Abstract
Background
Select patients with peri-ampullary cancers require concomitant colon resection (CR) during a pancreaticoduodenectomy (PD) for margin-negative resections. This study analysed the impact of concomitant CR on major morbidity (MM) and mortality.
Methods
National Surgical Quality Improvement Program (NSQIP) patients undergoing PD for peri-ampullary cancers were identified from 2005 to 2012. A 4: 1 propensity-score matched analysis isolated the impact of CR upon PD. Risk factors for 30-day MM and mortality were analysed to determine post-operative sequelae of PD+CR.
Results
From 10 965 PD and 159 PD+CR patients, 624 and 156, respectively, were selected for 4: 1 matched analysis. PD+CR resulted in a higher MM and mortality (50.0% and 9.0%) versus PD alone (28.8% and 2.9%, respectively, P < 0.001). Multivariate analysis identified risk factors for MM after PD: concomitant CR [odds ratio (OR)–3.19, P < 0.001], smoking (OR–1.92, P = 0.005), a lack of functional independence (OR–3.29, P = 0.018), cardiac disease (OR–2.39, P = 0.011), decreased albumin (per g/dl, OR–1.38, P = 0.033) and a longer operative time (versus median time, OR–1.56, P = 0.029). Independent predictors of mortality included concomitant CR (OR–3.16, P = 0.010), ventilator dependence (OR–13.87, P < 0.001) and septic shock (OR–6.02, P < 0.001).
Conclusions
CR was an independent predictor of MM and mortality after a PD. Patients requiring PD+CR should be identified pre-operatively, maximally optimized and referred to experienced surgeons at expert centres.
Introduction
A complete surgical resection with a pancreaticoduodenectomy (PD) is a required part of a potentially curative multimodality treatment plan for patients with pancreatic head adenocarcinoma.1,2 Despite improvements in neoadjuvant and adjuvant therapies, and surgical techniques, only 15% to 20% of patients are candidates for a resection owing to the presence of a distant tumour outside the confines of a resection, or because of locally advanced disease2–6. Select patients with peri-ampullary cancers require a multi-organ resection during a PD to achieve margin-negative resections7–12. A margin-negative, or R0, resection is critical as better oncological outcomes are associated with a successful R0 resection4,13–15. However, this aggressive approach with a multivisceral resection with PD may result in an increase in major morbidity and/or mortality, leading to delays in adjuvant therapy and even a decreased survival. Previous studies have shown that up to one-third of patients undergoing a PD have significant medical comorbidities and are less likely to be rescued from major post-operative complications, which suggests that surgeons should identify and optimize these comorbidities and utilize pre-habilitation to address functional deficits before an elective PD16.
Previous single-institution studies have suggested that PD can be safely combined with concomitant organ and/or vein resection without a significant difference in outcomes9–12,17. However, these studies contained highly selected patients operated upon at expert centres, which does not necessarily reflect general outcomes nationally. Within this context, this study sought to characterize the true major morbidity (MM) and mortality associated with PD+CR in a large national dataset. The primary aim was to compare national rates of post-PD MM and mortality in patients undergoing PD versus PD+CR. The secondary aims were to identify risk factors for morbidity and mortality in PD patients to isolate the independent impact of CR and to determine strategies surgeons may utilize to optimize correctable preoperative risk factors and improve patient outcomes.
Patients and methods
Patient and data collection
The American College of Surgeons National Surgical Improvement Program (ACS-NSQIP) is a nationally validated, risk-adjusted, outcomes-based programme to measure and improve the quality of surgical care using risk-adjusted clinical data from public and private hospitals. Analyses of NSQIP have demonstrated high quality and reproducible data18. Participating NSQIP hospitals and their researchers have full access to the Participant Use Files (PUF) containing de-identified data from all participating sites. As NSQIP patient information is de-identified, this study was exempt from University of Kentucky Institutional Review Board approval. All patients undergoing a PD for peri-ampullary cancers, with and without CR, from the 2005–2012 NSQIP database, were initially screened using current procedural terminology (CPT) codes 48150, 48152, 48153 and 48154. Patients with concurrent CR were identified by CPT codes 44140, 44141, 44143, 44144, 44145, 44146, 44147, 44150, 44156, 44157, 44158 and 44160 (Table S1). Patients undergoing a concurrent vein resection were identified by CPT codes 35221, 35251, 35281, 35531 and 35631.
Patients without a post-operative diagnosis of malignancy were excluded by applying International Classification of Diseases, 9th edition (ICD-9) codes to select only patients with peri-ampullary cancers as follows: 152, 155.1, 156,156.1, 156.2, 156.9, 157.0, 157.1, 157.2, 157.3, 157.4, 157.8, 157.9, 209.01, 209.3, 235.2, 235.5, 239 and 258.01.
Propensity-score matched analysis
All pre-, intra- and post-operative NSQIP variables were abstracted for analysis. Propensity score matching was used to mitigate the inherent selection bias of a retrospective database analysis. Using all preoperative variables, a propensity score was generated for each patient. Logistic regression was used to generate this score as a likelihood of receiving PD versus PD+CR. A 1:N matching macro assigned matched PD versus PD+CR cohorts based on the propensity score. Initially, this was performed as a 1:1 match; however, the size of the PD cohort was increased in a stepwise manner to maximize the available patients, while maintaining non-significant differences between the preoperative variables. This technique converged at a 4:1 match as optimal for this dataset.
