J Gastrointest Surg (2014) 18:340–347 DOI 10.1007/s11605-013-2388-x
2013 SSAT POSTER PRESENTATION
Clinical and Pathologic Features Influencing Survival in Patients Undergoing Pancreaticoduodenectomy for Pancreatic Adenocarcinoma Cynthia E. Weber & Eileen A. Bock & Michael G. Hurtuk & Gerard J. Abood & Jack Pickleman & Margo Shoup & Gerard V. Aranha
Received: 14 May 2013 / Accepted: 7 October 2013 / Published online: 23 November 2013 # 2013 The Society for Surgery of the Alimentary Tract
Abstract Objective The aim of the study was to determine the clinicopathological features that influence survival in patients with resected pancreatic ductal adenocarcinoma (PDA). Methods The study used a single institution retrospective review of patients undergoing pancreaticoduodenectomy (PD) for PDA from 1993 to 2010. Results Two hundred forty-six consecutive cases of resected PDA were identified: 128 males (52 %), median age 68 years. Median hospital length of stay was 8 days and 30-day mortality rate was 2.4 %. There were 101 (41.1 %) postoperative complications, 77 % of which were Dindo–Clavien Grade 3 or less. Overall survival was 85, 63, 25, and 15 % at 6 months, 1 year, 3 years, and 5 years, respectively, with a median survival of 17 months. Multivariate Cox proportional hazard modeling demonstrated lymph node ratio was negatively correlated with survival at all time points. Preoperative hypertension was a poor prognostic factor at 6 months, 3 years, and 5 years. The absence of postoperative complications was protective at 6 months whereas pancreatic leaks were associated with worse survival at 6 months. Abdominal pain on presentation, operative time, and estimated blood loss were also associated with decreased survival at various time points. Conclusion The strongest prognostic variable for short- and long-term survival after PD for PDA is lymph node ratio. Short-term survival is influenced by the postoperative course.
This study was presented at the 54th Annual Meeting of the Society for Surgery of the Alimentary Tract, Digestive Diseases Week 2013, Poster Presentation and the 47th Annual Pancreas Club Meeting, 2013, Poster of Distinction. C. E. Weber : E. A. Bock : M. G. Hurtuk : G. J. Abood : J. Pickleman : G. V. Aranha Department of Surgery, Division of Surgical Oncology, Loyola University, Maywood, IL, USA G. V. Aranha Department of Surgery, Division of Surgical Oncology, Hines VA Hospital, Hines,, IL, USA M. Shoup Department of Surgery, Division of Surgical Oncology, Cadence Health, Winfield, IL, USA G. V. Aranha (*) Division of Surgical Oncology, Loyola University Medical Center, 2160 S First Ave, Maywood, IL 60153, USA e-mail: [email protected]
Keywords Pancreaticoduodenectomy . Pancreatic ductal adenocarcinoma . Lymph node ratio
Introduction Pancreatic cancer is the fourth leading cause of cancer death in the USA. Contrary to the trends of many solid tumors, the incidence and the death rate for pancreatic cancer are slowly rising. The incidence of pancreatic cancer is rising by about 1.5 % per year and the death rate increased by 0.4 % between 2004 and 2008. Lifetime risk for pancreatic cancer in the USA is 1 in 78 (1.5 %). For the year 2013, it is estimated that 45,220 new cases of pancreatic cancer will be diagnosed and 38,460 individuals will die from their disease.1,2 Unfortunately, the majority of patients present with advanced stage disease. Thus, only about 20 % are eligible for surgical resection at the time of diagnosis.3–7 Despite recent improvements in systemic treatment strategies,
J Gastrointest Surg (2014) 18:340–347
341
aggressive surgical resection with negative margins remains the only possible chance for cure as systemic chemotherapy has yet to dramatically affect survival. Overall 5-year survival for all comers with pancreatic cancer remains dismal, at less than 5 %.3,8–10 In contrast, the median survival after surgical resection is likely somewhere between 13 and 28 months, with the 5-year survival rates for patients after surgical resection ranging from 4 to 24 % (see Table 1).3–18 The purpose of this study was to define the short- and longterm survival rates as well as characterize the postoperative course of a cohort of patients that underwent pancreaticoduodenectomy (PD) for pancreatic ductal adenocarcinoma (PDA) at a single institution between 1993 and 2010. We also sought to determine clinicopathological features that were associated with survival in our cohort of patients.
