, PPH accounted for 14 of the 32 postoperative deaths in all 945 patients. This mortality demonstrates the clinical severity of PPH. Our data indicate that early mortality and morbidity are highly correlated with the PPH grades proposed by the ISGPS. The duration of ICU stay can also be used as a measure of the morbidity associated with the postoperative course. In the present study, the median duration of the postoperative ICU stay was significantly prolonged in patients with PPH. Follow-up of our patients revealed that PPH also impairs the health of the patients in lifelong terms. None of the patients with grade B PPH reported rehospitalization. In contrast, 21% of the patients with grade C PPH reported rehospitalization as the result of abdominal pain, infections, endocrine insufficiency, and renal insufficiency. Of the latter group, 36% of the patients also reported subjective impairment in quality of life as the result of intraabdominal pain. This number of patients with pain is greater as expected, as a recent study of our
group showed persistent pain/worse pain only in 13,8% of patients with chronic pancreatitis after PD.15 Therefore, the IGSPS definition appears to be valid for discriminating between PPH grades. Our results are limited because PPH grade A was not evaluated. In conclusion, we observed a high correlation between the grades of PPH and the clinical course, associated morbidity, and mortality, which demonstrates the potential that this classification has to offer in clinical trials and audit. REFERENCES 1. Kleespies A, Albertsmeier M, Obeidat F, Seeliger H, Jauch KW, Bruns CJ. The challenge of pancreatic anastomosis. Langenbecks Arch Surg 2008;393:459-71. 2. Ho CK, Kleeff J, Friess H, Buchler MW. Complications of pancreatic surgery. HPB (Oxford) 2005;7:99-108. 3. Bassi C, Dervenis C, Butturini G, et al. Postoperative pancreatic fistula: an international study group (ISGPF) definition. Surgery 2005;138:8-13. 4. Wente MN, Bassi C, Dervenis C, et al. Delayed gastric emptying (DGE) after pancreatic surgery: a suggested definition by the International Study Group of Pancreatic Surgery (ISGPS). Surgery 2007;142:761-8.
620 Gr€ u tzmann et al
5. Wente MN, Veit JA, Bassi C, et al. Postpancreatectomy hemorrhage (PPH): an International Study Group of Pancreatic Surgery (ISGPS) definition. Surgery 2007; 142:20-5. 6. Welsch T, Borm M, Degrate L, Hinz U, Buchler MW, Wente MN. Evaluation of the International Study Group of Pancreatic Surgery definition of delayed gastric emptying after pancreatoduodenectomy in a high-volume centre. Br J Surg 2010;97:1043-50. 7. R€ uckert F, Kersting S, Fiedler D, et al. Chronic pancreatitis: early results of pancreaticoduodenectomy and analysis of risk factors. Pancreas 2011;40:925-30. 8. Yekebas EF, Wolfram L, Cataldegirmen G, et al. Postpancreatectomy hemorrhage: diagnosis and treatment: an analysis in 1669 consecutive pancreatic resections. Ann Surg 2007;246:269-80. 9. 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.
Surgery April 2012
10. Choi SH, Moon HJ, Heo JS, Joh JW, Kim YI. Delayed hemorrhage after pancreaticoduodenectomy. J Am Coll Surg 2004;199:186-91. 11. van Berge Henegouwen MI, Allema JH, van Gulik TM, Verbeek PC, Obertop H, Gouma DJ. Delayed massive haemorrhage after pancreatic and biliary surgery. Br J Surg 1995;82:1527-31. 12. Turrini O, Moutardier V, Guiramand J, et al. Hemorrhage after duodenopancreatectomy: impact of neoadjuvant radiochemotherapy and experience with sentinel bleeding. World J Surg 2005;29:212-6. 13. Wei HK, Wang SE, Shyr YM, et al. Risk factors for postpancreaticoduodenectomy bleeding and finding an innovative approach to treatment. Dig Surg 2009;26:297-305. 14. Sanjay P, Fawzi A, Fulke JL, et al. Late post pancreatectomy haemorrhage. Risk factors and modern management. JOP 2010;11:220-5. 15. R€ uckert F, Distler M, Hoffmann S, et al. Quality of life in patients after pancreaticoduodenectomy for chronic pancreatitis. J Gastrointest Surg 2011;15:1143-50.