Definitions and variables analysed
Pre-operative risk factors identified were American Society of Anesthesiology (ASA) class of ≥ 3, diabetes, 10% weight loss in 6 months, current smoker, current alcohol use, a lack of functional dependence, body mass index (BMI) (kg/m2), a history of chronic obstructive pulmonary disorder (COPD), dyspnea, a history of cardiac disease, a history of hypertension, albumin <4 g/dl, haematocrit (Hct) < 39% and pre-operative chemotherapy and radiation treatment. Intra-operative risk factors included a vein resection, total operation time (minutes, median), operative time > 370 min, peri-operative transfusion and return to the operating room (OR). Post-operative complications evaluated were wound infections/complications including superficial surgical site infection (SSI), deep SSI, organ space infection (OSI) and wound dehiscence. Other post-operative complications included pneumonia, failure to wean from the ventilator, re-intubation, cardiac arrest, a post-operative myocardial infarction (MI), sepsis, septic shock and the total length of stay (LOS). MM was defined as any of the following: pneumonia, re-intubation, ventilator dependence, renal insufficiency/failure, cardiac events, neurological events, sepsis/septic shock, return to the OR, dehiscence, organ space infection and venous thromboembolism. Mortality was defined as death within 30 days or the index hospitalization.
Statistical analyses
Univariate analyses were performed using non-parametric tests to compare risk factors for 30-day major morbidity and mortality in the two cohorts. Categorical variables were reported as number (n, %) and compared using a chi-squared or Fisher's Exact test as appropriate. Non-parametric continuous variables were reported with medians (range) and compared using the Mann-Whitney U-test or anova as appropriate. Multivariate analyses were performed using logistic regression. Candidate models were chosen using stepwise forward selection and backward elimination to isolate the impact of CR on PD. The MM model adjusted for all pre-operative variables, operative time variables and procedure type. The mortality model adjusted for all pre-operative variables, post-operative variables including morbidity variables and procedure type. Models were adjusted for multiple comparisons and subgroups. Analyses were conducted using SAS© (version 9.3; SAS Institute Inc., Cary, NC, USA). All tests were two-sided with significance defined as P < 0.05.
Results
Patients
ACS- NSQIP data extraction identified 10 965 patients undergoing PD and 159 patients undergoing PD+CR in total. Their clinical characteristics are detailed in Table 1. The two populations are closely, but not completely, similar in their preoperative characteristics. Significant intra-operative differences between the two groups included all intra-operative variables except the need for vein resection (P = 0.067).
Table 1.
Demographic and peri-operative variables in unmatched patients undergoing PD and PD+CR
| PD Only | PD +CR | Total | P-value | |
|---|---|---|---|---|
| (N = 10 965) | (N = 159) | (N = 11 124) | ||
| Demographic variables | ||||
| Race | ||||
| Caucasian | 7061 (64.4%) | 96 (60.4%) | 7157 (64.3%) | 0.294 |
| Non-Caucasian | 3904 (35.6%) | 63 (39.6) | 3967 (35.7%) | |
| Age (years, median) | 66.0 | 64.0 | 66.0 | 0.015 |
| 18–74 years | 58.0–74.0 | 54.0–73.0 | 58.0–74.0 | |
| 75 years and greater | 2631 (24.1%) | 34 (21.5%) | 2665 (24.1%) | 0.452 |
| Gender | ||||
| Male | 5767 (52.7%) | 95 (59.7%) | 5862 (52.8%) | 0.075 |
| Female | 5198 (47.3%) | 64 (40.3%) | 5262 (47.2%) | |
| ASA class 3 or higher | 7898 (72.0%) | 121 (76.1%) | 8019 (72.1%) | 0.256 |
| Diabetes | 2084 (20.2%) | 25 (16.9%) | 2109 (20.1%) | 0.32 |
| 10% weight loss in 6 months | 2137 (19.5%) | 39 (24.5%) | 2176 (19.6%) | 0.112 |
| Smoker | 2145 (19.6%) | 38 (23.9%) | 2183 (19.6%) | 0.172 |
| Alcohol use | 245 (3.1%) | 3 (2.6%) | 248 (3.1%) | >0.999 |
| Lack of functional independence | 263 (2.4%) | 4 (2.5%) | 267 (2.4%) | 0.794 |