Methods Following IRB approval, a retrospective review was conducted of a prospectively maintained database of all
patients undergoing PD for PDA at a single institution from 1993 through 2010. Patients who underwent PD for any other pathology were excluded from this study. Electronic medical records and paper charts were reviewed for demographic data, preoperative symptoms, physical exam findings, and patient comorbidities. Operative reports and anesthesia records were viewed to document intraoperative blood loss, intraoperative transfusions, and operative time. Final pathology information, including TNM stage, the presence or absence of perineural or lymphovascular invasion, as well as final margin status was documented from pathology reports. Postoperative information such as length of hospital stay, and the occurrence of postoperative complications, as graded by the Dindo–Clavien classification, was also recorded.19–21 Pancreatic leak was defined as a drain volume of greater than 50 cm3 of amylase-rich fluid or a drain amylase level more than three times the serum amylase on or after the third postoperative day after the intake of a general diet The patient was considered to have undergone a reoperation if they returned to the operating room within 30 days after the original surgery, either on the same admission or on readmission for a complication from the
Table 1 Retrospective studies of survival and prognostic factors in pancreatic cancer Author
Year
Patients undergoing resection (n)
Whipple (%)
Median survival (months)
5-year survival (%)
Prognostic factors on multivariate analysis
Wenger et al. Berger et al.
2000 2003
158 128
91 100
13.8 –
11.8 by LNR (12.0, 16.0, 0)
tumor size, margins lymph node ratio, CA 19–9, margins, adjuvant therapy
Howard et al.
2006
226
90
13
4.0
Sierzega et al.
2006
96
100
14.2
–
Winter et al.
2006
1,175
100
18
18.0
House et al.
2007
696
86
16.0
Pawlik et al. Slidell et al.
2007 2007
905 4,005
100 74
by N stage (27, 16) 17.4 13
tumor size, tumor grade, margins, postoperative complications tumor grade, N stage, margins, lymph node ratio tumor size, N stage, margins, tumor grade, COPD, bile leak, adjuvant therapy N stage, tumor grade, margins
16.1 6.8
Kazanjian et al.
2008
182
100
28.5
24.0
Ueda et al.
2008
140
71
14.5
12.3
Riediger et al. You et al. Bhatti et al. La Torre et al. Ferrone et al. Wentz et al.
2009 2009 2010 2011 2012 2012
204 219 84 101 499 169
85 100 100 100 84 89
18 14 22 19 19 15.1
15.0 14.2 13.0 17.0 19.0 –
age, tumor size, tumor grade, lymph node ratio tumor size, local extension, tumor grade, adjuvant therapy, marital status, N stage, lymph node ratio estimated blood loss, tumor grade, N stage, perineural invasion, operation before 1996 CA 19–9, tumor size, venous involvement, blood transfusion requirement, serum albumin on postoperative month number 1 margins, lymph node ratio, tumor grade tumor size, tumor grade lymph node ratio lymph node ratio, tumor grade, margins margins, N stage lymph node ratio, CA 19-9
342
original surgery. The 30-day readmission rate was determined by patient records for readmission to the institution as well as by viewing follow-up progress notes to ascertain whether or not the patient had been admitted to an outside institution within 30 days after surgery. Overall survival was determined by searching the Social Security Death Index for alive or deceased status and date of death. For all patients, a traditional PD was performed. In over half of cases (56 %), the pancreatic remnant was managed with a pancreaticogastrostomy (PG), with pancreaticojejunostomy representing the remainder of the cohort. The PG reconstruction by this primary surgeon has been extensively reported in the literature.22,23 Statistical analysis was performed using Stata12.0 (College Station, TX). For each survival endpoint (6 months, 1 year, 3 years, and 5 years), a univariate Cox proportional hazards analysis of each clinicopathological variable was performed. The significant variables for each survival endpoint were then combined and evaluated in a backward stepwise Cox proportional hazards model. The final multivariate model contains the significant variables determined by the backward stepwise selection. A p value of 5 Units (see Table 3). Pathologic description of the final specimen demonstrated perineural invasion in 83.3 % (205/232) and lymphovascular
