27.02 MELD-Na Score as a Predictor of Anastomotic Leak in Elective Colorectal Surgery

Posted on January 25, 2017

K. Coakley1, S. Sarasani1, T. Prasad1, S. Steele2, I. Paquette3, B. Heniford1, B. Davis1  2Case Western Reserve University School Of Medicine,Department Of Surgery,Cleveland, OH, USA 3University Of Cincinnati,Department Of Surgery,Cincinnati, OH, USA 1Carolinas Medical Center,GI And Minimally Invasive Surgery,Charlotte, NC, USA

Introduction:
In patients with cirrhosis awaiting liver transplantation The Model for End-Stage Liver Disease Sodium Model (MELD-Na) is extensively studied.  Because of the simplicity of the scoring system, there has been interest in applying MELD-Na to predict patient outcomes in the non-cirrhotic surgical patient, and has been shown to predict postoperative morbidity and mortality after elective colon cancer surgery.  Our aim was to identify the utility of MELD-Na to predict anastomotic leak in all types of elective colorectal cases.

Methods:

The ACS NSQIP Targeted Colectomy database was queried (2012 – 201) for all elective colorectal procedures in patients without ascites.  Leak rates were compared by MELD-Na score using Chi-square tests and multivariate logistic regression analysis.

Results:
We identified 44,540 elective colorectal cases (mean age, 60.5 years ±14.4, mean BMI 28.8±6.6, 52% female), of which 70% were colectomy and 30% proctectomy.  Laparoscopic approach accounted for 64.72% while 35.3% were open.  The overall complication and mortality rates were 21% and 0.7%, respectively, with a total anastomotic leak rate of 3.4%.    Overall, 98% had a preoperative MELD-Na score between 10-20.  Incremental increases in MELD-Na score (10-14, 15-19 and ≥20) were associated with an increased leak rate, specifically in proctectomies (3.9% vs 5.1% vs10.7% p<0.028).  MELD-Na score ≥20 had an increased leak rate when compared to those with MELD-Na 10-14 (OR 1.627; 95% CI (1.015, 2.607).  A MELD-Na increase from 10-14 into 15-19 increases overall mortality (OR 5.22; 95% CI 3.55, 7.671).    In all elective colorectal procedures, for every one-point increase in MELD-Na score, anastomotic leak (OR 1.04 95% CI (1.006, 1.07), mortality (OR 1.24; 95% CI, (1.20, 1.27) and overall complications (OR 1.10; 95% CI (1.09,1.12) increased.   MELD-Na was an independent predictor of anastomotic leak in proctectomies, when controlling for gender, steroid use, smoking, approach, operative time, preoperative chemotherapy and Crohns Disease (OR 1.06, 95% CI (1.002, 1.122)). 

Conclusion:
MELD-Na is an independent predictor of anastomotic leak in proctectomies.  Anastomotic leak risk increases with increasing MELD-Na in elective colorectal resections, as does 30-day mortality and overall complication rate.  As MELD-Na score increases to above 20, restorative proctectomy has a 10% rate of anastomotic leak. 

27.01 Transgastric pancreatic necrosectomy – expedited return to pre-pancreatitis health

Posted on January 25, 2017

M. M. Dua1, D. J. Worhunsky1, L. Malhotra1, W. G. Park1, J. A. Norton1, G. A. Poultsides1, B. C. Visser1  1Stanford University,Palo Alto, CA, USA

Introduction:  The best operative strategy for necrotizing pancreatitis remains controversial. Traditional surgical necrosectomy is associated with significant morbidity; operative debridement contributes to the substantial risk of pancreatic
and bowel fistulae, which are associated with recurrent hospitalizations and long-term support to manage pain or nutritional requirements. Minimally invasive endoscopic and percutaneous strategies typically require multiple procedures and a prolonged hospital course. We developed a transgastric approach to pancreatic necrosectomy to overcome the shortcomings of the other techniques described.

Methods:  Patients with walled-off, retrogastric pancreatic necrosis who underwent transgastric necrosectomy (TN) during 2009-2016 were retrospectively reviewed. Open TN is performed via an anterior gastrotomy to debride the pancreas through a wide cystgastrostomy in the posterior wall. Laparoscopic TN involves endoscopic insufflation of the stomach for placement of transgastric ports for operative debridement. The cystgastrostomy is left open in both types of TN to allow ongoing internal drainage of necrosis. Endpoints included postoperative complications and mortality.

Results: Forty-four patients underwent TN (9 open, 35 laparoscopic). Operative indications included persistent unwellness (n=26), infection (n=14), pseudoaneurysm hemorrhage failing embolization (n=3), and worsening sepsis (n=1). The median peroperative APACHE II score for the total cohort was 6 (0-27); however, disease severity was higher in the open TN group compared to the laparoscopic TN group (APACHE II score 12 vs 5, p = 0.03) resulting in a longer length of stay (LOS 11 vs 7 days, open vs laparoscopic, respectively, p = 0.01). Clinical outcomes for the total cohort are represented in the attached table.  A majority of the cohort (74%) experienced none (n=23) or minor (n=10) complications. Six patients had postoperative bleeding; 5 required embolization and there was one death. No patient required more than one operative debridement; five patients required percutaneous drainage for residual collections. There were no postoperative fistulae or wound complications.

Conclusion: The transgastric approach to pancreatic necrosectomy allows for effective debridement with a single definitive operation and minimizes the morbidity associated with prolonged drainage, fistulae and wound complications. When anatomically suitable, the transgastric approach (whether laparoscopic or open) is an effective strategy that expedites return to pre-pancreatitis health and offers significant benefits in the recovery of these patients.   

26.10 Early Identification of Deep Space Infection in Colorectal Surgery

Posted on January 25, 2017

J. R. Bergquist1,2, C. B. Storlie2, K. L. Mathis1, J. C. Boughey3, D. A. Etzioni4, E. B. Habermann2, R. R. Cima1  1Mayo Clinic,Division Of Colon And Rectal Surgery,Rochester, MN, USA 2Mayo Clinic,Robert D And Patricia E Kern Center For The Science Of Health Care Delivery,Rochester, MN, USA 3Mayo Clinic,Department Of Surgery,Rochester, MN, USA 4Mayo Clinic In Arizona,Colon And Rectal Surgery,Phoenix, AZ, USA

Introduction:  Key drivers of colorectal surgical-site-infection (C-SSI) occurrence are institution-specific, and early identification of patients who will develop C-SSI requiring readmission remains challenging. We developed an analytic tool which would utilize institution-specific data for C-SSI screening and treatment during index hospitalization. 

Methods:  Elective colorectal resections from institutional ACS-NSQIP datasets (2006-2014) at 2 locations were included. A Bayesian-Probit regression model with multiple-imputation (BPMI) via Dirichlet process handled missing data. The baseline for comparison was a multivariate logistic regression model (GLM) with indicator variables for missing data (e.g., adding a “missing” level to factors) and stepwise variable selection. Out-of-sample performance was evaluated with Receiver Operating Characteristic (ROC) and Net Reclassification Improvement (NRI) analysis of 10-fold cross-validated samples. Primary endpoint was C-SSI requiring hospital readmission. 