| BMI (kg/m2, median) | 26.2 | 24.5 | 26.2 | 0.007 |
| COPD | 493 (4.5%) | 2 (1.3%) | 495 (4.4%) | 0.049 |
| Dyspnea | 881 (8.0%) | 14 (8.8%) | 895 (8.0%) | 0.723 |
| Cardiac disease | 906 (8.3%) | 10 (6.3%) | 916 (8.2%) | 0.369 |
| Hypertension | 6071 (55.4%) | 75 (47.2%) | 6146 (55.2%) | 0.039 |
| Albumin <4 g/dl | 6901 (69.1%) | 113 (76.9%) | 7014 (69.2%) | 0.043 |
| Hct<39% | 6805 (63.3%) | 118 (75.2%) | 6923 (63.4%) | 0.002 |
| Chemotherapy | 306 (3.9%) | 4 (3.5%) | 310 (3.9%) | 0.999 |
| Radiation treatment | 323 (4.1%) | 2 (1.8%) | 325 (4.1%) | 0.333 |
| Perioperative variables | ||||
| Intraoperative characteristics | ||||
| Vein Resection required | 276 (2.5%) | 8 (5.0%) | 284 (2.6%) | 0.067 |
| Total operation time (min, median) | 359.0 | 426.0 | 360.0 | <0.001 |
| Operative time > 370 min | 6634 (60.5%) | 119 (74.8%) | 6753 (60.7%) | <0.001 |
| Perioperative transfusion | 1711 (15.7%) | 41 (26.3%) | 1752 (15.8%) | <0.001 |
| Return to OR | 735 (6.7%) | 20 (12.6%) | 755 (6.8%) | 0.003 |
| Wound infection/complication | ||||
| Superficial SSI | 1125 (10.3%) | 28 (17.6%) | 1153 (10.4%) | 0.003 |
| Deep SSI | 279 (2.5%) | 1 (0.6%) | 280 (2.5%) | 0.194 |
| Organ space infection | 1144 (10.4%) | 36 (22.6%) | 1180 (10.6%) | <0.001 |
| Wound dehiscence | 216 (2.0%) | 4 (2.5%) | 220 (2.0%) | 0.559 |
| Postoperative complications | ||||
| Pneumonia | 520 (4.7%) | 11 (6.9%) | 531 (4.8%) | 0.201 |
| Failure to wean from ventilator | 580 (5.3%) | 20 (12.6%) | 600 (5.4%) | <0.001 |
| Re-intubation | 562 (5.1%) | 15 (9.4%) | 577 (5.2%) | 0.015 |
| Cardiac arrest | 139 (1.3%) | 6 (3.8%) | 145 (1.3%) | 0.018 |
| Post-operative MI | 90 (0.8%) | 3 (1.9%) | 93 (0.8%) | 0.148 |
| Sepsis | 1117 (10.2%) | 35 (22.0%) | 1152 (10.4%) | <0.001 |
| Septic Shock | 488 (4.5%) | 17 (10.7%) | 505 (4.5%) | <0.001 |
| Total LOS (days, median) | 10.0 | 13.0 | 10.0 | <0.001 |
| Major morbidity | 2949 (26.9%) | 80 (50.3%) | 3029 (27.2%) | <0.001 |
| 30-day mortality | 304 (2.8%) | 14 (8.8%) | 318 (2.9%) | <0.001 |
PD, pancreaticoduodenectomy; CR, colon resection; ASA, American Society of Anesthesiology; BMI, body mass index; COPD, chronic obstructive pulmonary disorder; Hct, hematocrit; OR, operating room; SSI, surgical site infection; MI, myocardial infarction; LOS, length of stay.
A total of 780 patients met inclusion criteria for the 4:1 propensity-score matched analysis (PD=624, PD+CR=156, Table 2). With the macro-generated propensity score, the matched cohorts were well matched based on patient demographics and pre-operative risk factors including administration of chemotherapy and radiation. The only intra-operative variable that was statistically different between the groups was total operative time (P < 0.001).
Table 2.
Demographic variables of 4: 1 propensity-score matched population undergoing PD and PD+CR
| PD | PD+CR | Total | P-value | |
|---|---|---|---|---|
| (N = 624) | (N = 156) | (N = 780) | ||
| Race | ||||
| Caucasian | 377 (60.4%) | 95 (60.9%) | 472 (60.5%) | 0.913 |
| Non-Caucasian | 247 (39.6%) | 61 (39.1%) | 308 (39.5%) | |
| Age (years, median) | 64.0 | 64.5 | 64.0 | 0.699 |
| 18–74 years | 56.0–73.0 | 54.0–73.0 | 55.0–73.0 | |
| 75 years and greater | 129 (20.7%) | 34 (21.8%) | 163 (20.9%) | 0.758 |
| Gender | ||||
| Male | 375 (60.1%) | 94 (60.3%) | 469 (60.1%) | 0.971 |
| Female | 249 (39.9%) | 62 (39.7%) | 311 (39.9%) | |
| ASA class 3 or higher | 448 (71.8%) | 118 (75.6%) | 566 (72.6%) | 0.336 |
| Diabetes | 116 (19.9%) | 24 (16.6%) | 140 (19.2%) | 0.365 |
| 10% weight loss in 6 months | 126 (20.2%) | 38 (24.4%) | 164 (21.0%) | 0.253 |
| Smoker | 138 (22.1%) | 36 (23.1%) | 174 (22.3%) | 0.796 |
| Alcohol use | 14 (3.0%) | 3 (2.7%) | 17 (3.0%) | >0.999 |
| Lack of functional independence | 21 (3.4%) | 4 (2.6%) | 25 (3.2%) | 0.609 |
| BMI (kg/m2, median) | 25.3 | 24.7 | 25.3 | 0.99 |
| COPD | 24 (3.8%) | 2 (1.3%) | 26 (3.3%) | 0.111 |
| Dyspnea | 51 (8.2%) | 14 (9.0%) | 65 (8.3%) | 0.746 |
| Cardiac disease | 41 (6.6%) | 9 (5.8%) | 50 (6.4%) | 0.715 |