J Gastrointest Surg (2014) 18:340–347 Table 2 Patient clinical and pathological characteristics (n=246)
Variable Age (years) Median (range) Sex Male Female
No. of patients (%)
68 (30–88) 128 (52) 118 (48)
Lymphovascular invasion Present 133 (54.1) Absent 93 (37.8) Not documented 20 (8.1) Perineural invasion Present 205 (83.3) Absent 27 (11) Not documented 14 (5.7) Resection margin R0 161 (65.4) R1 76 (30.9) Not documented 9 (3.7) Lymph node ratio 0 61 (24.8) >0 to 0.2 114 (46.3) >0.2 to 0.4 33 (13.4) >0.4 30 (12.2) Not documented 8 (3.3)
invasion in 54.1 % (133/226). The median number of lymph nodes sampled was 19. The predominant T stage was T3, in 72 % (174/241) of patients, while T1, T2, and T4 were diagnosed in 6.6 % (16/241), 20.8 % (50/241), and 0.4 % (1/241) of patients, respectively, with N1 disease representing 76.7 % (184/240) of nodal status. Based on the number of positive lymph nodes and the number of total lymph nodes
Fig. 1 Kaplan–Meier curve of overall survival in our cohort of 246 patients who underwent pancreaticoduodenectomy for pancreatic ductal adenocarcinoma
J Gastrointest Surg (2014) 18:340–347
343
Table 3 Operative factors and postoperative outcomes 3
Intraoperative blood loss (cm ) Median (range) Operative time (hours) Median (range) Intraoperative blood transfusions (n; %) 0 1 to 5 >5 Not documented Length of stay (days) Median (range) Complication (n; %) None Any complication Dindo–Clavien Classification Grade 1 Grade 2 Grade 3a or 3b Grade 4a or 4b Grade 5 Data missing Delayed gastric emptying Pancreatic leak Reoperation (n; %) 30-day readmission (n; %)
800 (75–10,000) 6.25 (3.5–15.5) 152 (61.8) 86 (35) 7 (2.8) 1 (0.4) 8 (5–80) 145 (58.9) 101 (41.1) 19 (18.8) 47 (46.5) 12 (11.9) 13 (12.9) 6 (5.9) 4 (4) 28 (11.4) 16 (6.5) 7 (2.8) 23 (9.8)
found in the specimen, we computed a lymph node ratio. Among our patients, 24.8 % (61/238) of patients had a lymph node ratio of 0, 46.3 % (114/238) of patients had a lymph node ratio of >0 to 0.2, 13.4 % (33/238) of patients had a lymph node ratio of >0.2 to 0.4, and 12.2 % (30/238) of patients had a lymph node ratio of >0.4. Due to missing information, we were unable to calculate a lymph node ratio for 3.3 % (8/246) of patients. An R0, or microscopically negative resection, was achieved in 65.4 % (161/237) of patients (see Table 2). Postoperatively, the median length of stay was 8 days, with a range of 5 to 80 days. In the postoperative period, as classified by the Dindo–Clavien classification, the majority of patients experiencing complications were considered Grade 1 (18.8 % or 19/101) or Grade 2 (46.5 % or 47/101). Grade 3 complications occurred in 11.9 % (12/101) of patients. There were 12.9 % (13/101) that were classified as Grade 4. Among our cohort of 246 patients, we documented 28 cases of delayed gastric emptying, representing 11.4 %. With regards to anastomotic leaks, there were 16 (6.5 %) pancreatic leaks, 3 (1.2 %) biliary leaks, and 1 (0.4 %) gastrojejunostomy leak. A total of seven patients underwent reoperation within 30 days of the original surgery. Indications for reoperation included: complicated infections, persistent efferent limb obstruction, wound dehiscence and evisceration, and