Results: Among 2376 resections, deep/organ space C-SSI rate was 4.6% (N=108: Figure-patients 3,4). Among patients developing C-SSI, N=65(60.1%) were discharged prior to clinical diagnosis (Figure-patient 3). The tool identified N=15(23.1%) of these patients prior to discharge (3 requiring re-operation), with 10% false alarm rate. Among patients clinically diagnosed with C-SSI prior to discharge (patient 4), the tool identified C-SSI 4.5 (mean) days prior to clinical identification. Tool performance generated ROC=0.77 and NRI=21.7%, demonstrating high predictive accuracy. When applied to independent validation data (N=478 cases, N=20 SSI), the tool identified during hospitalization 40% of patients discharged then readmitted with C-SSI (ROC=0.75; NRI=8.4%). 

Conclusion: Identification of C-SSI prior to clinical presentation can facilitate early intervention, potentially reducing morbidity, re-admission, and re-operation. Our tool correctly identified a substantial proportion of patients who were discharged and readmitted with C-SSI in two independent datasets. This institutionally-generic analytic tool can improve outcomes and reduce costs associated with readmission and late C-SSI identification.  

 

26.09 Use of Dual Lumen VV ECMO for Neonatal and Pediatric Patients in a Tertiary Care Children’s Hospital

Posted on January 25, 2017

J. L. Carpenter1, Y. R. Yu1, D. L. Cass1, O. O. Olutoye1, J. A. Thomas2, C. Burgman2, C. J. Fernandes3, T. C. Lee1  1Texas Children’s Hospital,Department Of Surgery,Houston, TX, USA 2Texas Children’s Hospital,Critical Care Section, Department Of Pediatrics,Houston, TX, USA 3Texas Children’s Hospital,Neonatology Section, Department Of Pediatrics,Houston, TX, USA

Introduction:

Recent advances in extracorporeal membrane oxygenation (ECMO) have led to increased use of venovenous (VV) ECMO in the neonatal and pediatric patient population yet there is little data on outcomes related to this technology. Reported complications of dual lumen VV ECMO have included need for venoarterial (VA) conversion and cardiac perforation. We present the evolution and experience of neonatal and pediatric VV ECMO at a tertiary care institution.

Methods:

Records for NICU and PICU patients who received ECMO support from 01/2005 to 07/2016 were reviewed. Comparison groups included cannulation mode and indication for ECMO. Analyses of survival to discharge, complications (metabolic, hemorrhagic, neurologic, renal, cardiovascular, pulmonary, infectious, and mechanical), and decannulation rate were performed with χ2 tests. Kaplan Meier analysis was used to compare survival based on indication for therapy.

Results:

A total of 160 patients (105 NICU, 55 PICU), ages 0 days – 19 years, required 13 ± 11 days of ECMO. Indications were sepsis (8%) and cardiorespiratory failure (92%); of which 44% (n=64) had diaphragmatic hernia. VV ECMO was the primary cannulation mode in 83 patients with a survival of 64%. VA ECMO was used in 77 patients with 54% survival. Nine VV patients (11%) required VA conversion. VV cannulas were placed percutaneously in 45% of patients (n=37) with 16 placed via an existing central line. Ten VV patients were extubated to spontaneous respirations while on ECMO; three survived to discharge. Overall, 74% of patients (n=118) were successfully decannulated and 57% survived to discharge. Since 2010, the frequency of VV cannulation increased from 50% to 85% and the mortality rate was unchanged. VA ECMO was associated with a significantly higher rate of acute intra-cranial hemorrhage than VV (28% vs 9%, p=0.003). There were no differences in survival (p=0.52), complications (p=0.40), or re-operation rate (p=0.85) between VV and VA groups. There were no cardiac injuries with the use of double lumen VV cannulas. Survival by ECMO indication is shown in Figure 1. There is a significant difference in overall survival (p=0.002); septic patients had a median survival of 20 days, whereas patients with cardiorespiratory failure had a median survival of 129 days.

Conclusion:

VV ECMO cannulation is associated with a lower rate of intra-cranial hemorrhage and may be the preferred first-line mode of ECMO support for cardiorespiratory failure.  VV can be an effective mode of ECMO support in the both the NICU and PICU populations, though conversion to VA ECMO may occasionally be necessary.

 

26.08 The Effect of Resident Involvement on Perioperative Outcomes in Bariatric Surgeries

Posted on January 25, 2017

J. Kudsi1, K. Hayes1, R. Amdur1, P. Lin1, K. Vaziri1  1George Washington University,Department Of General Surgery,Washington, DISTRICT OF COLUMBIA, USA

Introduction:

The current surgical residency training is based on a model of graduated responsibility, giving greater responsibility based on an individual trainees’ ability. To assess the effect of this model on care of the bariatric surgery patient; we decided to study the impact of resident involvement in bariatric surgery stratified by level of training. The aim of this study is to assess the impact of resident involvement on perioperative outcome in bariatric procedures including sleeve gastrectomy (SG) and gastric bypass (GB). 
 

Methods:

Four-year retrospective review 2006-2010 of ACS-NSQIP database for 19,616 lap/open GB (87.8%), and 2,730 lap/open SG (12.2%). All concurrent procedures were excluded except: EGD, liver biopsy, wedge liver biopsy, and lap biopsy. Other exclusions included cases with both laparoscopic and open procedures, emergency cases, cases missing PGY, and those with PGY level higher than 5.  

Pre-treatment patient characteristics and outcomes were compared across levels of the resident variable Juniors PGY 1/2/3 (J) vs seniors PGY 4-5 (S) vs no residents (N) using chi-square for categorical variables and analysis of variance for continuous variables. 

13 composite outcomes were compared; wound (superficial surgical site infection, deep wound infection, organ space infection, dehiscence), pulmonary (pneumonia, prolonged intubation, reintubation), sepsis/septic shock, deep venous thrombosis/pulmonary embolism (DVT/PE), bleeding, cardiac (MI, cardiac arrest), renal (AKI, dialysis) and urinary tract infection (UTI), operative time>4h, Length of Stay (LOS)>3days, return to OR, and mortality. All confounding variables were controlled.
 

Results:

There were 19,616 GB (87.8%), and 2,730 SG (12.2%). Surgical assist distribution was: Junior 3,554 (15.9%), Senior 5,406 (24.2%), N 13,386 (59.9%). 

Cases that included Senior residents more often involved African American patients than cases treated by no resident or junior residents. Non-resident cases had slightly lower BMI, fewer non-independent patients, and less COPD, than cases that involved residents. Cases with Junior residents had the highest rate of dyspnea.

There was no difference in mortality. Senior residents had significantly worse outcomes compared to junior and non-residents in LOS>3 days (S 12.4%, J 10.9%, N 8.3%, P<.0001), wound complications (S 3.6%, J 2.8%, N 2.7% P.007), pulmonary complications (S 1.2%, J 0.7%, N 0.9 P .033), sepsis/ septic shock (S 1.3%, J 1.1%, N 0.8% P .0022), cardiac events (S 0.33, J 0.14, N 0.12, P 0.005) and OR time>4h (S 4.16, J 4.11, N 1.6, P < 0.0001). Junior residents had worse outcome in renal complications (S 0.46, J 0.51, N 0.23, P 0.006).

