58.01 Ultrasound is a Sensitive Adjunct to Plain Radiographs in Management of Necrotizing Enterocolitis

Posted on January 18, 2016

S. E. Horne3, S. M. Cruz1,3, S. Nuthakki2, P. E. Lau1,3, D. A. Lazar1,3, S. E. Welty2, O. O. Olutoye1,3 1Baylor College Of Medicine,Department Of Surgery,Houston, TX, USA 2Texas Children’s Hospital,Pediatrics,Houston, TX, USA 3Texas Children’s Hospital,Pediatric Surgery,Houston, TX, USA

Introduction:

Necrotizing Enterocolitis (NEC) is the most common gastrointestinal emergency of the preterm infant with an incidence of 5-10%. It is traditionally diagnosed with a combination of physical examination and plain radiographs of the abdomen. The diagnostic role for ultrasound in NEC is uncertain. We hypothesized that ultrasound (US) is as sensitive as plain radiographs in the diagnosis and management of NEC.

Methods:

The medical records of all infants with NEC in a single pediatric tertiary center from January 2006- January 2013 were reviewed. In order to factor in NEC’s rapid rate of changes in pathologic findings, patients that underwent US within four hours of abdominal XR were included in the analysis. Bell’s Criteria were utilized to stage each patient during his/her course. Clinical, radiologic, surgical and pathological findings were reviewed. Statistical analysis was performed using Student's t-test and Mann-Whitney U test for continuous variables and Fisher's exact for categorical variables.

Results:

During this period, 186 neonates were diagnosed with NEC, of which 26 met inclusion criteria. It was noted that US was done for confirmatory purposes in these 26 patients after plain radiographs did not agree with clinical findings. Plain radiographs and ultrasound were taken within an average of 2.46 ± 1.17 hrs of each other. At the time of XR and US, the Bell’s staging of the patients was Stage 1 in 27% of the cases (n=7), Stage 2 in 42% (n=11), and Stage 3 in 31% (n=8). There were 92% preterm infants, 38.5% had congenital anomalies (i.e. omphalocele, congenital neck mass, and Congenital High Airway Obstruction Syndrome), 19% had cardiac abnormalities, and 23% had patent ductus arteriosus. The survival rate of our NEC population was 65% (n=17). Surgical intervention was undergone in 65% (n=17) of our patients in which the median time between imaging findings and time of surgery was 2 (0-59) days. When comparing both modality for reliability in detecting intestinal ischemia and/or perforation with surgical findings, US had a sensitivity of 72% with a positive predictive value (PPV) of 93% while plain radiographs had a sensitivity of 42% with a PPV of 100%. In all cases that required surgical drainage (n=4), ultrasound findings of complex fluid collections guided the decision for bedside surgical drainage where plain radiograph did not suggest free fluid.

Conclusion:

In this study, ultrasound proved to be helpful in assessing the need for surgical interventions in neonates where diagnosis of advanced NEC is ambiguous. US appeared to be more sensitive in reliably detecting intestinal perforation/ischemia in comparison to plain radiographs. However, both plain radiographs and ultrasound play a key role in the diagnosis and should be considered conjunctively during the management and treatment of NEC.

58.02 Significance of Splenic Contrast Blush Among Blunt Injured Children

Posted on January 18, 2016

S. N. Acker1, D. A. Partrick1, L. R. Hill1, D. D. Bensard1,2 1Children’s Hospital Colorado,Pediatric Surgery,Aurora, CO, USA 2Denver Health Medical Center,Department Of Surgery,Denver, CO, USA

Introduction: Among adult patients who suffer blunt solid organ injury, the presence of contrast blush (CB) on initial screening CT scan is associated with a 22 to 24 times greater likelihood of failure of non operative management (NOM) than if no CB is present. Current Western Trauma Association guidelines recommend that all adults with the presence of a blunt splenic injury and contrast blush should be considered for angioembolization. Currently, rate of splenectomy and failure of NOM of blunt organ injury is used as a quality indicator to assess trauma centers caring for injured children. We hypothesized that children with a CB represent a higher risk group and the use of splenectomy rate alone may not be an accurate measure of quality.

Methods: We performed a retrospective review of all children admitted to either of two academic pediatric trauma centers following blunt trauma with any grade liver or spleen injury from 1/09-12/13. Data evaluated included presence of CB on initial CT, need for intervention, and timing of intervention.

Results:245 children were admitted with blunt liver or spleen injury. 45 children were excluded due to lack of radiology report. 183 children had no CB (91%); 17 children had either a definite CB or CB could not be ruled out (9%) and were included in the CB group. Of those with CB, 3 required splenectomy (18%); all three were taken directly from the ED to the OR due to hemodynamic instability that failed to respond to packed red blood cell (PRBC) transfusion. Three additional children in the CB group received a PRBC transfusion within 12 hours of injury. One child had an avulsed kidney and underwent laparotomy, one child received PRBC transfusion during an orthopedic procedure, and the last child was hypotensive and thus received a transfusion. Each of the remaining 11 children with CB remained hemodynamically stable throughout their hospital stay; none required intervention in the form of laparotomy, angioembolization (AE), or PRBC transfusion. No children in either group underwent AE. Children with a CB were more likely to require splenectomy than children with no CB (n=3 (18%) vs n=1 (1%); p<0.01). CB was associated with a higher injury grade than lack of CB (3.4 vs 2.7; p <0.001).

Conclusion:

Contrast blush is rare among children with blunt solid organ injury, but does identify a higher risk group. The presence of CB alone did not precipitate splenectomy, however children with CB were more likely to be hemodynamically unstable, requiring intervention. If splenectomy rates are to be utilized as a quality measure for pediatric trauma care, the presence of vascular blush should also be considered as it represents a 5x higher risk of need for emergent intervention. The need for splenectomy can be used as a quality indicator, however this should not include children with a vascular blush.

57.07 Solid Pseudopapillary Tumors of the Pancreas: An Analysis of 390 Adult Patients

Posted on January 18, 2016

L. M. Youngwirth1, D. P. Nussbaum1, E. Benrashid1, B. C. Gulack1, D. G. Blazer1 1Duke University Medical Center,Durham, NC, USA

Introduction: Primary solid pseudopapillary tumors of the pancreas are rare and typically present in young women. Data regarding the outcomes of adult patients with these tumors are limited. The purpose of this study was to determine the presentation, oncologic profile, and survival of this patient population utilizing a large national database.

Methods: The National Cancer Data Base (1998-2012) was queried for all adult patients with a diagnosis of a solid pseudopapillary tumor of the pancreas. Patient demographic, clinical, and pathologic characteristics at the time of diagnosis were determined. A binary logistic regression model was utilized to identify variables associated with resection. A Cox proportional hazards model was developed to identify factors associated with survival.