incomplete removal of JP. Two patients that died within 30 days after surgery underwent reoperation prior to their death. In addition to the above two patients, there were four other patients that died within 30 days of surgery, for a total of 6 (5.9 %) Grade 5 complications, or a 30-day postoperative mortality rate of 2.4 % (6/246) in this cohort. Using univariate Cox proportional hazard modeling, we analyzed patient demographics, preoperative symptoms and comorbidities, operative factors, final pathologic data, and postoperative complications individually to determine associations between these factors and survival at 6 months, 1 year, 3 years, and 5 years. For each time period, these significant univariate variables were then combined into a multivariate model and analyzed by a backward selection Cox proportional hazards model (see Tables 4, 5, 6, and 7). At 6 months, on univariate analysis, female gender was associated with a worse survival (hazard ratio (HR) 2.00, p= 0.046). In addition, the presence of preoperative hypertension was negatively correlated with survival (HR 2.54, p=0.016). Several operative and postoperative complications were statistically correlated with survival at this short-term survival time point. Estimated blood loss (HR 1.00, p=0.001) and operative time (HR 1.22, p=0.04) was statistically significant in the short term. The absence of any postoperative complication was protective, and these patients were more likely to be alive at 6 months (HR of 0.3, p=0.001). In contrast, the presence of a postoperative wound infection (HR 3.16, p=0.017) or postoperative pancreatic anastomotic leak (HR 3.50, p=0.005) were negatively correlated with survival at 6 months. Finally, lymph node ratio was negatively
Table 4 Factors influencing 6-month survival Parameters influencing survival
Hazard ratio (95 % CI)
Univariate analysis (6-month survival) Female gender 2.00 (1.01 to 3.94) Preoperative hypertension 2.54 (1.19 to 5.40) Estimated blood loss 1.00 (1.00 to 1.00) Operative time 1.22 (1.01 to 1.47) Lymph node ratio 4.65 (1.15 to 18.78) No postoperative complication 0.30 (0.15 to 0.60) Wound infection 3.16 (1.23 to 8.12) Pancreatic leak 3.50 (1.46 to 8.42) Multivariate Analysis (6-month survival) Female gender 2.49 (1.15 to5.39) Preoperative hypertension 2.62 (1.16 to 5.91) Estimated blood loss 1.00 (1.00 to 1.00) Lymph node ratio 15.88 (3.22 to 78.42) Pancreatic leak 2.90 (1.08 to 7.81) No postoperative complication 0.39 (0.18 to 0.87) *Denotes statistical significance at p
2013 SSAT POSTER PRESENTATION
Clinical and Pathologic Features Influencing Survival in Patients Undergoing Pancreaticoduodenectomy for Pancreatic Adenocarcinoma Cynthia E. Weber & Eileen A. Bock & Michael G. Hurtuk & Gerard J. Abood & Jack Pickleman & Margo Shoup & Gerard V. Aranha
Received: 14 May 2013 / Accepted: 7 October 2013 / Published online: 23 November 2013 # 2013 The Society for Surgery of the Alimentary Tract
Abstract Objective The aim of the study was to determine the clinicopathological features that influence survival in patients with resected pancreatic ductal adenocarcinoma (PDA). Methods The study used a single institution retrospective review of patients undergoing pancreaticoduodenectomy (PD) for PDA from 1993 to 2010. Results Two hundred forty-six consecutive cases of resected PDA were identified: 128 males (52 %), median age 68 years. Median hospital length of stay was 8 days and 30-day mortality rate was 2.4 %. There were 101 (41.1 %) postoperative complications, 77 % of which were Dindo–Clavien Grade 3 or less. Overall survival was 85, 63, 25, and 15 % at 6 months, 1 year, 3 years, and 5 years, respectively, with a median survival of 17 months. Multivariate Cox proportional hazard modeling demonstrated lymph node ratio was negatively correlated with survival at all time points. Preoperative hypertension was a poor prognostic factor at 6 months, 3 years, and 5 years. The absence of postoperative complications was protective at 6 months whereas pancreatic leaks were associated with worse survival at 6 months. Abdominal pain on presentation, operative time, and estimated blood loss were also associated with decreased survival at various time points. Conclusion The strongest prognostic variable for short- and long-term survival after PD for PDA is lymph node ratio. Short-term survival is influenced by the postoperative course.