Conclusion:

Bariatric procedures with senior resident assistance have worse outcomes when compared to junior or non-resident assistance. These results suggest further evaluation of the graduated responsibility model in bariatric operative education.
 

26.07 Hospital Variation in Bariatric Surgery Episode Costs in the Commercially-Insured

Posted on January 25, 2017

A. Kelsall1, A. Ghaferi2  1University Of Michigan,Medical School,Ann Arbor, MI, USA 2University Of Michigan,Department Of Surgery,Ann Arbor, MI, USA

Introduction:  Bundled payments are gaining momentum as a cost-containment measure that will have sweeping impacts on the provision of surgical care. Despite evidence that private payers are increasingly interested in bundled payment models, research on surgical episode cost variation for common procedures, such as bariatric surgery, has largely focused on Medicare beneficiaries and excluded the commercially-insured population. In this context, we examined hospital-level variation in bariatric surgery episode cost for commercially-insured patients in Michigan.

Methods: Using data from a state-wide collaboration between a major commercial insurer and hospitals in Michigan, we retrospectively identified patients undergoing bariatric surgery procedures- sleeve gastrectomy (sleeve) or Roux-en-Y gastric bypass (RYGB)- from January 2009 through October 2014. We included in the analysis only those hospitals that performed a minimum of 30 procedures during the study period (N=9035 procedures, 31 hospitals). We calculated price- and risk-adjusted payments from index admission to 30 days post-discharge. We divided hospitals into quintiles based on average episode cost and examined variation in four components of episode cost, namely index hospitalization, professional fees, post-acute care, and readmissions. 

Results: We found the average risk and price-adjusted payment for a bariatric surgery episode was $12,246. The highest-cost quintile averaged $1,519 (12%) more per episode than the lowest quintile.  Index hospitalization accounted for the largest share of episode payments (73% and 80% for RYGB and sleeve, respectively) and explained much of the variation between top and bottom quintiles (58.3% for RYGB, 35% for sleeve) (fig. 1). Professional fees accounted for a significant proportion of episode payments (19.4% and 13.4% for RYGB and sleeve procedures, respectively) and drove a roughly proportional share of variation between highest and lowest cost quintiles. Readmission and post-discharge payments accounted for disproportionate shares of the variation between quintiles. For example, in sleeve procedures, post-discharge payments accounted for 3.9% of total episode payments, but explained 22.6% of variation between highest and lowest cost quintiles.

Conclusion: Our findings demonstrate substantial variation in bariatric surgery episode costs in the commercially-insured population. While index hospitalization accounted for the largest share of episode costs, variation in other cost components (i.e., readmissions and post-discharge payments) explained a share of variation disproportionate to their contribution to overall cost, suggesting they are potential targets for quality and efficiency-improvement efforts under bundled payment models.
 

26.06 Symptomatic Hematomas Following Cervical Exploration: A Comparative Analysis over 40 Years

Posted on January 25, 2017

A. Jyot1, T. Pandian1, M. H. Zeb1, N. D. Naik1, A. Chandra1, F. J. Cardenas1, M. Mohan1, E. H. Buckarma1, D. R. Farley1  1Mayo Clinic,General Surgery,Rochester, MINNESOTA, USA

Introduction: Cervical hematoma is a highly dreaded complication of cervical exploration and poses a unique challenge due to the combination of its rarity and associated high mortality. Rising endocrine case volumes over the last decade underscore the need for better understanding of this life threatening condition. We previously reported on the incidence (0.31%) and outcomes of this complication from 1976-2000 (Study A). Given that many of these operations are now performed through smaller incisions and as outpatients, we aim to analyze the complication rate and possible changes in trends from 2001-2015 (Study B).

Methods: A retrospective case-control study including 10,138 patients undergoing thyroidectomy and parathyroidectomy from 2001- 2015 was conducted. Cases were matched 1-to-1 for gender, age, type and year of operation. Univariate analysis testing was performed to assess for baseline discrepancies between study groups followed by a conditional logistic regression to identify perioperative risk factors.

Results: Thirty-two hematomas requiring re-exploration were identified (Study B incidence =0.30%, Study A incidence=0.31%). There were 24 women and 8 men (mean age= 58.4±17.1 years), undergoing thyroidectomy (22), parathyroidectomy (8) and both procedures (2). No perioperative risk factors for developing a cervical hematoma were identified. Most hematomas (n=18, 56%) presented within 6 hours of wound closure, while 7 (22%) presented between 7and 24 hours and 7 (22%) beyond 24 hours. This was in contrast to study A where the most common time of presentation was beyond 6 hours (43%). Neck swelling was the most common presenting symptom (n=22, 69%), followed by neck pain (n=8, 25%), respiratory distress (n=6, 19%), dysphagia (n=6, 19%) and wound discharge (n=4, 13%). At re-exploration, 19 (60%) hematomas were found to be deep and 13 (40%) superficial to the strap muscles. The bleeding source was identified in 24 (75%) cases (11 arterial, 8 venous, 3 diffuse oozing and 2 with oozing and venous bleeding). In our study and control groups, vocal cord paralysis/voice change (25 vs. 22, p=0.171), followed by hypocalcemia (5 vs. 3, p=0.708), were common complications, however no complication reached statistical significance. Mean hospital stay was longer in the patients requiring cervical re-exploration (3.1 days vs. 1.6 days, p=0.005).

Conclusion: The frequency of cervical hematomas remains unaltered over 4 decades. Failure to define a high risk population in the current study highlights the need for meticulous hemostasis. With increasing outpatient neck surgery, scrutiny prior to dismissal and clear patient education regarding symptomatic cervical hematomas is imperative.

 

 

26.05 Exponential Decay Modeling Can Define Parameters of the Weight Loss Trajectory After Gastric Bypass

Posted on January 25, 2017

E. S. Wise1,2, J. Felton1, M. D. Kligman1  1University Of Maryland Medical Center,Department Of Surgery,Baltimore, MD, USA 2Vanderbilt University Medical Center,Department Of Surgery,Nashville, TN, USA

Introduction:

Laparoscopic Roux-en-Y gastric bypass (LRYGB) is a well-described operation that produces durable and clinically significant weight loss. While factors influencing future weight loss have been studied, temporal patterns of weight loss are less well described. We test the hypotheses that postoperative weight loss may conform to exponential decay, and subsequently, that three-month weight loss can help characterize a patient’s weight loss trajectory.

Methods:

A retrospective analysis of 1,097 consecutive LRYGB patients at a single institution over a ten year period provided the data necessary to generate postoperative weight loss curves. Using pre- and postoperative BMI data, the mean and standard deviation of postoperative BMI as a function of time was obtained, with multiple linear and nonlinear fits tested for optimal conformity. The highest and poorest performing patients at three month follow-up were stratified based on their cumulative rate of weight loss into two strata: <0.3% EBMIL/day (n = 102) and >0.5% EBMIL/d (n = 191). Exponential decay rate constants (λ) were generated for each group, allowing for optimized estimation of time until half of the weight loss is complete (t1/2) as well as plateau BMI (BMIf). Linear regression analysis was used to interrogate the association of λ, calculated at three months and normalized using a BMIf of 25 kg/m2 (λn, 3 mo.), to %EBMIL at 2-3 years, a surrogate for actual BMIf.

