hepatocellular carcinoma (HCC) in vivo. Materials and Methods:Male mice genetically deleted for the gene encoding GDF-15(Gdf15�/� mice) and wild-type controls were exposed to the hepa-tocarcinogen diethylnitrosamine (DEN). Mice were sacrificed at sixmonths of age and their livers dissected and processed for histology.Tumor number and size relative to total area liver area examinedwere determined. Results: At six-months of age, tumors wereidentified in 16 of 20 (80%) of Gdf15�/� mice and 16 of 19wild-type mice (84%) (N.S.). No significant difference in tumor-occupied area was observed in Gdf15�/� mice versus wild-typemice. In addition, no difference in invasiveness was observed inHCC arising in Gdf15�/� as compared to wild-type mice. Strongimmunohistochemical staining for GDF-15 was noted in smallHCC foci in wild type mice. In contrast a number of advanced HCCtumors in wild type mice were associated with a loss of GDF-15expression. Conclusions: Although highly expressed in associationwith multiple gastrointestinal cancers, and lost in some advancedHCC, genetic ablation of GDF-15 has no apparent effect on HCCtumor formation rate, growth rate or invasiveness in this model ofprogressive HCC development in vivo.

QS106. THE OPIOID GROWTH FACTOR AXIS IN HUMANTHYROID CANCER. David Goldenberg, Ian Zagon, Patri-cia J. McLaughlin; Penn State College of Medicine, Her-shey, PA

Background: Thyroid cancer is diagnosed in 33,550 persons an-nually and results in 1530 deaths per year. Patients with papil-lary or follicular thyroid cancers have excellent prognosis follow-ing surgery and radioactive iodine treatment. Undifferentiatedthyroid cancers are highly aggressive and resistant to currenttherapies. The native opioid growth factor (OGF), [Met5]-enkephalin, is a tonic inhibitory peptide that modulates cell pro-liferation and migration, as well as tissue organization, duringdevelopment, cancer, homeostatic cellular renewal, wound heal-ing, and angiogenesis. OGF action is mediated by the OGF recep-tor (OGFr). New treatments using biotherapies such as opioidgrowth factor (OGF) for thyroid cancers may prove beneficial.Methods: Patient surgical samples were collected at the time ofthyroidectomy and neck dissection and processed for receptorbinding assays or immunohistochemistry for OGF and OGFr.Results: Immunohistochemical detection of OGF and OGFr wasevident in the cytoplasm of all samples. Specific binding was detectedin all samples. OGFr Bmax values for thyroid cancers were found tobe 2-fold lower than non neoplastic thyroid tissues. Conclusions:Data demonstrates that a potent negative growth regulator ofcancer is present in both neoplastic and non neoplastic thyroidtissues. These findings support the hypothesis that the OGF-OGFr axis may serve as an effective biotherapy for thyroid can-cers and that OGFr levels may serve as a marker for thyroidtumor progression.

QS107. NON-INVASIVE MOLECULAR IMAGING BIOMARK-ERS OF THERAPEUTIC RESPONSE IN COLOREC-TAL CANCER. H. Charles Manning, Nipun B. Merchant,Allan Coe Foutch, Jack Viostko, Shelby K. Wyatt, M. KayWashington, Bonnie Lafleur, Ronald Baldwin, Darryl J.Bornhop, John C. Gore, Robert J. Coffey; Vanderbilt Uni-versity Medical Center, Nashville, TN

Introduction: The epidermal growth factor receptor (EGFR) isoverexpressed in more than 50% of colorectal adenocarcinomas(CRC) and is frequently associated with an aggressive, invasivephenotype. Therapeutic monoclonal antibodies targeting EGFR,including cetuximab (C225, Erbitux™), have shown consistentsingle agent activity in patients with CRC. However, no clinicalbiomarkers of therapeutic response have identified, with the soleexception of an acneform rash. We have utilized a preclinical

model of CRC to simultaneously characterize unique, but comple-mentary, non-invasive molecular imaging metrics as potentialbiomarkers of treatment response to cetuximab. Methods: Non-invasive imaging assessments of EGFR expression, apoptosis, andproliferation were evaluated in vitro and in vivo. Two optical(NIR) molecular imaging probes targeting either EGFR expres-sion (NIR800-EGF) or apoptosis (NIR700-Annexin-V) were syn-thesized and validated in cell lines and nude mice bearing DiFixenograft tumors. Additionally, cellular proliferation was as-sessed using the PET tracer [18F]-FLT. Multiparametric assess-ment of response to cetuximab treatment was accomplished byimaging NIR800-EGF, NIR700-Annexin-V, and [18F]-FLT con-comitantly in a cohort of mice bearing DiFi tumor xenografts.Results: The molecular imaging probe NIR800-EGF was shown tobe EGFR specific in cultured DiFi cells, and in vivo uptake in DiFixenografts exhibited a linear correlation with tumor burden (n �9, r2 � 0.881). In vitro and in vivo uptake of NIR700-Annexin-Vwas correlated with caspase 3/7 activity in cetuximab treated DiFicells and animals bearing DiFi xenograft tumors. In multipara-metric treatment response imaging, significantly decreasedNIR800-EGF uptake (P � 0.0001) was observed in cetuximabtreated animals compared to untreated cohorts. Additionally,treated animals exhibited significantly increased NIR700-Annexin-V uptake (P � 0.016) compared to untreated cohorts. Nostatistically significant difference between treated and untreatedanimals was observed by [18F]-FLT-PET imaging. All imagingresults were validated by immunohistochemistry Methods, includ-ing EGFR, phospho-EGFR, caspase 3, and KI-67 staining. Con-clusions: Non-invasive molecular imaging Methods have consid-erable potential as biomarkers of clinical prognosis and treatmentresponse. While a single imaging metric may be clinically infor-mative, our data presented here illustrate that utilizing multiple,complementary molecular imaging readouts of response concomi-tantly may provide additional information towards optimizingtherapeutic intervention unobtainable from a single metric.