Background. Although postpancreatectomy hemorrhage (PPH) is observed infrequently after pancreatic surgery, it remains a serious complication with a high rate of mortality. Recently, the International Study Group of Pancreatic Surgery (ISGPS) issued a new definition for PPH. To evaluate and validate this new definition, we analyzed data retrospectively from our center. Methods. Data from 945 patients who underwent pancreatic surgery in our department between October 1993 and December 2009 were identified retrospectively from our prospective database with regard to the occurrences of PPH. We graded the hemorrhages recorded in our database according to the ISGPS consensus definition. We assessed the clinical course, morbidity, mortality, and duration of hospital stay for patients with grade B and C PPHs in comparison with patients who underwent pancreatic resections without hemorrhage. Results. Grade B PPH after pancreatic surgery occurred in 16 patients (1.7%), and grade C PPH occurred in 38 patients (4.0%). Mortality was significantly increased in PPH grades B and C compared with control patients (25.9% vs 2.0%; P < .001) and contributed to nearly one-half of the mortality in the present series. Morbidity was also increased in patients with grade B (76.5%) and C (94.6%) PPH compared with control patients (59.6%; P < .001). Grade B and C PPH correlated significantly with the incidence of grade C postoperative pancreatic fistula (14.8% vs 1.9%), grade C delayed gastric emptying (18.5% vs 4.0%), and wound infection (38.9% vs 13.5%) compared with control patients. Conclusion. This is the first clinical evaluation of the ISGPS PPH definition. Our data indicate that the new definition correlates well with morbidity, mortality, and duration of hospital stay. The definition, therefore, seems suitable for clinical and scientific applications. (Surgery 2012;151:612-20.) From the Department of General, Thoracic and Vascular Surgery, University Hospital Carl Gustav Carus, Technische Universit€ a t Dresden, Dresden, Germany
PANCREATIC SURGERY has improved distinctly in the last 15 years. Mortality rates after pancreatic head resection in high-volume centers have been reduced from 20% to less than 5%1 and this reduced Robert Gr€ utzmann and Felix R€ uckert contributed equally to this work. Accepted for publication September 22, 2011. Reprint requests: Robert Gr€ utzmann, MD, PhD, Department of General, Thoracic and Vascular Surgery, University Hospital Carl Gustav Carus, Technische Universit€at Dresden, Fetscherstrasse 74, 01307 Dresden, Germany. E-mail: Robert.Gruetzmann@ uniklinikum-dresden.de. 0039-6060/$ - see front matter Ó 2012 Mosby, Inc. All rights reserved. doi:10.1016/j.surg.2011.09.039
612 SURGERY
mortality is attributed primarily to technical advances in operative techniques, perioperative care, and management of complications.2 Specific complications after pancreatic surgery include postoperative pancreatic fistula (POPF), delayed gastric emptying (DGE), and postoperative pancreatic hemorrhage (PPH). Although it is rare, PPH is one of the most dreaded complications of pancreatic surgery because of its high mortality. The differences in the frequencies of these complications are used commonly by authors of clinical trials to assess the outcomes and benefits of new techniques. The disparities in postoperative morbidity rates, however, are not only attributed to differences in the operative techniques and patient
Surgery Volume 151, Number 4
care but also to the absence of internationally accepted uniform definitions of complications. International classifications of POPF3 and DGE4 have been validated through clinical trials and now are accepted broadly in the field. Until recently, however, there were no international classifications for PPH. To better assess this complication of pancreatic operations, the International Study Group of Pancreatic Surgery (ISGPS) defined grades of PPHs in 20075 to provide objective and internationally comparable definitions for PPH. This classification was a prerequisite for improving the management of complications and the assessment of the outcomes of clinical trials.6 According to the ISGPS consensus definition, PPH is categorized into three grades: A, B, and C. The grading scheme considers the onset of bleeding, the location of the bleeding, and the overall clinical impact (Fig 1). Unfortunately, despite its existence, this classification still lacks