Results:

For the entire cohort, one-phase exponential decay provided the best fit for the weight loss function over time (n = 1,097, r = .43, λ (x 103) = 7.3, t1/2 = 95 days, BMIf = 31.4). Patients who performed poorly at three months (<0.3% EBMIL/d, n = 102, r = .49, lambda(x 103) = 6.4, t1/2 = 108 days, BMIf = 38.9) had a smaller λ and a higher BMIf than those who performed optimally (>0.5% EBMIL/d, n = 191, r = .62, λ (x 103) = 9.4, t1/2 = 74 days, BMIf = 26.6; Figure 1A). Normalized rate constants calculated from three-month weight loss (λn,3 mo.) demonstrated a significant correlation with %EBMIL2-3 years (n = 428, P < .001, r = .28, B = -4.1 %EBMIL per unit increase in normalized lambda; Figure 1B).

Conclusion:

We demonstrate that weight loss after LRYGB conforms to exponential decay. This finding necessarily submits that weight loss trajectory is governed by a patient-specific rate constant and a plateau BMI. Further studies are necessary to characterize the patient and institution-specific factors that contribute to these parameters. However, we find that patient performance at three months, as suggested by λn,3 mo, is a significant predictor of long term weight loss in accordance with our hypothesis.

 

26.04 Defining optimal opioid pain medication prescription length following common surgical procedures

Posted on January 25, 2017

R. E. Scully1, W. Jiang1, A. Schoenfeld1, M. A. Chaudhary1, S. Lipsitz1, P. Learn2, T. Koehlmoos2, A. Haider1, L. L. Nguyen1  1Brigham And Women’s Hospital,Center For Surgery And Public Health,Boston, MA, USA 2Uniformed Services University Of The Health Sciences,Bethesda, MD, USA

Introduction: Over-prescription of pain medications has been implicated as a driver of the burgeoning opiate epidemic. Accordingly, legislation limiting length of initial opioid prescription, typically to less than 7 days, has been recently passed in several states. The goal of the current project was to describe opioid pain medication prescription patterns following common surgical procedures and to determine the appropriateness of the prescription as indicated by the rate of prescription refills.

Methods: The Department of Defense Military Health System Data Repository (MDR) tracks care delivered to active, disabled, and retired members of the US armed forces and their dependents insured through TRICARE. The MDR was queried for individuals (age 18-64) who had undergone common surgical procedures between 2006 and 2014. Procedures chosen were cholecystectomy, appendectomy, hernia repair, anterior cruciate ligament reconstruction, rotator cuff tear repair, discectomy, mastectomy, and hysterectomy. Individuals with a prior diagnosis of chronic pain, substance dependence, or an opioid prescription within the 6 months preceding the index procedure were excluded. Refill was defined as repeat opioid prescription within 14 days of the end of the initial prescription. Adjusted risk of opioid prescription refill by number of days of initial prescription was modeled using a generalized additive model with spline smoothing.

Results: Among the 203,834 individuals included, the median length of initial opioid prescription was 4 days [Interquartile Range (IQR): 3-5 days] for general surgery procedures, 4 days [IQR 3-5 days] for women’s health procedures, and 6 days [IQR 5-9 days] for musculoskeletal procedures. When adjusted for clinical and demographic factors, the proportion of individuals requiring refill was low, regardless of length of initial prescription and type of procedure (Figure 1). The early nadir in probability of refill was at an initial prescription of 11 days for general surgery procedures (probability of refill = 7.87%), 14 days for women’s health procedures (probability of refill = 11.55%), and 19 days for musculoskeletal procedures (probability of refill = 1.36%).

Conclusions: An opiate prescription after surgery aims to balance adequate pain treatment while minimizing the duration of treatment.  The statistically-modeled optimal initial prescription length appears to be somewhat longer than 7 days for the surgical procedures included here. In practice, the optimal length of opiate prescription lies between the observed median prescription length and the early nadir, or 4 to 11 days for general surgery procedures, 4 to 14 days for women’s health procedures, and 6 to 19 days for musculoskeletal procedures.

26.03 Health Care Consumption and Sick Leave for Persistent Abdominal Pain after Cholecystectomy

Posted on January 25, 2017

S. Z. Wennmacker1, M. G. Dijkgraaf4, G. P. Westert3, J. P. Drenth2, C. J. Van Laarhoven1, P. R. De Reuver1  1Radboud University Medical Center,Surgery,Nijmegen, , Netherlands 2Radboud Univeristy Medical Center,Gastroenterology And Hepatology,Nijmegen, , Netherlands 3Radboud University Medical Center,Scientific Institute For Quality Of Healthcare (IQ Healthcare),Nijmegen, , Netherlands 4Academic Medical Center,Clinical Research Unit,Amsterdam, , Netherlands

Introduction: Annually, 800.000 cholecystectomies are performed in the United States and 22.000 in the Netherlands. Estimated costs of a cholecystectomy in the Netherlands are around 4000 euro’s. Gallbladder removal for symptomatic gallstones appears to be ineffective in terms of pain relief, in up to 40% of patients. Although several studies have reported on persistent abdominal pain after cholecystectomy, there is no literature on the actual burden of persistent pain to the health care system. The aim of this study is to determine health care consumption and the related costs in patients with persistent abdominal pain after cholecystectomy.

Methods: All 146 patients of a previous prospective multicenter cohort study who reported persistent abdominal pain 24 weeks after cholecystectomy between June 2012 and June 2014 were included in this study. Health care consumption was assessed in February 2016 using Patients experience of surgery questionnaire (PESQ), Medical Consumption Questionnaire (iMCQ) and patients’ medical records. Sick leave and productivity loss of (un)paid work were assessed by the Productivity Cost Questionnaire (iPCQ). Costs were calculated according the Dutch “Guideline for performing economic evaluations in health care” and reported in euro's.

Results: The response rate was 85% (124/146 patients), after a mean follow-up of 31.0 months after surgery (SD 6.5). A total of 55.6% (n=69) of patients had additional care for persistent abdominal pain after cholecystectomy; 30.6% received primary care, 37.1% received secondary care, 16% were admitted in the emergency department, and 8.9% of the patients were admitted to hospital. Diagnostic procedures were performed in 33.9% (n=42) of the patients, which revealed gallstone or surgery related causes in nine patients. In 20 patients another diagnosis was found. Additional treatment included use of medication in 17.7% (n= 22) of the patients (10% uses analgetics, 9.6% uses proton pomp inhibitors ). Additional interventions were performed in 7 patients (5.6%). Estimated mean medical costs for persistent abdominal pain since cholecystectomy were €1,239 (SD €3,573) per patient. Subsequent mean costs of sick leave and productivity loss of (un)paid work were €727 (SD €2,163) per patient.