| Hypertension | 329 (52.7%) | 72 (46.2%) | 401 (51.4%) | 0.142 |
| Albumin <4 g/dl | 420 (74.1%) | 113 (77.4%) | 533 (74.8%) | 0.41 |
| Hct<39% | 454 (72.8%) | 118 (75.6%) | 572 (73.3%) | 0.466 |
| Chemotherapy | 22 (4.8%) | 4 (3.6%) | 26 (4.5%) | 0.601 |
| Radiation treatment | 17 (3.7%) | 2 (1.8%) | 19 (3.3%) | 0.553 |
| Peri-operative variables | ||||
| Intra-operative characteristics | ||||
| Vein Resection required | 18 (2.9%) | 8 (5.1%) | 26 (3.3%) | 0.163 |
| Total operation time (min, median) | 360.5 | 420.0 | 370.0 | <0.001 |
| Operative time > 370 min | 374 (59.9%) | 117 (75.0%) | 491 (62.9%) | <0.001 |
| Perioperative transfusion | 117 (18.9%) | 39 (25.5%) | 156 (20.2%) | 0.071 |
| Return to OR | 59 (9.5%) | 19 (12.2%) | 78 (10.0%) | 0.31 |
| Wound infection/complication | ||||
| Superficial SSI | 62 (9.9%) | 28 (17.9%) | 90 (11.5%) | 0.005 |
| Deep SSI | 22 (3.5%) | 1 (0.6%) | 23 (2.9%) | 0.063 |
| Organ space infection | 69 (11.1%) | 35 (22.4%) | 104 (13.3%) | <0.001 |
| Wound dehiscence | 15 (2.4%) | 4 (2.6%) | 19 (2.4%) | >0.999 |
| Post-operative complications | ||||
| Pneumonia | 32 (5.1%) | 11 (7.1%) | 43 (5.5%) | 0.347 |
| Failure to wean from ventilator | 43 (6.9%) | 19 (12.2%) | 62 (7.9%) | 0.029 |
| Reintubation | 43 (6.9%) | 15 (9.6%) | 58 (7.4%) | 0.246 |
| Cardiac arrest | 10 (1.6%) | 6 (3.8%) | 16 (2.1%) | 0.107 |
| Post-operative MI | 6 (1.0%) | 3 (1.9%) | 9 (1.2%) | 0.394 |
| Sepsis | 59 (9.5%) | 34 (21.8%) | 93 (11.9%) | <0.001 |
| Septic Shock | 37 (5.9%) | 17 (10.9%) | 54 (6.9%) | 0.029 |
| Total LOS (days, median) | 10.0 | 13.0 | 11.0 | <0.001 |
| Major morbidity | 180 (28.8%) | 78 (50.0%) | 258 (33.1%) | <0.001 |
| 30-day mortality | 18 (2.9%) | 14 (9.0%) | 32 (4.1%) | <0.001 |
PD, Pancreaticoduodenectomy; CR, colon resection; ASA, American Society of Anesthesiology; BMI, body mass index; COPD, chronic obstructive pulmonary disorder; Hct, hematocrit; OR, operating room; SSI, surgical site infection; MI, myocardial infarction; LOS, length of stay.
Univariate analysis of matched cohorts
PD+CR resulted in a significantly higher MM and mortality (50.0% and 9.0%) versus PD alone (28.8% and 2.9%, respectively, P < 0.001) (Fig. 1). In the matched cohort (n = 156), 80 patients had a MM and 14 patients died (Table 3). Post-operative complications associated with mortality on univariate analysis in this group included re-intubation (42.9%, P < 0.001), cardiac arrest (35.7%, P < 0.001), septic shock (50.0%, P < 0.001) and MM (92.9%, P < 0.001).
Figure 1.
PD+CR resulted in higher major morbidity and mortality, 50.0% and 9.0%, respectively, versus PD alone 28.8% and 2.9%, respectively, P < 0.001. Major morbidity and mortality of pancreaticoduodenectomy (PD) with or without colon resection (CR)
Table 3.
Univariate analysis of major morbidity and mortality in patients with PD+CR (N = 156)
| Morbidity | P-value | Mortality | P-value | |
|---|---|---|---|---|
| N = 80 | N = 14 | |||
| Demographic variables | ||||
| Race | ||||
| Caucasian | 43 (53.8%) | 0.086 | 8 (57.1%) | 0.796 |
| Non-Caucasian | 37 (46.2%) | 6 (42.9%) | ||
| Age (years, median) | 63.0 | 0.25 | 72.0 | 0.097 |
| 18–74 years | 53.0–73.0 | 61.0–77.0 | ||
| 75 years and greater | 16 (20.3%) | 0.699 | 5 (35.7%) | 0.183 |
| Gender | ||||
| Male | 50 (62.5%) | 0.476 | 10 (71.4%) | 0.351 |
| Female | 30 (37.5%) | 4 (28.6%) | ||
| ASA class 3 or higher | 66 (82.5%) | 0.057 | 12 (85.7%) | 0.521 |
| Diabetes | 13 (18.1%) | 0.713 | 2 (16.7%) | >0.999 |
| 10% weight loss in 6 months | 17 (21.3%) | 0.334 | 7 (50.0%) | 0.044 |
| Smoker | 23 (28.8%) | 0.149 | 4 (28.6%) | 0.744 |
| Alcohol use | 3 (4.8%) | 0.249 | 0 | >0.999 |
| Lack of functional independence | 77 (96.3%) | 0.62 | 2 (14.3%) | 0.039 |
| BMI (kg/m2, median) | 25.1 | 0.611 | 23.4 | 0.164 |
| COPD | 1 (1.3%) | >0.999 | 0 | >0.999 |
| Dyspnea | 8 (10.0%) | 0.593 | 3 (21.4%) | 0.111 |
| History of cardiac disease | 8 (10.0%) | 0.098 | 1 (7.1%) | >0.999 |
| History of hypertension | 37 (46.3%) | 0.815 | 7 (50.0%) | 0.824 |