clinical validation. Therefore, in the present study, we reviewed our experience with PPH during postoperative courses after pancreatic surgery during the past 15 years to assess the utility of the new ISGPS definition in clinical settings and in clinical trials. During this time, our center performed 945 pancreatic resections. We analyzed the correlation between the new ISGPS consensus definition with clinical course, morbidity, mortality, and duration of stay. METHODS Patients and data collection. Patients who underwent pancreatic surgery between October 1993 and December 2009 in the Department of General, Thoracic and Vascular Surgery in the Carl Gustav Carus University Hospital, Dresden, were entered into a prospective electronic database. The database and medical records for each patient were analysed retrospectively. Patients were contacted by mail or by telephone and asked to participate in our survey. The survey data were complemented by the clinical notes of their physicians and surgeons. Information for deceased patients was obtained from family members or from the general practitioner. The patients were asked about rehospitalization after the initial operation, complications from PPH, and their general condition. Histologic findings were obtained from pathology reports and were classified according to the World Health Organization grading system. Operations. Pylorus-preserving and classical pancreatoduodenectomies and Whipple reconstructions were performed with a single jejunal limb. Pancreatojejunostomies were achieved with an end-to-side ‘‘dunking’’ technique by the use of
Gr€ u tzmann et al 613
two rows of interrupted suturing with absorbable material. Hepaticojejunostomies were performed by the use of a single row of transmural interrupted sutures with absorbable monofilament material (5-0 to 6-0) as described previously.7 For anastomoses involving small bowel (ie, duodeno- or gastrojejunostomy) we use hand-sewn single layer running suture with absorbable monofilamental suture. We performed left resections by suturing over the pancreatic duct with the use of nonabsorbable monofilament material. The parenchyma was closed with the use of absorbable monofilament material. Definitions. On the basis of existing data, we categorized postoperative hemorrhage bleedings according to the ISGPS consensus definition.5 The different grades were defined according to the time of onset of the hemorrhage, the location of the bleeding, the clinical impact and clinical condition of the patient, and the diagnostic and therapeutic consequences (Fig 1). An ‘‘early hemorrhage’’ was defined as hemorrhage within the first 24 hours. Mild bleeding was characterized as a small- or medium-volume blood loss (drop of hemoglobin concentration of <3 g/dL) with no or minimal clinical impairment, no need for invasive intervention (reoperation or interventional angiography), and successful conservative treatment (fluid resuscitation and blood transfusion of 2 to 3 units packed red blood cells). Severe bleeding was defined as a larger volume blood loss (decrease in hemoglobin concentration of >3 g/dL) and potentially lifethreatening clinical impairment with tachycardia, hypotension, and/or oliguria; treatment involved the need for blood transfusion (>3 units packed red blood cells) and/or invasive treatment (reoperation or interventional angiography). The database did not contain grade A PPH patients because minor episodes of hemorrhage that had no clinical impact were often not recorded. Mortality was defined as in-hospital and/or death within 30 days of the index operation. Wound infection, urinary tract infection, pancreatitis, pneumonia, and cholangitis were included as complications when there was the need for antibiotic treatment and/or when hospital stay was prolonged because of these complications. These complications, as well as grades B and C POPFs, grades B and C DGE, and anastomotic leaks were subsumed as ‘‘overall morbidity.’’ DGE and POPF were classified according to the definitions provided by the International Study Group on Pancreatic Fistula3 and the ISGPS.4 ‘‘Anastomotic leaks’’ were assessed either during relaparotomy or by medical imaging. For the purposes of this manuscript, ‘‘pancreatic surgery’’
614 Gr€ u tzmann et al
Surgery April 2012
Fig 1. Overview of the consensus definition for PPH, including grading and diagnostic and therapeutic consequences (adapted from Wente et al5).