Results: A total of 390 patients met inclusion criteria. The mean age at diagnosis was 40 years with the majority of tumors occurring in females (83.3%) and those of white race (68.8%). The mean tumor size was 6.7 cm and the majority of patients underwent surgical resection (87.4%). Of the 341 patients undergoing surgical resection, 2.9% underwent local excision, 11.1% underwent total pancreatectomy, 33.5% underwent pancreaticoduodenectomy, and 52.5% underwent partial pancreatectomy. Of the patients undergoing surgical resection, 8.5% had positive lymph nodes and 8.7% had positive margins. Five year survival was 49.7% for those who did not undergo resection compared to 95.0% for those who did undergo resection (p < 0.01). After adjustment, decreasing patient age (OR = 1.37, p < 0.01), decreasing tumor size (OR = 1.02, p < 0.01), and tumors located in the body or tail of the pancreas (OR = 2.38, p = 0.02) predicted resection. For patients undergoing resection, increasing patient age (HR = 1.49, p = 0.04) and treatment at a non-academic facility (HR = 3.23, p = 0.03) were associated with compromised survival after adjustment. Tumor size, lymph node status, and margin status were not independently associated with survival.

Conclusion: Patients who underwent resection had excellent overall survival. Of patients undergoing resection, increasing age and treatment at a non-academic facility were associated with compromised survival. Given these outcomes and the established importance of facility experience for pancreatic resection, adult patients with primary solid pseudopapillary tumors of the pancreas are ideally suited for referral to highly experienced centers.

57.08 Can Segmental Colectomy Be Utilized For Fulminant Clostridium Difficile Colitis? An Analysis Of 746 Patients

Posted on January 18, 2016

H. Aziz1, Z. Sun1, M. Adam1, J. Kim1, J. Migaly1, C. Mantyh1 1Duke University Medical Center,General Surgery,Durham, NC, USA

Introduction:
Clinical C. difficile infection ranges from asymptomatic carrier state to fulminant colitis. Total abdominal colectomy (TAC) is the standard of care for patients with fulminant colitis; however, small institutional series have suggested that segmental colectomy is a viable surgical alternative in selected patients. Our objective was to evaluate outcomes after segmental colectomy (SC) versus TAC for C. difficile colitis using a large risk-stratified national database.

Methods:
Patients undergoing SC or TAC between 2005-2013 for fulminant C. difficile colitis were identified from the American College of Surgeons National Surgical Quality Improvement Program (NSQIP). Short-term outcomes from SC vs TAC were compared after adjusting for age, race, ASA class, white blood cell count, preoperative sepsis, and ostomy creation. Moreover, adjusted analysis of outcomes of the sickest patients from these two cohorts was also performed.

Results:
Of the 746 patients undergoing colectomies for C. difficile colitis, 559 (75%) patients underwent TAC and 187 (25%) had SC. After adjustment for demographic characteristics and clinical disease severity, there were no differences between SC and TAC with regard to overall complications (OR 0.85, 95% CI 0.46-1.55, p=0.59), postoperative sepsis (OR 0.99, CI 0.56-1.77, p=0.99), hospital length of stay (-0.3 days, CI -0.9-0.5, p=0.48), or 30-day mortality (OR 0.94, CI 0.51-0.1.73, p=0.85). Among patients with sepsis or septic shock, overall complications (OR 0.78, 95% CI 0.3- 1.63, p=0.52), postoperative sepsis (OR 1.19, CI 0.63-2.27, p=0.59), hospital length of stay (-0.2 days, CI -0.4-0.1, p=.24), and 30-day mortality (0.67, CI 0.33-1.36, p=0.27) were not different between the two groups.

Conclusion:
This analysis from a multi-institutional national dataset of patients with C. difficile colitis demonstrates that segmental colectomy may be an acceptable alternative to total abdominal colectomy in patients eligible for a more limited resection for C.difficile colitis.

57.09 Diverticular Disease is Associated With Benign Renal and Hepatic Cysts

Posted on January 18, 2016

P. F. Wrafter1, T. M. Connelly3, B. C. Lucey4, A. Berg5, W. Koltun6, W. P. Joyce1 1Galway Clinic,Department Of Surgery,Galway, GALWAY, Ireland 3University College Hospital Galway,Department of Surgery,Galway, GALWAY, Ireland 4Galway Clinic,Department Of Radiology,Galway, GALWAY, Ireland 5Penn State University College Of Medicine,Division Of Biostatistics & Bioinformatics,Hershey, PA, USA 6Penn State Hershey Medical Center,Division Of Colon And Rectal Surgery,York, PA, USA

Introduction: Colonic diverticula have been linked with dysregulated collagen, dysfunctional matrix metalloproteinases, incisional hernia formation and genetic disorders with a collagen vascular component. Similarly, hepatic and renal cysts have been associated with defects in collagen deposition and matrix metalloproteinase overexpression. We aimed to determine a correlation between diverticular disease (DD) and cystic disease of the kidneys and liver.

Methods: Consecutive abdominal computed topography (CT) scans performed between January and July 2015 at our institution were prospectively studied. Patient demographics, the presence or absence of DD and the number of renal and hepatic cysts were recorded. Patients who had a sigmoid colectomy for pathology other than DD and scans in which DD and/or solid organs were not fully visualized due to patient or disease factors were excluded. R software was used for statistics. A subgroup analysis was performed on youthful DD patients (<55 years of age, n=32) vs older Controls (>55, n=213).

Results: After exclusion, 607 (369 non DD controls, 40.5% male and 238 DD patients, 50.8% male) were studied. Incidence of cystic disease in Controls vs DD patients was 22.5% vs 71.4% (p<0.00001). The mean number of cysts was 3.8 +/-3.9 (SD) vs 5.6+/-4.3 (SD, p=.0009).

Renal cysts were more common than hepatic cysts in both groups (88% of all Control cysts and 86% of all DD patient cysts) and were present in 18.7% of the Controls vs 53.4% of the DD cohort (p<.00001). Hepatic cyst prevalence was 2.4% vs 8.8% (p=.0008).

In the subgroup analysis, a similar result was found with cystic disease present in 29.1% of Controls >55 years old vs 56.2% of DD patients <55 years old (p=.004).

Conclusion: A significant association between benign intra-abdominal cystic disease and DD was demonstrated. This association was present overall and in the subgroup analysis of youthful DD patients who are more likely to have a genetic versus environmental disease aetiology compared with Controls >55. These findings suggest a global defect in connective tissue integrity in the majority of DD patients.

57.04 Incidence of Hepaticojejunostomy Stricture Following Hepaticojejunostomy

Posted on January 18, 2016

F. M. Dimou1, D. Adhikari1, H. Mehta1, D. Jupiter1, T. S. Riall1, K. Brown1 1University Of Texas Medical Branch,Galveston, TX, USA

Introduction: Operations requiring hepaticojejunostomy are uncommon and the true incidence of biliary stricture after hepaticojejunostomy is unknown. Our goal was to use population-based data to determine the timing, incidence, and management of stricture after hepaticojejunostomy for benign and malignant disease.