| Albumin <4 g/dl | 53 (70.7%) | 0.069 | 11 (78.6%) | >0.999 |
| Hct<39% | 57 (72.2%) | 0.38 | 10 (71.4%) | 0.749 |
| Chemotherapy | 4 (6.5%) | 0.124 | 0 | >0.999 |
| Radiation treatment | 2 (3.2%) | 0.499 | 0 | >0.999 |
| Peri-operative variables | ||||
| Intra-operative characteristics | ||||
| Vein Resection required | 4 (5.0%) | >0.999 | 2 (14.3%) | 0.148 |
| Total operation time (min, median) | 432.5 | 0.521 | 458.0 | 0.194 |
| Operative time > 370 min | 63 (78.8%) | 0.253 | 10 (71.4%) | 0.752 |
| Peri-operative transfusion | 24 (31.2%) | 0.171 | 4 (36.4%) | 0.481 |
| Return to OR | 20 (25.0%) | <0.001 | 7 (50.0%) | <0.001 |
| Wound infection/complication | ||||
| Superficial SSI | 17 (21.3%) | 0.225 | 1 (7.1%) | 0.467 |
| Deep SSI | 1 (1.3%) | >0.999 | 0 | >0.999 |
| Organ space infection | 36 (45.0%) | <0.001 | 6 (42.9%) | 0.088 |
| Wound dehiscence | 4 (5.0%) | 0.12 | 1 (7.1%) | 0.311 |
| Post-operative complications | ||||
| Pneumonia | 11 (13.8%) | <0.001 | 3 (21.4%) | 0.059 |
| Failure to wean from ventilator | 20 (25.0%) | <0.001 | 4 (28.6%) | 0.08 |
| Re-intubation | 15 (18.8%) | <0.001 | 6 (42.9%) | <0.001 |
| Cardiac arrest | 6 (7.5%) | 0.028 | 5 (35.7%) | <0.001 |
| Post-operative MI | 3 (3.8%) | 0.245 | 1 (7.1%) | 0.243 |
| Sepsis | 35 (43.8%) | <0.001 | 1 (7.1%) | 0.307 |
| Septic Shock | 17 (21.3%) | <0.001 | 7 (50.0%) | <0.001 |
| Total LOS (days, median) | 17.0 | <0.001 | 13.5 | 0.4 |
| Major morbidity | 80 (100.0%) | <0.001 | 13 (92.9%) | <0.001 |
| 30-day mortality | 13 (16.3%) | <0.001 | 14 (100.0%) | <0.001 |
PD, Pancreaticoduodenectomy; CR, colon resection; ASA, American Society of Anesthesiology; BMI, body mass index; COPD, chronic obstructive pulmonary disorder; Hct, hematocrit; OR, operating room; SSI, surgical site infection; MI, myocardial infarction; LOS, length of stay.
Multivariate analyses
Multivariate analysis identified the following independent risk factors for MM after PD: active smoking (OR–1.92, P = 0.005), the lack of functional independence (OR–3.29, P = 0.018), cardiac disease (OR–2.39, P = 0.011), a decreased albumin (per g/dl, OR–1.38, P = 0.033), a longer operative time (versus median time 370 min, OR–1.56, P = 0.029) and a concomitant CR (OR–3.19, P < 0.001, Table 4). Independent predictors of mortality included ventilator dependence (OR–13.87, P < 0.001), septic shock (OR–6.02, P < 0.001), and concomitant CR (OR-3.16, P = 0.010, Table 5).
Table 4.
Independent predictors of major morbidity after PD+CR
| Independent predictor | Odds Ratio | 95% Confidence Interval | P-value |
|---|---|---|---|
| Smoker | 1.92 | 1.22–3.03 | 0.005 |
| Lack of functional independence | 3.29 | 1.23–8.84 | 0.018 |
| Cardiac disease | 2.39 | 1.22–4.67 | 0.011 |
| Operative time >370 min | 1.56 | 1.05–2.34 | 0.029 |
| Decreased albumin | 1.38 | 1.03–1.85 | 0.033 |
| Concomitant colon resection | 3.19 | 2.00–5.10 | <0.001 |
PD, Pancreaticoduodenectomy; CR, colon resection.
Table 5.
Independent predictors of mortality after PD+CR
| Independent predictor | Odds Ratio | 95% Confidence Interval | P-value |
|---|---|---|---|
| Ventilator dependence | 13.87 | 5.08–37.87 | <0.001 |
| Septic shock | 6.02 | 2.29–15.87 | <0.001 |
| Concomitant colon resection | 3.16 | 1.32–7.55 | 0.001 |
*Based on propensity-score matched analysis with 780 total patients.
PD, Pancreaticoduodenectomy; CR, colon resection. Refer to Table 2.
Discussion
Contrary to past single-institution reports, this study, using a 4: 1 propensity-score matched analysis derived from the ACS-NSQIP database, determined that PD+CR resulted in a significantly higher MM and mortality than PD alone. CR was an independent risk factor for both MM and mortality with PD+CR patients having three times the likelihood of both MM and mortality compared with patients undergoing PD alone. This dramatically increased risk highlights the importance of identifying patients upfront who may need PD+CR so that they can be optimized pre-operatively and possibly referred to surgeons and cancer centres with adequate experience.
