Specific targeting of tumor vasculature by DT-VEGF fusion protein reduces angiogenesis and growth of pancreatic cancer

1000 mg/d (n =914) and diclofenac100 mg/d (n = 3510) with regardto UGI safety. Investigators were required to report all potential clinically significant UGI events and were a[[owud/ requested to follow local standards of care with regard to work-up and treatment of events. Events data were collected prospectively.An independentGastrointestinal Events Committee (GEC)reviewedall datain a blindedfashion. Eventswere categorizedas UGI ulcer complications (perforations, gastric outlet obstruction, bleeding)or symptomatic UGI ulcerations(predetined as in CLASS). A total of 144 cases were reviewed and adjudicated by the GEC. Results: See table. Conclusions: These data confirm CLASS results that celecoxib is associatedwith significantly fewer ulcer complications and symptomatic UGI ulcerations than conventional NSAIDs. The differences in event rates betweenSUCCESSand CLASS may relate to regional differences in surveillance or clinical practice. Sponsored by Pharmacia Corporation and Pfizer, Inc.

552 Stratilyiog the Risk of Clinical Upper Gt Events in NSAID Users: Results from a Double-Blind OutcomesStudy Loren Laine, U S C Sch of Medicine, Los Angeles, CA; Christopher J. Hawkey, Univ of Nottingham, Nottingham United Kingdom; Claire Bombardier, Univ of Toronto, Toronto Canada; Deborah Shapiro, Alise Re/c/n, Merck & Co, Inc, Rahway, NJ Studies rarely provide the rates of GI events in NSAID users stratified by different risk factors. This study determines the independentbaseline risk factors for clinical upper GI events with a multivariate analysis of a large outcomes study and provides rates of events for patients with these risk factors. METHODS:8076 rheumatoid arthritis (RA) patients -> 50 yrs (or --40 yrs and on steroids) were randomly assignedto rofecoxib 50 mg qd or naproxen500 mg bid, and followed for a median of 9 mos (range, 0.5 - 13 mos). The primary endpoint was clinical upper GI events (perforation, obstruction, bleeding, symptomatic ulcers), adjudicated by an independentcommittee using predetinedcriteria. Backwardstepwise regressionanalysis including factors from a univariateanalysiswith p -< 0.20 was performed. RESULTS:Independent risk factors included age -> 65 yrs, history of upper GI events, severe RA (ARA class 4), RA -> 25 yrs, baselinesteroid use, no baselineNSAID use, baselineH2-receptorantagonist use, and history of UGI symptoms. The annuaUzedincidences(rates of events per 100 patientyears) for the non-selective NSAID naproxenfor patients with these factors were: age -> 65 yrs, 8.6%; history of uncomplicated upper GI event, 13.5%; history of complicated upper GI event, 18.8%; severe RA, 14.3%; RA >- 25 yrs, 9.3%; baselinesteroid use, 5.7%; no baseline NSAID use, 7.6%; baselineH2-receptorantagonist use, 8.8%; history of upper GI symptoms, 8.8%; age >- 65 yrs plus baseline steroid use, 12.4%; age >- 65 plus history of upper GI event, 28.8%. A lower incidence of upper GI events was observed with rofecoxib than with naproxen in subgroups with and without each of these risk factors (range of relative risks = 0.31 - 0.68). The rates in a very low-risk group (age < 65, no steroid use, no prior upper GI events, plus no H. pylori infection) were 0.2% for rofecoxib and 1.9% for naproxen(relative risk = 0.12 (0.02 - 0.96)). CONCLUSIONS:Clinical upper GI events occurred at annualized incidences of > 12% in users of a nonselectiveNSAIDwith a past history of upper GI events, severe rheumatoid arthritis, or age - 65 plus steroid use. Older age plus a prior upper GI event increasedthe rate to > 28%. In contrast, the annualizedincidence of upper GI events in patients without standard risk factors for NSAID use (older age, past upper GI event, and steroid use) and without H. pylori infection was 1.9% with naproxen and only 0.2% with rofecoxib.