includes the following procedures according to the ISGPS definition: pylorus-preserving or classic pancreatoduodenectomy, pancreatic left resection, duodenum-preserving pancreatic head resection, central pancreatectomy, and total pancreatectomy. Diagnoses included ‘‘malignant tumors,’’ ‘‘chronic pancreatitis,’’ and other benign diseases, which were categorized under ‘‘benign tumors’’ (like benign duodenal disease, ampullary hyperplasia and others). Statistical analysis. Values in ratio scales are expressed as the means ± SD. The analysis was performed by use of the v2 test and Student t-test, and a P-value <.05 was considered significant. Statistical computations were performed with Excel (Microsoft, Redmond, WA) and PASW Statistics 18.0 for Windows (SPSS, Chicago, IL, USA). RESULTS Patient cohort. From October 1993 to December 2009, 945 patients underwent pancreatic surgery in our department. The patient group consisted of 544 men (57.6%) and 401 women (42.4%) and had a mean age of 57.9 (±12.9) years. The following procedures were performed in these patients: the classic pancreatoduodenectomy (PD) (262 patients, 27.6%), pylorus-preserving pancreatoduodenectomy (476 patients, 50.4%), total pancreatectomy (13 patients, 1.4%), a left resection (151 patients, 16.0%), central
pancreatectomy (32 patients, 3.4%), and duodenumpreserving head resection (11 patients, 1.2%). A total of 538 cases of malignant neoplasms (56.9%), 149 cases of benign neoplasms (15.8%), and 258 cases of chronic pancreatitis (27.3%) were observed in the patient group (Table I). Prevalence of PPH and grading. Of the 945 patients, 54 developed PPH (5.7%). Of these, 16 were classified as grade B PPH (1.7%), and 38 were classified as grade C PPH (4.0%). Of the 16 patients with grade B PPH, 10 (62.5%) had intraluminal bleeding with a mean onset of bleeding 204 hours postoperatively (±131 hours), and 6 (37.5%) had extraluminal bleeding with a mean onset of bleeding 24 hours postoperatively (±129 hours). Of the 38 patients with grade C PPH, 16 (42.1%) had intraluminal bleeding with a mean onset of bleeding of 216 hours postoperatively (±260 hours). Twenty-two patients with PPH grade C (57.9%) had extraluminal bleeding with a mean onset of bleeding of 144 hours postoperatively (±261 hours; Fig 2). No correlation was observed between the incidence of PPH and patient age, sex, or diagnosis (Table II). In addition, we did not observe any correlation of PPH with the duration of the operation or the amount of intraoperative blood loss (data not shown). We could show a tendency towards a greater incidence of PPH in classic PD and PPPD compared with the
Gr€ u tzmann et al 615
Surgery Volume 151, Number 4
Table I. Characteristics of the patient cohort Number of patients (%) n
Operation PD PPPD Total pancreatectomy Left resection Central pancreatectomy DPHR Total
Malignant neoplasm
262 476 13 151 32 11 945
162 294 12 59 11
(61.8) (61.8) (92.3) (39.1) (34.4) 0 538 (56.9)
Benign neoplasm 15 56 1 54 20 3 149
(5.7) (11.8) (7.7) (35.7) (62.5) (27.3) (15.8)
Chronic pancreatitis
Mortality
85 (32.5) 126 (26.5) 0 38 (25.2) 1 (3.1) 8 (72.7) 258 (27.3)
10 (3.8) 19 (3.9) 0 3 (2.0) 0 0 32 (3.4)
Shown are absolute numbers with percentages in parentheses. DPHR, Duodenum-preserving head resection; PD, classic pancreatoduodenectomy; PPPD, pylorus-preserving pancreatoduodenectomy.
PPH Grade C (n=38)
PPH Grade B (n=16)
Intraluminal (n=10) Sentinel Bleeding/ Clinical signs • Hematemesis/ Blood in the nasogastric tube output (n=8) • Black stool (n=1)
Initial endoscopic treatment n= 8 7 successfull 1 recurrent endoscopy (succesfull) Conservative treatment n=1 1 successfull
Extraluminal (n=6) Sentinel Bleeding/ Clinical signs • Hematemesis (n=1) • Blood in the abdominal drain output (n=1) • Hypotensions (n=4)
Relaparotomy n=5 4 successfull 1 with multiple relaparotomies (not successfull)
Intraluminal (n=16)
Extraluminal (n=22)
Sentinel Bleeding/ Clinical signs
Sentinel Bleeding/ Clinical signs
• Hematemesis/ Blood in the nasogastric tube output (n=8) • Black stool (n=7) • Septic shock (n=1)
• Hematemesis (n=4) • Blood in the abdominal drain output (n=8) • Hypotension (n=1) • Drop in Hb (n=6) • Hematochezia (n=2) • Septic shock (n=1)
Initial endoscopic treatment n= 16 4 successfull 10 recurrent endoscopies - 10 relaparotomy (6 successfull) 2 relaparotomies (successfull)
Initial endoscopic treatment n=1 1 relaparotomy neccessary (successfull)
15 Relaparotomies 7 successfull 8 with re-relaparotomie (2 successfull) 2 endoscopic treatments 1 not successfully 1 with relap (not successfull) 3 interventional angiographies 2 successfull 1 relarotomy (not succesfull) 2 percutaneous drainages 1 successfull 1 with relap (successfull)
Mortality 0%
Mortality 17% (n=1)
Mortality 25% (n=4)
Mortality 41% (n=9)
Fig 2. Incidence of grade B and C PPHs in the patient cohort, including the locations and clinical signs of bleeding, and the overall therapeutic consequence and outcome.