Although MM and mortality rates have improved over the past few decades for patients undergoing PD, this well-matched analysis of a national dataset determined that the addition of CR independently increased the risks of both MM and mortality. To date, there have been few studies that have evaluated outcomes for patients undergoing PD+CR. Several single-centre studies have suggested that MM and mortality were comparable between patients undergoing PD+CR versus PD alone, minimizing the clinical significance of adding CR to PD9–12,17. However, these studies had an inherent selection bias as they were single-institution retrospective studies that only reported patients chosen for surgery, leaving out of the analyses those who were anatomically unresectable or physiologically inoperable. This study used a large national database to enhance the statistical power for this rare combination operation, and propensity score matching to mitigate surgeon-patient selection bias. In a large observational study, there are often significant differences between characteristics in the treatment group versus the control group, and these must be adjusted to determine the valid treatment effect19–21. This study used propensity score matching to reduce this observational bias using a 4: 1 matched analysis. Comparing Table 1 (unmatched comparison with differences in preoperative variables) versus Table 2 (matched comparison with balanced preoperative variables) shows how propensity score matching reduces the inherent selection bias of a retrospective database.
Similar to the present study, many previous authors showed the association between morbidity increased with a longer operative time8,11. Hartwig et al. also found that an extended multivisceral resection of 2 or more additional organs as an independent risk factor for intra-abdominal complications or need for re-laparotomy, but did not find a significant difference in mortality after a PD8. Certainly, none of the independent risk factors (besides CR) in the multivariate models for MM and mortality were surprising. The key finding in the current study was that even adjusting for those independent risk factors, including neoadjuvant chemotherapy or radiation, CR significantly increased the risk of MM and mortality7,10,16,22,23. One reason CR may increase MM and mortality is its inherent potential risk for anastomotic leak, infected fluid collection and/or an abscess. We also determined that even in the propensity-matched group, OSI is almost two times higher in the PD+CR group when compared with PD alone (22.4% versus 11.1%, P < 0.001). SSI were also increased in the PD+CR group (17.9% versus 9.9%, P = 0.005). OSI was associated with MM (P < 0.001). However, septic shock was identified as an independent risk factor for mortality, which could be as a result of leak-related OSI after PD+CR. Although some authors may suggest using OSI as a surrogate for leak, previous authors demonstrated that OSI are too low in the NSQIP data file to be used as a surrogate for grade B and C pancreatic fistulae24. Similarly to our analysis, they suggest that procedure-specific variables should be used to extrapolate information regarding outcomes, but this is only applicable with OSI as these data are not available regarding bile leak and/or delayed gastric emptying.
Bhayani et al. utilized the ACS-NSQIP database to evaluated multi-organ resections with PD and determined that CR specifically was an independent predictor of MM, and also found that PD+CR had greater rates of wound, pulmonary, cardiac, thromboembolic, renal and septic complications25. However, those authors did not perform statistical measures to correct for observational and selection bias, and their populations for PD and PD+CR (n = 9 927 versus 273, respectively) were different from the present study. Furthermore, the present study determined that current smoking, a lack of functional independence, cardiac disease, operative time > 370 min, a decreased albumin and concomitant CR were independent predictors of increased MM, highlighting how the surgeon can anticipate which patients are at a greater risk for complications so that these patients can be optimized pre-operatively.14,15,26,27
Previous studies suggest that identification of pre-operative risk factors, especially in patients with multiple co-morbidities or those in the geriatric population, may help identify and offer the opportunity for correction of these risk factors prior to surgery16,28. The present study also identifies multiple risk factors for MM and mortality that can potentially be corrected pre-operatively to improve patient outcomes. Patients can be offered tobacco cessation interventions, pre-habilitation and aggressive immunonutrition, to improve MM rates. Patients who are anticipated to undergo PD+CR may be offered pre-operative counselling focused on the calculated increased risk of MM and mortality. Furthermore, when PD+CR is anticipated, this magnitude of surgery could be classified within the realm of borderline resectable anatomy and could be treated with neoadjuvant therapy, as opposed to surgery upfront, to increase the chance of multimodal therapy completion16. The time that is bought with this treatment sequencing strategy offers a window of opportunity to correct further or optimize risk factors for MM such as nutritional status, performance status and even cardiopulmonary issues.
Potential limitations of this study include its retrospective nature and the use of pooled data from a large national database. This general surgical database cannot account for outcome variables occurring after 30 days such as 90-day post-operative complications, which is increasingly reported for cancer surgery, the impact of complications on adjuvant therapy administration and long-term oncological outcomes. Also, despite using CPT codes to identify PD and PD+CR, the specific reason for CR cannot be proven as ACS-NSQIP does not provide surgical case information regarding colonic mural involvement versus mesenteric involvement alone as an indication for colectomy. However, most clinicians would probably acknowledge that the most common reasons are involvement of the middle colic artery and/or of the right or transverse colon mesentery12,29–31. To that end, we do provide case-specific information regarding the type of colectomy in Table S1 which shows that the most common CR performed with PD is a right colectomy with or without resection of the terminal ileum (CPT 44140 or 44160, respectively). ACS-NSQIP does not include final pathologic diagnosis. However, this study excluded patients without post-operative diagnoses of malignancy by applying ICD-9 codes to select only patients with peri-ampullary cancers. Given that the aim of this study was to analyse short-term outcomes of PD versus PD+CR, the data obtained from ACS-NSQIP applied to the primary research question. Despite these limitations, this study contributes to the current surgical oncology literature as it represents the largest and most current evaluation of the risk factors for 30-day MM and mortality in patients undergoing PD+CR.