Methods: We used 5% Medicare claims data (1996 to 2011) to identify patients ≥66 years who underwent an operation requiring a hepaticojejunostomy (alone or as part of a larger operation). Hepaticojejunostomy stricture was identified by diagnosis codes for stricture (ICD-9 code 576.2) and/or PTC drain placement occurring >3 months after the initial operation. A cumulative incidence curve was used to describe timing of stricture diagnosis. The use of imaging and intervention were evaluated. In the cumulative incidence curve, patients were censored when they died (no longer at risk for stricture) or were lost to follow-up (no further Medicare claims). A Cox proportional hazards model was constructed to identify factors associated with stricture diagnosis.

Results: 3,374 patients underwent an operation requiring a hepaticojejunostomy. The 2- and 5-year survival for the cohort was 57% and 43%. The mean age at the time of surgery was 75.3±6.2 years. 1,729 (51.2%) patients had a malignant diagnosis. Overall, 403 patients developed a stricture after surgery. Taking into account death and loss to follow-up, the cumulative incidence of stricture was 12.5% at 2 years and 17.4% at 5 years. Mean time to stricture formation was 16.8±21.6 months (median=8.5 months). 51.9% (N=209) of patients who developed a stricture had a percutaneous transhepatic catheter (PTC) placed. Of the 403 patients with a stricture diagnosis, 233 (57.8%) were for complications related to stricture. The most common reason for stricture-related admission was cholangitis (N=94). Only 18 of the 403 patients (4.5%) required definitive reoperation. Based on a Cox proportional hazards model, only the presence of a preoperative endostent (HR 1.67; 95% CI 1.35, 2.07) predicted stricture formation; preoperative PTC (HR 1.29; 95% CI 0.70, 2.37) did not.

Conclusion: In patients who survive, strictures occur with high frequency after an operation requiring hepaticojejunostomy and should be followed with serial liver function tests. Preoperative stent placement is associated with future stricture formation in patients who undergo hepaticojejunostomy. Even though the majority can be managed non-operatively, stricture diagnosis remains burdensome requiring frequent rehospitalizations, follow-up, and procedures.

57.05 Comparative Analysis of Laparoscopic vs Robotic Distal Pancreatectomy: Is Robotic Surgery Superior?

Posted on January 18, 2016

T. B. Tran1, D. J. Worhunsky1, J. N. Leal1, G. A. Poultsides1, J. A. Norton1, B. C. Visser1, M. M. Dua1 1Stanford University,Stanford, CA, USA

Introduction: In recent years, several institutional reports of robotic distal pancreatectomy (RDP) claim superiority over laparoscopic distal pancreatectomy (LDP) given its ability to overcome technical limitations associated with traditional laparoscopic surgery. However, many of these comparisons include RDP cohorts comprised of smaller subset of surgeons that adopt the robotic platform or 2 attending surgeon teams versus LDP cohorts from larger groups of institutional surgeons across several disciplines. We hypothesize that this difference in surgical teams is a primary contributing factor to the superior outcomes of RDP.

Methods: We performed a retrospective analysis of all minimally invasive spleen-preserving distal pancreatectomies from 2008 to 2015. All cases were performed with the intent to preserve the spleen. Operative characteristics, perioperative outcomes, and 90-day morbidity were compared between LDP and RDP. Subgroup analysis of institutional LDP versus RDP and single-surgeon LDP versus RDP was also performed to control for surgeon variability.

Results: A total of 95 patients underwent minimally invasive distal pancreatectomy with intent to preserve the spleen: 71 (74.7%) LDP and 24 (25.3%) RDP. Conversion rates to open distal pancreatectomy were equivalent (1 RDP and 3 LDP; both 4.2%). Rates of any complication were higher in the LDP group (42.2% vs. 17.4; p=0.046) but specific complications including pancreatic leak, postoperative bleeding, wound infection, intra-abdominal abscess, and splenic infarction were similar between the groups. Although operative time was higher in patients who underwent RDP (mean 287.1 vs. 217.1 min; p=0.001), hospital length of stay was significantly shorter in the RDP group (mean 2.4 vs. 3.6 days; p< 0.001). RDP was associated with more than two-thirds odds reduction in postoperative complications (OR 0.30 95% CI 0.092-0.980; p=0.046). However, subgroup analysis demonstrated that differences between LDP and RDP diminish upon evaluation of single-surgeon outcomes (Table). The previous RDP risk reduction of postoperative complications was not significant in subgroup analysis of a single surgeon who is facile in both RDP and LDP (OR 0.39, 95% CI 0.959-1.041; p=0.98).

Conclusion: RDP is an acceptable minimally invasive approach with postoperative outcomes equivalent to that of LDP in carefully selected patients. Although RDP may offer the benefits of improved range of motion and dexterity over LDP, the novel robotic approach is not necessarily better than LDP as often claimed in institutional comparisons. Robotic technology should be viewed as another tool to facilitate the transition from open surgery rather than a replacement of laparoscopic technique.

57.06 The Interplay of Pancreas Texture & Postoperative Serum Amylase Predicts Pancreatic Fistula Risk

Posted on January 18, 2016

N. S. McCall1, C. B. Chen1, M. J. Pucci1, S. Doane1, J. M. Winter1, C. J. Yeo1, H. Lavu1 1Thomas Jefferson University,Department Of Surgery,Philadelphia, PA, USA

Introduction: Postoperative pancreatic fistula (PF) remains one of the most significant complications after pancreaticoduodenectomy (PD). Soft (normal) pancreas texture is a well-established risk factor for PF. Recently, studies have suggested that post-PD serum hyperamylasemia (PPHA) may also be a risk factor. In this study, we evaluate the relationship of pancreas texture and post-operative serum amylase levels in determining PF risk in patients who have undergone PD.

Methods: This retrospective cohort study utilizing the pancreatic surgery database at Thomas Jefferson University evaluated all patients who underwent PD from 2009 to 2014. The highest postoperative serum amylase level from postoperative day (POD) 0 to POD 5 was obtained. An ROC curve analysis obtained a threshold value to define PPHA compared to normal serum amylase levels. Chi-square analyses and odds ratios evaluated the relationship between pancreas texture, serum amylase level, and the development of PF.

Results: Data from 525 consecutive patients were analyzed after excluding 80 due to a lack of information on pancreas texture. Serum amylase levels on POD2-POD5 were not found to be predictive of PF occurrence; thus, the remaining analysis focused on the highest amylase value between POD0 and POD1. The serum amylase threshold value of 165 IU/I yielded greatest accuracy from the ROC curve analysis (AUC: 0.75; Sensitivity: 0.70; Specificity: 0.72). Demographics and clinical information were similar between patients with PPHA and those with normal serum amylase. Soft pancreata were more likely to have PPHA (p < 0.001) and PF (p < 0.001) than hard pancreata. PPHA was associated with increased rates of PF (p < 0.001), intra-abdominal abscess (p < 0.001), delayed gastric emptying (p = 0.04), wound infection (p < 0.001), hepaticojejunostomy or duodenojejunostomy leakage (p = 0.03), cardiac complications (p = 0.02), pulmonary complications (p < 0.001), as well as increased length of stay (p < 0.001). Fifty-nine patients with soft pancreata developed PF, with 49 (83%) having PPHA. Overall, patients with a soft pancreas and PPHA had the highest PF rate (34.5%), and those with a hard pancreas and normal serum amylase had the lowest rate (5.9%)(Odds ratio: 8.35; 95% CI: 4.4-15.9).

