SS

AT

Ab

stra

cts

after dilatation. Four patients without dilatation had surgery including resection of angulatedbowel (n=2) or pouch excision with end ileostomy (n=2). In 1 patient, symptoms due toafferent limb syndrome relieved after conservative management. The last patient was lostto follow-up. CONCLUSION: A combined assessment of endoscopy and abdominal imagingis important to establish the diagnosis of afferent limb syndrome. Endoscopic or surgicalintervention is often needed.

W1649

Elastica Staining for Venous Invasion Results in Superior Prediction of CancerSpecific Survival in Colorectal CancerCampbell S. Roxburgh, Donald C. McMillan, John H. Anderson, Ruth F. McKee, Paul G.Horgan, Alan K. Foulis

Introduction: Venous invasion is an important high risk feature in colorectal cancer. However,prevalence of venous invasion in published studies ranges from 10-90%. To resolve disparityamongst reporting pathologists in our institution, methods to improve detection of vascularinvasion were sought. Elastica stains highlight elastin fibres present in the adventitia of bloodvessels providing a more objective method of detection of venous invasion. As a resultelastica stains have been used routinely on colorectal specimens in our institution since2002. The aim of the present study was to examine the impact of elastica staining on thevalue of venous invasion as a predictor of cancer specific survival, following curative resectionof colorectal cancer. Methods: 419 patients underwent curative elective colorectal cancerresection between 1997-2006. Patients were grouped prior to (1997-2001(cohort 1, n=194)) and following introduction of elastica staining (2003-2006 (cohort 2, n=225)). In athird cohort 53 patients between 1997-2000, both H&E alone and elastica H&E techniqueswere applied allowing a direct comparison within a single cohort. Results: Clinico-patholo-gical characteristics and 3-year survival rates were similar in both groups (cohort 1 vs cohort2). Rate of detected venous invasion increased from 18% to 58% following introduction ofelastica staining (p<0.001). The 3-year cancer survival rate associated with absence of venousinvasion was 84% in cohort 1, compared with 96% in cohort 2 (P<0.01). Elastica stainingimproved the prognostic value of venous invasion; the area under the receiver operatorcurve rising from 0.59 (P=0.040;1997-2001) to 0.68 (P<0.001;2003-2006) using cancermortality as an endpoint. Within cohort 2, the absence of venous invasion was associatedwith a 3 year cancer survival of 96% in both node negative and node positive disease. Adirect comparison between H&E alone and elastica H&E was made in 53 patients (cohort3). The area under the receiver operator curve rose from 0.58, P=0.293 (H&E alone) to0.74, P=0.003 (elastica H&E). Discussion: Increased detection of venous invasion withelastica staining, compared with H&E alone, provides superior prediction of cancer survivalin colorectal cancer. This relationship was seen in the comparison of two consecutive cohortsand in a direct comparison in a single cohort. When stained with elastica, the absence ofvenous invasion is associated with an excellent 3 year survival whatever the tumor's lymphnode status. Based on these results, elastica staining should be incorporated into the routinepathological assessment of venous invasion in colorectal cancer.

W1650

Grade of Ischemia on Lower Endoscopy is the Reliable Predictor of Need forOperative Intervention for Ischemic Colon in Critically Ill PatientsCatherine G. Velopulos, Matthew Hughes, Baharak Moshiree, Edward M. Copeland,Sanda Tan

Introduction: Diagnosis of ischemic colitis in critically ill patients has been plagued by lackof standardization of modalities. The utility of lower endoscopy has not been previouslyreviewed, particularly in regard to prevention of unnecessary operative intervention. Inaddition, no study compares the diagnostic yield of flexible sigmoidoscopy (FS) versuscolonoscopy (COL). Purpose: This project looks at the utility of lower endoscopy alone andin conjunction with other tests in predicting the need for operative intervention, specificallyaddressing FS versus COL. Methods: The endoscopy database was reviewed for lowerendoscopy performed in the intensive care units in our institution over the last five yearsin cases where ischemic colitis was suspected, and then correlated with clinical course,laboratory values, and need for operative intervention. Results: 64 COL and 24 FS metcriteria. Five patients (5.7%) had Grade III ischemia verified as non-viable bowel at laparo-tomy. 50 (56.8%) had Grade I or no ischemia. 7 proceeded to laparotomy, one for refractorybleeding, and one for duodenal perforation; none had ischemia identified at the time oflaparotomy. Of 33 (37.5%) diagnosed with Grade II ischemia, 12 proceeded to laparatomy.One had necrotic bowel. Two additional unstable patients had segments of bowel that werenoted to be viable, but these were resected to avoid a second-look laparotomy given theirinstability. The other 9 patients proceeded to laparotomy for bleeding or need for decompres-sion, not for ischemia. No lab tests correlated with endoscopy or operative findings. Inparticular, lactic acid did not predict presence or degree of ischemia. 2/64 (3%) COL revealedpotentially significant ischemia that would have been missed on FS; one required operativeintervention. 1/24 (4%) FS had a clinically significant right-sided lesion that was laterdiscovered. To achieve 0%, the number needed to treat would equal 25. Conclusion: Gradeof ischemia on lower endoscopy is the reliable predictor of need for operative interventionin critically ill patients, with equivalence of COL and FL. In the absence of another indicationfor intervention, Grade I or II ischemia can be managed expectantly. The only way to reliablydiagnose the degree of ischemia is through the use of lower endoscopy, with the majorityof negative laparotomies predicted pre-operatively.