Conclusions
Contrary to previous single-institution studies, this propensity-score matched analysis using the ACS-NSQIP database showed that adding CR to PD independently increased the magnitude of surgery as reflected in the three-fold likelihood of post-PD MM and mortality. To improve surgical outcomes, patients who require PD+CR should be identified pre-operatively using high-resolution imaging, maximally optimized nutritionally and physiologically, and ideally referred to experienced surgeons at expert centres.
Funding sources
None.
Conflicts of interest
None to declare.
Supporting information
Additional Supporting Information may be found in the online version of this article:
Table S1. Current Procedural Technology (CPT).
References
- 1.Cameron JL, Riall TS, Coleman J, Belcher KA. One thousand consecutive pancreaticoduodenectomies. Ann Surg. 2006;244:10–15. doi: 10.1097/01.sla.0000217673.04165.ea. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Fernandez-del Castillo C, Morales-Oyarvide V, McGrath D, Wargo JA, Ferrone CR, Thayer SP, et al. Evolution of the Whipple procedure at the Massachusetts General Hospital. Surgery. 2012;152(3 Suppl 1):S56–63. doi: 10.1016/j.surg.2012.05.022. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Harrison LE, Brennan MF. Portal vein involvement in pancreatic cancer: a sign of unresectability? Adv Surg. 1997;31:375–394. [PubMed] [Google Scholar]
- 4.Kwon D, McFarland K, Velanovich V, Martin RC., 2nd Borderline and locally advanced pancreatic adenocarcinoma margin accentuation with intraoperative irreversible electroporation. Surgery. 2014;156:910–920. doi: 10.1016/j.surg.2014.06.058. [DOI] [PubMed] [Google Scholar]
- 5.Halperin DM, Varadhachary GR. Resectable, borderline resectable, and locally advanced pancreatic cancer: what does it matter? Curr Oncol Rep. 2014;16:366. doi: 10.1007/s11912-013-0366-9. [DOI] [PubMed] [Google Scholar]
- 6.Kubota K, Sato T, Watanabe S, Hosono K, Kobayashi N, Mori R, et al. Covered self-expandable metal stent deployment promises safe neoadjuvant chemoradiation therapy in patients with borderline resectable pancreatic head cancer. Dig Endosc. 2014;26:77–86. doi: 10.1111/den.12049. [DOI] [PubMed] [Google Scholar]
- 7.Cirocchi R, Partelli S, Castellani E, Renzi C, Parisi A, Noya G, et al. Right hemicolectomy plus pancreaticoduodenectomy vs partial duodenectomy in treatment of locally advanced right colon cancer invading pancreas and/or only duodenum. Surg Oncol. 2014;23:92–98. doi: 10.1016/j.suronc.2014.03.003. [DOI] [PubMed] [Google Scholar]
- 8.Hartwig W, Hackert T, Hinz U, Hassenpflug M, Strobel O, Buchler MW, et al. Multivisceral resection for pancreatic malignancies: risk-analysis and long-term outcome. Ann Surg. 2009;250:81–87. doi: 10.1097/SLA.0b013e3181ad657b. [DOI] [PubMed] [Google Scholar]
- 9.Kimchi ET, Nikfarjam M, Gusani NJ, Avella DM, Staveley-O'Carroll KF. Combined pancreaticoduodenectomy and extended right hemicolectomy: outcomes and indications. HPB. 2009;11:559–564. doi: 10.1111/j.1477-2574.2009.00087.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Nikfarjam M, Sehmbey M, Kimchi ET, Gusani NJ, Shereef S, Avella DM, et al. Additional organ resection combined with pancreaticoduodenectomy does not increase postoperative morbidity and mortality. J Gastrointest Surg. 2009;13:915–921. doi: 10.1007/s11605-009-0801-2. [DOI] [PubMed] [Google Scholar]
- 11.Sasson AR, Hoffman JP, Ross EA, Kagan SA, Pingpank JF, Eisenberg BL. En bloc resection for locally advanced cancer of the pancreas: is it worthwhile? J Gastrointest Surg. 2002;6:147–157. doi: 10.1016/s1091-255x(01)00063-4. discussion 57-8. [DOI] [PubMed] [Google Scholar]
- 12.Suzuki Y, Fujino Y, Tanioka Y, Sakai T, Ajiki T, Ueda T, et al. Resection of the colon simultaneously with pancreaticoduodenectomy for tumors of the pancreas and periampullary region: short-term and long-term results. World J Surg. 2004;28:1007–1010. doi: 10.1007/s00268-004-7438-9. [DOI] [PubMed] [Google Scholar]
- 13.Frampton AE, Gall TM, Krell J, Ahmad R, Jiao LR. Is there a ‘margin’ for error in pancreatic cancer surgery? Future Oncol. 2013;9:31–4. doi: 10.2217/fon.12.175. [DOI] [PubMed] [Google Scholar]
- 14.Turrini O, Ewald J, Barbier L, Mokart D, Blache JL, Delpero JR. Should the portal vein be routinely resected during pancreaticoduodenectomy for adenocarcinoma? Ann Surg. 2013;257:726–30. doi: 10.1097/SLA.0b013e318269d23c. [DOI] [PubMed] [Google Scholar]
- 15.Carrere N, Sauvanet A, Goere D, Kianmanesh R, Vullierme MP, Couvelard A, et al. Pancreaticoduodenectomy with mesentericoportal vein resection for adenocarcinoma of the pancreatic head. World J Surg. 2006;30:1526–1535. doi: 10.1007/s00268-005-0784-4. [DOI] [PubMed] [Google Scholar]