Conclusion: Accurate estimation of PF risk is achievable by incorporation of pancreas texture and immediate post-operative serum amylase levels. Analysis of their interplay segregates patients into well-defined risk groups. The presence of PPHA on POD0 or POD1 increases PF risk, even more so than soft pancreatic texture alone. The minimal PF rate in patients with hard pancreata and normal POD 0-1 serum amylase levels identifies a cohort unlikely to develop PF.

56.10 Factors Associated with Intraoperative Evaluation for Sentinel Node Biopsy in Breast Cancer Patients

Posted on January 18, 2016

S. Z. Bhagwagar1, T. Hughes2, G. Babiera3, A. B. Chagpar1 1Yale University School Of Medicine,Surgery,New Haven, CT, USA 2McPherson Hospital,Surgery,McPherson, KS, USA 3University Of Texas MD Anderson Cancer Center,Surgical Oncology,Houston, TX, USA

Introduction: Sentinel lymph node biopsy (SLNB) has been accepted as a minimally invasive means of assessing axillary lymph nodes in breast cancer patients. While many surgeons had used intraoperative evaluation (IOE) either with frozen section (FS) or touch imprint cytology (TIC) to dictate whether to complete an axillary lymph node dissection (ALND) in the same operative setting, the widespread acceptance of the ACOSOG Z-0011 trial has lessened the need to do so in patients with 1-2 positive nodes undergoing a lumpectomy. We sought to determine, in a contemporary context, what surgeons’ practices were vis-à-vis the use of IOE, and factors affecting the same.

Methods: An online survey of surgeons was performed using the American College of Surgeons Communities’ Platform. Surgeons were asked about their clinical setting as well as their current use of IOE for SLNB. Nonparametric statistical analyses were performed using SPSS Version 21.0. In order to control for multiple comparisons, a Bonferroni correction was applied to the threshold for statistical significance, resulting in a significant p-value < 0.006 (0.05/9).

Results: 270 surgeons responded to the survey. To the question "In patients undergoing SLNB, do you get IOE?", 71 (27.1%) stated that they did IOE on all patients, 104 (39.7%) stated that they did so only on mastectomy patients, and 87 (33.2%) reported that they did not use IOE at all. Factors correlated with the use of IOE included geographic location (urban/suburban/rural), and proportion of practice that was breast-related (p<0.001 for both). In addition, surgeons who felt that isolated tumor cells did not constitute a positive lymph node and those who felt that the risk of lymphedema after ALND was > 10% were more likely to use IOE selectively in mastectomy patients (p=0.003 and p=0.005, respectively). Surgeon age, duration in practice, practice type (academic/hospital employed/private practice), and perception of the risk of lymphedema after SLNB did not influence surgeons’ use of IOE. Furthermore, whether surgeons used FS or TIC for IOE did not affect whether they utilized IOE always or selectively.

Conclusion: While nearly 70% of surgeons use IOE, the majority do so selectively in mastectomy patients only. Surgeons with solely breast-related practices in urban areas were more likely to use IOE only on mastectomy patients; those who felt that a positive SLN was defined as tumor deposits > 0.2 mm and who felt the risk of lymphedema was > 10% after ALND were also more likely to do so.

57.01 Hospital Costs for 90 Days Before and After Ventral Hernia Repair using Biologic vs. Synthetic Mesh

Posted on January 18, 2016

C. Totten1, N. D. Ward2, D. L. Davenport1, J. Roth1 1University Of Kentucky,Department Of Surgery,Lexington, KY, USA 2University Of Kentucky,College Of Medicine,Lexington, KY, USA

Introduction: Ventral hernia repair (VHR) using biologic mesh (BioM) has higher hospital costs during operative hospitalization than synthetic mesh (SynM). One study suggests BioM follow-up costs may be less making overall costs equivalent. This study compared 90-day pre- and post-VHR hospital costs in BioM versus SynM patients matched on infection risk.

Methods: This retrospective database study matched patient perioperative risk data in our local National Surgery Quality Improvement Database with resource utilization data from our cost accounting system for 385 VHR patients. Costs and hospital encounters were analyzed for 90 days before surgery and 90 days after discharge (180 days). Clinical, financial, and infectious outcomes were compared in unmatched SynM (n=303) and BioM (n=82) groups. Propensity scores were used to match 40 SynM and BioM case pairs with similar wound class, concomitant procedures, recurrent hernia status, approach, inpatient vs. outpatient, and American Society of Anesthesiology (ASA) class. Matched group outcomes were compared.

Results: BioM patients in the unmatched group were older, had higher ASA and wound class, more frequent concomitant procedures, and open inpatient procedures performed on recurrent hernias. Wound occurrences were more frequent in unmatched BioM patients (19.5% vs. 5.3%) as were 180-day costs ($28,800 vs $14,000). Matching was successful for mostly clean wound class patients (34/40 in the BioM group; 35/40 SynM) and a few with concomitant procedures (2/40 in the BioM; 3/40 SynM). In these low risk patients, wound occurrences were similar (BioM, 15%; SynM, 10%) as were readmissions (BioM 33%; SynM 23%), but 180-day costs remained higher ($31,844 vs. $15,527). There were no differences in 180-day diagnostics, emergency room, ICU, floor, pharmacy or therapeutic costs (Figure). 180-day O.R. services and supply costs were higher in the BioM group ($21,106 vs. $7,067).

Conclusion: BioM is utilized more in higher risk hernia repairs involving higher wound class and ASA scores, concomitant procedures, and recurrent hernias. Clinical outcomes following low risk VHR are similar with both SynM and BioM, however SynM is more cost effective including 90 days pre- and post-operation hospital costs.

57.03 The Model for End-Stage Liver Disease Predicts Outcomes for Patients Undergoing Cholecystectomy

Posted on January 18, 2016

S. Dolejs1, E. P. Ceppa1, J. Kays1, B. L. Zarzaur1 1Indiana University School Of Medicine,Indianapolis, IN, USA

Introduction: Surgeons are commonly asked to perform elective operations on patients with cirrhosis. Cirrhosis has consistently been shown to be a predictor of worse perioperative outcomes, but the magnitude of that risk is poorly understood as the current literature on these patients is predominantly small retrospective series. The Model for End-Stage Liver Disease (MELD) has been validated as a predictor of perioperative outcomes. The American College of Surgeons National Surgical Improvement Program (ACS-NSQIP) provides a unique opportunity to study a large population of patients with high MELD scores for many commonly performed operations to better quantify the perioperative risks of these patients.