This study was presented at the 54th Annual Meeting of the Society for Surgery of the Alimentary Tract, Digestive Diseases Week 2013, Poster Presentation and the 47th Annual Pancreas Club Meeting, 2013, Poster of Distinction. C. E. Weber : E. A. Bock : M. G. Hurtuk : G. J. Abood : J. Pickleman : G. V. Aranha Department of Surgery, Division of Surgical Oncology, Loyola University, Maywood, IL, USA G. V. Aranha Department of Surgery, Division of Surgical Oncology, Hines VA Hospital, Hines,, IL, USA M. Shoup Department of Surgery, Division of Surgical Oncology, Cadence Health, Winfield, IL, USA G. V. Aranha (*) Division of Surgical Oncology, Loyola University Medical Center, 2160 S First Ave, Maywood, IL 60153, USA e-mail: [email protected]
Keywords Pancreaticoduodenectomy . Pancreatic ductal adenocarcinoma . Lymph node ratio
Introduction Pancreatic cancer is the fourth leading cause of cancer death in the USA. Contrary to the trends of many solid tumors, the incidence and the death rate for pancreatic cancer are slowly rising. The incidence of pancreatic cancer is rising by about 1.5 % per year and the death rate increased by 0.4 % between 2004 and 2008. Lifetime risk for pancreatic cancer in the USA is 1 in 78 (1.5 %). For the year 2013, it is estimated that 45,220 new cases of pancreatic cancer will be diagnosed and 38,460 individuals will die from their disease.1,2 Unfortunately, the majority of patients present with advanced stage disease. Thus, only about 20 % are eligible for surgical resection at the time of diagnosis.3–7 Despite recent improvements in systemic treatment strategies,
J Gastrointest Surg (2014) 18:340–347
341
aggressive surgical resection with negative margins remains the only possible chance for cure as systemic chemotherapy has yet to dramatically affect survival. Overall 5-year survival for all comers with pancreatic cancer remains dismal, at less than 5 %.3,8–10 In contrast, the median survival after surgical resection is likely somewhere between 13 and 28 months, with the 5-year survival rates for patients after surgical resection ranging from 4 to 24 % (see Table 1).3–18 The purpose of this study was to define the short- and longterm survival rates as well as characterize the postoperative course of a cohort of patients that underwent pancreaticoduodenectomy (PD) for pancreatic ductal adenocarcinoma (PDA) at a single institution between 1993 and 2010. We also sought to determine clinicopathological features that were associated with survival in our cohort of patients.
Methods Following IRB approval, a retrospective review was conducted of a prospectively maintained database of all
patients undergoing PD for PDA at a single institution from 1993 through 2010. Patients who underwent PD for any other pathology were excluded from this study. Electronic medical records and paper charts were reviewed for demographic data, preoperative symptoms, physical exam findings, and patient comorbidities. Operative reports and anesthesia records were viewed to document intraoperative blood loss, intraoperative transfusions, and operative time. Final pathology information, including TNM stage, the presence or absence of perineural or lymphovascular invasion, as well as final margin status was documented from pathology reports. Postoperative information such as length of hospital stay, and the occurrence of postoperative complications, as graded by the Dindo–Clavien classification, was also recorded.19–21 Pancreatic leak was defined as a drain volume of greater than 50 cm3 of amylase-rich fluid or a drain amylase level more than three times the serum amylase on or after the third postoperative day after the intake of a general diet The patient was considered to have undergone a reoperation if they returned to the operating room within 30 days after the original surgery, either on the same admission or on readmission for a complication from the
Table 1 Retrospective studies of survival and prognostic factors in pancreatic cancer Author
Year
Patients undergoing resection (n)
Whipple (%)
Median survival (months)
5-year survival (%)
Prognostic factors on multivariate analysis
Wenger et al. Berger et al.
2000 2003
158 128
91 100
13.8 –
11.8 by LNR (12.0, 16.0, 0)
tumor size, margins lymph node ratio, CA 19–9, margins, adjuvant therapy
Howard et al.
2006
226
90
13
4.0
Sierzega et al.
2006
96
100
14.2
–
Winter et al.
2006
1,175
100
18
18.0
House et al.
2007
696
86
16.0
Pawlik et al. Slidell et al.
2007 2007
905 4,005
100 74
by N stage (27, 16) 17.4 13
tumor size, tumor grade, margins, postoperative complications tumor grade, N stage, margins, lymph node ratio tumor size, N stage, margins, tumor grade, COPD, bile leak, adjuvant therapy N stage, tumor grade, margins
16.1 6.8
Kazanjian et al.