Conclusion: Due to persistent abdominal pain after cholecystectomy, 55% of the patients needed additional health care, and one third of the patients underwent additional diagnostic procedures. Postoperative medical costs and costs of sick leave and productivity loss in patients with persistent abdominal pain are up to 50% of the initial costs of the cholecystectomy.

 

26.02 Timed Stair-Climbing is a Surrogate for Sarcopenia Measurements for Predicting Surgical Outcomes

Posted on January 25, 2017

S. Baker1, M. Waldrop1, J. Swords1, T. Wang1, M. J. Heslin1, C. M. Contreras1, S. Reddy1  1University Of Alabama at Birmingham,Surgical Oncology,Birmingham, Alabama, USA

Introduction: Estimating sarcopenia by measuring psoas muscle density (PMD) has been advocated as a method to accurately predict post-operative morbidity. This method is cumbersome and not feasible for a busy surgeon to use in practice. We have previously demonstrated that a simple timed stair climb (TSC) outperforms the ACS NSQIP Surgical Risk Calculator in predicting morbidity and hospital length of stay. The aim of the present study is to determine whether the TSC is a marker of axial muscle strength and can be used to replace PMD measurements in predicting morbidity.

Methods: From March 2014 to May 2015, 298 patients attempted TSC prior to undergoing elective abdominal surgery. PMD was measured using pre-operative CT scans obtained within 30 days of surgery at the L3 vertebra. Ninety-day complications were assessed using the Accordion Severity Grading System. Multivariable analysis was performed to identify pre-operative risk factors associated with operative morbidity.

Results: A grade 2 or higher complication occurred in 72 (24.2%) of patients with 8 (2.7%) deaths. There was an inverse relationship between PMD and TSC (Figure, P<0.0001) and a direct relationship between TSC and complications (Figure, P=0.04). On multivariable analysis only the decreasing PMD (P=0.018) and increasing TSC (P=0.026) were predictive of post-operative morbidity.  Area under the receiver operating characteristic curves demonstrated no difference between TSC and PMD in predicting complications (AUC 0.72 vs. 0.70, P=0.49).  Both TSC and PMD were superior to the ACS NSQIP Risk Calculator (AUC 0.55, both P<0.05).

Conclusions: Both TSC and PMD are excellent predictors of post-operative morbidity in this population. TSC appears to be a surrogate for axial muscle strength measured by PMD. TSC is an easy tool to administer in lieu of PMD when considering patient outcomes.

26.01 Surgeons Overestimate Post-operative Complications and Death.

Posted on January 25, 2017

K. Pei1, J. Healy1, K. A. Davis1  1Yale University School Of Medicine,Surgery,New Haven, CT, USA

Introduction:

 

Assessing post-operative morbidity and mortality is largely based on experience and published statistics. Projections of complications and death have critical implications for counseling patients preoperatively, particularly for challenging patients. We hypothesize that resident and attending surgeons overestimate complications and death after surgery for complex surgical patients. 

Methods:

General surgery residents and attending surgeons at an urban, tertiary, academic medical center were invited to participate in an online assessment.  Seven complex clinical scenarios were presented to participants via anonymous, online modules.   For each scenario, participants estimated the likelihood of any morbidity, mortality, surgical site infection, pneumonia, and cardiac complications on a 0-100% scale. Scenarios were representative of a diverse General Surgery practice including colectomy, duodenal ulcer repair, inguinal hernia repair, perforated viscus exploration, small bowel resection, cholecystectomy, and mastectomy.   Participant responses were compared to risk adjusted outcome measures by the American College of Surgeons National Surgical Quality Improvement Project (NSQIP) online calculator.   Responses were reported as means with 95% confidence intervals, differences in participant responses and NSQIP estimates were reported as absolute percentage differences of the mean.  This study was approved by the institutional Human Investigation Committee.

Results:

101 Residents and 48 Attending Surgeons (trained in General Surgery) were invited. Overall response rate was 73.8% for all participants.  For all 7 clinical scenarios, there was no significant difference between resident and attending estimates of morbidity or mortality.  There was significant variation in estimates among participants with wide 95% confidence intervals.  Overall, the mean percentages of the estimates were 25.8-30% over NSQIP estimates for morbidity and mortality.

Conclusion:

General surgery residents and attending surgeons did not significantly differ in their estimate of post-operative complications and death; however, both groups grossly overestimated risks in complex surgical patients.  These results demonstrate broad variance in and near universal overestimation of predicted surgical risk when compared to national, risk adjusted models.

25.10 UAB30, a novel retinoid, decreases tumor burden in a model of disseminated medulloblastoma

Posted on January 25, 2017

E. F. Garner1, G. K. Friedman3, V. R. Atigadda2, L. Nan3, B. P. Moore3, T. Etminan3, L. L. Stafman1, J. E. Stewart1, D. D. Muccio2, C. J. Grubbs4, E. A. Beierle1  1University Of Alabama at Birmingham,Division Of Pediatric Surgery,Birmingham, Alabama, USA 2University Of Alabama at Birmingham,Department Of Chemistry,Birmingham, Alabama, USA 3University Of Alabama at Birmingham,Division Of Pediatric Hematology And Oncology,Birmingham, Alabama, USA 4University Of Alabama at Birmingham,Department Of Surgery,Birmingham, Alabama, USA

Introduction:  Medulloblastoma (MB) is the most common malignant brain tumor in children.  There are four molecular subtypes of MB: WNT, SHH, Group 3 and Group 4, with Group 3 tumors having the most dismal prognosis. Our laboratory has previously shown that the novel retinoid, 9-cis-UAB30 (UAB30), inhibited MB proliferation in vitro and prolonged animal survival in a localized human patient-derived xenograft (PDX) MB mouse model. We hypothesized that UAB30 would also be effective in limiting the metastatic spread in a disseminated MB mouse model.

Methods:  We used a mouse PDX model with bioluminescent imaging to follow the spread of disease. The group 3 MB PDX cell line D341 labeled with GLuc-GFP was used for the in vivo study.  5 x 105 cells were stereotactically injected into the right lateral ventricle of 5-week-old athymic nude mice. The mice were randomized into three groups to receive control chow (n=8), 13-cis-retinoic acid (RA)-treated chow (n=8), or UAB30-treated chow (n=8). Treatment was initiated on the day of injection. Bioluminescence imaging was performed on days 7, 11, 14, 18, and 20 post-tumor injection using the IVIS Lumina III (Perkin Elmer). The animals were euthanized on day 20 when they met IACUC parameters. Tumor burden was quantified using total flux (photons/sec) in both the brain and spine. Student’s t-test was used to compare the mean bioluminescence between groups with p ≤ 0.05 considered significant. 

Results: Control mice had significantly greater tumor burden (9.87 x 107 photons/sec) in the brain at day 20 when compared to RA-treated (3.61 x 107 photons/sec; p = 0.02) and UAB30-treated (3.84 x 107 photons/sec; p = 0.03) mice (Fig. 1A and 1B). The UAB30-treated mice had significantly less spinal metastasis compared to the control mice at day 18 and day 20 (p = 0.02 and p= 0.04, respectively) (Fig. 1A and 1C). The RA-treated mice initially had less spinal metastasis than control mice at day 18 (p =0.05), but this difference was not significant at day 20 (Figure 1 C). 