Methods: The ACS-NSQIP database from 2005-2013 was used to study patients undergoing cholecystectomy. All patients with CPT codes identifying cholecystectomy were included, and the cohort was divided into laparoscopic or open cholecystectomy. Patients without all laboratory values needed to calculate the MELD were excluded. ICD-9 diagnoses related to choledocholithiasis and preoperative dialysis patients were excluded given the potential for confounding elevation of the MELD score. Bivariate data analysis was performed and logistic regression modeling was conducted to calculate risk-adjusted 30-day outcomes. Logistic regression modeling was performed with 26 preoperative validated variables included in ACS-NSQIP.

Results: A total of 63,464 patients were included in the study. Unadjusted mortality steadily increased as the MELD score increased in the laparoscopic (MELD=6-10, 0.2%; 11-15, 1.1%; 16-20, 3.2%; >20, 5.8%) and open groups (MELD=6-10, 1.5%; 11-15, 3.7%; 16-20, 8.6%; >20, 17.9%; P<0.0001). Unadjusted morbidity similarly increased with MELD score increases in the laparoscopic (MELD=6-10, 3.8%; 11-15, 9.9%; 16-20, 16.3%; >20, 22.8%) and open groups (MELD=6-10, 18.7%; 11-15, 28.2%; 16-20, 40.7%; >20, 57.8%; P<0.0001). MELD score acts as a progressive and independent predictor of morbidity and mortality after logistic regression modeling as shown in the attached table.

Conclusion: The MELD score is an objective and easy to calculate scoring system that independently predicts postoperative morbidity and mortality in patients undergoing cholecystectomy. Open cholecystectomy is associated with significantly more morbidity and mortality than laparoscopic cholecystectomy across all MELD groups.

56.07 Comparison of Perioperative Hemodynamic Management for Pheochromocytoma and Paraganglioma

Posted on January 18, 2016

J. D. Smith1, J. H. Hammond1, A. C. Niemann1, A. T. Reid1, S. A. DeBolle1, D. T. Hughes1 1University Of Michigan,Department Of Surgery,Ann Arbor, MI, USA

Introduction: Pheochromocytomas (PCC) and paragangliomas (PG) are catecholamine secreting neuroendocrine tumors. Most PCCs secrete both epinephrine and norepinephrine, while PGs secrete only norepinephrine. Our hypothesis is that PCCs and PGs unique catecholamine secreting profiles alters perioperative hemodynamics and management.

Methods: A retrospective analysis was performed of PCC and PG patients undergoing surgical treatment at the University of Michigan from 2006-2015. Non-secreting and extra-abdominal PG and PCC were excluded. Demographics, vital signs, lab data, pre-operative α-blocking, β-blocking or Ca2+ channel blocking medication use, operative hemodynamic data, perioperative use and dose of vasoactive medications, and post-operative hemodynamic data were compared between PCC and PG. Statistical significance was determined with Fisher’s exact tests for categorical variables and Mann-Whitney-U tests for continuous variables. p values <0.05 were considered significant.

Results: 62 PCC and 15 PG patients were compared; there was no significant difference in gender or age between groups. PGs had elevated normetanephrine but normal metanephrine levels. Laparoscopic resection was completed in 20.0% of PG of cases and 67.8% of PCC (p=0.0011). The median operative time for PG was 238min but only 153min for PCC (p=0.0027). Estimated blood loss in PG was 900mL vs 75mL in PCC (p=0.0013). Median fluid administration was 5744mL for PG and 4210 for PCC (p=0.0228). PCC patients had greater median minimum diastolic BP at 81mmHg vs 74mmHg in PG (p=0.0175). PCC patients had greater maximum diastolic BP, 88.5mmHg vs 82mmHg (p=0.0043). The median final dose of preoperative phenoxybenzamine for PG was 85mg/day and only 60mg/day for PCC (p=0.1965). 60% of PG and 66.1% of PCC patients used β-blockers (p=0.7649), and 20.0% of PG patients and 35.5% of PCC patients were placed on preoperative Ca2+ channel blockers (p=0.3605). No significant differences in the use of vasoactive medications were found with 46.7% of PG and 48.4% of PCC receiving MgSO4 (p=1.000), 40.0% of PG and 59.7% of PCC having β-blockers (p=0.2468) and 80.0% of PG and 75.8% of PCC receiving other vasoactive medications (p=1.000). A significantly higher percentage of PG patients, however, experienced all types of post-operative hemodynamic instability (HDI) events, with 46.7% of PG but only 12.9% of PCC patients experiencing HDI (p=0.0072, see Fig. 1).

Conclusion: Despite differing biochemical profiles, PCC and PG have similar preoperative blood pressure blockade requirements and respond to the medications similarly. PG patients, however, are significantly more likely to experience post-operative hemodynamic instability compared to PCC patients.

56.08 BRAF Mutational Status Predicts Papillary Thyroid Carcinoma Recurrence in Long Term Follow Up

Posted on January 18, 2016

A. Fingeret1, K. Economopoulos1, Y. Chen1, P. Sadow2, R. Hodin1, S. Parangi1, C. Lubitz1 1Massachusetts General Hospital,Endocrine Surgery Division,Boston, MA, USA 2Massachusetts General Hospital,Department Of Pathology,Boston, MA, USA

Introduction: Papillary thyroid carcinoma (PTC) incidence is increasing, yet disease specific survival has remained largely unchanged. BRAF mutational status has been inconsistently correlated with recurrence. Furthermore, with the impetus to observe low risk tumors without resection, accurate risk stratification is paramount.

Methods: An unselected historical cohort of patients (from 2000 – 2007) who underwent at least total thyroidectomy for PTC was analyzed to model recurrence by BRAF mutational status. BRAF status was determined by DNA extraction and PCR-based single nucleotide extension genotyping of paraffin embedded primary tumor specimens. Univariate analysis was performed with Wilcoxon rank sum and log rank test. A multivariable cox proportional hazards model was used to determine clinicopathologic factors associated with time to recurrence (primary outcome) and second recurrence (secondary outcome). Recurrence was defined as pathologically confirmed PTC or distant metastases treated with radioactive iodine (RAI) greater than 90 days from index operation.

Results: The recurrence incidence within our 605 patient cohort was 77 (12.7%). Of these, BRAF mutational status was determined for 402 (66.4%) with 254 (63.2%) containing BRAFV600E. The median overall follow up was 8.3 years (IQR 6.6 – 10.3) with 7.7 years (6.0 – 9.3) for BRAFV600E and 8.6 years (IQR 7.0 – 11.0) for wild type.There were no differences in clinicopathological risk factors between patients with and without BRAF testing. The incidence of recurrence among BRAFV600E patients was 38 (17.6%) compared to 11 (8.0%) among BRAF wild type (p = 0.02). Of the initial and repeat recurrences, 76 (98.7%) and 16 (80%) were locoregional, respectively. In the multivariable cox regression analysis controlling for clinicopathologic risk factors (Figure 1), the likelihood of PTC recurrence was increased by BRAF mutation status (hazard ratio 2.4, p = 0.03), tumor size by AJCC T stage (HR 1.9, p = 0.01), and regional lymph node metastases at index operation (HR 1.9, p =0.01). The median time to recurrence overall was 1.9 years (IQR 0.68 – 4.1), 2.1 years (IQR 1.1 – 4.3) for BRAFV600E and 2.8 years (IQR 0.6 – 5.3) for wild type. Twenty patients developed a second recurrence; in this subset of patients the incidence of BRAFV600E was also significantly correlated with time to recurrence (p = 0.04). Eleven (1.8%) patients died, with four (0.7%) deaths attributable to PTC.