S-896SSAT Abstracts

W1651

Primary Sclerosing Cholangitis and Extra Intestinal Manifestations in PatientsWith Ulcerative Colitis and Ileal Pouch-Anal AnastomosisHans H. Wasmuth, Gerd Trano, Birger H. Endreseth, Astrid Rydning, Arne Wibe, HelgeE. Myrvold

Objective The aim of this study was to assess complications and functional outcomes inpatients having ileal pouch-anal anastomosis for ulcerative colitis with or without primarysclerosing cholangitis or other extra intestinal manifestations; in particular if primary scleros-ing cholangitis is a risk factor for pouchitis Material and methods During the period 1984- 2007 289 patients with ulcerative colitis underwent proctocolectomy and construction ofileal pouch-anal anastomosis. All patients were followed up at our outpatient clinic and datarecorded prospectively. Mean follow-up was 12 years. Eleven patients had primary sclerosingcholangitis, nine had pyoderma gangrenosum, and twelve had arthitis or ankylosing spondyl-itis. Results Early complications like postoperatively bleedings, anastomotic leakages, pelvicsepsis and fistulas were similar for patients with or without extra intestinal manifestations.Patients with primary sclerosing cholangitis had more episodes of pouchitis, 5.25 vs. 2.72(p = 0.048), and more chronic pouchitis, 4/11 vs. 17/260 (p < 0.001) compared to patientswithout adjunct disease. They also had more day time incontinence, 0.55 vs. 0.07 (p =0.007) episodes per week, respectively. The frequency of bowel movements was similar inall groups. Neoplasm of the colon was more frequent in patients having primary sclerosingcholangitis, 4/11 vs. 4/260 in ulcerative patients (p < 0.001). Patients with pyodermagangrenosum or arthitis/ankylosing spondylitis had outcomes similar to patients withoutadjunct diseases. Conclusion An association between primary sclerosing cholangitis andchronic/severe pouchitis was found. There was no association between pouchitis and pyod-erma gangrenosum or arthritis/ ankylosing spondylitis. Long term functional outcome orsurgical complications did not differ between the groups. Primary sclerosing cholangitis isa risk factor for chronic pouchitis.

W1652

Extent of Response of Rectal Cancer to Neoadjuvant Chemoradiation PredictsOutcomeKelly M. McLean, Jonathan M. Hernandez, Jill Weber, Farhaad C. Golkar, Lauren Lange,David Shibata

Background: Combined modality chemoradiation therapy (CMT) administered prior tosurgery is the preferred approach for locally advanced rectal cancer (RC). Reports conflictregarding the prognostic significance of degree of response to CMT. This study evaluatesthe association between extent of clinical and pathologic response and clinicopathologicvariables and patient outcomes. Method: From 1998-2008, 157 patients with biopsy-provenStage II and III RC received neoadjuvant 5-FU based chemotherapy and radiation followedby radical resection. Clinicopathologic and survival data were collected by retrospectivereview of an IRB-approved database. Comparisons were performed using Chi-squared,Fisher's Exact Test or Student's t-test. Survival analyses used the Kaplan-Meier method.Results: Our population consisted of 62 women and 95 men with a median age of 61 years(range 29-85) and a median follow up of 31.4 months (range 1-116). 70 patients (44%)demonstrated clinical complete response (clinCR) while 38 (27%) had a pathologicCR(pathCR). 38 of 70 (55%) clinCR and 15 of 87 (17%) clin partial response (clinPR)patients demonstrated pathCR. Patients with clinCR were more likely to have sphincterpreservation (88% vs. 71%; p=0.01) and improved disease-free survival (DFS; p=0.04)compared to patients with clinPR. Degree of clin response did not impact overall survival (OS)or disease-specific survival (DSS). PathCR compared to pathPR showed a lower recurrence rate(5% vs. 22%; p=0.02) and, when adjusted for disease-specificity, improved DFS (p=0.03)and DSS (p=0.05). Conclusion: Our study confirms that pathCR of RC treated by CMT isassociated with superior DFS and DSS as compared to pathPR. Although clinical responsedoes not necessarily correlate with pathologic response, determination of clinCR has pro-gnostic value associated with improved DFS and increased sphincter preservation.

W1653

A Novel Data-Driven Staging of Colorectal CancerElena Manilich, Victor W. Fazio, Feza H. Remzi

Purpose : We define prognostic model that predicts survival of patients with colorectalcancer after a radical potentially curative resection. This study uses a data-driven approachto identify highly predictive cancer characteristics that involve TNM as well as non-TNMfactors. A novel prognostic methodology, random survival forest, accounts for the complexinterplay among clinical and histologic features. Methods : Survival data of 2,534 colon and2,380 rectal cancer patients undergoing a radical resection between 1969 and 2003 were