Exposure(pt.yr) UlcercomplicaUons (annualizedrate) Ulcer complications ÷ symptomaticulcers (annualizedrate)

Celecoxib

NSAlDs

2031 2/2031 (0.1%) 18/2031 (0.0%)

10t4 7/1014 (0.7%)* 2311014 (2.0%)*

Odds RaUo

95%oCI

7.02

1.34-69.27

2.23

1.12-4.48

*p=O.O03vs celecoxib 555 Significantly Improved Upper Gastrointestinal (UGI) Tolerability with Celecoxib, a COX-2 Specific Inhibitor, Comparedwith Conventional NSAIDs. The SUCCESS I Trial Jay L. Goldstein, Univ of Illinois at Chicago, Chicago, IL; NaurangAgrawal, Duke Univ Medical Ctr, Durham, NC; Glenn M. E/sen, Vanderbilt Univ Medical Ctr, Nashville, TN; William Stenson, Washington Univ Sch of Medicine, St Louis, UO; AI E. Bello, Pharmacia, Skokie, IL; John G. Fort, Susan P. Boots, Pharmacia, Peapack,NJ Background:The CelecoxibLong-termArthritis SafetyStudy (CLASS),a prospectiveoutcomes study conductedin North America, demonstrateda significant reduction in UGI ulcer complications and improved tolerability. To extend our understanding of the UGI safety advantages of celecoxib over conventional nonsteroidal anti-inflammatory drugs (NSAIDs), in day to day practice, a naturalistic study was conducted worldwide. Method: SUCCESS I, a large, 12week, multinational, double-blind, randomized trial in 13,274 osteoarthritis patients, was conducted in 39 countries. There were 6547 patients from Europe/Africa, 2756 from North America, 2889 from Latin America, and 1082 from Asia/Pacific.Celecoxib200 mg/d (n = 4421) and 400 mg/d (n = 4429) was compared with naproxen 1000 mg/d (n-~ 914) and diclofenac 100 mg/d (n=3510) with regard to GI tolerability. Results: The frequency of all reported GI adverse events (AEs), the 3 most frequently reported UGI AEs, and withdrawals due to GI AEs are shown in the table. Conclusions: UGI symptoms, more commonly associated with NSAID use, were consistently and significantly lower with celecoxib. The percent reduction ranged from 18.6% to 29.4%. Thesedata confirm the superior GI tolerability of celecoxibvs conventional NSAIDs and establish that this difference is clinically meaningful in terms of discontinuation of therapy for such symptoms. Sponsored by Pharmacia Corporation and Pfizer, Inc.

553 Lack of Correspondenceof Risk Factors and Clinical GI Outcomesfor Patients on Nonsteroidal AofHnflammatory Drugs (NSAIDs): Analysis From the Celecoxib Longterm Arthritis Safety Study (CLASS). Glenn M. E/sen, Vanderbilt Univ Medical Ctr, Nashville, TN; On Behalf Of The Class Investigators, Clin Researchand Development,Pharmacia Inc, Skokie, IL Objective: ConventionalNSAIDs are associatedwith the developmentof symptomatic ulcers and ulcer complications. Such serious gastrointestinal (GI) toxicity may occur more frequently in individuals with identifiable risk factors as noted in the MUCOSAtrial. Datafrom the CLASS study were analyzedto identity patientsat risk for NSAID-relatedserious GI toxicity. Methods: Univariate and multivariate analysis of a number of potential risk factors for symptomatic ulcers and ulcer complications was performed in patients receiving conventional NSAIDs. Results: Univariate analysis identified the following as significant risk factors (p-75 y (3.7), history of ulcer disease(2.7), history of GI bleeding(3.4), aspirin use (2.3), cardiovascular disease (1.6), and Helicobacter pylori seropositivity (2.0). The following were not significant risk factors: arthritis type, arthritis duration, patient global assessment of arthritis, alcohol use, tobacco use, corticosteroid use, anticoagulant use, and gender. Multivariate analysis identified age ->75 y, history of ulcer disease, and aspirin use as the most important dsk factors (relative risk of 3.3, 2.6, and 2.1, respectively). Risk of serious GI toxicity rose in a nonadditive fashion with the number of risk factors. Patients with B, 1, and 2 or more risk factors had crude rates of symptomatic ulcers and ulcer complications of 0.6%, 1.4%, and 4.1% on NSAIDs. Although the risk of an event increased with the number of risk factors, most events occurred in patients with 0 or 1 risk factor (65% of aft events). Patients on celecoxib were most at risk if they were on concurrent low-dose aspirin (relative risk of 3.7 by univariate analysis), which, as an NSAID, was associated with the risk factors above. Conclusions: This study confirms the NSAID risk factor analysis results of the MUCOSAtrial and that risk increases nonadditivelywith the number of risk factors. However, most events in patients treated with conventional NSAIDs occur in patients at low or no risk. Sponsored by Pharmacia Corporation and Pfizer, Inc.