Conclusion: Adjusting for known clinicopathologic risk factors for PTC recurrence, BRAF V600E mutation remains predictive over a long follow up.

56.09 Impact of Breast Cancer Subtype on Conversion to Breast Conservation with Neoadjuvant Chemotherapy

Posted on January 18, 2016

C. Capriccioso1, J. Brennan1, B. Heiden1, A. Hage1, A. Zheutlin1, M. S. Sabel1 1University Of Michigan,Surgery,Ann Arbor, MI, USA

Introduction: The metric of success for neoadjuvant chemotherapy (NACT) in breast cancer has traditionally been a complete pathologic response (pCR). Consequently, NACT is considered less often in luminal A patients, where a pCR is unlikely. However, a pCR is not always necessary to allow avoidance of mastectomy. We sought to examine the impact of NACT on BCT eligibility based on histologic subtype, particularly among patients with the luminal A subtype.

Methods: Our IRB-approved prospective breast cancer database was queried for patients who underwent NACT prior to curative surgery. Charts were reviewed for clinicopathologic features, pathologic response, surgical therapy and BCT eligibility before and after NACT as determined by the surgeon. Patients with inflammatory breast cancer or eligible for BCT prior to NACT were excluded when calculating the conversion rate. Patients were categorized by molecular subtype based on estrogen receptor, progesterone receptor and Her-2/neu overexpression: luminal A (ER or PR positive, Her-2/neu negative), luminal B (ER or PR positive, Her-2/neu positive), HER2- enriched (ER/PR negative, Her-2/neu positive), or basal –like (triple negative).

Results: Of the 417 patients identified, 139 (33%) were luminal A, 76 (18%) were luminal B, 69 (17%) were HER2-enriched and 133 (%) were triple negative. Rates of pCR were highest (55%) among HER2-enriched patient and were extremely low (6%) among luminal-A patients. The breast conservation rate was relatively similar across subtypes, ranging from 29% to 39%. However, the BCT rate was influenced by NACT use among initially BCT-eligible patients, and by BCT-eligible patients after NACT opting for bilateral mastectomy. Among non-inflammatory patients determined to be ineligible for BCT prior to NACT, conversion to BCT-eligibility (regardless of subsequent surgery) was highest (nearly 2/3rd) for HER2-enriched and triple-negative patients. While the BCT conversion rate was lower for hormone receptor positive, HER2-negative patients (41%), there was still a substantial impact on surgical outcomes.

Conclusion: For NACT, pCR rates or BCT rates present an incomplete assessment of this therapy by overlooking the impact on surgical decision making. A substantial percentage of luminal A patients can avoid mastectomy with NACT, despite a significantly lower pCR rate. Measuring BCT rates after NACT does not adequately measure the impact of NACT on BCT eligibility, as NACT may be used in BCT-eligible patients, and after NACT, BCT-eligible patients may opt for unilateral or bilateral mastectomies. Surgical response rate, or conversion to BCT eligibility, as assessed by the surgeon, should be measured prospectively in future trials of neoadjuvant chemotherapy.

56.04 Outcomes of Minimally Invasive vs Open Esophagectomy for Esophageal Cancer: an NCDB Analysis

Posted on January 18, 2016

A. Hanna1, M. Chuong1, S. Bentzen1, N. Hanna1 1University Of Maryland,School Of Medicine,Baltimore, MD, USA

Introduction: Esophageal cancer represents a growing public health burden in the United States, with 5 year survival rates at 20 to 30%. Surgery is the mainstay of localized esophageal cancer and minimally invasive esophagectomy (MIE) presents as an alternative approach to open esophagectomy (OE). While no randomized control trial has clarified its utility, MIE has been shown in small series to be better in hospital stay parameters, and equivalent in postoperative morbidity and mortality as well as survival. The generalizability from these studies, however, is not clear as treatment selection bias becomes a significant problem in applying their results in clinical practice.

Methods: The National Cancer Database was used to identify stage I-III esophageal adenocarcinoma and squamous cell carcinoma cancer patients 18 – 90 years old who underwent either MIE or OE from 2010 to 2012. Several factors were identified that differed between those undergoing MIE vs OE such as insurance type, the rate of receiving chemotherapy and radiation, grade, stage and the type of surgical procedure (total vs partial esophagectomy). Propensity score analysis (PSA) with 1:1 ‘nearest neighbor’ matching based on all demographic and disease factors was therefore used to control for this treatment selection bias. The main outcomes analyzed were 30 day mortality, 30-day unplanned readmissions, 1 year mortality, and overall survival. All statistical analyses used either a logisitic or Cox regression.

Results: Out of a total of 8659 patients in the database who had surgery in 2010 or later, 1004 MIE and 3026 OE met inclusion criteria. After PSA matching, both groups contained 1004 patients and were balanced in all demographic and disease factors, including the propensity of having a MIE. The number of lymph nodes sampled and rate of negative margins were not statisitically different in both the original matched cohorts. After multivariate analysis, surgery type (MIE vs OE) was not a significant factor in 30-day mortality, 30-day unplanned readmissions, or 1 year mortality in both the unmatched and matched cohorts. While it was also not a significant predictor of overall survival in the unmatched cohort, the use of MIE showed improved overall survival in the matched cohort (HR = 0.85, P < 0.01). Other factors that predicted for improved survival in both the unmatched and matched group include treatment at an academic research program, private insurance, a younger patient with less comorbidities, smaller tumors, lower grade and stage of cancer, no lymphovascular invasion, fewer positive lymph nodes, negative margins, and the receipt of neoadjuvant chemoradiation.

Conclusion: In the surgical management of esophageal cancer in the United States, a minimally invasive approach confers neither a benefit nor a risk compared to an open approach in short term outcomes and may actually provide a small but not insignificant survival benefit long term.

56.05 Expression of VEGFA and VEGFR correlates with Disease Free and Overall Survival in Colorectal Cancer

Posted on January 18, 2016

L. J. Fernandez1, A. L. Olex2, A. R. Wolen2, D. A. Fenstermacher2, M. G. Dozmorov2, K. Takabe1,3 1Virginia Commonwealth University,Department Of Surgery, Division Of Surgical Oncology,Richmond, VA, USA 2Virginia Commonwealth University,Department Of Biostatistics,Richmond, VA, USA 3Virginia Commonwealth University,Department Of Biochemistry & Molecular Biology,Richmond, VA, USA

Introduction: Angiogenesis is known to be a critical factor in progression of many cancers including colorectal cancer. Bevacizumab, an antibody against VEGFA (a well known mediator of angiogenesis) is used for the treatment of colorectal cancer, mostly in the advanced and recurrent setting. To date, there is no objective data that demonstrate a relationship between angiogenetic factors and survival. Using The Cancer Genome Atlas (TCGA) database, we sought to determine if the level of expression of angiogenesis-related genes in colorectal adenocarcinoma (COAD) tumors is associated with survival.