- 16.Tzeng CW, Katz MH, Fleming JB, Lee JE, Pisters PW, Holmes HM, et al. Morbidity and mortality after pancreaticoduodenectomy in patients with borderline resectable type C clinical classification. J Gastrointest Surg. 2014;18:146–155. doi: 10.1007/s11605-013-2371-6. discussion 55-6. [DOI] [PubMed] [Google Scholar]
- 17.Temple SJ, Kim PT, Serrano PE, Kagedan D, Cleary SP, Moulton CA, et al. Combined pancreaticoduodenectomy and colon resection for locally advanced peri-ampullary tumours: analysis of peri-operative morbidity and mortality. HPB. 2014;16:797–800. doi: 10.1111/hpb.12263. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Shiloach M, Frencher SK, Jr, Steeger JE, Rowell KS, Bartzokis K, Tomeh MG, et al. Toward robust information: data quality and inter-rater reliability in the American College of Surgeons National Surgical Quality Improvement Program. J Am Coll Surg. 2010;210:6–16. doi: 10.1016/j.jamcollsurg.2009.09.031. [DOI] [PubMed] [Google Scholar]
- 19.Rubin DB. Estimating causal effects from large data sets using propensity scores. Ann Intern Med. 1997;127:757–763. doi: 10.7326/0003-4819-127-8_part_2-199710151-00064. [DOI] [PubMed] [Google Scholar]
- 20.Austin PC, Small DS. The use of bootstrapping when using propensity-score matching without replacement: a simulation study. Stat Med. 33:4306–4319. doi: 10.1002/sim.6276. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Greevy R, Lu B, Silber JH, Rosenbaum P. Optimal multivariate matching before randomization. Biostatistics. 2004;5:263–275. doi: 10.1093/biostatistics/5.2.263. [DOI] [PubMed] [Google Scholar]
- 22.Tzeng CW, Aloia TA, Vauthey JN, Chang GJ, Ellis LM, Feig BW, et al. Morbidity of staged proctectomy after hepatectomy for colorectal cancer: a matched case-control analysis. Ann Surg Oncol. 2013;20:482–490. doi: 10.1245/s10434-012-2620-z. [DOI] [PubMed] [Google Scholar]
- 23.Cho SW, Tzeng CW, Johnston WC, Cassera MA, Newell PH, Hammill CW, et al. Neoadjuvant radiation therapy and its impact on complications after pancreaticoduodenectomy for pancreatic cancer: analysis of the American College of Surgeons National Surgical Quality Improvement Program (ACS-NSQIP) HPB. 2014;16:350–356. doi: 10.1111/hpb.12141. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Parikh JA, Beane JD, Kilbane EM, Milgrom DP, Pitt HA. Is American College of Surgeons NSQIP organ space infection a surrogate for pancreatic fistula? J Am Coll Surg. 2014;219:1111–1116. doi: 10.1016/j.jamcollsurg.2014.08.006. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Bhayani NH, Enomoto LM, James BC, Ortenzi G, Kaifi JT, Kimchi ET, et al. Multivisceral and extended resections during pancreatoduodenectomy increase morbidity and mortality. Surgery. 2014;155:567–574. doi: 10.1016/j.surg.2013.12.020. [DOI] [PubMed] [Google Scholar]
- 26.Banz VM, Croagh D, Coldham C, Taniere P, Buckels J, Isaac J, et al. Factors influencing outcome in patients undergoing portal vein resection for adenocarcinoma of the pancreas. Eur J Surg Oncol. 2012;38:72–79. doi: 10.1016/j.ejso.2011.08.134. [DOI] [PubMed] [Google Scholar]
- 27.Miyazaki M, Ito H, Kimura F, Shimizu H, Togawa A, Ohtsuka M, et al. Hepatic vein reconstruction using autologous vein graft for resection of advanced hepatobiliary malignancy. Hepatogastroenterology. 2004;51:1581–1585. [PubMed] [Google Scholar]
- 28.Dale W, Hemmerich J, Kamm A, Posner MC, Matthews JB, Rothman R, et al. Geriatric assessment improves prediction of surgical outcomes in older adults undergoing pancreaticoduodenectomy: a prospective cohort study. Ann Surg. 2014;259:960–965. doi: 10.1097/SLA.0000000000000226. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Zhang J, Leng JH, Qian HG, Qiu H, Wu JH, Liu BN, et al. En bloc pancreaticoduodenectomy and right colectomy in the treatment of locally advanced colon cancer. Dis Colon Rectum. 2013;56:874–880. doi: 10.1097/DCR.0b013e3182941704. [DOI] [PubMed] [Google Scholar]
- 30.Edwards MJ, Nakagawa K, McMasters KM. En bloc pancreaticoduodenectomy and colectomy for duodenal neoplasms. South Med J. 1997;90:733–735. doi: 10.1097/00007611-199707000-00017. [DOI] [PubMed] [Google Scholar]
- 31.Kapoor S, Das B, Pal S, Sahni P, Chattopadhyay TK. En bloc resection of right-sided colonic adenocarcinoma with adjacent organ invasion. Int J Colorectal Dis. 2006;21:265–268. doi: 10.1007/s00384-005-0756-z. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Table S1. Current Procedural Technology (CPT).