other operative procedures. However, this was not statically significant, maybe because of low case numbers of left resections, pancreatectomy, and central pancreatectomy (Table II). Location of the hemorrhage and outcome analysis. The main bleeding sites were the pancreatic cut surfaces, the suture lines at the pancreatojejunostomy site (n = 11), the suture line of the duodenojejunostomy after PPPD, and the suture line of the gastrojejunostomy after the classic PD (n = 10;
Fig 3). Patients with grade B PPH with intraluminal bleeding had no mortality. Clinical signs of the bleeding included hematemesis or blood in the nasogastric tube output in 8 patients, and black stool in one patient. Eight patients were treated successfully with endoscopy, and one patient was treated by transfusion of packed red blood cells. Of the patients in the grade B group with extraluminal bleeding, one patient died (17%). Clinical signs of the bleeding included hematemesis
616 Gr€ u tzmann et al
Surgery April 2012
Table II. Characteristics of the patient cohort No. of patients PPH
Sex Male Age, yr Diagnosis Malignant neoplasm Benign neoplasm Chronic pancreatitis Procedure PD PPPD Total pancreatectomy Left resection Central pancreatectomy DPHR Mortality Portal vein resection ICU stay, days POS, days
Control n = 891
B n = 16
C n = 38
P value
509 (57.2) 58.0 (±12.9)
10 (58.8) 57.6 (±13.6)
25 (67.6) 57.7 (±11.3)
n.s. n.s.
21 (55.3) 8 (21.0) 9 (23.7)
n.s.
7 25 2 4
n.s.
508 (57.1) 138 (15.4) 245 (27.5) 250 441 11 146 32 11 18 88 4.0
(28.1) (49.5) (1.2) (16.4) (3.6) (1.2) (2.0) (9.9) (±5.9)
18.9 (±13.5)
9 (56.3) 3 (18.7) 4 (25.0) 5 (31.2) 10 (62.6) 0 1 (6.2) 0 0 1 (6.2) 4 (25.0) 4.2 (±4.0)*
(18.4) (65.8) (5.3) (10.5) 0 0 13 (34.2) 2 (5.3) 21.2 (±28.3)y
20.5 (±8.7)*
40.2 (±35.0)y
<.001 n.s. n.s.* <.001y n.s.* <.001y
*t-test comparing grade C PPH and controls. yt-test comparing grade C PPH and controls. Total numbers are shown with percentages in parentheses. Statistical analyses were performed by the use of Pearson’s v2 test and Student’s t-test. DPHR, Duodenum-preserving head resection; n.s., not significant; PD, classic pancreatoduodenectomy; POS, postoperative stay; PPPD, pylorus-preserving pancreatoduodenectomy.