QS108. HYALURONON SYNTHASE-3 INHIBITION DE-CREASES COLON CANCER CELL GROWTH INVIVO. Eric Lai1, Rahul Singh1, Yali Zhao2, Gillian Howell2,Ashwani Rajput1, Kelli Bullard Dunn1; 1University atBuffalo/SUNY and Roswell Park Cancer Institute, Buffalo,NY; 2Roswell Park Cancer Institute, Buffalo, NY

Background: Hyaluronan (HA) and its biosynthetic enzymes(hyaluronan synthases; HAS1, �2, and �3) have been implicatedin cancer growth and progression. We have previously shown thatHAS3 is upregulated in metastatic SW620 colon cancer cells, andthat HAS3 and HA mediate anchorage-independent cellulargrowth in vitro, a correlate of tumor growth in vivo. We thereforehypothesized that inhibition of HAS3 expression and HA produc-tion in SW620 colon cancer cells would decrease tumor formationin a mouse model of primary tumor growth. Methods: HAS3expression was inhibited by transfecting SW620 cells with smallinterfering RNA (siRNA). A commercially available siRNA se-quence for HAS3 was purchased and customized with plasmidspecific primers. A scrambled siRNA served as a control. To selectspecific clones, transfected cells were plated at 300 cells/per 10cmdish. The resulting colonies were isolated with plastic rings andselected for testing. HAS3 expression was characterized by semi-quantitative RT-PCR (band intensities were normalized to corre-sponding GAPDH bands). Pericellular HA retention was assessedby a particle exclusion assay (PEA); in this assay, a matrix to cellratio of 1.00 equals no matrix. Stably transfected cells expressinglittle HAS3 and HA were then injected subcutaneously into theflanks of Balb/c nude mice (7�106 cells/injection, one injection permouse; HAS3 silenced �10, HAS3 scrambled [control] �10). Tu-mors were measured with calipers and their volume was calcu-lated using following equation: (radiusmin*radiusmin*radiusmax)*�/6.

310 ASSOCIATION FOR ACADEMIC SURGERY AND SOCIETY OF UNIVERSITY SURGEONS—ABSTRACTS

After 35 days, the mice were euthanized and tumors were har-vested and weighed. Values were compared using Student’s-t-test.A p value � 0.05 was considered statistically significant. Results:RT-PCR confirmed that SW620 cells transfected with siRNAagainst HAS3 expressed less HAS3 than cells transfected with ascrambled siRNA sequence (band intensities were 0.65 � 0.06 inthe HAS3 silenced cells vs. 1.21� 0.05 in the HAS3 scrambledcells; p�0.0024). PEA then showed that the HAS3 silenced cellsretained less pericellular HA than HAS3 scrambled cells; (matrixto cell ratio was 1.13 � 0.009 in the HAS3 silenced cells vs. 1.32� 0.01 in the HAS3 scrambled cells; p�0.0001). After injectioninto mice, subcutaneous tumor volume was significantly de-creased in the HAS3 silenced tumors after day 13 (figure). Finally,HAS3 silenced tumors weighed less than HAS3 scrambled tumors(mean weight of HAS3 silenced tumors was 0.83 � 0.15 grams vs.2.19 � 0.37 grams in the HAS3 scrambled tumors; p �0.0030).Conclusion: Inhibition of HAS3 expression and HA retention inSW620 colon cancer cells via siRNA transfection decreases sub-cutaneous tumor growth in mice. These data support the conten-tion that HAS3 and pericellular HA are critical factors in coloncancer cell growth.