Methods: All available clinical and RNA-seq data from COAD samples were obtained from the TCGA database. RNA-seqV2 data for VEGFA, VEGFR1, VEGFR2 (markers of angiogenesis) and CD31 (marker for vascular endothelial cells) was normalized for analysis using R; Gene-expression was compared between normal and tumor samples for our genes of interest. A cutoff was identified for each gene of interest to categorize each sample as either low or high. All the available clinical data in the database was analyzed by hand to maximize the number of patients with disease free (DFS) and overall survival (OS) data points. Kaplan-Meier curves were used to plot DFS and OS and statistical significance was calculated. Only the gene-expression in the tumor was used for survival analysis.

Results: There were 314 COAD samples with RNA-seq data; 41 corresponded to normal tissue and 273 to tumor samples. Of the 273 patients with tumor RNA-seq, 240 patients also had usable OS data available and 233 had usable DFS data available. There was a significant difference in expression of VEGFA (p=1.352e-04), VEGFR1 (p=7.613e-04) and CD31 (4.713e-09) between colon cancer and normal colon. VEGFA and VEGFR1 were highly expressed in tumors versus normal tissue, implying higher angiogenesis, while CD31 was higher in normal tissue. These expression differences are in agreement with the notion that normal colon mucosa is more vascular than the tumor, and tumors have higher angiogenetic activity. When comparing tumors based on their gene expression we found that there was worse DFS in patients with tumors that express higher levels of VEGFA (p=0.00498), VEGFR1 (p=0.002), VEGFR2 (p=0.03411), and CD31 (p=0.03619). Also, there was worse OS for patients with higher expression of VEGFA (p=0.00502). Our results are in agreement with the notion that tumors that express high levels of pro-angiogenic factors and receptors, as well as generation of vessels are associated with poor survival.

Conclusion: Tumors with increased angiogenesis, represented by high expression of VEGFA and VEGF receptors, have worse disease-free and overall survival. This is in agreement with the improved survival that was been seen in patients with metastatic colorectal cancer when treated with Bevacizumab.

56.06 Efficacy of Adjuvant Chemotherapy for Small Bowel Adenocarcinoma: A Propensity Score-matched Analysis

Posted on January 18, 2016

B. L. Ecker1, M. T. McMillan1, J. Datta1, D. L. Fraker1, G. C. Karakousis1, R. E. Roses1 1University Of Pennsylvania,Philadelphia, PA, USA

Introduction: The role of adjuvant chemotherapy (AC) in the treatment of small bowel adenocarcinoma is poorly defined. Previous analyses have been limited by small sample size and have failed to demonstrate a survival advantage.

Methods: Resected AJCC pathologic stage I-III small bowel adenocarcinoma patients receiving AC (n=1,637) or surgery alone (SA) (n=2,753) were identified in the National Cancer Data Base (1998–2011). Cox regression identified covariates associated with overall survival (OS). AC and SA cohorts were matched (1:1) by propensity scores based on the likelihood of receiving AC on multivariable analysis or hazard ratio on Cox modeling. Kaplan-Meier estimates of OS were compared.

Results: Omission of adjuvant chemotherapy conferred an increased risk of death (HR 1.36 95% CI 1.22–1.52, p<0.001). After propensity-score matching, median OS was superior in patients with AJCC stage III disease receiving AC compared with SA (n=1,217, 41.4 vs. 26.5 m, p<0.001), regardless of tumor location (duodenum: n=545, 36.3 vs. 25.7 m, p=0.002; jejunoileal: n=469, 47.9 vs. 30.0 m, p=0.015). Adjuvant chemotherapy was not associated with improved OS in patients with AJCC stage II disease (n=927, 116.7 vs. 92.1 m, p=0.410) regardless of tumor location (duodenal: n=235, 110.3 vs. 117.9 months, p=0.859; jejunoileal: n=450, 152.3 vs. 137.3 months, p=0.389), or the presence of putative high risk features including: poorly differentiated histology (n=254, 93.4 vs. 94.4 m, p=0.766), T4 tumor classification (n=291, 68.6 vs. 43.8 m, p=0.060), or a positive resection margin (n=62, 77.7 vs. 33.6 months, p=0.379). There was no OS benefit to adjuvant chemotherapy receipt for AJCC stage I patients (n=64, 116.6 vs. 102.1 months, p=0.843).

Conclusions: AC is associated with improved OS in AJCC stage III patients. The marked absolute OS advantage in patients with stage II disease and T4 tumors, or after positive margin resection, may justify use of AC in selected patients with earlier stage disease.

56.01 Blood Transfusion is an Independent Predictor of Morbidity and Mortality after Hepatectomy

Posted on January 18, 2016

A. N. Martin1, M. J. Kerwin1, F. E. Turrentine1, T. W. Bauer1, R. B. Adams1, G. J. Stukenborg1, V. M. Zaydfudim1 1University Of Virginia,Department Of Surgery,Charlottesville, VA, USA

Introduction: The impact of transfusion on morbidity and mortality in patients undergoing liver resection has been highly debated. Multiple studies have described worse outcomes associated with blood transfusion among patients selected for hepatectomy. However, independent effects between transfusion and extent of resection have not been well described. We hypothesize that blood transfusion is an independent factor that affects outcomes in patients undergoing hepatectomy.

Methods: Patients at our institution who underwent hepatectomy between 2006 and 2013 were identified and linked with the ACS-NSQIP PUF dataset. Individual abstracted metrics included specific diagnosis and extent of liver resection. Major hepatectomy was defined as resection of > 3 liver segments. Mortality was defined as death occurring within 30 days of the date of surgery. Composite overall morbidity was defined as the occurrence of surgical site infection, wound disruption, pneumonia, unplanned intubation, pulmonary embolism, > 48 hour ventilator, renal failure, urinary tract infection, stroke or cerebral vascular accident, cardiac arrest, myocardial infarction, deep venous thrombosis, or systemic sepsis. The NSQIP estimated probabilities of morbidity and mortality were used to include an overall adjustment for confounding effects, including age and preoperative comorbidities. Multivariable models were used to estimate the effects of extent of resection, estimated probabilities of morbidity and mortality, and use of blood transfusion on 30-day morbidity and mortality.