(n = 1), blood in the abdominal drain output (n = 1), and hypotension (n = 4). All patients underwent relaparotomy; however, one patient received endoscopy before the relaparotomy because the bleeding was initially suspected in the upper gastrointestinal tract. Patients with grade C PPH exhibited significantly greater mortality rates, as expected. Of the patients with PPH grade C, 4 of 16 patients with intraluminal bleeding died (25%). The clinical signs of bleeding were mostly hematemesis or blood in the nasogastric tubes output (n = 8) and melena (n = 7). Although all patients in this group underwent initial endoscopic treatment, only four of these patients were treated successfully. Ten patients underwent multiple endoscopies and relaparotomy and exhibited a high rate of mortality (n = 4, 40%). Two patients underwent successful reoperation immediately after the initial endoscopy. In the patients with PPH grade C and extraluminal bleeding, the mortality rate was even greater (n = 9, 41%). The clinical signs of bleeding were not consistent (Fig 2). Relaparotomies were required in 15 patients, 6 of whom died. Because of the complex
underlying causes of hemorrhage in this group, we also performed additional therapeutic interventions in a high percentage of these patients (two endoscopic treatments, three interventional angiographies). Two postoperative pancreatic fistulas in his group were treated by computed tomographyguided respiratory or ultrasound-guided percutaneous drainage. We observed significantly greater overall mortality in patients with postoperative hemorrhage compared with patients without bleeding (25.9% vs 2.0%; P < .001; Table II). Altogether, PPH accounted for 14 of the 32 postoperative deaths in the entire patient cohort. Our data also indicated a significantly greater duration of stay in the intensive care unit (ICU) for the patients with PPH. Patients without postoperative hemorrhage stayed in the ICU for only 4.0 days (±5.9), whereas the patients with PPH remained in the ICU for an average of 14.3 days (±21.3; P <.001). The PPH grades correlated strongly with DGE and POPF. The incidence of grade C DGE was 5.1 times greater in patients with grade C PPH than in the patients without PPH (P < .001). The occurrence of grade C POPF was 11 times greater in
Surgery Volume 151, Number 4
Gr€ u tzmann et al 617
Fig 3. Bleeding sites after pancreatic surgery in our patient cohort. (1) Pancreatic cut surface and suture line of pancreaticojejunostomy sites/jejunum (n = 11). (2) Suture line of duodenojejunostomy after PPPD or gastrojejunostomy after PD (n = 10). (3) Area of resection/diffuse intra-abdominal bleeding (n = 9). (4) Stomach intraluminal bleeding (n = 7). (5) Branches of the superior mesenteric artery (n = 4). (6) Splenic hilum/spleen (n = 3). (7) Common hepatic artery (n = 2). (8) Portal vein (n = 2). (9) Gastroduodenal artery (n = 2). (10) Hepaticojejunostomy (n = 2). (11) Upper mesenterial vein (n = 1). (12) Liver (n = 1). The pie charts show the mortality rates (red section) for the four most frequent bleeding locations, and the size of the diagram correlates with the case number. The bleeding sites in the figure are numbered according to the sequence shown above and are highlighted in magenta.
patients with grade C PPH than in patients without PPH (P < .001). The incidence of relaparotomy was 11.2 times greater in the grade C PPH patient group than in the patients without PPH (P < .001). The grade-dependent rates of other parameters are presented in Table III. Long-term results. Fourteen of the patients with PPH died postoperatively. Of the remaining patients, 16 had died by the time our study was conducted, but 24 patients with PPH were still alive. Follow-up for these patients occurred at 91 (±44) months after operation. Complete responses were obtained from 20 (83%) of the 24 living patients. None of the 6 patients with grade B PPH required additional hospitalization. One patient (16.6%) reported persistent diffuse abdominal pain after PPH. Of the 14 patients with grade C PPH, 3 (21%) patients required addition hospitalisation because of complications, which can be attributed to PPH. Six patients with PPH (43%) needed narcotics because of persistent intra-abdominal pain after PPH (Supplementary data). DISCUSSION The use of different and nonstandardized definitions of PPH after pancreatic surgery5 have increased the difficulty in comparing the results
from clinical trials. The need for an internationally accepted consensus definition of postoperative haemorrhage led to the suggested definition compiled by the ISGPS in 2007.5 The present study is the largest retrospective study to analyze PPH in a high-volume center. Our results show that the ISGPS consensus definition appears to be suitable for both clinical and scientific applications. Grading was achieved by reclassification of data from our prospective data base. The prevalence of grades B and C PPH in the patient cohort using this definition was 5.7%. A weak point of the new PPH grading system is the definition of grade A bleeding that subsumes mild bleedings within the first 24 as ‘‘small or medium volume blood loss (drop of hemoglobin concentration of <3 g/dL) with no or minimal clinical impairment, no need for invasive intervention, and successful conservative treatment.’’ Treatment of patients after operation on intensive care unit may also cause a decrease in Hb levels, resulting in false-positive classifications. Because there is no immediate clinical consequence, these kinds of hemorrhages were also not recorded in our prospective database. However, we found 16 patients with grade B PPH and 38 patients with grade C PPH.