2008
182
100
28.5
24.0
Ueda et al.
2008
140
71
14.5
12.3
Riediger et al. You et al. Bhatti et al. La Torre et al. Ferrone et al. Wentz et al.
2009 2009 2010 2011 2012 2012
204 219 84 101 499 169
85 100 100 100 84 89
18 14 22 19 19 15.1
15.0 14.2 13.0 17.0 19.0 –
age, tumor size, tumor grade, lymph node ratio tumor size, local extension, tumor grade, adjuvant therapy, marital status, N stage, lymph node ratio estimated blood loss, tumor grade, N stage, perineural invasion, operation before 1996 CA 19–9, tumor size, venous involvement, blood transfusion requirement, serum albumin on postoperative month number 1 margins, lymph node ratio, tumor grade tumor size, tumor grade lymph node ratio lymph node ratio, tumor grade, margins margins, N stage lymph node ratio, CA 19-9
342
original surgery. The 30-day readmission rate was determined by patient records for readmission to the institution as well as by viewing follow-up progress notes to ascertain whether or not the patient had been admitted to an outside institution within 30 days after surgery. Overall survival was determined by searching the Social Security Death Index for alive or deceased status and date of death. For all patients, a traditional PD was performed. In over half of cases (56 %), the pancreatic remnant was managed with a pancreaticogastrostomy (PG), with pancreaticojejunostomy representing the remainder of the cohort. The PG reconstruction by this primary surgeon has been extensively reported in the literature.22,23 Statistical analysis was performed using Stata12.0 (College Station, TX). For each survival endpoint (6 months, 1 year, 3 years, and 5 years), a univariate Cox proportional hazards analysis of each clinicopathological variable was performed. The significant variables for each survival endpoint were then combined and evaluated in a backward stepwise Cox proportional hazards model. The final multivariate model contains the significant variables determined by the backward stepwise selection. A p value of 5 Units (see Table 3). Pathologic description of the final specimen demonstrated perineural invasion in 83.3 % (205/232) and lymphovascular
J Gastrointest Surg (2014) 18:340–347 Table 2 Patient clinical and pathological characteristics (n=246)
Variable Age (years) Median (range) Sex Male Female
No. of patients (%)
68 (30–88) 128 (52) 118 (48)
Lymphovascular invasion Present 133 (54.1) Absent 93 (37.8) Not documented 20 (8.1) Perineural invasion Present 205 (83.3) Absent 27 (11) Not documented 14 (5.7) Resection margin R0 161 (65.4) R1 76 (30.9) Not documented 9 (3.7) Lymph node ratio 0 61 (24.8) >0 to 0.2 114 (46.3) >0.2 to 0.4 33 (13.4) >0.4 30 (12.2) Not documented 8 (3.3)
invasion in 54.1 % (133/226). The median number of lymph nodes sampled was 19. The predominant T stage was T3, in 72 % (174/241) of patients, while T1, T2, and T4 were diagnosed in 6.6 % (16/241), 20.8 % (50/241), and 0.4 % (1/241) of patients, respectively, with N1 disease representing 76.7 % (184/240) of nodal status. Based on the number of positive lymph nodes and the number of total lymph nodes
Fig. 1 Kaplan–Meier curve of overall survival in our cohort of 246 patients who underwent pancreaticoduodenectomy for pancreatic ductal adenocarcinoma
J Gastrointest Surg (2014) 18:340–347
343
Table 3 Operative factors and postoperative outcomes 3
Intraoperative blood loss (cm ) Median (range) Operative time (hours) Median (range) Intraoperative blood transfusions (n; %) 0 1 to 5 >5 Not documented Length of stay (days) Median (range) Complication (n; %) None Any complication Dindo–Clavien Classification Grade 1 Grade 2 Grade 3a or 3b Grade 4a or 4b Grade 5 Data missing Delayed gastric emptying Pancreatic leak Reoperation (n; %) 30-day readmission (n; %)
800 (75–10,000) 6.25 (3.5–15.5) 152 (61.8) 86 (35) 7 (2.8) 1 (0.4) 8 (5–80) 145 (58.9) 101 (41.1) 19 (18.8) 47 (46.5) 12 (11.9) 13 (12.9) 6 (5.9) 4 (4) 28 (11.4) 16 (6.5) 7 (2.8) 23 (9.8)