QS109. THE OMEGA-3 POLYUNSATURATED FATTY ACIDEPA DECREASES THE GROWTH OF PANCREATICCANCER CELLS BY COX-2 DEPENDENT AND INDE-PENDENT MECHANISMS. Hitoshi Funahashi, ElianeAngst, Makoto Satake, Sascha Hasan, Howard A. Reber,Oscar J. Hines, Guido Eibl; David Geffen School of Medicineat UCLA, Los Angeles, CA

Background and Aim: There is a large body of data indicatingthat omega-3 polyunsaturated fatty acids (PUFA), commonlyfound in fish oils, attenuate the development and inhibit thegrowth of human cancers. Omega-3 PUFAs are substrates forcyclooxygenase (COX) enzymes and the beneficial effects ofomega-3 PUFAs are thought to stem largely from an increasedproduction of anti-inflammatory prostaglandins, e.g. prostaglan-din E3 (PGE3). In previous studies we have demonstrated that incontrast to omega-6 PUFAs the omega-3 PUFA eicosapentaenoicacid (EPA) decreased the growth of human pancreatic cancer(PaCa) cells in vitro. However, the exact underlying mechanismsremained unclear. The aim of the present study was therefore tofurther characterize the mechanisms of the EPA-mediated de-crease in PaCa cell growth in vitro. Methods and Results: Thehuman PaCa cell lines BxPC-3 (COX-1 and COX-2 positive) andMIA PaCa-2 (COX-1 positive but COX-2 negative) were culturedin serum-free medium in the presence of EPA (0-100 �M) for 24hours. EPA significantly reduced the growth of both cell lines as

determined by cell count and MTT assay. The effect of EPA onBxPC-3 cell growth was partially attenuated by a selective COX-2but not a COX-1 inhibitor. The effect of the COX-2 inhibitor on theEPA-mediated reduction of BxPC-3 cell growth was partially re-versed by the addition of PGE3. Moreover, the effect of EPA onBxPC-3 cell growth was partially inhibited by antagonists of thePGE receptors EP2 and EP4. The expression of EP2 and EP4

receptors in PaCa cells was confirmed by real-time PCR. In con-trast, the growth-inhibitory effect of EPA in COX-2 negative MIAPaCa-2 cells was unaffected by selective COX-1 and �2 inhibitorsas well as by EP2 and EP4 receptor antagonists. However, pre-incubation of MIA PaCa-2 cells with GW9662, an antagonist of thenuclear receptor PPAR-�, completely reversed the effect of EPA onMIA PaCa-2 cell growth. Furthermore, EPA stimulated the tran-scriptional activity of PPAR-� as determined by reporter assays inMIA PaCa-2 cells transfected with a luciferase construct driven byPPAR-� response elements. Conclusions: Our studies demon-strate that the omega-3 PUFA EPA decreases the growth of pan-creatic cancer cell lines by COX-2 dependent and -independentmechanisms. The COX-2 independent mechanisms of EPA-induced growth inhibition were mediated by activation of thenuclear receptor PPAR-�. These data provide the basis for evalu-ating the therapeutic efficacy of omega-3 PUFAs in preclinicalmodels of pancreatic cancer.

QS110. FLUORESCENCE EXPRESSING VIRUSES EN-HANCE THE DETECTION OF RARE TUMOR CELLSIN HUMAN WHOLE BLOOD. Kaitlyn Kelly, SandraFong, Melissa Lee, Yuman Fong; Memorial Sloan KetteringCancer Center, New York, NY

Background: Cytologists are often asked to detect rare cancercells in biologic specimens for diagnosis and staging of cancer,though the limits of detection even for the best cytologist isapproximately 1 malignant cell in 20,000 normal cells. Manyviruses are being engineered to selectively infect tumors as part ofgene therapy strategies. We wondered if a herpes virus, NV1066,carrying the green fluorescence protein gene (GFP) can be used toselectively infect tumor and allow for enhanced detection of raretumor cells in blood. Methods: The human gastric adenocarci-noma cell line OCUM cells were placed in various concentrations(including controls with no cancer cells) in human whole blood.The specimens were then mixed with 1e6 NV1066 particles orheat inactivated virus (control) and incubated at 38°C for 4 to 12hours. The specimens were then examined under a fluorescentmicroscope for detection of GFP expression. Blinded reading offluorescence detection was performed by two MD’s as well as twohigh school students, and results correlated. Similarly, 10 mlaliquots of human whole blood were spiked with various dilutionsof OCUM cells (ranging from 1e6 to 1e2), incubated with NV1066for 4 hours, and sorted for GFP expression with flow cytometry.Sorted cells were counterstained with a red fluorescent antibodyto CEA confirming that they were, in fact, the seeded OCUM cells.Result: There was 100 percent specificity for detection of cancercells, in that only specimens with cancer cells exposed to live virusexpressed green fluorescence. Detection of cancer was found inspecimens with as little as one cancer cell in 5,000,000 normalcells, indicating that limits of detection are 250 times better thanconventional cytology. The processing involved simple incubationwithout the technical demands of immunohistochemistry. Concor-dance between trained medical professionals and high school stu-dents was 100%, indicating that the fluorescence read-out wasunequivocal. Further, fluorescence assisted cell sorting (FACS)allowed for rapid isolation of GFP-expressing cells. All green cellswere found to coexpress red fluorescence indicating that all virus-infected cells were cancer cells positive for CEA. (Figure) Conclu-sion: Fluorescence enhanced viral detection for cancer is a simplemethod that exploits many of the genetic engineering efforts of

311ASSOCIATION FOR ACADEMIC SURGERY AND SOCIETY OF UNIVERSITY SURGEONS—ABSTRACTS