NSAlDs (n--4394)

Celecoxib (n=8800)

% Reduction

p Value

All GI AES UGI AEs Abdominalpain Dyspepsia Nausea

21.0% 15.6% 6.2% 5.9% 3.4%

16.7% 11.9% 4.8% 4.8% 2.4%

20.4% 23.7% 22.5% 18.6% 29.4%

<0.00t <0.001 <0.001 <0.008 <0.001

W'dhdrawals dueto GI AE

6.8%

5.2%

23.5%

<0.001

561 Specific Targeting of Tumor Vasculofure by DT-VEGF Fusion Protein Reduces Angiogenests and Growth of Pancreatic Cancer Hubert G. Hotz, Gactrointestinal Surg, UCLA Sch of Medicine, Los Angeles, CA; Parkash S. Gill, Rizwan Uasood, Dept of Medicine and Pathology, USC Sch of Medicine, Los Angeles, CA; Birgit Hotz, Gastrointestinal Surg, UCLA Sch of Medicine, Los Angeles, CA; Heinz J. Buhr, Thomas Foitzik, Dept of Surg, UK Benjamin Franklin, Berlin Germany; Oscar J. Hines, Gastrointestinal Surg, UCLA Sch of Medicine, Los Angeles, CA; Howard A. Reber, Gastrointestinal Surg, Los Angeles, CA

554 Significant Reduction in Serious Upper Gastrointestinal (UGI) Eventswith Celecoxib, a COX-2 Specific Inhibitor, Compared with Conventional NSAIDs. The SUCCESSI Trial Jay L. Goldstein, Univ of Illinois at Chicago, Chicago, IL; Glenn M. E/sen, Vanderbilt Univ Medical Ctr, Nashville, TN; William Stenson, Washington Univ Sch of Medicine, St Louis, MO; Naurang Agrawal, Duke Univ Medical Ctr, Durham, NC; AI E. Bello, Pharmacia, Skokie, IL; John G. Fort, Susan P. Boots, Pharmacia, Peapack,NJ

Anti-angiogenesisis a noveltreatment strategyfor pancreaticcancer(PaCa).Activatedendothelial cells of the tumor vasculatureabundantlyexpress receptorsfor vascularendothelialgrowth factor (VEGF),a key mediator of angiogenesis.The aim of the study was to specificallytarget and damagethe vasculature of PaCa by fusing the VEGFisoform VEGF~ to diphtheria toxin (DT), which inhibits the protein synthesis of target cells. Methods: VEGFlesfusion protein containing the enzymaticand translocation domains of DT was produced with GST in vector pGEX-KG and expressed in E. col/SG12036. In vitro: Two human PaCacell lines (AsPC-1, poorly differentiated; HPAF-2,moderatelydifferentiated) and human endothelialcells (HUVEC) were exposedto increasing concentrations (1 - 10000 ng/ml) of DT-VEGF.Cell proliferation was assessed after 3 days by cell count and Ul-r-assay. In vivo: 1 mm~ fragments of sc. PaCadonor tumors were implanted into the pancreasof nude mice which receivedeither OTVEGF (200 pg/kg, every other day) or vehicle (Con) ip. for 14 weeks. Volume of primary tumor (TU-Vol), metastaticspread(Met-Score),and animal weight were determinedat autopsy. Microvessel density (MVD) was analyzed in CD31-stainedtumor sections. Results: In vitro:

Background: The Celecoxib Long-term Arthritis Safety Study (CLASS), a North American prospectiveoutcomes study, demonstrateda significant reduction in UGI ulcer complications and improved tolerability. To extend our understanding of the UGI safety advantages of celecoxibover conventional NSAIDs,a naturalistic study was conducted worldwide. Methods: SUCCESS I, a large, 12-week, multinational, prospective, double-blind randomized trial in 13,274 osteoarthritis patients, was conducted in 39 countries. There were 6547 patients from Europe/Africa, 2756 1rum North America, 2889 from Latin America, and 1082 from Asia/ Pacific. Celecoxib200 mg/d (n = 4421) and 400 mg/d (n = 4429) was comparedwith naproxen

A-105

Proliferation of the PaCacells was significantly inhibited at high concentrations of DT-VEGF (>- 1000 ng/ml). In contrast, DT-VEGFeffectively decreasedthe growth of HUVECat 10 ng/ ml. In viva: Table (p
AsPC/Con AsPC/DT-VEGF

HPAFICon

TU-Vol.(turn3) Met.Score(pts) Survival(n/n) WeigM(gram) MVD/0,74mrn2

1386~155 16 ~1.3 1/ 8 28.2:t:0.8 69.8+5.1

3920~495 7 2±1 3 4/ 8 30.0t0 7 85.7±7.2

652:L73" 83±1.1" 6 / 8* 27.3±09 26.6~3.8"

HPAF/DT.VE( 927±122* 0.75±0.16" 718 29.6~1.3 39.6~4.3"

562 The Value Of LaparoscopyIn The Management Of Non-Pancreatic Pedampullary Tumors Art D, Brooks, Memorial Sloan-Kettering Cancer Ctr, New York, NY; Michael Mallis, Philadelphia Coil of Osteopathic Medicine, Philadelphia,PA; Murray F, Brennan, Kevin C. Conlon, Memorial Sloan-KetteringCancer Ctr, New York, NY Background: Laparoscopy has proven valuable for the identification of radiologically occult advanceddiseasein patients with potentially reaectablepancreaticedenocaroinoma.Its value in other periampollary tumors has not been reported. This study was designedto determine the value of laparoscopy in the managementof these tumors. Methods: One hundred thirty nine patientswith radiologicallyresectablenon-pancreaticperiampullarytumors ware identified were from a prospective pert-pancreatic laparoscopy database of over 1000 procedures performed between 8/1993 and 7/2000. Criteria for radiolngic unresectablifity were major arterial involvement, long segment portal vein/mesentefic vein involvement, or metastatic disease. Criteria for laparoscopic unresectabilify included histologically proven peritoneal or hepaticmetastases,distant nodal involvement,arterialinvolvement,and local extensionoutside the resection field. Univariate analysis was conducted using Chi square for association and log rank analysisfor survival. Results: The median age at operation was 70 years (range 3187) and 55% were male. An adequate laparoscopy was performed in 127 (91%). Of these patients, laparoscopyidentified an additional 10 (8%) with unresectabledisease.Laparoscopy identified 4 patientswith liver metastases,4 with peritonealspread,and 2 with other metastatic disease. Of the 117 patients with laparoscopic resectable disease, 109 (93%) went on to resection. Laparoscopymissed 3 patients with liver metastases, 2 with vascular invasion, 1 with peritonealspread,t with local extensionand 1 with benigndisease.In the group of patients having inadequatestaging laparoscopy, 3 patients (25%) had liver metastasesdiscovered at exploratorylaparutorny. Patientswith resectablediseasewere treated by pancreaticoduodenectomy (89, 75%), ampullectomy (12, 10%), duodenal resection (10, 9%), or bile duct excision (7, 6%). Pathologic examination revealedmalignancy in 122 patients. Median post operative survival was 48.5 months (38-59) in resectedpatientsand 9.8 months (6-13) in non-reaected patients. Discussion: The addition of diagnostic laparoscopyto dynamic CT scanning results in an 8% decreasein the likelihood of finding unresectablediseaseat laparotomy. In contrast to patients with pancreatic adenocarcinoma,we believethat laparoscopy should be utilized in a selectivefashion for pre-operativestaging of patients suspectedof having non-pancreatic peri-ampullary tumors.