Conclusion: In this PDX model of disseminated MB, treatment with UAB30 significantly delayed progression of tumor burden in the brain and spine compared to untreated mice.  These data suggest that this novel retinoid should be further explored as a potential therapeutic for MB. 

 

25.09 Modifications to T Cell Receptor Enhance Anti-Tumor Activity Against Metastatic Melanoma

Posted on January 25, 2017

E. H. Wood1, K. Calabrese2, D. Murray2, K. Foley2, K. Nagato2, M. Nishimura1,2  1Loyola University Medical Center,Surgery,Maywood, ILLINOIS, USA 2Loyola University Chicago,Oncology Institute,Chicago, IL, USA

Introduction:
Adoptive cell transfer of T cells engineered to express receptors with specific anti-tumor effects has shown great promise in the treatment of metastatic melanoma. However, exogenously introduced T cell receptors (TCR) must compete with endogenous receptors for surface expression resulting in blunted anti-tumor activity. Our goal is to create TCR modifications that will enhance receptor pairing resulting in increased surface expression and improved tumor recognition.  

Methods:
Using TCR 1383I, a receptor against the melanoma antigen tyrosinase, the following modifications were made: codon optimization, introduction of a disulfide bridge into the constant region, introduction of a cytoplasmic leucine zipper, and murinization of the constant regions. These modifications have all been demonstrated in the literature to improve receptor pairing; however, they have never been tested side by side in the same TCR. Peripheral blood lymphocytes were transduced with each modification. Surface expression and peptide recognition were studied using multicolor flow cytometry, alanine scanning, and intracellular and extracellular cytokine release assays (IFN-γ  TNF-α , IL-2, IL-4, IL-22, IL-17A, and CD107a).

Results:
While the disulfide bridge and codon optimized TCRs demonstrated enhanced surface expression compared to the wild type receptor, there was no improvement in peptide recognition and anti-tumor effects. Only the murinized TCR significantly improved receptor pairing and surface expression. This receptor dramatically increased antigenic peptide recognition, and also altered the complement of cytokines produced by the lymphocytes. T cells expressing the murinized TCR not only secreted more cytokine but exhibited increased polyfunctionality compared to all other modifications and wild type. We utilized alanine scanning to delineate the interaction between the tyrosinase peptide and each TCR modification. The disulfide bridge, codon optimized, and leucine zipper modified TCRs demonstrated the same peptide recognition patterns as the wild type receptor, however the murinized receptor appeared to be less restricted. 

Conclusion:
TCR modifications were created in an attempt to enhance receptor pairing and T cell functionality. In the context of TCR 1383I, only the murinized modification consistently improved pairing, T cell surface expression, and anti-tumor functionality. TCRs require a CD3 coreceptor to initiate its signaling cascade and murine TCRs have been shown to have higher affinity for CD3 allowing our target TCR to outcompete endogenous receptor expression. The modification may have also created a conformational change allowing better recognition of target antigens. Future studies are needed to assess peptide-TCR interactions on a molecular level, to examine cross-reactivity against other peptides, and to establish an in vivo model.
 

25.08 Circulating Tumor Cell Phenotypic and Genotypic Profiling for Hepatocellular Carcinoma

Posted on January 25, 2017

C. M. Court1,5, S. Hou1,3, S. Sho1,5, P. Winograd1,5, Q. Li1, S. Sadeghi4, R. S. Finn4, F. W. Samuel3,6, B. V. Naini6, F. M. Kaldas1,2, R. W. Busuttil1,2, J. S. Tomlinson1,5, H. R. Tseng3, V. G. Agopian1,2  1University Of California – Los Angeles,Surgery,Los Angeles, CA, USA 2University Of California – Los Angeles,Liver Transplantation,Los Angeles, CA, USA 3University Of California – Los Angeles,Molecular And Medical Pharmacology,Los Angeles, CA, USA 4University Of California – Los Angeles,Hematology/Oncology,Los Angeles, CA, USA 5VA Greater Los Angeles,Surgery,Los Angeles, CA, USA 6University Of California – Los Angeles,Pathology,Los Angeles, CA, USA

Introduction: Treatment options available to patients with hepatocellular carcinoma (HCC) include potentially curative surgical therapy (resection and liver transplantation) for early stage patients and molecular targeted therapy for advanced/metastatic patients. Currently, treatment selection is based on clinical and radiological staging of disease, partly due to the lack of available biomarkers to inform treatment decisions. Circulating tumors cells (CTCs) are one such potential biomarker with proven efficacy in many cancers, but have not been well characterized in HCC. We sought to develop a novel, blood-based assay capable of detecting HCC CTC phenotypes and genotypes with prognostic importance that may impact treatment selection.

Methods: Utilizing the microfluidic NanoVelcro platform which allows for multichannel immunocytochemistry as well as single cell isolation and sequencing, we prospectively evaluated 4 milliliters of blood from 69 patients (50-HCC, 11-benign liver disease, 8-healthy controls) and developed assays to enumerate and characterize CTC phenotype (epithelial-type cytokeratin (CK+) CTCs, mesenchymal-type Vimentin (Vim+) CTCs, programmed death-ligand 1 (PDL1+) CTCs) and identify CTC genotype by sequencing for actionable somatic mutations associated with targeted drug therapy. CTC number and phenotype was correlated with clinicopathologic data. 

Results: CK+ CTCs were detected in 96% of HCC patients (Mean=6.3, range=1-27) and discriminated HCC and non-HCC patients (≥2 CTCs, sensitivity=90.0%, specificity=89.5%, AUROC=0.946, p<0.001). Mesenchymal-type Vim+ CTCs were detected nearly exclusively in advanced (stage III/IV) patients (Mean = 6.5, range = 1-20), with excellent discrimination between early, transplant-eligible and advanced stage, transplant-ineligible patients (AUROC = 0.970, p < 0.001; Fig). In advanced HCC patients, PDL1+ CTCs were detected in 5/8 (63%) of patients, and were noted to decrease following treatment with nivolumab, a PD-1 inhibitor. Using our single CTC isolation and sequencing method, we were able to perform gene panel sequencing in CTCs from 10 patients. In two of those patients, potentially actionable mutations (PIK3CA and TP53) were found in CTC-DNA. 

Conclusion:CTCs are a promising serum biomarker in HCC with potential treatment implications for both early and late stage patients. Vim+ CTCs accurately discriminated early and late stage HCC patients and may improve transplant candidate selection. For advanced stage patients, both CTC phenotype (PDL1+) and genotype analysis may allow for selection of specific targeted therapies. Longitudinal follow-up with evaluation of cancer-specific outcomes is necessary to establish whether CTCs may improve current treatment algorithms.