Results: Among 522 patients in the study, 40 (8%) patients required a blood transfusion. 172 (33%) underwent major hepatectomy. Indications for surgery included metastatic neoplasm (n=229, 44%), primary hepatic neoplasm (n=108, 21%), primary extra-hepatic biliary neoplasm (n=23, 4%), and non-malignant indications (n=162, 31%). 86 (16%) patients had a postoperative morbidity; 8 (2%) patients died. After adjustment for significant effect of NSQIP estimated probability of morbidity (OR 1.34, 95% CI 1.08-1.66, p=0.009), blood transfusion was significantly associated with postoperative morbidity (OR 4.00, 95% CI 1.99-8.04, p<0.0001). Extent of resection (OR 1.35, 95% CI 0.79-2.29, p=0.277) was not significantly associated with morbidity. Transfusion was also associated with a significant increase in risk of mortality (OR 19.4, 95% CI 4.02-93.96, p<0.0002). Extent of resection was not associated with mortality in the multivariable model.

Conclusion: Blood transfusion is a highly statistically significant independent predictor of morbidity and mortality after hepatectomy. In patients selected for hepatectomy, extent of resection is not associated with greater risk of postoperative complications.

56.02 Pancreas Fistula Risk Prediction: Implications For Hospital Costs And Payments

Posted on January 18, 2016

D. E. Abbott1, C. Tzeng2, M. T. McMillan3, M. P. Callery4, T. S. Kent4, J. D. Christein5, S. W. Behrman6, D. P. Schauer1, M. H. Eckman1, J. J. Sussman1, C. M. Vollmer3 1University Of Cincinnati,Cincinnati, OH, USA 2University Of Kentucky,Lexington, KY, USA 3University Of Pennsylvania,Philadelphia, PA, USA 4Beth Israel Deaconess Medical Center,Boston, MA, USA 5University Of Alabama-Birmingham,Birmingham, Alabama, USA 6University Of Tennessee,Memphis, TN, USA 7University Of Tennessee,Memphis, TN, USA

Introduction: Disease site-specific risk adjustment and predictors of outcome are becoming increasingly useful clinically. However, as payment models for surgical care evolve, it is unclear how such risk stratification may impact patient selection, resource allocation, hospital costs and reimbursements. We sought to determine whether a validated clinical risk score for postoperative pancreatic fistula (POPF) — the most morbid complication following

Methods: A multi-institutional cohort of 1193 patients undergoing PD, POPF risk-adjusted using the validated 10-point Fistula Risk Score (FRS), was matched by clinical outcome with patients undergoing PD at a separate center, from which hospital cost and reimbursement data were utilized. A decision analytic model was constructed to detail the impact of POPF risk, perioperative mortality, complication rates, and readmission rates on hospital costs, payments, and net profits. Probabilistic sensitivity analyses were performed to understand how alterations in clinical variables affected fiscal outcomes.

Results: Clinically significant POPF (ISGPF B/C) rates for negligible (0), low (1-2), moderate (3-6) and high-risk (7-10) FRS subsets were 0% (n= 164), 5.2% (n= 343), 16.3%% (n= 603), and 28.9% (n=83), respectively. Per-patient hospital cost for negligible-risk patients was $37,855. Low-, moderate-, and high-risk patients had incrementally higher hospital costs of $38,125 (0.7% above negligible-risk), $41,128 (+8.6%), and $41,983 (+10.9%), respectively. Similarly, hospital payment for negligible-risk patients was $42,685/patient, with incrementally higher payments for low-risk ($43,265; +1.4%), moderate-risk ($45,439; +6.5%) and high-risk ($46,564; +9.1%) patients. Net profit (payment minus cost), however, was similar across incremental risk groups: $4,380, $5,140, $4,311 and $4,581, respectively. The lowest 30-day readmission rates — associated with highest net profit — were also found for negligible/low-risk patients (10.5% and 11.1%), respectively, compared with readmission rates of moderate/high-risk patients (15% and 15.7%, respectively). Net profit per hospital day declined from negligible- to high-risk cohorts – most markedly in patients exceeding median lengths of stay (Figure 1).

Conclusion: Similar to actual POPF rates, financial outcomes following PD can be predicted using a validated clinical risk score for pancreatic fistula. Such accurate risk prediction may help hospitals and payers alike plan for resource allocation, cost controls and reimbursements matched to individual patient risk and acuity.

56.03 Evolution and Impact of Lymph Node Dissection during Pancreaticoduodenectomy for Pancreatic Cancer

Posted on January 18, 2016

M. F. Eskander1, S. W. De Geus1, G. G. Kasumova1, S. Ng1, W. Al-Refaie3, G. Ayata2, J. F. Tseng1 1Beth Israel Deaconess Medical Center,Department Of Surgery,Boston, MA, USA 2Beth Israel Deaconess Medical Center,Department Of Pathology,Boston, MA, USA 3MedStar Georgetown University Hospital,Department Of Surgery,Washington, DC, USA

Introduction: Lymph node examination during pancreaticoduodenectomy (PD) is a joint venture between surgery and pathology. Previous studies have suggested that frequently, insufficient lymph nodes are evaluated, leading some patients with N1 disease to misclassified as N0. We examine trends in lymph node dissection over time and investigate whether changes may have impacted overall survival and the long-term prognostic value of lymph node status.

Methods: The National Cancer Data Base was queried for patients diagnosed with non-metastatic pancreatic adenocarcinoma between 1998 and 2012 who underwent classic PD with antrectomy. Patients with unspecified number of lymph nodes examined were excluded. Cochran-Armitage trend analyses were performed to test for linear trends in binomial proportions of node examination and N classification over time. Kaplan-Meier curves and Cox proportional hazards models were used to assess the impact of lymph node status on overall survival for patients diagnosed over 2-year intervals from 1998-2011.

Results: For PDs in which at least one node was surgically examined, the median number of examined lymph nodes nearly doubled from 8 (IQR 5-13) in 1998 to 16 (IQR 11-23) in 2012. The rate of N1 disease increased from 61.3% to 66.4% (p<0.0001). (Figure) Over all years, median LNR remained between 0.10-0.13. 7.3% of resected patients in 1998 had no lymph nodes examined compared to only 2.4% in 2012 (p<0.0001). Median survival was significantly different between N1 and N0 patients (all p<0.0001) at all time intervals, and survival in both subgroups improved. In successive multivariate models adjusted for gender, race, T stage, margin status, and tumor grade, N0 vs. N1 status was consistently protective for overall survival (p<0.0001), but there was no change in the magnitude of its hazard ratio over time (overall HR 0.718; 95% CI 0.696-0.741). Likewise, lymph node status continued to be a significant predictor of overall survival when models were stratified by receipt of chemoradiotherapy vs. chemotherapy only vs. neither (HR 0.690; CI 0.669-0.713) and when stratified by age group (HR 0.709; CI 0.687-0.732).

Conclusion: In contrast to previous years, contemporary patients have an adequate number of nodes examined during standard PD. This, along with with rising rates of N1 cancer detection, suggests more accurate classification of lymph node status and prognosis after resection. However, the strength of lymph node status as a prognosticator for overall survival has not changed. More research is needed regarding the repercussions of increased lymph node dissection and the optimal standards for nodal harvest and evaluation in the modern era.

« Previous Page
Next Page »