found in the specimen, we computed a lymph node ratio. Among our patients, 24.8 % (61/238) of patients had a lymph node ratio of 0, 46.3 % (114/238) of patients had a lymph node ratio of >0 to 0.2, 13.4 % (33/238) of patients had a lymph node ratio of >0.2 to 0.4, and 12.2 % (30/238) of patients had a lymph node ratio of >0.4. Due to missing information, we were unable to calculate a lymph node ratio for 3.3 % (8/246) of patients. An R0, or microscopically negative resection, was achieved in 65.4 % (161/237) of patients (see Table 2). Postoperatively, the median length of stay was 8 days, with a range of 5 to 80 days. In the postoperative period, as classified by the Dindo–Clavien classification, the majority of patients experiencing complications were considered Grade 1 (18.8 % or 19/101) or Grade 2 (46.5 % or 47/101). Grade 3 complications occurred in 11.9 % (12/101) of patients. There were 12.9 % (13/101) that were classified as Grade 4. Among our cohort of 246 patients, we documented 28 cases of delayed gastric emptying, representing 11.4 %. With regards to anastomotic leaks, there were 16 (6.5 %) pancreatic leaks, 3 (1.2 %) biliary leaks, and 1 (0.4 %) gastrojejunostomy leak. A total of seven patients underwent reoperation within 30 days of the original surgery. Indications for reoperation included: complicated infections, persistent efferent limb obstruction, wound dehiscence and evisceration, and
incomplete removal of JP. Two patients that died within 30 days after surgery underwent reoperation prior to their death. In addition to the above two patients, there were four other patients that died within 30 days of surgery, for a total of 6 (5.9 %) Grade 5 complications, or a 30-day postoperative mortality rate of 2.4 % (6/246) in this cohort. Using univariate Cox proportional hazard modeling, we analyzed patient demographics, preoperative symptoms and comorbidities, operative factors, final pathologic data, and postoperative complications individually to determine associations between these factors and survival at 6 months, 1 year, 3 years, and 5 years. For each time period, these significant univariate variables were then combined into a multivariate model and analyzed by a backward selection Cox proportional hazards model (see Tables 4, 5, 6, and 7). At 6 months, on univariate analysis, female gender was associated with a worse survival (hazard ratio (HR) 2.00, p= 0.046). In addition, the presence of preoperative hypertension was negatively correlated with survival (HR 2.54, p=0.016). Several operative and postoperative complications were statistically correlated with survival at this short-term survival time point. Estimated blood loss (HR 1.00, p=0.001) and operative time (HR 1.22, p=0.04) was statistically significant in the short term. The absence of any postoperative complication was protective, and these patients were more likely to be alive at 6 months (HR of 0.3, p=0.001). In contrast, the presence of a postoperative wound infection (HR 3.16, p=0.017) or postoperative pancreatic anastomotic leak (HR 3.50, p=0.005) were negatively correlated with survival at 6 months. Finally, lymph node ratio was negatively
Table 4 Factors influencing 6-month survival Parameters influencing survival
Hazard ratio (95 % CI)
Univariate analysis (6-month survival) Female gender 2.00 (1.01 to 3.94) Preoperative hypertension 2.54 (1.19 to 5.40) Estimated blood loss 1.00 (1.00 to 1.00) Operative time 1.22 (1.01 to 1.47) Lymph node ratio 4.65 (1.15 to 18.78) No postoperative complication 0.30 (0.15 to 0.60) Wound infection 3.16 (1.23 to 8.12) Pancreatic leak 3.50 (1.46 to 8.42) Multivariate Analysis (6-month survival) Female gender 2.49 (1.15 to5.39) Preoperative hypertension 2.62 (1.16 to 5.91) Estimated blood loss 1.00 (1.00 to 1.00) Lymph node ratio 15.88 (3.22 to 78.42) Pancreatic leak 2.90 (1.08 to 7.81) No postoperative complication 0.39 (0.18 to 0.87) *Denotes statistical significance at p