25.07 Mutant-Allele Tumor Heterogeneity Scores Correlate With Neoadjuvant Therapy Response in Rectal Cancer

Posted on January 25, 2017

A. Greenbaum2, S. Ness4, T. Bocklage3, J. Lee1, A. Rajput2  1University Of New Mexico HSC,Epidemiology, Biostatistics And Preventative Medicine/Internal Medicine,Albuquerque, NM, USA 2University Of New Mexico HSC,Surgery,Albuquerque, NM, USA 3University Of New Mexico HSC,Pathology,Albuquerque, NM, USA 4University Of New Mexico HSC,Internal Medicine,Albuquerque, NM, USA

Introduction:   Neoadjuvant chemoradiation is the standard of care for locally advanced adenocarcinoma of the rectum.  It is currently unknown which patients will respond to therapy.  We aimed to determine if Mutant-Allele Tumor Heterogeneity (MATH) scores, a novel bioinformatics tool, can predict response to neoadjuvant treatment in locally advanced rectal tumors.

Methods:    We performed high read-depth (“deep”) sequencing of >400 cancer-relevant genes on a group of 13 patients with locally advanced rectal adenocarcinoma.  Normal and tumor DNA were extracted from formalin-fixed, paraffin-embedded tissues.   DNA samples were analyzed using the Ion Ampliseq Comprehensive Cancer Panel™ assay. Sequencing was performed on the Ion Proton Next-Generation Sequencing™ instrument.   Mutant allele frequencies were determined and a calculated MATH score was used to quantify tumor heterogeneity.  Response to chemo therapy was determine by primary resection pathology report.

Results:  A total of 13 patients with locally advanced rectal cancer (T3/4 or N1/2) were analyzed.  The boxplot in Figure 1 shows the range of calculated MATH scores by neoadjuvant therapy response category. Four patients were noted to have complete response, 7 had minimal/moderate and 2 demonstrated poor response. Tumor heterogeneity (as shown in MATH scores) was found to be significantly different amongst the 3 response groups (p=0.026), with higher MATH scores correlating with poorer response to treatment.

Conclusion:  The novel approach of applying the shape of a whole bioinformatics data set to analyze tumor heterogeneity may provide a useful biomarker for locally advanced rectal cancer.  MATH scores may allow a means of predicting response to neoadjuvant chemoradiation therapy.

 

25.05 Pomalidomide enhanced antitumor effects of gemcitabine and nab-Paclitaxel in pancreatic cancer cells

Posted on January 25, 2017

N. SAITO1, Y. Shirai1, T. Horiuchi1, H. Sugano1, R. Iwase1, K. Haruki1, Y. Fujiwara1, K. Furukawa1, H. Shiba1, T. Uwagawa1, T. Ohashi2, K. Yanaga1  1The Jike University School Of Medicine,Department Of Surgery,Minato-ku, TOKYO, Japan 2The Jikei University School of Medicine,Division Of Gene Therapy, Research Center For Medical Sciences,Minato-ku, TOKYO, Japan

Introduction:  NF-κB plays an important role in chemoresistance. Although gemcitabine and nab-paclitaxel therapy (GN) has been effective for pancreatic cancer, the therapeutic efficacy is attenuated by anticancer agents-induced activation of NF-κB. Meanwhile, pomalidomide is a novel immunomodulatory drug derived from thalidomide. Since thalidomide is a NF-κB inhibitor in digestive cancer, we hypothesized that pomalidomide also inhibits NF-κB activation, and enhances antitumor effects of GN for pancreatic cancer cell lines. 

Methods:  In vitro, we used human pancreatic cancer cell lines (MIA PaCa-2, PANC-1). We compared the antitumor effect of pomalidomide plus GN (GNP) with GN. Concentration of NF-κB, cell proliferation, cell cycle, and induction of apoptosis were evaluated by each assay. In vivo, we created xenograft orthotopic pancreatic cancer model (BALB/c with PANC-1). The animals were treated with oral polmalidomide five times a week and i.p. injection of GN once a week for 5 weeks. We evaluated sequential tumor volume by MRI and conclusive tumor weight and volume. We assessed the in vivo protein and apoptosis levels examined in vitro.

Results: 

Pomalidomide suppressed GN-induced NF-κB activation (MIA PaCa-2; GN : GNP = 9.57 ± 0.47 : 7.09 ± 0.20 ng/mg; p<0.01, PANC-1; GN : GNP = 69.58 ± 5.18 : 32.34 ± 13.61 ng/mg; p<0.05). Cell viability in GNP was significantly lower than that in GN (MIA PaCa-2; GN : GNP = 51.6 ± 9.0 : 24.1 ± 7.7 % ; p?0.01 , PANC-1; GN : GNP = 70.4 ± 7.5 : 41.7 ± 2.7 % ; p?0.01).

Moreover, in GNP, the levels of apoptotic protein (cleaved caspase 8, cleaved caspase 3, cleaved PARP) were higher than those in GN. Similar to nuclear NF-κB concentrations, phosphorylated IκBα was lower in GNP than that in GN. In addition, pomalidomide suppressed the expression levels of VEGF of pancreatic cancer in a dose-dependent manner. 

In vivo, the tumor weight (GN : GNP = 538 ± 36 : 385 ± 88 mg; p<0.05) and tumor volume (GN : GNP = 587 ± 51 : 313 ± 89 mm3; p=0.01) were significantly lower in GNP than those in the GN after five weeks of treatment. Moreover, immunohistochemical staining revealed down-regulation of VEGF and Ki-67 in GNP. 

Conclusion: Pomalidomide inhibited NF-κB activation and enhanced the antitumor effects of GN on pancreatic cancer cells.
 

25.04 Loss of miR-155 Upregulates WEE1 in Metastatic Melanoma

Posted on January 25, 2017

J. A. Campbell1, I. Huffnagle1, G. P. Robertson2, C. R. Pameijer1  1Penn State Hershey Medical Center,Department Of Surgery,Hershey, PA, USA 2Penn State Hershey Medical Center,Department Of Pharmacology,Hershey, PA, USA

Introduction:  WEE1 is a protein kinase in the BRAF pathway and is over-expressed in melanoma. It is involved in regulating cell cycle progression and is involved in tumor progression in several malignant tumors. In melanoma, experimentally induced down-regulation of WEE1 has been shown to inhibit cell growth in vitro and in vivo. Like many other proteins, WEE1 expression is regulated by microRNA. MicroRNAs are small, non-coding segments of RNA that are involved in post transcriptional regulation of gene expression and have been implicated in oncogenesis. Melanoma tumors have been shown to have very low levels of microRNA-155 (miR-155), although the downstream effect of this low expression is unknown. We evaluated a group of melanoma patient tumor samples for miR-155 expression, clinical outcome and possible mechanism of action of miR-155.

Methods:  RNA was extracted from melanoma patient tumor samples. miRNA microarray analysis was performed, with confirmation by qRT-PCR. Melanoma cell lines transfected with miR-155 were used for a mouse experimental metastasis model. Luciferase reporter assay was used to show interaction between miR-155 and the 3’UTR of WEE1. Western blotting was used to show the effect of miR-155 transfection on the protein levels of WEE1 in melanoma cells.