618 Gr€ u tzmann et al
Surgery April 2012
Table III. Characteristics of the patient cohort No. of patients (%) PPH Control n = 891 Morbidity* Bleeding Pancreatitis H-J leak P-J leak Pneumonia Urinary tract infect DGE Grade A Grade B Grade C POPF Grade A Grade B Grade C Postoperative cholangitis Wound infection Relaparotomy
534 (60.0) 0 48 (5.4) 21 (2.4) 51 (5.7) 37 (4.2) 52 (5.8)
B n = 16 13 16 1 1 2 2
(81.2) (100) (6.2) (6.2) (12.5) (12.5) 0
186 (20.9) 67 (7.5) 36 (4.0)
3 (18.7) 3 (18.7) 2 (12.5)
100 47 17 14 120 59
4 (25.0) 2 (12.5) 0 0 5 (31.2) 6 (37.5)
(11.2) (5.3) (1.9) (1.6) (13.5) (6.6)
C n = 38
P value
(92.1) (100) (23.7) (15.8) (34.2) (28.9) (15.8)
<.001 <.001 <.001 <.001 <.001 <.001 .043
8 (21.0) 5 (13.2) 8 (21.0)
<.001
35 38 9 6 13 11 6
10 4 8 2 16 28
(26.3) (10.5) (21.0) (5.3) (42.1) (73.7)
<.001
n.s. <.001 <.001
*Morbidity included wound infections, urinary tract infections, grade B and C POPF, grade B and C DGE, pancreatitis, anastomosis insufficiencies, pneumonia, and cholangitis. Statistical analyses were performed using Pearson’s chi-squared test. DGE, Delayed gastric emptying; H-J, hepaticojejunostomy; P-J, pancreatojejunostomy; PPH, postpancreatectomy haemorrhage; POPF, postoperative pancreatic fistula.
Previous groups analyzing postoperative hemorrhage after pancreatic surgery have reported occurrences ranging between 2.5% and 9.3%.8-13 The differences in these numbers are mostly attributable to the different definitions of PPH used. One other study in which authors used the ISGPS definition demonstrated that 10.4% of their patient cohort developed PPH.14 In our study, patients with grade B PPH had greater incidences of intraluminal bleeding, and grade C PPH patients exhibited greater occurrences of extraluminal bleeding. Early extraluminal bleeding in grade B patients was often the result of inadequate hemostasis at the time of operation. Bleeding in this location is relatively easy to manage and has a relatively low mortality rate. Extraluminal bleeding in grade C patients correlated with anastomotic leaks at the bilio and pancreatico-enteric anastomoses and were located mainly in the retroperitoneal operative field or peripancreatic vessels. Grade C intraluminal bleedings usually occurred from the pancreatic parenchyma at the site of gland transaction. Contrary to PPH that occurred early postoperatively (first 24 hours), these types of PPH,
particularly extraluminal bleeding, are much more difficult to handle because of the underlying pathogenesis and have greater mortality rates. The aggressive therapeutic efforts required to treat these PPH are mirrored by the associated morbidity. Although 6.7% of patients without PPH underwent relaparotomy, 38% of patients with grade B PPH and 74% of patients with grade C PPH required relaparotomy, this difference was statistically significant. A suggested diagnostic and therapeutic algorithm is depicted in Fig 4. In the present study, other complications, including wound infection, urinary tract infection, grades B and C POPFs, grades B and C DGE, pancreatitis, anastomosis leaks, pneumonia, and cholangitis, were also associated with PPH. Sixty percent of patients without PPH exhibited complications, whereas 81% and 92% of the patients with grades B and C PPH developed complications; indeed, nearly all types of complications were more common in patients with grades B and C PPH. One of the most important findings was that PPH contributed to nearly one-half of the mortality in the present study. The mortality rate in patients with grades B and C PPH was 3.1 and 17.1 times greater,
Gr€ u tzmann et al 619
Surgery Volume 151, Number 4
Fig 4. Proposed algorithm for diagnosis and treatment of PPH.