564 PI-3'Kinase and NF-KB Induced Chemoresistance in Human Pancreatic Cancer Shimul A. Shah, Michael W. Potter, Mohir H. Hedeshian,UMass Medical Sch, Worcester, MA; Ravi S. Chad, Mark P. Callery, UMass Medical Sch, Worcester, MA Introduction: Since TNF-a and some chemotherapeuticagents activate both apoptosis and NF-KB dependentanti-apoptotic genes, they may neutralizetheir own anti-tumor effects. The cell signaling mechanismsfor such chemoresistanceare not clear, but may involve PI-3'Kinase (PI3K). To clarify this, we examinedwhether cross-signalingbetweenPI3Kand NF-KBenhances the anti-tumor effect of TNF-a in human pancreaticcancercells. Methods: Ouiescantpancreatic cancer cells (Panc-1, MiaPaCa-2)withTNF-a, Ly294002 (PI3K inhibitor), alone or combined, were restimulated with mitogan (10% FCS) to induce cell cycle entry. Proliferation (MTT), cell cycle progression (APOBrDU incorporation and FACS analysis), and apoptosis (PARP cleavage;caspase-3activation - Western blot [WB])were measured.Akt activation (Akt kinase assay) and IKBa degradationwere determinedby WB. Translocationof NF-KBinto the nucleus was examined by EMSA, while a NF-KB/luciterasereporter gane was used to quantify NF-KB dependent gene expression. Statistical analysis - two*tailed t-test (p
565 Ee Bloc Resection for Locally Advanced Cancer of the Pancreas: Is it Worthwhile? Aaron R. Sasson, John P. Hoffman, Eric Ross, Fox Chase Cancer Ctr, Philadelphia,PA; Steven A. Kaoan,Temple Univ Sch of Medicine, Philadelphia,PA; James F. Pingpank Jr., Burton L Eisenberg, Fox Chase Cancer Ctr, Philadelphia, PA Objective: The benefit of radical surgical resection of contiguously involved structures for locally advancedpancreaticcancer is not known. The aim of this study was to examinepatient outcome after extended pancreatic resection for locally advanced cancer of the pancreas. Methods: We retrospectively reviewedthe records of 116 patients with edenocarcinomaof the pancreas who underwent extirpative pancreatic surgery between 1987 and 2000. Of the 116 patients,39 (34%) required resectionof surrounding structures (Group I), and 77 patients (64%) had standard pancreatic resections (Group II). In all cases, all macroscopic disease was excised. In Group I, a total of 43 contiguously involved structures were resected;vascular 25 (58%), colon 12 (28%), adrenal 3 (7%), liver 2 (5%), stomach 1 (2%), and multiple 3 (7%). Excisionof regional blood vessels included: superior mesantericvein and/or portal vein in 16, hepaticartery in 5, celiac axis in 4, and middle colic vessels in 3. Results: No difference between Group I and Group li was detected for the following parameters:age, sex, history of previous operation, estimuteit blood loss, or hospital stay. For the entire cohort, the morbidity and mortality were 40% and 1.7%, respectively,and were similar betweenthe two groups. Adjuvanttherapy was administeredto over 90% of patients in both groups. However, patients in Group I were more likely to have received neoedjuvant therapy (74% vs. 42% (p= .001)). Total pancreatectomyand distal pancreatectomywere more often performed in Group I (p = .003). Additionally, the mean operativetime was longer (8.3 hours compared to 7 hours (p=.0006)). Both groups had similar rates of microscopic positive margins and involved lymph nodes, as well as total number of lymph nodes removed.The median survival for patients in Group I was 24 months and in Group II was 17 months (p =.28). in comparing patients in Group I who had vascular resections vs. colon resections, the median survival was 26 and 16 months, respectively (p=.06). Conclusion: We were unable to detect a difference in outcome for patientswith locally advancedcancersrequiring extendedpancreatic resections comparedto patientswith standard resections.Although not statistically significant, patients with vascularstructures resectedhad a longer mediansurvival than patients requiring colon resections.En bloc resection of involved surrounding structures, to completelyextirpate all macroscopic disease,may be of benefit in selectedpatients with locally advanceddisease.