Results: In patient tumor samples miR-155 was lost in patients who develop metastatic disease. In the mouse model, transfection of cells from two metastatic melanoma cell lines with miR-155 mimic decreased the metastatic potential of melanoma cells, with a significant difference in the number of lung metastases between miR-155 and control mice (figure 1a). Transfection of melanoma cells with miR-155 reduced WEE1 protein levels (figure 1b) while inhibition of endogenous miR-155 up-regulated WEE1 expression.  Luciferase reporter assay demonstrated that miR-155 interacts with the WEE1 3’UTR and impaired gene expression in melanoma cells.

Conclusion: miR-155 is lost in patients who develop metastatic melanoma. Loss of miR-155 allows for upregulation of WEE1 in melanoma. Our in vivo studies showed significant decrease in metastasis with cells over expressing miR-155 suggesting that in melanoma miR-155 affects metastasis through WEE1 kinase.

25.03 Use of Circulating Microvesicles, Exosomes, as a Biomarker to Track Disease Burden in Melanoma

Posted on January 25, 2017

J. A. Cintolo-Gonzalez1, W. Michaud1, S. Cohen1, D. Plana1, D. J. Panka2, R. J. Sullivan3, G. M. Boland1  1Massachusetts General Hospital,Surgery,Boston, MA, USA 2Beth Israel Deaconess Medical Center,Hematology/Oncology,Boston, MA, USA 3Massachusetts General Hospital,Hematology/Oncology,Boston, MA, USA

Introduction:
Exosomes are extracellular microvesicles, which contain a variety of nucleic acids (DNA, messenger RNA (mRNA), microRNA (miRNA)) and proteins. Exosomes can be analyzed from archived serum/plasma samples making them ideal biomarkers to study tumor-based changes. We examined the feasibility and accuracy of exosomal mRNA analysis to assess disease burden in patients undergoing surgical resection for metastatic melanoma.

Methods:

Melanoma cell lines (A375, RPMI 7951, SK-MEL-30, and MeWo) were purchased directly from ATCC. Serial tumor and blood samples were collected from melanoma patients under IRB approved protocols.  Exosomes were isolated using combined filtration and ultracentrifugation. Exosomal RNA was isolated using the exoRNA serum/plasma kit (Qiagen Inc). 

Quantitative PCR (qPCR) was used to assess concordance of gene expression between paired exosomes and cell lines.  Paired exosomal and tumor mRNA from patient samples underwent whole transcriptome sequencing using Affymetrix Whole Transcriptome Pico Array.  Patient plasma-derived exosomes were analyzed via qPCR at pre-resection and post-resection time points.

Results:
Paired exosomes and their parental cell lines demonstrated concordance in gene expression as assessed by qPCR.  Specifically, in cell lines harboring a BRAF V600E mutation, BRAF V600E was detected by qPCR in cell lines and paired exosomes.   Likewise, whole transcriptome analysis of patient-derived exosomes and paired tumors demonstrated 80% concordance of gene expression. In patients undergoing resection of BRAF V600E mutant metastatic lesions, there was a significant decrease in exosomal BRAF V600E mRNA in patients rendered NED or with minimal residual disease but not in patients undergoing palliative resection with multiple residual sites of disease (Figure 1).  Of the patients undergoing definitive resection, patient 410, who showed ongoing, albeit decreased, exosomal BRAF V600E expression, recurred 3 months following resection.

Conclusion:
Exosomal mRNA reflects parental tumor mRNA expression and accurately predicts the presence of residual disease after resection in melanoma.  These findings support the use of exosomes as a biomarker for assessing tumor burden after surgery. Work is ongoing to assess the utility of exosomal RNA to assess minimal residual disease and risk of recurrence in high risk melanoma patients.

25.02 PI3K/mTOR Inhibition Suppresses Pancreatic Cancer by Reprogramming Cancer-Associated Fibroblasts

Posted on January 25, 2017

J. L. Williams2, I. A. Elliott1, A. H. Nguyen1, C. Matsumura1, R. Ghukasyan1, P. A. Toste1, J. R. Capri3, S. G. Patel1, L. Li1, N. Wu1, C. G. Radu3, T. R. Donahue1,3  1David Geffen School Of Medicine, University Of California At Los Angeles,Department Of Surgery,Los Angeles, CA, USA 2Harbor-UCLA Medical Center,Department Of Surgery,Torrance, CA, USA 3David Geffen School Of Medicine, University Of California At Los Angeles,Department Of Molecular And Medical Pharmacology,Los Angeles, CA, USA

Introduction:  Pancreatic ductal adenocarcinoma (PDAC) is a highly lethal disease, with an overall survival of less than one year.  Contributing to this poor prognosis is PDAC’s characteristically dense stroma, which is comprised predominantly of cancer-associated fibroblasts (CAFs). CAFs promote tumor growth, metastasis, and treatment resistance because when activated by environmental factors such as cancer cell (CC) exposure, hypoxia, and cytotoxic chemotherapy, they secrete protumorigenic cytokines, growth factors, and extracellular matrix components.  However, elimination of CAFs also increases tumor growth and invasion.  Therefore, rather than ablation, reprogramming of CAFs to an inactive or quiescent state is a potential PDAC treatment strategy.  In this study, we aimed to determine the effect of PI3K/mTOR inhibition, a pathway known to be involved in fibroblast activation, on PDAC tumorigenicity.

Methods:  Immortalized CAFs were treated with NVP-BEZ235 (BEZ), a dual PI3K/mTOR inhibitor, and exposed to gemicitabine (GEM), hypoxia (1% O2), or PDAC CC conditioned media (CM) for 48 hours.  After treatment, markers of CAF activation were determined via reverse transcription polymerase chain reaction (RT-PCR), Western blot, and mass spectrometry proteomics analysis of CAF CM.  Additionally, PDAC CCs (Mia PaCa-2 and PANC-1) exposed to CM from treated CAFs were tested with in vitro viability, migration, and invasion assays. PDAC CCs were also co-injected with treated CAFs in vivo, and tumor size was assessed.  A 3D co-culture model was used to assess the effect of BEZ treatment on CCs grown together with CAFs.

Results: After treatment with BEZ, gene expression of known markers of CAF activation, including alpha-smooth muscle actin (αSMA) and type I collagen, were decreased, even after exposure to activating stimuli.  Similarly, αSMA protein levels were reduced in BEZ-treated CAFs. CM from BEZ-treated CAFs contained lower levels of extracellular matrix components, proinflammatory cytokines, and growth factors. CCs exposed to CM from BEZ-treated CAFs were less viable, migratory, and invasive in vitro. Pre-treatment of PDAC CAFs with BEZ restricted PDAC growth in our in vivo co-implantation model (Figure 1).  In CC-CAF 3D co-culture, combination treatment of BEZ with GEM was more effective at suppressing CC growth than GEM treatment alone.

Conclusion: Inhibition of PI3K/mTOR decreases CAF activation, resulting in reduced PDAC CC tumorigenesis in vitro and in vivo. Additionally, combination treatment with BEZ, a dual PI3K/mTOR inhibitor, and GEM synergistically inhibits PDAC CC growth in the presence of CAFs. These data suggest that the addition of BEZ may improve the effectiveness of cytotoxic chemotherapy in PDAC.
 

« Previous Page
Next Page »