respectively, compared with patients without PPH. In the analyzed patient cohort, PPH accounted for 14 of the 32 postoperative deaths in all 945 patients. This mortality demonstrates the clinical severity of PPH. Our data indicate that early mortality and morbidity are highly correlated with the PPH grades proposed by the ISGPS. The duration of ICU stay can also be used as a measure of the morbidity associated with the postoperative course. In the present study, the median duration of the postoperative ICU stay was significantly prolonged in patients with PPH. Follow-up of our patients revealed that PPH also impairs the health of the patients in lifelong terms. None of the patients with grade B PPH reported rehospitalization. In contrast, 21% of the patients with grade C PPH reported rehospitalization as the result of abdominal pain, infections, endocrine insufficiency, and renal insufficiency. Of the latter group, 36% of the patients also reported subjective impairment in quality of life as the result of intraabdominal pain. This number of patients with pain is greater as expected, as a recent study of our
group showed persistent pain/worse pain only in 13,8% of patients with chronic pancreatitis after PD.15 Therefore, the IGSPS definition appears to be valid for discriminating between PPH grades. Our results are limited because PPH grade A was not evaluated. In conclusion, we observed a high correlation between the grades of PPH and the clinical course, associated morbidity, and mortality, which demonstrates the potential that this classification has to offer in clinical trials and audit. REFERENCES 1. Kleespies A, Albertsmeier M, Obeidat F, Seeliger H, Jauch KW, Bruns CJ. The challenge of pancreatic anastomosis. Langenbecks Arch Surg 2008;393:459-71. 2. Ho CK, Kleeff J, Friess H, Buchler MW. Complications of pancreatic surgery. HPB (Oxford) 2005;7:99-108. 3. Bassi C, Dervenis C, Butturini G, et al. Postoperative pancreatic fistula: an international study group (ISGPF) definition. Surgery 2005;138:8-13. 4. Wente MN, Bassi C, Dervenis C, et al. Delayed gastric emptying (DGE) after pancreatic surgery: a suggested definition by the International Study Group of Pancreatic Surgery (ISGPS). Surgery 2007;142:761-8.
620 Gr€ u tzmann et al
5. Wente MN, Veit JA, Bassi C, et al. Postpancreatectomy hemorrhage (PPH): an International Study Group of Pancreatic Surgery (ISGPS) definition. Surgery 2007; 142:20-5. 6. Welsch T, Borm M, Degrate L, Hinz U, Buchler MW, Wente MN. Evaluation of the International Study Group of Pancreatic Surgery definition of delayed gastric emptying after pancreatoduodenectomy in a high-volume centre. Br J Surg 2010;97:1043-50. 7. R€ uckert F, Kersting S, Fiedler D, et al. Chronic pancreatitis: early results of pancreaticoduodenectomy and analysis of risk factors. Pancreas 2011;40:925-30. 8. Yekebas EF, Wolfram L, Cataldegirmen G, et al. Postpancreatectomy hemorrhage: diagnosis and treatment: an analysis in 1669 consecutive pancreatic resections. Ann Surg 2007;246:269-80. 9. 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.
Surgery April 2012
10. Choi SH, Moon HJ, Heo JS, Joh JW, Kim YI. Delayed hemorrhage after pancreaticoduodenectomy. J Am Coll Surg 2004;199:186-91. 11. van Berge Henegouwen MI, Allema JH, van Gulik TM, Verbeek PC, Obertop H, Gouma DJ. Delayed massive haemorrhage after pancreatic and biliary surgery. Br J Surg 1995;82:1527-31. 12. Turrini O, Moutardier V, Guiramand J, et al. Hemorrhage after duodenopancreatectomy: impact of neoadjuvant radiochemotherapy and experience with sentinel bleeding. World J Surg 2005;29:212-6. 13. Wei HK, Wang SE, Shyr YM, et al. Risk factors for postpancreaticoduodenectomy bleeding and finding an innovative approach to treatment. Dig Surg 2009;26:297-305. 14. Sanjay P, Fawzi A, Fulke JL, et al. Late post pancreatectomy haemorrhage. Risk factors and modern management. JOP 2010;11:220-5. 15. R€ uckert F, Distler M, Hoffmann S, et al. Quality of life in patients after pancreaticoduodenectomy for chronic pancreatitis. J Gastrointest Surg 2011;15:1143-50.