563 Increased Frequency Of Heparanase Gene Expression In Human Pdman/And Metastatic Pancreatic Cancer Anthony W. Kim, Xiulong Xu, Paolo Gattuso, Richard A. Prinz, RUSH-PresbyterianSt Luke's Medical Ctr, Chicago, IL INTRODUCTION:Extracellularmatrix (ECM) degradationis an essentialstep that allows tumor cells to penetratea tissue barrier and becomemetastatic.Heparanase(HPR) is an endoglycosF dase that specifically degradesthe heparansulfate of proteoglycan,a chief component of the ECM. Previousstudies demonstratedthat HPR expressioncould be detectedin hepatocellular, breast, and colon carcinomas.This study tests whether HPR is expressedin pancreaticcancer. METHODS: 82 pancreatic tumor samples and 7 normal pancreas samples collected from surgical specimens were examinedfor HPR gone expression by in situ hybridization with an antisense RNA probe. Tumors included 70 adenocarcinomasand twelve metastases from edenocarcinomas. The metastases included three implants on the small intestine, three on the peritoneum, two on the gallbladder, one on the spleen, one on the stomach, and one on the thigh. Two primary adenocarcinomasknown to be positive by in situ hybridization were additionally immunostained using immunofluorescent techniques. HPR expressionwas comparedwith the clinicopathologicalfeatures of the pancreaticcancers.RESULTS:The percentage of HPR expression in pancreatic adenocarcinomaand its metastaseswas 72% (59/82). 50 of 70 primary pancreatic adenocarcinomas(71%) and 9 of the 12 metastases(75%) were HPR positive. Neitherage nor gender was a predictor of HPR expression.Neithertumor stage nor lymph node involvement were independentpredictors of HPR expression. None of the 7 normal pancreas specimens were HPR positive. Both specimens that were HPR positive on immunofluorascence were also positive when studied by in situ bybddization. CONCLUSION: HPR is expressedwith high frequency in pancreaticedenocarcinomaand its metastases,but not in normal pancreatic tissue. HPR may contribute to the highly invasiveand early metastatic characteristicsof pancreaticcancer. HPRexpressionappearsto be independentof age,gender, tumor stage, and lymph node involvement.

Ligand Activation Of Alternatively Spliced FGFR-1 Modulates Pancreatic Cancer Cell Malignancy Selwyn M. Vickers, Zhi-Oiang Huang, Leeann Macmillan-Crow, John A. Thompson, Univ of Alabama at Birmingham, Birmingham, AL Pancreatic cancer continues to be a devastating tumor (28,000 new cases/year; 5%-5 year survival). Pancreatictumors frequently (90%) overexpress fibroblast growth factor liganda (FGF-1, FGF-2)and alternativelyspliced high affinity receptors(FGFR-t/5). (FGFR-1a previously found in normal pancreatictissue). Methods: In vitro: to study the significanceof this observation, PANC-1cells were stably transfectedvia the pMEXneovector containing FGFR-la (PAN@ 1a) or FGFR-1/5(PANC-1/5) isoforms. Cells were treated with 150 ~M peroxynitrite, 25 and 100 /~g/ml of 5-FU. Cells were evaluated for growth inhibition, apoptosis (by fluorascein diacetate/pmpidium iodide and annexin V staining, caspase3 activation) and for BcI-XL/BAX expression (by Western Blot). In vivo: 5x10e cells of each isoform were injected into nude

A-106